article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the crewther relation @xcite was originally derived as a remarkable consequence of conformally invariant field theory .
it provides a nontrivial relation for three fundamental constants of a gauge theory with zero @xmath0 function and zero fermion masses : @xmath6 where @xmath7 is the anomalous constant governing @xmath8 decay @xcite , @xmath9 is the coefficient of the bjorken sum rules for the polarized deep - inelastic electron scattering @xcite , and @xmath10 is the isovector part of the cross - section ratio for the electron - positron annihilation into hadrons @xcite . in qcd @xcite , the logarithmic derivative of the running coupling , the @xmath0 function , is nonzero and negative ;
thus the strong coupling @xmath11 becomes small at short distances due to asymptotic freedom , allowing the perturbative calculation of high - momentum transfer processes . since
physical qcd is non - conformal , the crewther relation will be explicitly broken by the @xmath0-dependent terms arising from pqcd radiative corrections .
however , one can define a generalized crewther relation ( gcr ) " by incorporating perturbative qcd ( pqcd ) approximants to each observable , where each observable is evaluated at corresponding physical scales @xcite ; i.e. , @xmath12 where @xmath13 , @xmath1 is the non - singlet adler function , and @xmath2 is the coefficient in the bjorken sum rule for polarized deep - inelastic electron scattering . the @xmath14-term is introduced in order to compensate for the conformal symmetry breaking ( csb)-terms from both @xmath1 and @xmath15 . using the renormalization group equation ( rge ) , it can be written in the form @xmath16 where the @xmath0-function , @xmath17 ; the coefficients @xmath18 are independent of the @xmath19 .
thus @xmath14 , as constructed in eq.([eq : csbbeta ] ) , is a linear function of the @xmath19-terms . the @xmath20 and @xmath21 terms are universal , whereas all high - order @xmath22-terms are scheme - dependent
an essential requirement of the renormalization group is that relations between physical observables can not depend on a theoretical convention such as the choice of the renormalization scheme .
this property is called renormalization group invariance " ( rgi ) @xcite .
it is usually argued that the gcr only becomes renormalization scheme - and - scale independent at infinite order , leaving renormalization scheme - and - scale dependence at any finite order @xcite .
for example , as will be shown later , the linear property of @xmath14 persists only when both @xmath1 and @xmath2 are renormalized using the @xmath23-scheme .
the utility of the gcr ( [ eq : gcr ] ) at finite order in pqcd is thus greatly constrained .
does there exist a general scheme - independent gcr for pqcd theory at any fixed order ?
in fact , as we shall show in this article , a fundamental scheme - independent scale - fixed gcr can be achieved at each finite order by applying the principle of maximal conformality " ( pmc ) @xcite .
the pmc is designed to eliminate renormalization scheme - and - scale ambiguities simultaneously .
the pmc has a rigorous theoretical foundation and satisfies the essential properties of rgi @xcite .
it provides the underlying principle for the brodsky - lepage - mackenzie ( blm ) method @xcite , and it reduces to the standard gell mann - low method @xcite in the @xmath24 abelian limit @xcite . as in qed , separate renormalization scales and effective numbers of quark flavors ( @xmath25 ) appear for each skeleton graph , reflecting their differing virtualities at each perturbative order .
the coefficients of the resulting pqcd series match the coefficients of the corresponding conformal theory with @xmath26 , ensuring the scheme - independence of the pmc predictions at any fixed order .
the usual divergent @xmath27 renormalon terms in the pqcd series do not appear .
furthermore , the pmc predictions have negligible dependence on the choice of the initial renormalization scale .
an important concept in qcd is the effective charge " @xcite running couplings which can be defined from any perturbatively calculable observable . in the case of the gcr
, we will introduce two effective couplings defined by the non - singlet adler function @xmath1 and the bjorken sum rules @xmath2 , @xmath28 where @xmath29 is the mass scale of the observable the kinematic value at which it is measured .
substituting eqs.([eq : adns ] , [ eq : ag1 ] ) into eq.([eq : gcr ] ) , we obtain @xmath30[1-\widehat{\alpha}_{g_1}(q ) ] & = & 1+\delta_{\rm csb}. \label{eff : gcr}\end{aligned}\ ] ] an effective coupling is by definition pqcd calculable ; the @xmath0-pattern for its perturbative series at each order is a superposition of all of the @xmath19-terms which govern the evolution of the lower - order @xmath11 contributions appearing at this particular order .
more explicitly , the @xmath0-patterns for the pqcd approximants of these two effective couplings @xmath31 and @xmath32 are @xmath33 and @xmath34 where the @xmath35 are conformal coefficients and @xmath36 are non - conformal ones . thus we obtain @xmath37a_s^2 + \bigg[r^d_{3,0}-r^c_{2,0}r^d_{1,0}-r^c_{1,0}r^d_{2,0}-r^c_{3,0 } \nonumber\\ & & + ( 2r^d_{3,1}-r^c_{2,1 } r^d_{1,0}-r^c_{1,0 } r^d_{2,1}-2 r^c_{3,1})\beta_0+(r^d_{2,1}-r^c_{2,1})\beta_1+(r^d_{3,2}-r^c_{3,2})\beta_0 ^ 2 \bigg]a_s^3 \nonumber\\ & & + \bigg[r^d_{4,0}-r^c_{3,0 } r^d_{1,0}-r^c_{2,0 } r^d_{2,0}-r^c_{1,0 } r^d_{3,0}-r^c_{4,0}+\left(3r^d_{4,1}-2 r^c_{3,1 } r^d_{1,0}-r^c_{2,1 } r^d_{2,0}-r^c_{2,0 } r^d_{2,1}-2 r^c_{1,0 } r^d_{3,1}-3 r^c_{4,1}\right)\beta_0 \nonumber\\ & & + \left(-r^c_{2,1 } r^d_{1,0}-r^c_{1,0 } r^d_{2,1}-2 r^c_{3,1}+2 r^d_{3,1}\right)\beta_1+\left(r^d_{2,1}-r^c_{2,1}\right)\beta_2 + \frac{5}{2}(r^d_{3,2}-r^c_{3,2})\beta_1 \beta_0 \nonumber\\ & & + \left(3r^d_{4,2}-r^c_{3,2 } r^d_{1,0}-r^c_{2,1 } r^d_{2,1}-r^c_{1,0 } r^d_{3,2}-3 r^c_{4,2}\right)\beta_0 ^ 2+(r^d_{4,3}-r^c_{4,3})\beta_0 ^ 3 \bigg]a_s^4+\mathcal{o}(a_s^5 ) .
\label{eq : csbexpan}\end{aligned}\ ] ] the coefficients of @xmath38 ( @xmath39 ) , the coefficients of @xmath40 ( @xmath41 ) , etc . , also follow from the pmc degeneracy relations among different orders @xcite , as shown explicitly by eqs.([eq : cbetapattern ] , [ eq : dbetapattern ] ) .
the degeneracy relations explain why one only needs to introduce one new parameter at each new order in the perturbative series ( [ eq : csbbeta ] ) for @xmath14 .
the above formulae are general and applicable for any renormalization scheme
. however , if one uses dimensional regularization , specifically , the @xmath23-scheme , the form of @xmath14 will be greatly simplified .
for example , the conformal coefficients of @xmath1 and @xmath2 in the @xmath23-scheme can be derived from the pqcd series derived in refs.@xcite which have been computed up to four - loop level .
they satisfy the following relations @xmath42 there are also simple relations among the non - conformal coefficients when one uses the @xmath23-scheme , @xmath43 comparing with eq.([eq : csbbeta ] ) , we obtain @xmath44 this confirms the linear features of @xmath14 using the @xmath23-scheme .
the pmc degeneracy relations among different orders ensure the linear property of @xmath14 ; one can then derive the coefficients @xmath45 and @xmath46 from the dependence of the @xmath47 on @xmath25 with the help of the degeneracy relations .
it should be noted that the simplifying linear properties of @xmath14 are not obtained in general .
we will illustrate this by using the generalized dimensional regularization @xmath48-scheme introduced in ref.@xcite . in the @xmath48-scheme one
defines the uv regularization by subtracting an extra arbitrary constant @xmath49 : @xmath50 .
the conventional @xmath23-scheme corresponds to @xmath51 .
the resulting @xmath0-function is the same for all @xmath48-schemes .
thus a scale transformation between different @xmath48-schemes corresponds simply to a scale displacement , such as @xmath52 @xcite , where @xmath53 stands for the scale of the @xmath23-scheme .
the displacement relation implies @xmath54 where @xmath55 and @xmath56 .
the conformal - breaking term under the @xmath48-scheme thus becomes @xmath57 = -\sum_{i=2}^{\infty}\sum_{k=1}^{i-1 } k^{\rm ns}_k[\delta ] \beta^{\cal r}_{i-1-k } a_s^i , \label{d : eq : csbbeta}\end{aligned}\ ] ] in which the first three coefficients are @xmath58 & = & k^{\rm ns}_1 , \\
k^{\rm ns}_2[\delta ] & = & k^{\rm ns}_2 + 2 \delta \beta_0 k^{\rm ns}_1 , \\ k^{\rm ns}_3[\delta ] & = & k^{\rm ns}_3 + 3 \delta \beta_0 k^{\rm ns}_2 + 3
( \delta \beta_1 + \delta^2 \beta_0 ^ 2 ) k^{\rm ns}_1.\end{aligned}\ ] ] here @xmath59 $ ] corresponds to the case of the @xmath23-scheme . obviously ,
for any other @xmath60-scheme , the @xmath61-terms explicitly break the linear property of @xmath14 . in other words ,
the scheme - dependence of the fixed - order pqcd series breaks the generality of @xmath14 and hence the generality of the gcr ( [ eq : gcr ] ) .
in contrast , as we shall show , the pmc sets the optimal scale for each pqcd approximant at each order by absorbing every @xmath0 term of the pqcd series into its respective running coupling , thus providing scheme - independent fixed - order pqcd predictions .
in particular , we will obtain a generalized scheme - independent crewther relation ( gsicr ) for qcd .
as an illustration , we will derive the pmc predictions for @xmath1 and @xmath4 .
the non - singlet adler function is defined as @xcite , @xmath62 where @xmath63 , @xmath53 encodes the renormalization scale .
the coefficient @xmath64 is the non - singlet part of the polarization function for a flavor - singlet vector current .
the running behavior of @xmath65 is controlled by @xmath66 where @xmath67 .
the adler function depends on the anomalous dimension of the photon field @xmath68 ; its perturbative coefficients @xmath69 using the @xmath23-scheme up to four - loop level can be found in ref.@xcite .
thus we obtain @xmath70 .
\label{eq : dexpression}\ ] ] since @xmath71 at any fixed order , the pqcd approximant @xmath1 is a local rgi quantity . it should be emphasized that the anomalous dimension @xmath72 is associated with the renormalization of the qed coupling , it ensures that @xmath1 satisfy the local rgi , but not the standard rgi @xcite .
this explains why the qed anomalous dimension @xmath73 , which appears in the definition of the adler function , can not be used to set the pqcd renormalization scales for @xmath1 .
up to the four - loop level using conventional scale - setting.,scaledwidth=48.0% ] the dependence of the ratio @xmath74 on the momentum scale @xmath29 which is predicted using conventional scale - setting is illustrated in fig.([fig0 ] ) .
one observes a small deviation of the ratio from unity , which can not be diminished even by including more loop corrections . at @xmath75 gev ,
the ratio is @xmath76 for the two - loop corrections , which changes to @xmath77 for the three- and four - loop corrections .
this small derivation occurs because the non - zero @xmath78 starts at @xmath79 -order , as shown by eq.([eq : csbbeta ] ) .
the scheme - dependent conformal - breaking term @xmath14 accounts for the scheme - dependence of @xmath74 as predicted by the gcr ( [ eq : gcr ] ) ; it also explains why the ratio deviates from unity .
up to four - loop level after applying pmc scale - setting.,scaledwidth=48.0% ] and @xmath80 versus the kinematic scale @xmath29 of @xmath81 and @xmath82.,scaledwidth=48.0% ] the scheme dependence of the generalized crewther relation is eliminated when one applies pmc scale - setting .
the pmc utilizes the rge recursively to identify the @xmath0-pattern of any pqcd expansion .
eq.([eq : dexpression ] ) shows that , except for the anomalous coefficient @xmath73 , which is associated with the renormalization of @xmath83 , the only remaining terms in @xmath1 are the pqcd rge @xmath47-terms .
after shifting the nonconformal @xmath47 terms into the scales of qcd coupling , one recovers the conformal series and determines the optimal scales of the running coupling at each order . thus , after applying the pmc @xcite , we obtain @xmath84 similarly , after applying pmc scale - setting to @xmath85 , we obtain @xmath86 here @xmath87 is introduced for convenience .
it has the same scheme - independent conformal coefficients as those of @xmath1 , as ensured by the relations ( [ eq : confrelation ] ) .
the first three pmc scales @xmath88 or @xmath89 can be derived from the pqcd series for @xmath1 and @xmath4 given in refs.@xcite . the dependence of the pmc scales on @xmath29 are presented in fig.([figscale ] ) .
the conformal coefficients of the scheme - independent @xmath90 and @xmath91 are identical , but their pmc scales differ : @xmath92 ; thus , as is the case for the conventional prediction shown in fig.([fig0 ] ) , one also has a small deviation from unity , as shown in fig.([fig ] ) : @xmath93 this result shows that by applying pmc scale - setting , one can systematically improve the pqcd predictions for @xmath81 and @xmath85 , although , due to the conformal breaking of qcd , they can not be exactly equal .
we shall demonstrate in the next section that a new scheme - independent gcr can be achieved by using the effective coupling approach .
following eqs.([eq : cbetapattern ] , [ eq : dbetapattern ] ) , the effective coupling @xmath31 can be expanded in terms of @xmath32 , @xmath94 here the @xmath95 are the @xmath0-terms for the @xmath96-scheme , which can be related to the @xmath23 coefficients by using @xmath97 . the coefficients @xmath98 are functions of @xmath99 , whose expressions are given in refs.@xcite together with the conformal coefficients @xmath100 .
after applying pmc scale - setting , we obtain a new scheme - independent gcr ( gsicr ) up to @xmath101-order , @xmath102 at the known four - loop level ( @xmath103 ) , the first three pmc scales are @xmath104+\mathcal{o}(\widehat{\alpha}_{g_1}^3 ) , \nonumber\\ \ln\frac{q_2 ^ 2}{q^2 } & = & -\frac{r_{3,1}}{r_{2,0}}+\widehat{\alpha}_{g_1}(q)\beta_0\left(\frac{3r_{4,2}}{2r_{2,0 } } -\frac{3{r_{3,1}}^2}{2{r_{2,0}}^2}\right)+\mathcal{o}(\widehat{\alpha}_{g_1}^2 ) , \nonumber\\
\ln\frac{q_3 ^ 2}{q^2 } & = & -\frac{r_{4,1}}{r_{3,0}}+\mathcal{o}(\widehat{\alpha}_{g_1}).\end{aligned}\ ] ] eq.([eqgcr ] ) shows that the relation between @xmath105 and @xmath106 is independent of the renormalization scheme used to renormalize them .
together with the optimal scales set by the rge , the gsicr is both scheme - and - scale independent at any fixed order .
such scale - fixed relations between physical observables are also called commensurate scale relations " ( csrs ) @xcite ) ( [ eqgcr ] ) also holds .
however , these csrs are derived without determining whether the @xmath25-terms at high orders pertain to the rge or not ; thus one can not be certain that one obtains the correct running behavior at each order . ] . in the conformally invariant limit where all @xmath19-terms tend to zero
, we have @xmath107 and we recover the scheme - independent conformal crewther relation @xcite @xmath108[1-\widehat{\alpha}_{g_1}(q)]=1.\ ] ] in conventional analyses , the @xmath14-term is introduced into the original gcr , eq.([eq : gcr ] ) , to collect all of pqcd conformal breaking terms .
the resulting @xmath14-term is scheme dependent and only in the @xmath23-scheme is it a linear expansion of @xmath19-functions .
however , by applying the pmc , we achieve a new fundamental scheme - independent gsicr , eq.([eqgcr ] ) , which is scheme - and - scale independent at any fixed order .
the pmc provides a systematic method to set the renormalization scale of high - energy processes , thus solving the conventional renormalization scheme - and - scale ambiguities .
there is negligible dependence in the choice of the initial renormalization scale .
the scheme - independent gcr given in eq.([eqgcr ] ) provides a remarkable direct connection between deep inelastic neutrino - nucleon scattering and hadronic @xmath109 annihilation .
similar scheme - independent gcrs can also be derived for many other physical observables , such as : \i ) the connection of the adler function @xmath110 to the gross - llewellyn smith ( gls ) sum rule @xmath111 @xcite , @xmath112 and \ii ) the csr connection between the gls sum rule and the @xmath109 annihilation ratio @xmath113 , @xmath114 the pmc scales @xmath115 and @xmath116 can be determined using the standard pmc scale - setting procedure .
as is the case of the scheme - independent generalized crewther relation , these scheme - and - scale independent commensurate scale relations also provide fundamental , high precision tests of nonconformal qcd . + * acknowledgement * : this work was supported in part by the national natural science foundation of china under grant no.11275280 and no.11625520 , and the department of energy contract no.de-ac02-76sf00515 .
slac - pub-16845 .
pitt - pacc-1612 .
we end with a comment on the use of the two - fold " perturbative expansion , which has been suggested for determining the @xmath0-pattern of a pqcd approximant @xcite . using this procedure ,
a physical observable @xmath117 can be approximated by the perturbative expansion @xmath118 where the coefficients @xmath119 are free from @xmath19-terms .
the two - fold perturbation expansion generates the same @xmath0-pattern as that of pmc @xmath120-scheme at each order @xcite .
however , in comparison with the pmc expansion , such as eqs.([eq : cbetapattern ] , [ eq : dbetapattern ] ) , the degeneracy relations introduced by the two - fold perturbation expansion is different from the pmc ones , even at order @xmath121 .
the degeneracy relations introduced by the pmc are required by the conformality of the final series @xcite ; they show that the @xmath0-pattern for the pqcd series at each order is a superposition of the @xmath19-terms which govern the evolution of all of the lower - order @xmath11 contributions .
conversely , they determine the correct running behavior of @xmath11 at each order .
in contrast , the @xmath0-pattern of the two - fold perturbation expansion can only yield an approximate running behavior at the contributing order ; in effect it is equivalent to using the following approximate solution of the rge to set the scale of the running coupling : @xmath122 where @xmath123 is some initial scale . using the scale displacement equation ,
one obtains the approximate @xmath0-pattern and degeneracy relations of eq.([eq : twofoldbeta ] ) .
s. j. brodsky and x. g. wu , `` scale setting using the extended renormalization group and the principle of maximum conformality : the qcd coupling constant at four loops , '' phys .
d * 85 * , 034038 ( 2012 ) [ phys .
d * 86 * , 079903 ( 2012 ) ] .
h. y. bi , x. g. wu , y. ma , h. h. ma , s. j. brodsky and m. mojaza , `` degeneracy relations in qcd and the equivalence of two systematic all - orders methods for setting the renormalization scale , '' phys .
b * 748 * , 13 ( 2015 ) .
g. cveti@xmath129 and a. l. kataev , `` adler function and bjorken polarized sum rule : perturbation expansions in powers of @xmath130 conformal anomaly and studies of the conformal symmetry limit , '' phys .
d * 94 * , 014006 ( 2016 ) . h. h. ma , x. g. wu , y. ma , s. j. brodsky and m. mojaza , `` setting the renormalization scale in perturbative qcd : comparisons of the principle of maximum conformality with the sequential extended brodsky - lepage - mackenzie approach , '' phys .
d * 91 * , 094028 ( 2015 ) . | the principle of maximal conformality ( pmc ) provides a systematic way to set the renormalization scales order - by - order for any perturbative qcd process .
the resulting predictions are independent of the choice of renormalization scheme , a requirement of renormalization group invariance .
the crewther relation , which was originally derived for conformal theory , provides a remarkable connection between two observables when the @xmath0 function vanishes : one can show that the product of the bjorken sum rule for spin - dependent deep inelastic lepton - nucleon scattering times the adler function , defined from the cross section for electron - positron annihilation into hadrons , has no pqcd radiative corrections .
the generalized crewther relation " relates these two observables for physical qcd with nonzero @xmath0 function ; specifically , it connects the non - singlet adler function ( @xmath1 ) to the bjorken sum rule coefficient for polarized deep - inelastic electron scattering ( @xmath2 ) at leading twist .
a scheme - dependent @xmath3-term appears in the analysis in order to compensate for the conformal symmetry breaking ( csb ) terms from perturbative qcd . in conventional analyses ,
this normally leads to unphysical dependence in both the choice of the renormalization scheme and the choice of the initial scale at any finite order .
however , by applying pmc scale - setting , we can fix the scales of the qcd coupling unambiguously at every order of pqcd .
the result is that both @xmath1 and the inverse coefficient @xmath4 have identical pqcd coefficients , which also exactly match the coefficients of the corresponding conformal theory .
thus one obtains a new generalized crewther relation for qcd which connects two effective charges , @xmath5 , at their respective physical scales .
this identity is independent of the choice of the renormalization scheme at any finite order , and the dependence on the choice of the initial scale is negligible .
similar scale - fixed commensurate scale relations also connect other physical observables at their physical momentum scales , thus providing convention - independent , fundamental precision tests of qcd . |
the magnetic molecules [ mn@xmath0o@xmath0(ch@xmath6coo)@xmath7(h@xmath8o)@xmath9@xmath102ch@xmath6cooh@xmath104h@xmath8o , abbreviated mn@xmath0-ac @xcite , and [ ( c@xmath11h@xmath12n@xmath6)@xmath11fe@xmath1(@xmath13-o)@xmath8(@xmath14-oh)@xmath0]br@xmath16(h@xmath8o)br@xmath108h@xmath8o , in short fe@xmath1br@xmath1 @xcite , have been the focus of extensive studies since it was discovered that they exhibit the rare phenomenon of macroscopic quantum tunneling ( mqt ) @xcite .
as has now been well established @xcite , both of these compounds have a net total spin s = 10 , and can be grown as high quality single crystals @xcite .
the evidence for mqt consisted of the following observations : ( a ) below a certain temperature , known as the blocking temperature , t@xmath17 ( 2.7 k for mn@xmath0-ac and 1 k for fe@xmath1br@xmath1 ) , their magnetic hysteresis loops exhibited sharp steps at regular intervals ( about 0.46 tesla ( t ) for mn@xmath0-ac , and 0.24 t for fe@xmath1br@xmath1 ) , when the field was applied along the easy axes @xcite , and ( b ) the magnetization relaxation rate became temperature independent at low temperatures @xcite .
furthermore , this quantized hysteretic behavior was found also for very dilute samples , such as frozen into organic solvents @xcite .
this observation implies that the hysteresis loop is a property of every single molecule , rather than that of a macroscopic domain , hence they have been described as single molecule magnets ( smm s ) @xcite .
it can thus be expected that these smm s hold the potential for becoming an integral part of a molecular - size memory device@xcite .
mn@xmath0-ac has also been proposed as a potential candidate for a quantum computing element@xcite . to advance our understanding of these materials for possible applications ,
it is important to understand their electrical conductivity behavior .
although these materials appear to be insulating , they are single crystalline materials with a unique configuration of large molecular units containing transition metal ions and polarizable subunits , nested in a bridging network
. one might thus expect some sort of semiconducting behavior , albeit with high resistivity .
interestingly , information about the electrical conductivity is not yet available for either compound , despite the fact that mn@xmath0-ac and fe@xmath1br@xmath1 have been studied by dielectric relaxation @xcite , far - infrared absorption under applied magnetic fields @xcite , raman scattering @xcite , micro - hall techniques @xcite , micro - squid magnetometry @xcite , epr @xcite , nmr @xcite , specific heat@xcite and magnetization measurements @xcite , neutron scattering @xcite , and optical absorption @xcite .
we note , however , that from optical absorption measurements oppenheimer _ et al .
_ @xcite have deduced optical excitation band gaps , @xmath5 , of 1.1 and 1.75 ev for the minority ( inner tetrahedron ) and majority ( crown ) spin systems in mn@xmath0-ac , respectively .
these values were considered comparable with corresponding theoretical estimates of 0.45 and 2.08 ev by pederson and khanna @xcite and of 0.85 and 1.10 ev by zeng _
_ @xcite . in the present investigation
we have carried out electrical conductivity measurements on single crystals of both mn@xmath0-ac and fe@xmath1br@xmath1 over the temperature range of 77 k to 300 k. confirmatory measurements were made using _ ac _
dielectric techniques .
the results show that both compounds exhibit fairly clear semiconducting behavior ( 200 - 300 k ) with distinctly different transport activation energies .
it should be noted that in an intrinsic semiconductor , the activation energy ( or gap ) measured via conductivity is to be compared with 1/2 @xmath5 , and thus for mn@xmath0-ac the agreement with the optical data are satisfactory .
we also describe photoconductivity over the visible range and x - ray damage investigations on mn@xmath0-ac to further probe the nature of electrical transport in these materials .
the measured photoconductivity exhibits a significant wavelength dependence .
long black rectangular crystals of mn@xmath0-ac were synthesized following the procedure of lis @xcite .
the crystals typically grew to dimensions of about 0.6 x 0.6 x 2.8 mm@xmath18 .
high quality single crystals of fe@xmath1br@xmath1 were prepared by the method described in the literature @xcite .
the fe@xmath1br@xmath1 crystals grew as dark brown orthorhombic plates of about 4.0 x 6.0 x 0.5 mm@xmath18 .
the samples have been routinely monitored for quality by nmr , x - ray diffraction , and magnetization measurements .
_ dc _ resistance measurements were conducted under either a constant voltage or a constant current mode using a conventional four probe technique . a high input impedance ( 2@xmath19 ohm ) electrometer
was used to measure the voltage drop across the sample when constant current was applied .
currents were typically in the 0.1 to 10 na range , and voltages were generally 100 v or less .
the current - voltage characteristics were periodically checked to verify ohmic behavior .
_ ac _ conductance measurements were made with a standard _
ac _ impedance bridge technique .
a capacitive electrode configuration was made by painting two flat parallel surfaces of a sample with conductive ( silver or graphite ) paste .
the capacitive and dissipative signals were detected by a lock - in amplifier with an excitation frequency of about 8 khz . in all cases
the measurements were made under vacuum in a temperature controlled probe .
photoconductivity measurements were made on mn@xmath0-ac using a he - ne laser for red ( 632.8 nm ) and argon laser for blue ( 488 nm ) and green ( 514 nm ) light .
the light intensity was calibrated before each measurement .
photocurrent was measured using a lock - in amplifier while the sample was under _ direct current _ bias and illuminated by chopped light .
a more detailed description of the experiment is published elsewhere@xcite . for the x - ray experiments ,
mn@xmath0-ac crystals were irradiated with 40 kv , 40 ma , cu @xmath20 radiation at room temperature in order to observe the effects of defects .
the temperature dependence of the resistance @xmath21 of a mn@xmath0-ac sample is shown in fig . [ temp](a ) for a 4-terminal , constant current configuration .
the resistivity values are on the order of @xmath22 @xmath23 cm at room temperature , and increase rapidly in an activated manner upon cooling .
below about 200 k , ohmic equilibrium is lost due to the high resistance values .
thus the arrhenius analysis was limited to temperatures above this temperature . over the 200 - 300 k range ,
@xmath24 exhibits a linear dependence as a function of 1/t as shown in the inset , characteristic of a semiconducting system with a well defined band gap where @xmath25 , and @xmath2 is the thermal activation energy . from the slopes of curves of the inset
, @xmath2 is estimated to be 0.38 @xmath3 0.05 ev .
[ temp](b ) shows the temperature dependence of the current for the constant voltage ( 50 v ) bias condition . as the resistance of the sample increases with decreasing temperature , the current rapidly decreases , and is unmeasurable below @xmath26 210 k. the corresponding @xmath24 vs @xmath27 curve is shown in the inset .
the linear dependence yields a value of @xmath2 = 0.36 @xmath3 0.05 ev .
[ bridge ] shows the @xmath24 vs @xmath27 curve of a mn@xmath0-ac sample obtained by the impedance bridge technique .
the sample was cooled from room temperature to 200 k. the solid line corresponds to @xmath2 = 0.36 @xmath3 0.05 ev .
the resistance again shows activated behavior but the linear relation is not so clear as in the _ dc _ resistance case .
a significant observation was that the mn@xmath0-ac crystals lose solvent upon heating above 300 k. one sample was heated to 350 k and then cooled down to 200 k. the plot of @xmath24 vs @xmath27 yielded two separate straight lines as can be noted from fig .
the activation energy at the higher temperature range was 0.35 @xmath3 0.05 ev , while the @xmath2 was 0.18 @xmath3 0.05 ev for the lower temperatures , after heating .
thus , care must be taken not to heat the samples above 300 k or so .
photoconductivity ( pc ) was measured on mn@xmath0-ac using the _ ac _ component of the photocurrent for chopped laser light illumination .
this was done by biasing the sample with different values of _ dc _ current @xcite .
[ photo ] shows the dependence of the pc on the intensity ( power ) of the laser radiation at the three wavelengths used , 632.8 nm ( red ) , 514 nm ( green ) , and 488 nm ( blue ) .
pc is seen to increase with photon energy .
the increase is about a factor of eight when going from 632.8 nm to 488 nm .
clearly this enhancement must relate to the creation of charge carriers by the photons , or to the increase in temperature due to light absorption , or both .
a simple thermal mechanism is not supported by the earlier uv - visible absorption data of oppenheimer _ et al . _
@xcite on mn@xmath0-ac .
the spectra show a gradual increase in absorption with photon energy and the absorption edge was estimated to be about 1.1 ev
. however , over the 632.8 to 488 nm range , the absorption was nearly ( within a factor of two ) constant , while the pc increases by a factor of eight .
these considerations argue against a major role of thermal heating in the mechanism of the observed pc enhancement . the effect is thus ascribed to an enhancement of charge carriers due to optical absorption . recently , hernandez _ et al . _ @xcite observed an increase in the magnetization tunneling rate of mn@xmath0-ac caused by defects in the lattice as a result of x - ray irradiation and heat treatments . in order to probe the possible role of defects in the transport properties of mn@xmath0-ac ,
we have also carried out an x - ray irradiation study . as the irradiation dose was increased from 2 hr to 20 hrs , the overall resistivity of the sample increased , but a plot of @xmath24 vs @xmath27 for the different exposure times showed the activation energies remained fairly constant .
the radiation - induced defects thus seem to act as trapping sites for the carriers .
this effect is further discussed in section iv .
preliminary temperature dependent conductivity measurements have also been carried out on single crystals of fe@xmath1br@xmath1 .
figure [ fe8res ] shows a typical @xmath28 vs. @xmath27 plot .
the slope of the line yields a value of @xmath2 = 0.73 @xmath3 0.1 ev , which is seen to be significantly higher than that of mn@xmath0-ac ( 0.37 @xmath3 0.05 ev ) .
this is discussed later in terms of the bonding of fe@xmath1br@xmath1 and mn@xmath0-ac .
at present no optical data are available for fe@xmath1br@xmath1 for comparison with the conductivity measurements .
table i summarizes the conductivity results in comparison to the optical data @xcite and theoretical calculations @xcite .
the main result of this study of the electrical transport in these smm - type single crystalline materials is that they exhibit thermally activated conductivity characteristic of a gapped semiconductor over the range of 200 - 300 k. photoconductivity measurements also support their description as gapped semiconductors .
however , the precise nature of the carrier transport is difficult to determine , since generally measurements over many orders of magnitude in temperature are necessary to establish the functional dependence of @xmath21 . in order to understand the conduction pathway , we examined the connectivity between the neighboring mn@xmath0-ac and fe@xmath1br@xmath1 clusters . figures [ mn12 ] and [ fe8 ] show several transport scenarios are possible .
first , as a result of the crystalline lattice , one can expect that there will be a band structure with gaps . at present , only the electronic structure of the clusters has been computed@xcite . for a band - gapped semiconductor ,
one expects the resistivity to vary as @xmath29 , where @xmath2 is related to the optical gap @xmath5 by @xmath30@xcite .
however , due to the complexity of the crystal structure , it is possible that impurities and/or disorder play significant roles in the charge transport .
for example , thermally activated hopping between impurity sites can also give similar temperature dependence .
.comparison of @xmath5 from conductivity and optical data , and theoretical calculations . [ cols="^,^,^,^",options="header " , ] if conduction is through a distribution of impurity sites , variable range hopping ( vrh ) should dominate .
however , the following considerations argue against a vrh behavior .
furthermore , the resistivity ( @xmath31 @xmath23 cm ) is very high for a typical vrh conduction system@xcite . moreover , application of the mott formula for vhr ( @xmath32 , with @xmath33 , where d is the dimensionality ) yielded t@xmath34 @xmath35 3@xmath36 k ( for d=3 ) .
the high sensitivity of the vrh model to defects can be tested by introducing them artificially , by ion implantation , or in the present case , by x - ray irradiation . since the @xmath37 values obtained from the mott formula indicate an extremely small density of impurity sites , i.e. @xmath38 , a small number of additional defects should decrease @xmath37 significantly .
however , our experimental results on irradiated samples ( sec . iii.c . ) indicate that @xmath37 and @xmath2 are insensitive to the creation of defects .
hence the x - ray investigation supports the idea of intrinsic semiconductor - like conduction in mn@xmath0-ac , and by inference , also for fe@xmath1br@xmath1 .
while we have not been able to arrive at any detailed picture of the conduction pathways , we offer the following possibilities based on the structure and bonding characteristics of both lattices .
the pathway for mn@xmath0-ac is based upon simple coulombic interactions between mn@xmath39 ions , and the polar molecules which lie between two mn@xmath0-ac clusters .
a water molecule bound to a mn@xmath39 on the outer crown lies 2.67 @xmath40 away from an unbound acetate ligand , which is , in turn , 2.77 @xmath40 away from a symmetrically equivalent , unbound acetate ligand adjacent to the closest mn@xmath0-ac cluster as seen in figure [ mn12 ] .
the fe@xmath1br@xmath1 conduction pathway is illustrated in figure [ fe8 ] .
the proposed pathway between two fe@xmath1br@xmath1 clusters is through an n - h bond in the 1,4,7-triazacyclononane .
the conduction pathway thus extends from the n - h to a water ( 2.4 @xmath40 ) , to a br@xmath41 ( 2.3 @xmath40 ) , to a water ( 2.1 @xmath40 ) , and finally to a hydrogen ( 2.4 @xmath40 ) directly connected to the 1,4,7-triazacyclononane on the adjacent fe@xmath1br@xmath1 .
it should be noted that the proposed mn@xmath0-ac conduction pathway is much more direct than that for fe@xmath1br@xmath1 .
this is consistent with the higher activation energy found for fe@xmath1br@xmath1 .
we have found that both mn@xmath0-ac and fe@xmath1br@xmath1 exhibit a gapped semiconductor - like behavior in their electrical transport properties . the limited temperature range over which the resistance was measurable , and over which the materials are stable ,
restricts a knowledge of the precise functional form of r(t ) .
nevertheless , complementary photoconductivity and x - ray irradiation studies support a model where the transport is governed by a well - defined energy gap .
the @xmath2 s have been determined to be 0.37 @xmath3 0.05 and 0.73 @xmath3 0.1 ev for mn@xmath0-ac and fe@xmath1br@xmath1 respectively .
assuming an intrinsic semiconducting behavior , they lead to @xmath5 values of 0.74 @xmath3 0.10 ev for mn@xmath0-ac and 1.5 @xmath3 0.2 ev for fe@xmath1br@xmath1 . for mn@xmath0-ac , the agreement is seen to be reasonably good with the optical band gaps for minority ( inner tetrahedron ) spins , and the theoretical estimates by pederson and khanna @xciteas well as zeng _
additional optical and theoretical data are needed for fe@xmath1br@xmath1 . at present , calculations exist only for the molecular band - gaps , but not for the entire lattice .
hence , we can only speculate that the inter - cluster ligand bridges may play an important role in the conduction mechanism .
further computations on the full crystal band structure are thus desirable .
we would like to thank dr .
x. wei for his assistance in the optical measurements .
this work is supported by nsf - dmr 023532 , darpa , and nsf / nirt - dmr 0103290 .
the national high magnetic field laboratory is supported through a cooperative agreement between the national science foundation and the state of florida .
a. b. sushkov , b. r. jones , j. l. musfeldt , y. j. wang , r. m. achey and n. s. dalal , _ phys .
b _ * 63 * , 214408 ( 2001 ) ; a. b. sushkov , j. l. musfeldt , y. j. wang , r. m. achey , and n. s. dalal _ phys . rev . b _ * 66 * , 144430 ( 2002 ) .
d. arcon , j. dolinsek , t. apih , r. blinc , n. s. dalal , and r. m. achey , _ phys .
b _ * 58 * , 2941 ( 1998 ) ; j. dolinsek , d. arcon , r. blinc , p. vonlanthen , h. r. ott , r. m. achey and n. s. dalal , _ europhys . lett . _
* 42 * , 691 ( 1998 ) . | resistivity measurements are reported for single crystals of mn@xmath0-acetate and fe@xmath1br@xmath1 . both materials exhibit a semiconductor - like , thermally activated behavior over the 200 - 300 k range .
the activation energy , @xmath2 , obtained for mn@xmath0-acetate was 0.37 @xmath3 0.05 ev , which is to be contrasted with the value of 0.55 ev deduced from the earlier reported absorption edge measurements and the range of 0.3 - 1 ev from intramolecular density of states calculations , assuming @xmath4= @xmath5 , the optical band gap .
for fe@xmath1br@xmath1 , @xmath2 was measured as 0.73 @xmath3 0.1 ev , and is discussed in light of the available approximate band structure calculations .
some plausible pathways are indicated based on the crystal structures of both lattices . for mn@xmath0-acetate ,
we also measured photoconductivity in the visible range ; the conductivity increased by a factor of about eight on increasing the photon energy from 632.8 nm ( red ) to 488 nm ( blue ) .
x - ray irradiation increased the resistivity , but @xmath2 was insensitive to exposure . |
we acknowledge support from the french national agency : labex action ( anr-11-labx-01 - 01 ) and placore ( anr - bs10 - 0007 ) , and from the conseil regional de bourgogne ( pari - photcom ) .
dd and gcf thank the european cost action mp1403 nanoscale quantum optics . | hybrid molecular - plasmonic nanostructures have demonstrated their potential for surface enhanced spectroscopies , sensing or quantum control at the nanoscale . in this work , we investigate the strong coupling regime and explicitly describe the hybridization between the localized plasmons of a metal nanoparticle and the excited state of a quantum emitter , offering a simple and precise understanding of the energy exchange in full analogy with cavity quantum electrodynamics treatment and dressed atom picture .
both near field emission and far field radiation are discussed , revealing the richness of such optical nanosources .
optical microcavities can store light for a long time allowing efficient light - matter interaction with important applications in quantum technologies , low threshold laser @xcite , supercontinuum laser @xcite or indistinguishable single photon source @xcite .
it relies on the extremely high quality factor of the cavity mode but at the price of diffraction limited sizes .
that is why strong efforts have be done since a decade to transpose cavity quantum electrodynamics ( cqed ) concepts to nanophotonics and plasmonics @xcite .
particular attention has been devoted to the strong coupling regime @xcite since it offers the possibility of a control dynamics of the light emission , as _
e.g. _ photon blockade @xcite or coherent control @xcite . in this letter
, we build an effective hamiltonian that fully transposes the cqed description to an hybrid plasmon - quantum emitter nanosource .
we demonstrate it can be exactly described in full analogy with cqed representation .
specifically , the coupled plasmon - emitter system behaves like an emitter in a multimodal lossy cavity .
we notably determine the structure of the emitter states dressed by the plasmon modes .
we consider the hybrid system displayed in fig .
[ mnp ] .
a two level system ( tls ) quantum emitter is located close to a metal nanoparticle ( mnp ) .
the optical transition is characterized by the frequency @xmath0 , the dipole moment @xmath1 and the operator @xmath2 . for the sake of clarity
, we consider a tls emitter coupled to spherical mnp since the localized surface plasmon ( lsp ) modes involved in the coupling process are well identified and the hybridization of the emitter and mnp modes will be unambiguously demonstrated .
.,scaledwidth=35.0% ] the hamiltonian of the coupled system writes @xmath3\label{hamil}\end{aligned}\ ] ] the first term refers to the tls energy and we have phenomelogically introduced the decay rate of the excited state @xmath4 in the second term.the third term describes the total energy of the electromagnetic field where @xmath5 ( @xmath6 ) is the lsp polaritonic vector field operator associated to the creation ( annihilation ) of a quantum of electromagnetic mode in presence of the mnp .
the last term describes the emitter - field interaction under the rotating - wave approximation .
the electromagnetic field has to be quantized by taking into account the dispersing and absorbing nature of the metal @xcite .
the electromagnetic mode dispersion and absorption are governed by the real and imaginary parts of the metal dielectric constant @xmath7 , that satisfy the kramers - kronig relations .
as a situation , we assume a drude - like behaviour @xmath8@xcite .
we use @xmath9 , @xmath10 ev and @xmath11 mev for silver @xcite .
the quantization can be performed by introducing a noise polarization operator expressed in terms of the creation ( annihilation ) operators @xmath12 ( @xmath13 ) @xcite .
the electric field operator can be expressed as @xmath14 with @xmath15^{\dag},\nonumber\\ \mathbf{\hat{e}}(\mathbf{r},\omega)= & i\sqrt{\frac{\hbar}{\pi\epsilon_0}}k_0 ^ 2 \int d{\mathbf{r } ' } \sqrt{\varepsilon_i(\mathbf{r}',\omega ) } { \mathbf g}({\mathbf r},{\mathbf r}',\omega)\hat{{\mathbf f}}({\mathbf r'},\omega),\end{aligned}\ ] ] where @xmath16 and @xmath17 is the green s tensor .
it contains all the information about the field response of the mnp . in the following ,
we investigate the optical response of the emitter - mnp system .
we assume an emitter initially in its excited state @xmath18 and the lsp field in the ground state ( vacuum ) .
the wave function of the hybrid system in the interaction picture writes at time @xmath19 @xcite @xmath20 where @xmath21 corresponds to the emitter in its excited state and no lsp mode excited whereas @xmath22 corresponds to the emitter in its ground state and a single excited lsp mode of energy @xmath23 .
the elementary excitation of a lsp is defined through the action of the bosonic vector field operator on the vacuum state @xmath24 .
the dynamics of the probability amplitudes @xmath25 and @xmath26 are derived from the time - dependent schrdinger equation @xcite . as a first step
, the coupling between the emitter and the mnp can be studied in the near field through the polarization spectrum @xmath27 @xcite @xmath28 @xmath29 is the dyadic component along the direction @xmath30 of the dipolar emitter ( @xmath31 ) . note that the free - space contribution of the green s tensor is included in the transition frequency @xmath0 ( lamb shift ) and decay rate @xmath4 ( weisskopf - wigner theory ) .
therefore , only the scattering part of the green s tensor appears in eq .
( [ polarisation_finale ] ) .
the polarization spectrum characterizes the near field emission properties of the coupled system .
it is also necessary to define the signal radiated in the far field zone .
following ref .
@xcite , the spectrum recorded at the detector position @xmath32 expresses @xmath33 nm , including all the mnp modes ( n=25 ensures the convergence ) .
the detector is located at ( r=@xmath34,@xmath35 ) .
the vertical line refers to the emitter emission frequency ( @xmath36 ev ) .
bottom line : recorded signal as a function of the detector position for a sphere radius @xmath37 nm ( c ) and @xmath38 nm ( d ) .
the three curves are calculated at the three peaks of the far field spectrum @xmath39 .
blue curves : @xmath40 ev ( c ) or @xmath41 ev ( d ) .
black solid curves : @xmath42 ev ( a ) or @xmath43 ev ( b ) .
black dashed curves : @xmath44 ev ( c ) or @xmath45 ev ( d ) .
[ spectre],title="fig:",scaledwidth=47.0% ] + nm , including all the mnp modes ( n=25 ensures the convergence ) .
the detector is located at ( r=@xmath34,@xmath35 ) .
the vertical line refers to the emitter emission frequency ( @xmath36 ev ) .
bottom line : recorded signal as a function of the detector position for a sphere radius @xmath37 nm ( c ) and @xmath38 nm ( d ) .
the three curves are calculated at the three peaks of the far field spectrum @xmath39 .
blue curves : @xmath40 ev ( c ) or @xmath41 ev ( d ) .
black solid curves : @xmath42 ev ( a ) or @xmath43 ev ( b ) .
black dashed curves : @xmath44 ev ( c ) or @xmath45 ev ( d ) .
[ spectre],title="fig:",scaledwidth=47.0% ] both expressions can be applied in the weak and strong coupling limits since no markov approximation has been made . for comparison purpose
, we consider an emitter with @xmath46 mev and dipole moment@xmath47 d , radially oriented , as in ref .
@xcite , see fig .
[ spectre]a .
the polarization ( near field ) spectrum @xmath48 presents a split of @xmath49 mev , revealing the strong coupling regime .
the signal recorded in the far field is radically different , see fig .
[ spectre]b ) .
we observe three peaks : the dominant one at 2.8 ev is associated to the lsp dipolar resonance ( denoted lsp@xmath50 ) , known to be strongly radiative .
the two others peaks show reminiscence of the mode splitting , with one of them dominant ( near @xmath51 ev ) and the last one hardly observable ( near 3 ev ) .
these 3 peaks can present similar amplitudes for large particles .
we also represent the radiation diagram in fig .
[ spectre]c , d for two particle radii .
we recover the dipolar angular emission for all the wavelength [ @xmath52 except at the emission energy close to @xmath53 ev for which forward scattering occurs for the largest particle ( solid black curve , fig .
[ spectre]d ) .
this reveals the role of the quadrupolar mode ( lsp@xmath54 ) in the coupling process @xcite .
more understanding of the emitter - mnp coupling process is achievable using the effective model we recently developed @xcite .
in particular , it makes a complete analogy with cqed description , paving the way towards direct transposition of cavity controlled dynamics at the nanoscale .
to this aim , the hybrid emitter - mnp system is described as a quantum emitter coupled to a reservoir of @xmath55 lsp modes structured by the local density of states .
the interaction hamiltonian can be written as @xmath56\mathbf{d}_{eg}^\star \;\label{lien}\end{aligned}\ ] ] @xmath57 quantifies the coupling between the emitter and the mnp @xmath58 mode .
it is expressed in terms of the green s dyad , linking the preceding description with the following effective model .
the excitation of a single lsp of order @xmath59 ( lsp@xmath60 ) obeys @xmath61 with the bosonic operator @xmath62 , @xmath63 is the electric field associated to mode @xmath59 .
each resonance follows a lorentzian profile so that the coupling constant with a given mode can be represented by @xmath64 where @xmath65 is the coupling strength of the emitter to the mnp @xmath58 mode . @xmath66 and @xmath67 are the mode resonance frequency and width , respectively . @xmath66 and @xmath67
depend on the mnp material and size whereas the coupling strength @xmath65 depends also on the distance to the surface .
we calculated that the coupling strength to a given mode fastly decay with distance , but can overcome the joule losses in the mnp ( @xmath68 ) for separation distances below few nanometers , suggesting the feasibility of strong coupling .
specifically , we observe that high order modes play a significant role . finally , the effective hamiltonian is obtained by tracing out the continuous degrees of freedom of the modes in order to establish a set of @xmath55 discrete modes . in the tensor product basis @xmath69
, its matrix representation is @xcite @xmath70 where @xmath71 is the detuning from the resonance .
this effective hamiltonian provides a very practical representation of the hybrid configuration .
the emitter couples to each lsp@xmath60 mode with the coupling strength @xmath65 .
the losses @xmath67 reflect the population leakage from the excited state @xmath72 to the ground state @xmath73 . in order to interpret the degeneracy breaking in the strong coupling regime ( fig.[spectre]a , c ) ,
let us first consider the interaction between the quantum emitter and one single lsp mode of the mnp .
detailed analysis reveals that the third mode ( lsp@xmath74 ) presents the main contribution to the coupling process .
therefore , we approximate the effective hamiltonian by @xmath75 if we neglect the loss rates @xmath76 and @xmath77 , the diagonalization of the effective hamiltonian leads to the dressed state of the hybrid emitter - lsp@xmath74 system with angular frequencies @xmath78 .
mode contribution only .
the emitter is resonant with the third mode ( @xmath79 ev , green line ) .
the blue and black lines characterize the frequencies @xmath80 and @xmath81 of the two dressed states calculated from the approximated effective hamiltonian [ eff_hamil_third ] .
( b ) energy diagram of hybrid nanosource.[degen_3],scaledwidth=23.0% ] if the emission is resonant with the lsp@xmath74 mode ( @xmath82 so that @xmath83 ) , the energy splitting is @xmath84 mev , that is close to the splitting observed in the polarization spectrum calculated in fig.[degen_3]a ) ( @xmath85 mev ) , where only the lsp@xmath74 mode is considered . taking into account the dissipation of the modes ,
the angular frequencies of the dressed states become @xmath86 $ ] where @xmath87 are the complex eigenvalues of the effective hamiltonian ( eq . [ eff_hamil_third ] ) .
we recover @xmath85 mev , as expected . however , the energy splitting is still low compared to the one observed in fig .
[ spectre]a ) ( @xmath49 mev ) , that takes into account all the lsp modes of the mnp .
we therefore consider all the 25 lsp modes in the effective hamiltonian ( eq . [ eff_hamil ] ) .
its diagonalisation leads to @xmath88 dressed states with angular frequencies @xmath89 \ , , ( m=1,\ldots , 26)$ ] ( @xmath90 are the eigenvalues ) .
for such dissipative systems , we have to define right and left eigenvectors @xmath91 and @xmath92 , respectively , satisfying @xmath93 and @xmath94 , @xmath95 .
for hamiltonian of the form ( [ eff_hamil ] ) , one can simply connect them as follows @xcite @xmath96 where @xmath97 and @xmath98 gives the weight of each mode @xmath99 or @xmath100 .
we have now all the ingredients to interpret the polarization spectrum in the strong coupling regime ( see fig .
[ degen_multi ] ) .
the mode hybridization , deduced from the diagonalization of the hamiltonian is depicted in fig .
[ degen_multi]b ) .
we indicate the main lsp modes involved for each dressed state .
we observe that the energy of the dressed states @xmath101 and @xmath102 exactly match the two peaks in the polarization spectrum .
these dressed states mainly result from the hybridization of the excited level of the emitter with either the lsp@xmath54 and lsp@xmath74 ( @xmath101 ) or lsp@xmath103 to lsp@xmath104 ( @xmath102 ) modes of the mnp .
in addition , the shoulder visible in the polarization spectrum near @xmath105 ev originates from the dressed states @xmath106 .
note that the @xmath107 and @xmath101 states present a large contribution of the tls or radiative lsp@xmath108 modes that radiate in the far field zone ( see fig .
[ spectre]b and [ spectre]c , d ) . on the contrary
, the dressed state @xmath102 appears as a dark mode in agreement with the far field spectrum calculated in fig .
[ spectre]b ) .
a ) .
black lines indicate the 26 hybrid modes .
( b ) energy diagram of the hybrid system deduced from the full effective hamiltonian diagonalization .
a thicker line indicates a stronger weight @xmath97 of the atom ( @xmath109 , left part ) or @xmath98 of the lsp@xmath60 mode ( @xmath72 , right part of the diagram ) . in ( a )
the green line corresponds to the emission frequency of the emitter ( @xmath36 ev ) leading to the strong coupling .
the blue ( magenta ) line refers to the frequency @xmath110 ( @xmath111 ) of the dressed state @xmath101 ( @xmath102 ) .
the cyan line near @xmath112 ev indicates the frequency of the @xmath106 dressed state.[degen_multi],title="fig:",scaledwidth=25.0% ] a ) .
black lines indicate the 26 hybrid modes .
( b ) energy diagram of the hybrid system deduced from the full effective hamiltonian diagonalization .
a thicker line indicates a stronger weight @xmath97 of the atom ( @xmath109 , left part ) or @xmath98 of the lsp@xmath60 mode ( @xmath72 , right part of the diagram ) . in ( a )
the green line corresponds to the emission frequency of the emitter ( @xmath36 ev ) leading to the strong coupling .
the blue ( magenta ) line refers to the frequency @xmath110 ( @xmath111 ) of the dressed state @xmath101 ( @xmath102 ) .
the cyan line near @xmath112 ev indicates the frequency of the @xmath106 dressed state.[degen_multi],title="fig:",scaledwidth=28.0% ] it is worthwile to note that the strong coupling regime can be achieved at the single molecule level thanks to cumulative effect of coupling to several @xmath113 modes .
another possibility would be to increase the number @xmath55 of emitters coupled to the mnp .
indeed , the effective hamiltonian presents a similar structure than atom in a cavity so that we also expect a rabi splitting proportional to @xmath114 . taking benefit of both the number of involved lsp modes in the coupling process and increasing the number of molecules would relax the strong coupling conditions .
d ( strong coupling ) , @xmath115 d ( showing a quasi - exponential decay ) and in vacuum .
[ dynamics],scaledwidth=30.0% ] the effective model also unravels the dynamics of the strongly coupled system .
indeed , the wavefunction writes at time @xmath19 : @xmath116 with @xmath117 if we assume an emitter initially in its excited state and no lsp mode populated .
the evolution of the populations obey @xmath118 for the excited state of the emitter and @xmath119 for the @xmath58 lsp mode .
figure [ dynamics ] presents the populations dynamics .
although strongly damped , a clear rabi oscillation is visible revealing reversible ultrafast energy transfert with a period @xmath120 ps , as expected .
the energy transfer between the emitter and the mnp is mainly governed by the lsp@xmath54 and lsp@xmath74 modes , the other being poorly populated .
optimized configurations , such as nanoprism that facilitates the strong coupling regime @xcite would permit to improve the energy transfer efficiency .
additionnally , we check that we recover a fast exponential decay in the weak coupling regime ( for @xmath115 d ) , in agreement with the fermi s golden rule . to summarize
, we have described the optical response of the hybrid mnp - quantum emitter nanosource in analogy with a cqed description and dressed atom picture . specifically , we clarified the nature of the dressed states in the strong coupling regime .
since the effective hamiltonian parameters are easily extracted from the green s tensor of the plasmonic nanostructures , this formalism can be generalized to more complex system as _
e.g. _ plasmonic nanostructures of arbitrary shape , three - level system in @xmath121 configuration or adding an external driving field .
additionnally , this description offers a simple and very intuitive understanding of the spectroscopic properties of the hybrid nanosource .
finally , although relying on a different paradigm ( mode confinement instead of mode lifetime ) , this formalism permits a direct transposition of cqed concept to the nanoscale and constitutes therefore a powerful tool to propose and design original nanophotonics or plasmonics devices . |
there is a unique relationship between the eos of cold dense matter and the sequence of compact star configurations in the mass - radius ( m - r ) diagram provided by the tolman - oppenheimer - volkoff equations .
it can be used to quantify the likelihood of eos models by bayesian analyses using a selection of mass and/or radius measurements as priors @xcite . in this work
we discuss a family of hybrid neutron stars described by an eos with a first - order phase transition that is obtained from a maxwell construction between hadronic and quark matter branches .
our approach to the hybrid eos captures important physical features : * excluded volume corrections in the hadronic eos , * multi - quark interactions for quark matter eos .
the excluded volume corrections at the hadronic level take into account the inner structure of the nucleons which results in pauli blocking due to quark exchange interactions .
this can be seen as a precursor of the quark delocalization which is the expected mechanism of the deconfinement transition @xcite that is still intensified by the simultaneous partial chiral restoration @xcite .
the excluded volume modification is implemented for baryon densities exceeding saturation @xmath1 @xmath2 ( the density in the interior of atomic nuclei ) by considering the available volume @xmath3 for the motion of nucleons at a given density @xmath4 as defined in @xcite @xmath5~ , & \textmd{if } & n > n_{\textmd{sat}}~ , \end{array } \right .
\label{vex}\ ] ] where @xmath6 is the van - der - waals excluded volume that is related to the nucleon hard - core radius @xmath7 .
note that in the form ( [ vex ] ) assigning positive ( negative ) values for @xmath8 leads to a stiffening ( softening ) of the original eos for which we adopt dd2 @xcite with parameters from @xcite . here
, we will consider only positive values of excluded volume and introduce for them the dimensionless parameter @xmath9 $ ] , varying it in the range from @xmath10 to @xmath11 .
we consider the multi - quark interactions as they were introduced by @xcite and succesfully applied to massive hybrid star twins in @xcite .
it is expected that quark matter will still be strongly interacting at the energy densities present in neutron star interiors .
those fall in the range where in lattice qcd simulations at finite temperatures the transition from hadronic to quark matter is obtained .
therefore the introduction of the 8-quark vector channel interaction may be justified . in accordance with refs .
@xcite , we denote the dimensionless coupling constant of this channel by @xmath12 and vary it in the range from @xmath10 to @xmath13 . increasing the value of this parameter
has the effect of stiffening the eos at higher densities , which allows to support hybrid stars as massive as 2 m@xmath14 and thus fulfilling the constraint from recent observations @xcite , see fig .
[ ahp_scheme ] . [ cols="^,^ " , ]
the results of our ba for the two cases defined above are shown in the two panels of fig .
[ fig : flowprs ] . for the case
a , due to the constraint of a large radius , the stiffest possible eos among the accessible parameters was selected as the most probable hybrid star eos .
in particular , largest values of the parameters @xmath15 and @xmath12 that would correspond to hybrid star branches connected with hadronic ones , in the classification of @xcite . for case
b , the high - mass twin star solution is preferred which corresponds to a disconnected hybrid star branch ( `` third family '' ) and disselects large values of the @xmath12 parameter .
herewith we could demonstrate the usefulness of the bayesian analysis method . it allows to check the discriminating power of a certain experimental result before the actual measurement was made . in this sense the discovery potential of future programmes for compact star observations , like nicer or ska could be quantified .
to test the interesting possibility that a strong first - order phase transition might occur in massive neutron stars of @xmath16 @xcite leading to the appearance of a `` third family '' of hybrid stars ( mass twins ) at this mass , one should perform radius measurements for these massive neutron stars @xcite .
this possibility offers bright prospects for future observational campaigns and bears the chance to `` prove '' the existence of a critical point @xcite in the qcd phase diagram from astrophysics .
we acknowledge fruitful discussions with m. c. miller and j. trmper .
we thank s. typel and s. benic for providing us with their eos data . d.b . , h.g . and
. profited from support of networking activities by the cost action mp1304 `` newcompstar '' .
this work was supported by the ncn `` opus '' programme under contract umo-2014/13/b / st9/02621 .
d.e.a - c .
, a.a . and h.g .
received support from the bogoliubov infeld and the ter - antonian smorodinsky programmes .
acknowledges jinr grant no .
14 - 602 - 01 .
9 a. w. steiner , j. m. lattimer and e. f. brown , astrophys .
j. * 722 * , 33 ( 2010 ) .
g. rpke , d. blaschke and h. schulz , phys .
d * 34 * , 3499 ( 1986 ) .
d. blaschke , h. grigorian and g. rpke , in preparation ( 2015 ) .
s. typel , to appear in the 2015 epja topical issue on `` exotic matter in neutron stars '' .
s. typel and h. h. wolter .
a * 656 * , 331 ( 1999 ) .
s. typel , g. rpke , t. klhn , d. blaschke and h. h. wolter , phys .
c * 81 * , 015803 ( 2010 ) .
s. benic .
j. a * 50 * , 111 ( 2014 ) .
s. benic , d. blaschke , d. e. alvarez - castillo , t. fischer and s. typel , astron .
astrophys . *
577 * , a40 ( 2015 ) .
p. demorest , t. pennucci , s. ransom , m. roberts and j. hessels , nature * 467 * , 1081 ( 2010 ) .
j. antoniadis , p. c. c. freire , n. wex , t. m. tauris , r. s. lynch , m. h. van kerkwijk , m. kramer and c. bassa , v. s. dhillon , t. driebe , j. w. t. hessels , v. m. kaspi , v. i. kondratiev , n. langer , t. r. marsh , m. a. mclaughlin , t. t. pennucci , s. m. ransom , i. h. stairs , j. leeuwen , j. p. w. verbiest , d. g. whelan , science * 340 * , 6131 ( 2013 ) .
d. e. alvarez - castillo , a. ayriyan , d. blaschke and h. grigorian , arxiv:1506.07755 s. bogdanov , astrophys .
j. * 762 * , 96 ( 2013 ) .
d. blaschke , h. grigorian , d. e. alvarez - castillo and a. ayriyan , j. phys .
. ser . * 496 * , 012002 ( 2014 ) .
m. g. alford , s. han and m. prakash , phys .
d * 88 * , 083013 ( 2013 ) .
d. blaschke , d. e. alvarez - castillo and s. benic , pos * cpod 2013 * , 063 ( 2013 ) .
d. e. alvarez - castillo and d. blaschke , arxiv:1304.7758 [ astro-ph.he ] .
a. ayriyan , d. e. alvarez - castillo , d. blaschke , h. grigorian and m. sokolowski , phys .
nucl . * 46 * , 854 ( 2015 ) .
d. e. alvarez - castillo and d. blaschke , phys .
* 46 * , 846 ( 2015 ) . | we suggest a new bayesian analysis ( ba ) using disjunct m - r constraints for extracting probability measures for cold , dense matter equations of state ( eos ) .
one of the key issues of such an analysis is the question of a deconfinement transition in compact stars and whether it proceeds as a crossover or rather as a first order transition .
we show by postulating results of not yet existing radius measurements for the known pulsars with a mass of @xmath0 that a radius gap of about 3 km would clearly select an eos with a strong first order phase transition as the most probably one .
this would support the existence of a critical endpoint in the qcd phase diagram under scrutiny in present and upcoming heavy - ion collision experiments . |
within the last several years there has been a wealth of papers published in this journal regarding the motion of an orbiting body on a cylindrically - symmetric surface , such as a funnel or a circular spandex fabric , residing in a uniform gravitational field . some of the interest in this topic can be attributed to fact that the orbital dynamics of an object on such a two - dimensional ( 2d ) surface offers a non - trivial and interesting application of lagrangian dynamics . with the help of a camera capable of generating short video clips and one of the aforementioned surfaces
, one can easily extend this theoretical study to the experimental domain .
this combined theoretical and experimental approach offers comprehensive undergraduate research possibilities as the relevant physics is accessible to any student who has taken an advanced dynamics course.@xcite perhaps some of the interest in this topic arises from the fact that an orbiting body on a 2d surface is often used as a conceptual analogy when describing particle orbits about massive objects in general relativity . in einstein s theory of general relativity ( gr ) , gravity is described as the warping of space and time due to the presence of matter and energy .
although in newtonian theory gravity is understood as the force of one massive object on another , in gr gravity emerges as a physical phenomenon due to the fact that a massive object warps the spacetime around it . according to gr , the planets move along geodesics , or free - particle orbits , in the non - euclidean geometry about the sun .
hence , the planets move in their elliptical - like orbits aroung the sun not due to a ` force ' per se but rather due to the fact that the spacetime around the sun is curved by its presence .
in a typical introductory gr course , the spacetime external to a static , spherically - symmetric massive object is often extensively studied as it offers one of the simplest , exact solutions to the field equations of gr .
when geodesics about this object are then encountered , the conceptual analogy of a marble rolling on a taut elastic fabric warped by a massive compact object , such as a bowling ball , is often embraced in the discussion . when the rolling marble is far from the massive object , the elastic fabric is flat and the marble moves in a straight line at a constant speed in agreement with newton s first law .
when the rolling marble approaches the massive object , as the analogy goes , it moves in a curved path not due to a force acting on the marble , but rather due to the fact that the space in which it resides ( the 2d surface ) is warped by the object .
this analogy offers the student of gr a way of visualizing a warped spacetime and should not be confused with embedding diagrams.@xcite however , with this being said , it has been shown that there exists no 2d cylindrically - symmetric surface residing in a uniform gravitational field that can generate the exact newtonian orbits of planetary motion , except for the special case of circular orbits.@xcite this null result was shown to hold when considering the fully general relativistic treatment of planetary orbits around non - rotating , spherically - symmetric objects.@xcite white and walker first explored marbles orbiting on a warped spandex surface by applying newtonian mechanics to a marble residing on a cylindrically - symmetric spandex fabric subjected to a central mass .
there they arrived at an expression describing the approximate shape of the spandex surface and found that a marble orbiting upon this surface obeys the relation @xmath2 for circular orbits in the small slope regime .
this fascinating result is highly reminiscent of kepler s third law for planetary motion , but with the powers transposed.@xcite lemons and lipscombe then showed that the shape of the spandex surface can equivalently be found by minimizing the total potential energy of the spandex fabric through the use of the calculus of variations .
the total potential energy considered includes the elastic potential energy of the spandex surface and the gravitational potential energy of the central mass .
the calculus of variations approach proves advantageous over the newtonian treatment as the results of white and walker emerge as a special case of a more general treatment , where a pre - stretch of the fabric can be included in the analysis.@xcite this work was later expanded upon by middleton and langston , where the authors considered circular orbits on the spandex fabric in _ both _ the small and large slope regimes .
there they generalized the aforementioned calculus of variations treatment by including the previously - neglected gravitational potential energy of the spandex surface to the total potential energy of the spandex / central mass system .
they showed that the mass of the spandex fabric interior to the orbital path influences the motion of the marble and can even dominate the orbital characteristics .
they also showed that the modulus of elasticity of the spandex fabric is not constant and is itself a function of the stretch.@xcite later , nauenberg considered precessing elliptical - like orbits with small eccentricities on cylindrically - symmetric funnels with various shape profiles . by employing a perturbative method , an approximate solution to the orbital equation of motion
was obtained where the precession parameter was found to be determined uniquely by the slope of the cylindrically - symmetric surface.@xcite this perturbative method was then used to find the 2d surfaces that are capable of generating the stationary elliptical orbits of newtonian gravitation and the precessing elliptical - like orbits of gr , for orbits with small eccentricities.@xcite here , we re interested in understanding elliptical - like orbits of a marble on a warped spandex fabric .
our paper is outlined as follows . in sec .
[ eqn ] , we first consider the motion of a rolling marble on an arbitrary , cylindrically - symmetric surface and present the orbital equation of motion . for elliptical - like orbits with small eccentricities , we solve this equation of motion perturbatively and find a valid solution when the precession parameter obeys an algebraic relation involving the slope of the surface .
we then present the equation that describes the slope of an elastic surface , which arises from minimizing the total potential energy of the spandex / central mass system . in sec .
[ orbits ] and sec .
[ orbits2 ] , we explore the elliptical - like orbits of the rolling marble in the small and large slope regimes , respectively .
we arrive at an expression describing the angular separation between successive like - apsides in each regime and compare with our experimental results.@xcite
lastly , in appendix [ ads ] , we outline the general relativistic treatment of obtaining elliptical - like orbits with small eccentricities about a static , spherically - symmetric massive object in the presence of a vacuum energy . after arriving at an expression describing the angular separation between successive like - apsides
, we compare it to its counterpart for the marble on the spandex surface in both the small and large slope regimes .
we begin this manuscript by presenting the equations of motion that describe a spherical marble of uniform mass density rolling on a cylindrically - symmetric surface in a uniform gravitational field .
these equations can be obtained by first constructing the lagrangian that describes a rolling marble , which is constrained to reside upon a cylindrically - symmetric surface , in cylindrical coordinates @xmath3 .
this lagrangian includes both the translational and rotational kinetic energy of the rolling marble and the gravitational potential energy of the orbiting body .
the resultant lagrangian can then be subjected to the euler - lagrange equations , which yield the equations of motion of the form ( 1+z^2)+zz^2-r^2+gz&=&0,[roft ] + & = & , [ phi ] where a dot indicates a time - derivative , @xmath4 , and @xmath5 is the conserved angular momentum per unit mass.@xcite it is noted that these equations of motion have been derived in full detail elsewhere.@xcite in arriving at eqs .
( [ roft ] ) and ( [ phi ] ) , it is noted that the scalar approximation for an object that rolls without slipping was employed .
a full vector treatment of the intrinsic angular momentum of a rolling object on a cylindrically - symmetric surface can be found elsewhere.@xcite notice that one can easily decouple the above differential equations by inserting eq .
( [ phi ] ) into eq .
( [ roft ] ) .
the resulting differential equation can then be solved for @xmath6 , at least in principle , once the equation describing the cylindrically - symmetric surface , @xmath7 , is specified .
here we are interested in arriving at an expression for the radial distance in terms of the azimuthal angle , @xmath8 . using the chain rule and eq .
( [ phi ] ) , we construct a differential operator of the form [ oper ] = . by employing eqs .
( [ phi ] ) and ( [ oper ] ) , eq .
( [ roft ] ) can be transformed into an orbital equation of motion of the form [ rofphi ] ( 1+z^2)+(zz-(1+z^2))()^2-r+zr^4=0 .
this dynamical equation of motion describes the radial distance of the orbiting marble in terms of the azimuthal angle , for a given cylindrically - symmetric surface specified by @xmath7 , and equates to a non - linear differential equation that can be solved perturbatively .
as here we are interested in studying elliptical - like orbits , we choose an approximate solution for the radial distance of the orbiting body to be of the form [ sol ] r()=r_0(1- ( ) ) , where @xmath9 is the eccentricity of the orbit and @xmath10 is the precession parameter . for an orbiting body
whose radial distance is described by eq .
( [ sol ] ) , @xmath11 represents the average radial distance of the orbital motion .
we note that in what follows , @xmath9 will be treated as small and used as our expansion parameter , whereas the parameters @xmath10 and @xmath11 will be determined by eq .
( [ rofphi ] ) .
the precession parameter , @xmath10 , can be further understood through the relation [ delphi ] , where @xmath12 represents the angular separation between two like - apsides ( the angle between two consecutive apocenters or two consecutive pericenters ) of the orbital motion.@xcite notice that for @xmath13 , eq .
( [ sol ] ) equates to the perturbative approximation of the _ exact _ solution of newtonian gravitation , whose orbits equate to the well - known conic sections , to first - order in the eccentricity.@xcite when the azimuthal angle advances by @xmath14 , the orbiting body will find itself at the same radial distance and will precisely retrace its preceding orbital path .
hence , for @xmath13 , eq .
( [ sol ] ) approximately describes closed or stationary elliptical orbits with small eccentricities .
when @xmath15 , @xmath16 and eq .
( [ sol ] ) approximately describes precessing elliptical - like orbits with small eccentricities .
notice that for @xmath17 , eq . ( [ delphi ] ) implies that the angular separation between two consecutive like - apsides will be _ less than _ @xmath14 and the apsides will ` march backwards ' in the azimuthal direction over successive orbits .
contrarily , for @xmath18 , the angular separation between two consecutive like apsides will be _ greater than _ @xmath14 and the apsides will ` march forward ' in the azimuthal direction .
inserting eq .
( [ sol ] ) into eq .
( [ rofphi ] ) and keeping terms up to first - order in @xmath9 , we find a valid approximate solution to the orbital equation of motion when the algebraic expressions ^2&=&gr_0 ^ 3z_0[ell ] + ^2&=&[nu ] are satisfied , where @xmath19 and @xmath20 are radial derivatives of @xmath21 evaluated at the average radial distance , @xmath11 . in arriving at the above expressions , it is noted that the slope and the radial derivative of the slope were expanded about the average radial distance of the respective orbit .
the above expressions were presented in a previous work@xcite and then later used to find the 2d surfaces that generate newtonian and general relativistic orbits with small eccentricities.@xcite equation ( [ ell ] ) specifies the necessary angular momentum per unit mass needed , at a given average radius and for a given slope of the surface , for circular or elliptical - like orbits to occur .
equation ( [ nu ] ) yields a prediction for the precession parameter in terms of the slope and the radial derivative of the slope of a given surface at a given average radial distance . in the subsection that follows we present the equation that describes the slope of a cylindrically - symmetric spandex fabric , subjected to a central mass placed upon its surface , in a uniform gravitational field . this differential equation yields approximate expressions for the slope in the small and large slope regimes , which can be inserted into eq .
( [ nu ] ) to yield a theoretical prediction for the precession parameter in each regime . using eq .
( [ delphi ] ) , the approximate solution for the angular separation between two consecutive like - apsides for a marble orbiting on this spandex surface can be found in both the small and large slope regimes .
the equation that describes the slope of a cylindrically - symmetric spandex surface residing in a uniform gravitational field can be obtained through the use of the calculus of variations , as the surface will take the shape that _ minimizes _ the total potential energy of the central mass / elastic fabric system .
the total potential energy can be written as an integral functional , where we take into account the elastic potential energy of the spandex fabric and the gravitational potential energy of the surface of uniform areal mass density @xmath22 and of the central object of mass @xmath23 .
this resultant functional can then be subjected to euler s equation , which yields a differential equation that describes the slope of the spandex fabric.@xcite this resultant differential equation can be integrated and yields the expression [ shape ] rz(1-)=(m+_0r^2 ) , where we defined the parameter @xmath24 , where @xmath25 is the modulus of elasticity of the elastic material .
we note that this differential equation that describes the slope of the elastic surface was previously derived in an earlier work of one of the authors.@xcite notice that eq .
( [ shape ] ) describes the slope of the spandex fabric at radial distances @xmath26 , where @xmath27 is the radius of the spandex surface , and is dependent upon the modulus of elasticity , the mass of the central object , and the mass of the spandex fabric interior to this radial distance , which is explicitly seen through the @xmath28 term . equation ( [ shape ] ) corresponds to a quartic equation , which can be solved exactly for the slope , @xmath29 , but can not be integrated analytically to yield the shape of the surface , @xmath7 .
the exact expression for @xmath30 is rather awkward to work with but can be conveniently approximated in the small and large slope regimes .
these regimes correspond to where the surface is nearly horizontal and nearly vertical , respectively .
the theoretical work outlined in this section offers an interesting and non - trivial application of lagrangian dynamics and of the calculus of variations for an advanced undergraduate student . for the student who wishes to forego these derivations and focus on the experimental aspects of this project , eqs .
, , and can easily serve as a starting point . in the next section
we explore the behavior of the slope of the spandex surface in the small slope regime and , using eq .
( [ nu ] ) , arrive at an expression for the precession parameter .
interestingly , this theoretical prediction for the precession parameter can then easily be compared to experiment by directly measuring the angular separation between successive like - apsides and then comparing to the theoretical prediction of eq .
( [ delphi ] ) .
we first explore the behavior of the angular separation between consecutive like - apsides in the regime where the slope of the spandex surface is small and the surface nearly horizontal .
this physically corresponds to the region near the perimeter of the spandex fabric when a small central mass ( or no central mass ) has been placed upon the surface . in this small slope regime
where @xmath31 , we can expand the square root in the denominator of eq .
( [ shape ] ) in powers of @xmath32 . keeping only the lowest - order non - vanishing contribution and then solving eq .
( [ shape ] ) for @xmath30 , we obtain an approximate expression describing the slope of the elastic surface of the form [ smallslope ] z(r)()^1/3(m+_0r^2)^1/3 . inserting eq .
( [ smallslope ] ) into eq .
( [ nu ] ) yields an approximate expression for the precession parameter when the slope of the surface is evaluated at the average radial distance , @xmath11 , of the orbiting body . plugging this expression for @xmath10 into eq .
( [ delphi ] ) and solving for @xmath12 , we obtain an expression predicting the value of the angular separation between consecutive apocenters ( or between consecutive pericenters ) , which takes the form [ smallsol ] = 220^^1/2 ^ -1/2 . upon inspection of eq .
( [ smallsol ] ) , it is noted that the value of this angular separation between successive like - apsides is dependent upon the mass of the central object , @xmath23 , and the mass of the spandex fabric interior to the average radial distance of the elliptical - like orbit , @xmath33 .
these two competing terms are on equal footing when @xmath34 . for our spandex fabric , we measured the uniform areal mass density to be @xmath35 kg / m@xmath36 and found that the average radial distance of the elliptical - like orbits in the small slope regime is @xmath37 m. using these measured values , we find that these two terms are on equal footing for a central mass of about @xmath38 kg . hence ,
when the mass of the central object is less than @xmath39 kg , the mass of the spandex fabric interior to the average radial distance of the elliptical - like orbit becomes the dominant factor in determining the angular separation between consecutive like - apsides , as described by eq .
( [ smallsol ] ) . when the mass of the central object is much smaller than the mass term of the spandex fabric , eq .
( [ smallsol ] ) yields a limiting behavior for the angular separation of the form [ smallm ] _ m_0r_0 ^ 2197^^1/2 .
conversely , when the mass of the central object dominates over the mass term of the spandex fabric , eq .
( [ smallsol ] ) yields a limiting behavior for the angular separation of the form [ smallsigma ] _
m_0r_0 ^ 2220^^1/2 , where , physically , one must be careful to ensure that the elliptical - like orbits are still occurring in the small slope regime with this relatively large central mass .
upon comparison of the above approximate expressions , notice that eq .
( [ smallm ] ) predicts that the angular separation increases with increasing average radial distance whereas eq .
( [ smallsigma ] ) predicts a decreasing angular separation with increasing average radial distance , for a given central mass .
further notice that eq .
( [ smallsol ] ) predicts a minimal angular separation of @xmath40 , which occurs for orbits with vanishingly small average radial distances when the surface is void of a central mass . to experimentally test our theoretical prediction for the angular separation between successive apocenters and pericenters in the small slope regime
, we needed a cylindrically - symmetric spandex fabric , capable of supporting elliptical - like orbits of a marble upon its surface , and a means of precisely determining the polar coordinates of the orbiting marble .
a 4-ft diameter mini - trampoline was stripped of its trampoline surface and springs .
a styrofoam insert was fashioned to exactly fit the interior of the trampoline frame , which collectively formed a flat , table - like structure . with the styrofoam insert in place
, a nylon - spandex fabric was draped over the trampoline frame , where we ensured that no pre - stretch or slack was introduced .
the fabric was then securely fastened to the frame by means of a ratchet strap and the styrofoam insert was removed , allowing the fabric to hang freely .
kg , in the small slope regime .
each circle marks the position of the rolling marble , in intervals of 1/60 of a second , as it orbits the central mass .
the three orbits imaged here were found to have angular separations between successive apocenters of @xmath41 and eccentricities of @xmath42 , respectively.,width=566 ] it is noted that the shape of the freely hanging spandex fabric can easily be obtained . when the spandex fabric is void of a central mass the slope of the surface is small , hence , the approximation for the slope given by eq .
( [ smallslope ] ) is valid everywhere on the surface . setting @xmath43 in eq .
( [ smallslope ] ) and integrating we obtain the height of the surface , which is given by the expression [ shape0 ] z(r)_m=0=(2_0)^1/3(r^4/3-r^4/3 ) , where @xmath44 m is the inner radius of the trampoline frame and we chose the zero point of our coordinate system to coincide with @xmath45 . by measuring the height of the spandex fabric at the center of the apparatus , which was found to be @xmath46 m , and by knowing the areal mass density of the spandex fabric , we experimentally obtain a value of @xmath47 m / kg .
having an experimental value for @xmath48 is necessary as we wish to compare our predicted values for the angular separation between any successive like - apsides to the experimentally measured values .
a camera capable of taking video clips at a rate of 60 frames per second was mounted above the spandex surface .
a plumb line , strung from the camera s perched location , was used to align the trampoline frame below and to determine the precise center of the fabric where the central mass should be placed .
a ramp was mounted near the perimeter of the trampoline and orientated in the tangential direction so elliptical - like orbits with varying eccentricities , and hence varying average radial distances , could be generated by simply varying the initial speed of the marble .
several short video clips of elliptical - like orbits were then generated for a range of central masses and eccentricities .
the video clips were imported into tracker,@xcite a video - analysis and modeling software program , where the marble s radial and angular coordinates could readily be extracted one frame at a time ( see fig .
[ fig : smallorbit ] ) . the apocenters and pericenters of the orbiting marble
were identified by scrolling through the instantaneous radii and determining the maximum and minimum values , along with the accompanying angular coordinates . to remain within the small slope regime , only the first five orbits of
a given run were considered , which yielded a maximum of ten data points per run .
these data were then imported into excel for analysis .
, versus the average radial distance , @xmath11 , of elliptical - like orbits in the small slope regime for central masses ranging from @xmath43 kg to @xmath49 kg .
, width=566 ] the angular separation between successive like - apsides could easily be obtained from the measured angular coordinates .
the average radial distance , @xmath11 , was found by averaging over the measured instantaneous radii over one orbit ; one data point was hence generated from these two quantities .
it is noted that the radius of a successive apocenter or pericenter was found to decrease ( dramatically in the small slope regime ) due to frictional losses and drag . to account for this loss ,
the average radial distance of two successive apocenters or pericenters was calculated and used in the analysis in determining the eccentricity of an orbit.@xcite elliptical - like orbits with eccentricities falling within the range of @xmath50 were kept for analysis whereas orbits with eccentricities outside of this adopted range were systematically discarded . using our aforementioned measured value of @xmath48 , the theoretical curves of the angular separation between successive like - apsides , as described by eq .
( [ smallsol ] ) , can be plotted versus the average radial distances , @xmath11 , of an elliptical - like orbit for a given central mass .
these theoretical curves can then be compared to the measured values of @xmath12 , both are collectively displayed in fig .
[ fig : smallslope ] . upon examination of fig .
[ fig : smallslope ] , notice that the measured values of @xmath12 tend to reside below the theoretical curves , with larger discrepancies occurring for smaller central masses .
the authors hypothesize that this is predominantly due to an increased frictional loss for an orbiting marble about smaller central masses . as the central mass decreases , the stretching of the spandex fabric decreases , and the warping of the spandex fabric due to the orbiting marble s weight increases .
this generates a scenario were the marble is having to effectively climb out of its own gravitational
well , more so then for a larger central mass , and encounters an increased frictional loss resulting in the marble reaching its apocenter earlier than it would without this loss .
this can be compared to 2d projectile motion where the projectile is subjected to a retarding force . for a projectile subjected to a large retarding force
, one finds that it reaches a smaller peak height in a shorter horizontal distance than it would for a smaller retarding force .
further note that the measured values for @xmath12 tend to increase for increasing average radial distance for zero central mass and decrease for increasing average radial distances for non - zero central mass , inline with the theoretical predictions . in the next section ,
we explore the behavior of the angular separation between successive apocenters in the large slope regime .
we find that within this regime , the angular separation takes on values _
_ 360@xmath1 for very small radial distances and large central masses .
this corresponds to the apsides marching forward in the azimuthal direction over successive orbits .
interestingly , elliptical - like orbits around non - rotating , spherically - symmetric massive objects in general relativity are also found to have angular separations between successive like - apsides taking on values greater than 360@xmath1 . a comparison between these gr orbits and the orbits on a spandex fabric will be discussed in the appendix .
we now wish to explore the behavior of the angular separation between successive apocenters in the large slope regime .
this regime corresponds physically to the region deep inside the well , namely , on the spandex fabric where @xmath11 is small , when a large central mass has been placed upon the surface . in this large slope regime ,
where @xmath51 , we can again expand the square root in the denominator of eq .
( [ shape ] ) by first pulling @xmath30 out of the square root and then expanding in powers of @xmath52 . keeping only the lowest - order contribution in this expression and then solving eq .
( [ shape ] ) for @xmath30 , the slope takes on the approximate form [ largeslope ] z(r)1 + , where we dropped the areal mass density term as it is insignificant in this regime . now inserting eq .
( [ largeslope ] ) into eq .
( [ nu ] ) and then plugging this expression for @xmath10 into eq .
( [ delphi ] ) , we arrive at an expression describing the angular separation between successive apocenters in the large slope regime , which takes the form [ largesol ] = 360^ , where we evaluated @xmath53 at the average radial distance , @xmath11 , of the elliptical - like orbits .
notice that this expression , like its counterpart in the small slope regime , also has two terms competing for dominance .
however , as the above expression is only valid for @xmath51 , with @xmath30 approximately given by eq .
( [ largeslope ] ) , eq .
( [ largesol ] ) is nt valid when @xmath54 . for a large central mass and a _ very _ small average radial distance , when the orbiting marble is deep within the well , eq .
( [ largesol ] ) yields a limiting behavior for the angular separation of the form [ largesolapprox ] _
notice that when @xmath55 , eq .
( [ largesolapprox ] ) predicts that @xmath56 .
hence , as @xmath57 approaches and then becomes larger than the @xmath58 , successive apocenters are predicted to transition from marching backwards to marching forwards in the azimuthal plane over successive orbits .
we now wish to compare our theoretical prediction for the angular separation in the large slope regime , given by eq .
( [ largesol ] ) , to our experimentally measured values.@xcite if the modulus of elasticity was truly constant for the spandex fabric , then the experimentally obtained value of @xmath47 m / kg in the small slope regime could be used here .
this , however , is not the case and has been discussed extensively elsewhere.@xcite although the modulus of elasticity has been shown to be a function of the stretch , it is approximately constant in each regime . therefore , the analysis presented in this manuscript is valid when applied to each regime separately .
, versus the average radial distance , @xmath11 , of elliptical - like orbits in the large slope regime for central masses that range from @xmath59 kg to @xmath60 kg .
, width=566 ] to obtain @xmath48 in this large slope regime , we note the slope of the spandex surface , given by eq .
( [ largeslope ] ) , can be integrated , which yields the height of the spandex fabric as a function of the radius and the central mass .
performing this integration and then evaluating at a small fixed radius @xmath61 m on the spandex fabric , we obtain the expression [ alphals ] = ( m - m_0 ) , where , here , we set @xmath62 where @xmath63 corresponds to the smallest central mass probed .
this expression gives a scaled height of the fabric as a function of the central mass , @xmath23 , for a given radius of the fabric , @xmath64 , in the large slope regime .
as all of the quantities in eq .
( [ alphals ] ) save @xmath48 can be measured directly , the slope of the best fit line of a plot @xmath65 vs @xmath66 yields the value of the parameter @xmath48 . by orientating our camera to obtain a side view of the fabric s surface , we subjected our spandex fabric to the same range of central masses used in generating particle orbits in this large slope regime .
we then imported this video of the shape profile of the spandex fabric into tracker and measured the heights of the fabric at the aforementioned fixed radius as a function of the central mass @xmath23 .
the slope of this best fit line yields a value of @xmath67 m / kg . the experimental setup and procedure for probing the large slope regime of the spandex fabric was quite similar to that of the small slope regime , with a few notable exceptions .
as here we wish to probe small radial distances for large central masses , we positioned a small spherical object of radius @xmath68 cm in the center of the fabric .
a sling , capable of supporting several kilograms of mass , was fashioned and fastened to the spherical object from below the fabric with the help of an adjustable pipe clamp .
this setup allowed us to probe a range of large central masses while maintaining a constant diameter circular well .
the average radii of successive orbits do not decrease as drastically when the orbiting body is deep within the well as compared to orbits in the small slope regime , hence many orbits can potentially be analyzed for any given run .
the angular separations , @xmath12 , were found and the corresponding average radial distances , @xmath11 , were obtained .
again , ellliptical - like orbits with eccentricities found to lie within the range @xmath50 were kept and further analyzed , those not in this defined range were systematically discarded . using our experimentally determined value of @xmath67 m / kg , the theoretical curves of the angular separation , @xmath12 , described by eq .
( [ largesol ] ) , can be plotted versus the average radial distance , @xmath11 , of an elliptical - like orbit for a given central mass along with the measured values and are collectively displayed in fig .
[ fig : largeslope ] .
notice that as the average radial distance decreases , the angular separation increases .
the horizontal dashed line displayed in fig .
[ fig : largeslope ] corresponds to @xmath69 .
hence , for data points lying above this line , the apocenter marched forward in the azimuthal direction when compared to that of the previous orbits .
we note that when the average radial distances of the elliptical - like orbits were greater than @xmath70 cm , the measured values of @xmath12 were found to become substantially smaller than those shown in fig . [
fig : largeslope ] . as our theoretical prediction for @xmath12 is only valid in the large slope regime , we neglected the data points beyond this radial distance . for radial distances smaller than @xmath71 cm
, the measured values of @xmath12 were found to be quite sporadic . for orbits with small average radial distances ,
the angular velocity becomes quite large and hence only a handful of particle locations are imaged ( @xmath72 ) for a given orbit ; this yields an angular separation of @xmath73 between imaged locations and gives rise to a fairly large uncertainty in the angular measurement . here
we display the data points with average radial distances @xmath74 .
in this manuscript , we arrived at expressions describing the angular separation between successive like - apsides for a marble rolling on a cylindrically - symmetric spandex surface in a uniform gravitational field in both the small and large slope regimes .
we found that a minimal angular separation of @xmath0 was predicted for orbits with small radial distances when the surface is void of a central mass .
we also showed that for a marble orbiting with a small radius about a large central mass , corresponding to when the marble is deep within the well , the angular separation between successive like - apsides takes on values greater than 360@xmath1 .
for these orbits , successive apocenters ` march forward ' in the azimuthal direction in the orbital plane , reminiscent of particle orbits about spherically - symmetric massive objects as described by general relativity .
these theoretical predictions were shown to be consistent with the experimental measurements .
although the orbital characteristics of a marble on a warped spandex fabric fundamentally differ from those of particle orbits about a spherically - symmetric massive object as described by general relativity , the authors contend that embracing the conceptual analogy between the two is insightful for the student of gr .
hence , this proposed theoretical / experimental undergraduate research project can be extended into the realm of general relativity , where a study of the elliptical - like orbits about a spherically - symmetric massive object with small eccentricities can be compared and contrasted to the theoretical work contained within this manuscript . in the appendix ,
we study the physics of the analogy and find that some striking similarities exist between these elliptical - like orbits .
although differing in functional form , we show that the angular separation between like - apsides for marbles orbiting in the large slope regime and for particles orbiting near the innermost stable circular orbit both diverge for vanishingly small radial distances in a somewhat similar manner .
we also show that the areal mass density of the spandex fabric plays the role of the vacuum energy density in gr , in regards to their competing relationship with their respective central mass , in the expressions for the angular separation .
in this section , we outline the general relativistic treatment of obtaining elliptical - like orbits with small eccentricities about a spherically - symmetric massive object in the presence of a constant vacuum energy density , or equivalently , a cosmological constant .
we arrive at the necessary conditions for elliptical - like orbits to occur in this spacetime and compare and contrast the expression for the angular separation between consecutive like - apsides to that obtained for a rolling marble on the warped spandex fabric in _ both _ the small and large slope regimes .
although these expressions differ in their functional forms , there are some striking similarities . here
we argue that a thorough study of the physics of the analogy , a marble orbiting on a spandex surface , can aid in the learning of free particle orbits , or geodesics , in general relativity .
the addition of the cosmological constant to the field equations of general relativity can be traced back to einstein himself .
one of his first applications of this new theory of gravitation , which describes gravity as the warping of space and time due to the presence of matter and energy , was to the field of cosmology .
einstein added the cosmological constant to the relativistic equations only after realizing that these equations fail to offer a static cosmological solution to describe the seemingly static universe of his time .
when the universe was later discovered to be expanding , the cosmological constant became an unnecessary addition and einstein missed the opportunity of predicting an expanding universe , as his cosmological constant - free equations seemed to be implying . with the much later discovery of a late - time _ accelerated _ expansion of the universe , the importance of a cosmological constant has once again arisen , but with the cosmological constant now recast into the form of a vacuum energy density .
mathematically , a cosmological constant , @xmath75 , which finds itself on the curvature side of the field equations of gr , can be equivalently reformulated into a constant vacuum energy density , @xmath76 , which resides on the matter / energy side .
these quantities are related through the algebraic relation [ rholambda ] _
0=(c^2/8 g ) , where @xmath77 is newton s universal constant and @xmath78 is the speed of light.@xcite the reason for the relabeling of the cosmological constant emerges in quantum field theory , where zero - point fluctuations , corresponding to particles and anti - particles coming into and out of existence , are predicted .
these vacuum fluctuations give rise to a nonzero vacuum energy density .
it is noted that the cosmological constant , and therefore the constant vacuum energy density , can be of either sign .
observational evidence , however , suggests that we live in a universe with a small but nonzero , _ positive _ vacuum energy density whose presence is needed to explain the observed accelerated expansion of the universe .
using the schwarzschild coordinates @xmath79 , the equations of motion describing an object of mass @xmath80 orbiting about a non - rotating , spherically symmetric massive object of mass @xmath23 , in the presence of a cosmological constant , in general relativity are of the form + -+-c^2r&=&0,[releq ] + & = & [ relphi ] , where in this section a dot indicates a derivative with respect to proper time and @xmath5 is the conserved angular momentum per unit mass.@xcite it is noted that the last two terms in eq .
( [ releq ] ) correspond to relativistic terms which are not present in newtonian gravitation . by setting these two relativistic terms equal to zero ,
one obtains the equations of motion of newtonian theory ; these yield a solution for the radial distance that describes orbits that equate to the conic sections.@xcite the @xmath81 term equates to a general relativistic correction term to newtonian gravitation , present even for a vanishing cosmological constant , and offers a small-@xmath53 modification to newtonian theory .
this correction term demands that the closed elliptical orbits of newtonian theory are replaced with precessing elliptical - like orbits , which will become apparent in the paragraphs that follow . for a nonzero cosmological constant , notice that the last term in eq .
( [ releq ] ) offers a large-@xmath53 modification to the orbits . as in our treatment of elliptical - like orbits on a warped spandex surface ,
we are interested in obtaining the solution for the radial distance of the orbiting body in terms of the azimuthal angle . using the chain rule and eq .
( [ relphi ] ) , we construct a differential operator of the form [ relop ] = , where @xmath82 is the proper time . by employing eqs .
( [ relphi ] ) and ( [ relop ] ) , eq .
( [ releq ] ) can be transformed into an orbital equation of motion of the form [ releqphi ] -()^2+r^2-r+-r^5=0 .
the solution to this orbital equation of motion describes the possible particle orbits about a non - rotating , spherically - symmetric massive object in the presence of a constant vacuum energy density .
we wish to examine elliptical - like orbits with small eccentricities and first examine the case of a vanishing cosmological constant . inserting the perturbative solution of the form [ relsol ] r()=r_0(1- ( ) ) into eq .
( [ releqphi ] ) , setting @xmath83 , and keeping terms up to first - order in @xmath9 , we find a valid solution when the constants and parameters obey the relations ^2&=&gmr_0(1-)^-1,[relell2norho ] + ^2&=&1-[relnu2norho ] .
these algebraic expressions correspond to the conditions necessary for elliptical - like orbits with small eccentricities about a non - rotating , spherically - symmetric massive object , void of vacuum energy , and have been discussed elsewhere.@xcite it is noted that when the general relativistic term @xmath84 is set equal to zero , we recover the stationary elliptical orbits of newtonian gravitation , characterized by @xmath13 .
notice that @xmath10 becomes complex when @xmath85 ; this implies that elliptical - like orbits are _ only _ allowed for radii larger than this limiting value .
this threshold radius , @xmath86 , is known as the innermost stable circular orbit .
further notice that for radii smaller than @xmath87 , both the angular momentum per unit mass , @xmath5 , and the parameter @xmath10 become complex .
this implies that there are no circular orbits , stable or unstable , allowed for an object orbiting about a non - rotating , spherically - symmetric massive object for these small radii .
using eqs .
( [ delphi ] ) and ( [ relnu2norho ] ) , the angular separation between successive apocenters takes the form [ reldelphinorho ] = 360^(1-)^-1/2 , where we used our aforementioned definition of @xmath86 .
note that as the average radius , @xmath11 , of an elliptical - like orbit with a small eccentricity takes on values that approach the innermost stable circular orbit , @xmath86 , the angular separation between successive apocenters increases without bound . to further explore this limiting behavior ,
we define [ smallr ] r_0r_isco+r , where @xmath88 is a vanishingly small radial distance .
inserting eq .
( [ smallr ] ) into eq .
( [ reldelphinorho ] ) and expanding for small @xmath53 , we find the limiting behavior for the angular separation to be of the form [ rll6gm ] _
r6gm / c^2360^ , near the innermost stable circular orbit .
recall that the functional form of the angular separation for a marble orbiting on the warped spandex fabric in the large slope regime took on a limiting behavior of the form [ rllam ] _
r_0 m . upon comparison of eqs .
( [ rll6gm ] ) and ( [ rllam ] ) , we note that both expressions diverge in the limit of vanishingly small distances @xmath11 and @xmath53 , however , their functional forms differ .
@xmath89 for the orbiting object about the non - rotating , spherically - symmetric massive object near the innermost stable circular orbit whereas @xmath90 for the rolling marble on the spandex fabric in the large slope regime for small radial distances . in the vanishingly small @xmath11 regime
the slope of the spandex surface diverges , which can be seen upon inspection of eq .
( [ largeslope ] ) , when the central mass is assumed to be point - like .
we further note that the parameter @xmath24 plays the role of @xmath91 in the particle orbits of general relativity .
both constants set the scale of their respective theories and determine the value of the angular separation for a given central mass and average radial distance .
we wish to push the conceptual analogy further between particle orbits in general relativity and rolling marble orbits on a warped spandex fabric in the small slope regime by considering the functional form of the angular separation for a nonzero cosmological constant .
inserting eq .
( [ relsol ] ) into eq .
( [ releqphi ] ) , with @xmath92 , and again keeping terms up to first - order in @xmath9 , we find that eq .
( [ relsol ] ) is a valid approximate solution when ^2&=&gr_0(1-)^-1(m-2r_0 ^ 3_0),[relell2 ] + ^2&=&1 - -6(1-)[relnu2 ] , where we eliminated @xmath75 in favor of @xmath76 by using eq .
( [ rholambda ] ) . now inserting eq .
( [ relnu2 ] ) into eq .
( [ delphi ] ) and solving for @xmath12 , we obtain the relativistic expression for the angular separation between consecutive like - apsides , which takes the form [ reldelphi ] = 360^(1-)^-1/2 ^ -1/2 .
although it is tempting to identify @xmath93 as the mass of the vacuum interior to the average radial distance of the elliptical - like orbit , as would be the case in classical physics , one must be careful when calculating volume in the non - euclidean spacetime of general relativity . recall that the functional form of the angular separation between successive like - apsides for a marble orbiting on a warped spandex fabric in the small slope regime is of the form [ smallsol2 ] = 220^^1/2 ^ -1/2 . comparing the denominators of the last terms in eqs .
( [ reldelphi ] ) and ( [ smallsol2 ] ) , we find that the areal mass density , @xmath22 , of the spandex fabric plays the role of a _ negative _ vacuum energy density , @xmath94 , of spacetime . in each respective theory
, the mass of the central object competes for dominance with the mass term associated with the vacuum / spandex fabric . as observational evidence implies that we live in a universe with a nonzero ( albeit tiny ) , _ positive _ vacuum energy density
, one can envision a spandex fabric with a negative areal mass density when the conceptual analogy of particle orbits in general relativity and rolling marble orbits on a warped spandex fabric is employed .
this fictional spandex fabric would feel a repulsive force from the uniform gravitational field of the earth and tend to hang _
up _ towards the ceiling rather than down towards the floor .
this contribution to the total warping of the spandex surface gives rise to a repulsive force on the orbiting marble , effectively competing with the attractive force of the central mass contribution .
the term apsides refers to the maximum and minimum radial distances of an orbiting body .
the apocenter ( pericenter ) refers to the point on the elliptical orbit that is farthest from ( closest to ) the central mass.@xcite the exact solution to newtonian gravitation is of the form @xmath95 , where @xmath11 is known historically as half of the latus rectum . for small eccentricity this expression can be expanded and equates to eq .
( [ sol ] ) to first - order in @xmath9 for @xmath13 . for an apocenter - pericenter - apocenter orbit ,
having radii @xmath99 and @xmath100 , respectively , the average apocenter distance was first calculated through @xmath101 and then the eccentricity was calculated through @xmath102 .
likewise , for a pericenter - apocenter - pericenter orbit , having radii @xmath103 and @xmath104 , respectively , the average pericenter distance was first calculated through @xmath105 and then the eccentricity was calculated through @xmath106 . in the large slope regime , where an abundance of elliptical - like orbits are generated for a given run , we choose to neglect the angular separation between successive pericenters .
this differs from the small slope regime where we measured the angular separation between successive apocenters and between successive pericenters . | we present a theoretical and experimental analysis of the elliptical - like orbits of a marble rolling on a warped spandex fabric .
we arrive at an expression describing the angular separation between successive apocenters , or equivalently successive pericenters , in both the small and large slope regimes .
we find that a minimal angular separation of @xmath0 is predicted for orbits with small radial distances when the surface is void of a central mass .
we then show that for small radii and large central masses , when the orbiting marble is deep within the well , the angular separation between successive apocenters transitions to values greater than 360@xmath1 .
we lastly compare these expressions to those describing elliptical - like orbits about a static , spherically - symmetric massive object in the presence of a constant vacuum energy , as described by general relativity . |
the reaction mechanism leading to the knockout of a single proton by a real photon has been the subject of some debate recently .
some nonrelativistic analyses @xcite suggest that the direct knockout ( dko ) contribution may be very small compared to the data .
accordingly it was concluded that meson exchange current ( mec ) contributions must be the main mechanism responsible for the observed cross sections .
similar conclusions were reported earlier for the nonrelativistic analyses carried out by miller _
@xcite for ground state transitions for the reaction @xmath4n at 60 and 72 mev , and by ireland _
@xcite for @xmath5 reactions on several nuclei for @xmath6 near 60 mev ( the same nuclei involved in the discussion reported in ref .
the above conclusions do not seem consistent with the findings by ryckebusch _
these authors find mec effects to be relatively small for ground state transitions .
the above statements illustrate the existing difficulty of arriving at a consensus within the nonrelativistic framework as to the extent of contributions from processes beyond simple direct knockout to ground state transitions in @xmath7 reactions .
although the differing views stated above appear to be somewhat dependent on the nuclear models used in the nonrelativistic calculations , they are , however , symptomatic of our incomplete understanding of the nature of the reaction mechanism for photonuclear reactions .
these results are quite different from those of a recent relativistic analysis of johansson _ et al .
_ @xcite who find that , for an incident photon energy of 60 mev , the dko contribution accounts for most of the observed data , with no indication of any systematic sharp deviation from the data at this energy . in this paper
we extend the analysis reported in ref .
@xcite , for a photon energy of 60 mev , to a much wider range of data .
we consider several data sets for the @xmath5 reaction [ and some data on the inverse reaction @xmath8 on a number of target nuclei and covering a range of photon energies extending well into the @xmath9-resonance region .
the spectroscopic factors and wave functions used in the calculations are fixed at the values obtained from a parallel analysis of the @xmath0 reaction on the same target nuclei .
the objective of this study is to use this type of constrained analysis to gain some insight into the role of the dko mechanism and to see if a consistent description of the available data is possible .
section [ rel_obs ] outlines the relativistic calculations for the direct knockout contribution to the @xmath5 reaction .
results of the calculations and details of the comparisons with data are given in section [ disc ] .
our conclusions are given in section [ concl ] .
the differential cross section due to the direct knockout contribution to the @xmath5 reaction has been given previously @xcite but we provide it here again for ease of reference .
the relativistic expression for the differential cross section leading to a specific final state of the residual nucleus can be written as @xmath10 where @xmath11 and @xmath12 are the spin projections of the bound and continuum protons .
we denote the four - momentum of the final proton @xmath13 and the four - momentum of the incident photon as @xmath14 .
the four - vector @xmath15 is the photon polarization vector with two polarization states @xmath16 , and summation is implied over repeated greek indices .
the recoil factor @xmath17 is given in any frame by @xcite @xmath18 the four - momentum of the recoil nucleus is denoted by @xmath19 .
the function @xmath20 is @xmath21 where the wave functions of the continuum and bound nucleons , denoted @xmath22 and @xmath23 respectively , are solutions of the dirac equation containing appropriate potentials @xcite .
the @xmath24 matrix @xmath25 , operating on the nucleon spinors , is given by @xmath26 .\end{aligned}\ ] ] the ingredients of the model are basically the same as used in an analysis of light to medium weight nuclei at 60 mev @xcite : the bound state protons are described by solutions of a dirac equation containing the relativistic hartree potentials of blunden and iqbal @xcite , while the final state continuum proton is described by solutions of a dirac equation containing complex phenomenological optical potentials obtained from fits to proton elastic scattering data @xcite .
given these potentials , the only parameters left to determine are the spectroscopic factors .
for the light nuclei : @xmath27b , @xmath3c and @xmath28o , we have obtained the spectroscopic factors by fitting the results of our @xmath0 model @xcite to available data .
the @xmath29pb data are not suitable for analysis using this model because of the lack of coulomb distortions for the incident and final electrons .
the spectroscopic factors used in this case are those of udias _ et al .
_ @xcite who have performed a relativistic analysis of the @xmath29pb data . in the following
we show the results of our calculations compared to the experimentally determined cross sections for several nuclei covering a wide energy range .
we have performed calculations for the @xmath5 reaction on several target nuclei , over a wide range of energies .
these are compared to existing data in order to assess the extent to which the direct knockout mechanism contributes to the observed cross sections .
the ingredients of the calculations have all been determined elsewhere and since there are no adjustments made , the results can be considered as predictions of the model . in the graphs to be discussed below
there are curves corresponding to several different calculations .
the description of the calculations represented by each of these curves is as follows : 1 . dashed curve energy- ( e- ) dependent parameterization of the dirac optical potentials specific to a single nucleus @xcite while the bound state wave function is obtained through a dirac - hartree calculation @xcite ; 2 . dotted curve
e - dependent parameterization of the dirac optical potentials specific to a single nucleus and the binding potential has a woods - saxon form ; 3 . solid curve energy- and mass- [ ( e+a)- ] dependent parameterization of the dirac optical potentials and the same dirac - hartree bound state wave function as in curve 1 above
dot - dashed curve curve 3 divided by a factor of 2.0 to bring the model calculations close to the data .
all the figures shown below use this designation of curves .
the first three are simply for calculations using a variety of existing potential models in order to provide some feeling for the sensitivity of the results to variations in these ingredients .
the dot - dashed curve is only relevant to graphs shown for the @xmath3c target .
a considerable amount of data are available for this target .
we have made comparisons of our relativistic dko model calculations with these data , concentrating mainly on ground state transitions .
these comparisons are shown in figs . 1 - 3
. data for four of the energies shown in fig . 1 : @xmath6 = 49.0 , 58.4 , 67.8 and 78.5
mev were reported by springham _
_ @xcite and are obtained using the tagged photon facility at mainz .
the absolute magnitude of their cross sections was obtained by normalizing the data at each energy to data taken by mathews _
@xcite for the @xmath30 reaction , data which include both the ground and first excited state of the residual @xmath31b nucleus .
the data of aschenauer _ et al . _
@xcite , at photon energies of @xmath6 = 45.0 and 54.0 mev were obtained at the max - lab at the university of lund .
these data were normalized completely independently of any previous experiment , and found to be consistent with existing data within systematic errors .
the data at @xmath6 = 73.5 mev from rauf @xcite were also obtained at the max - lab .
these data were normalized to previous measurements including those of mathews
_ @xcite . the most obvious feature apparent in fig .
1 is that the calculations tend to form a narrow band lying above the data .
note that there is not much sensitivity to reasonable variation of the ingredients of the model .
the light dot - dashed curve in all the figures shows the solid curve divided by a factor of 2 , and this brings the curve close to the data in all cases .
the calculated curves have the correct shapes and the variation of magnitude with incident photon energy also seems to be correct , but the curves lie consistently above the data by a factor of 2 .
figure 2 shows the differential cross section as a function of photon energy for four different proton angles .
the experimental data are taken from ruijter _
_ @xcite ; the experiments were also performed at the max - lab at the university of lund .
absolute normalization of these data was obtained independent of any other experiment and the results were found to be consistent with a large amount of other data .
again the calculated curves lie above the data by close to a factor of 2 .
we see that for photon energies above 40 mev the energy and angular dependence are quite well reproduced by our model , but results lie above the data by a factor of 2 . in an attempt to compare to other data as well
, we consider two experiments in which the first excited state of @xmath31b at 2.12 mev ( 1/2@xmath32 ) could not be resolved from the ground state .
the experiment of mori _ et al . _
@xcite was performed at the laboratory of nuclear science at tohoku university .
the experiment of harty _ et al . _
@xcite was performed at mainz .
the differential cross sections of both experiments were normalized without reference to any other experimental results and found to be consistent with other data . in order to compare to some data which do not contain the first excited state , ground state data from other sources
were multiplied by a factor of 1.27 by both groups .
this factor is assumed to account for population of the first excited state in @xmath31b being @xmath3327% as probable as population of the ground state over the range of energies considered in both experiments .
we adopt this factor in what follows . in the left - hand column of fig .
3 we show the measured energy distributions of mori _ et al . _ @xcite as the circular data points .
the triangular points are taken from the data of harty _ et al .
the angles at which these data were taken do not coincide with the angles of the experiment by mori _
et al._. for this reason the triangles shown at 66.0@xmath34 represent cross sections which have been averaged for proton angles of 63.3@xmath34 and 68.4@xmath34 , while on the graph labeled 91.1@xmath34 we show data averaged for 88.5@xmath34 and 93.5@xmath34 . note
that the 88 mev data point of harty _ et al .
_ lies almost on top of the 87.8 mev data point of mori _ et al . _
showing the consistency between the two data sets .
the curves are calculated assuming that the recoil nucleus is left in its ground state , and then multiplied by 1.27 .
the curves again lie above the data by a factor of 2 at low photon energies , but at the higher energies of the mainz experiment the calculations seem to move closer to the data . the right - hand column of fig .
3 shows the angular distributions obtained by harty
_ et al . _
@xcite compared to calculations as discussed in the previous paragraph .
as the energy of the incident photons increases , the calculations seem to move from lying above the data by about a factor of 2 , to falling within the error bars for the data at the highest energy .
of course , because of the large error bars at larger proton angles the trend is not definitive , but it is suggestive .
the picture that is emerging for the status of the comparisons for the @xmath3c case can be further clarified by looking at the data available for @xmath35 reactions on this nucleus as well as those leading to its formation as a residual nucleus .
the data are those obtained recently by bright _
@xcite at uppsala .
figure 4 shows comparisons to the data at proton energies of 98 and 176 mev for ground state transitions to @xmath3c and @xmath36n residual nuclei .
the former reaction is the inverse of the @xmath5 reactions discussed above .
the comparisons for the @xmath3c residual nucleus are shown on the right hand side of the figure . using the same wave functions and spectroscopic factors as in fig . 1
, we find that the calculations for @xmath37 = 98 mev lie slightly above the data at all angles except for the last point at @xmath38140@xmath34 . at 176 mev
the calculations are closer to the data except for the large angles .
thus the data for the inverse reaction confirm the behavior alluded to above ; at lower energies the calculations seem to overestimate the cross sections . the comparisons on the left - hand side of fig . 4 present a somewhat different picture .
using the maximum value for the spectroscopic factor , the relativistic calculations for radiative capture on @xmath3c are close to or below the data . with a more realistic value of the spectroscopic factor
the calculations will be further reduced in magnitude .
this situation is in clear contrast to the cases discussed above .
it should be noted , however , that with a spectroscopic factor in the range 0.51.0 ( the maximum possible value is 1.0 in this case ) , the contributions of the knockout mechanism to the reaction are substantial in the region of lower missing momenta .
it is worthwhile to point out here that there are also unresolved difficulties for the @xmath3c target in the @xmath0 reaction , in addition to the difficulties discussed in the current work .
the data from nikhef @xcite are for kinematics with a fixed final proton kinetic energy of @xmath37 = 70 mev and nonrelativistic calculations shown in that paper can not reproduce the shape of the distribution in missing momentum . in particular
when the calculations are scaled to fit the peak for positive missing momenta , the calculations fall below the data for negative missing momenta .
our relativistic calculations show exactly this behaviour and the spectroscopic factor that we have used in this work is obtained by matching to the positive missing momentum peak .
a proposed solution to this problem was to adjust the ratio of transverse to longitudinal response functions , and when this ratio was adjusted to @xmath11.3 @xcite the shape of the missing momentum distribution was reproduced .
this problem was considered further by van der steenhoven @xcite who found no justification for this enhancement factor .
the newer data from mainz on this nucleus , reported by blomqvist _
_ @xcite have a higher final proton kinetic energy of @xmath39 90 mev . in this case , with the increase in normalization of the data by a factor of 1.19 @xcite , both nonrelativistic and our relativistic calculations can describe the shape of the measured missing momentum distribution using the spectroscopic factor as obtained from matching to the positive missing momentum peak of the nikhef data .
this behaviour is consistent with the current results for the @xmath5 reaction on this nucleus : at low final proton energies model calculations differ from the data and as the final proton energy increases the calculations move closer to experimental results .
the data for proton knockout from the @xmath40 orbital in @xmath3c seem to indicate that at low energies a simple shell model description is not adequate to explain the data .
a more complete description might possibly involve inclusion of the deformed nature of the ground state wave function through a configuration mixing picture .
data for both the @xmath41 and @xmath5 reactions has been obtained for a @xmath27b target by de bever @xcite at two different energies .
the data for knockout of a @xmath40 proton leading to the ground state of @xmath42be are shown in fig .
spectroscopic factors were obtained by scaling the model calculations to the @xmath41 momentum distribution ( or reduced cross section ) data for the @xmath37 = 70 mev case .
the other @xmath41 and @xmath5 curves were then calculated without any adjustment of the parameters .
it should be noted that the optical potentials used here are parameterized using proton elastic scattering data on targets from @xmath3c to @xmath29pb . as a result of a lack of proton elastic scattering data on @xmath27b
, we simply extrapolate the ( e+a)-dependent potentials for use with a target lighter than @xmath3c .
the @xmath5 calculations for this nucleus are quite sensitive to changes in the potentials used to generate the nuclear wave functions . in spite of this , and the fact that @xmath27b is not a closed shell nucleus , the dko clearly produces results in the neighbourhood of the data . figure 6 shows the differential cross section as a function of proton angle for knockout of a valence proton from an @xmath28o target , leading to the ground state of @xmath43n .
the photon energy range is the largest available , with eight energies in the range 60 mev @xmath44 361 mev .
the data come from three sources : miller _ et al .
_ @xcite provide data points at energies of 60 an 72 mev , while data shown at 60 , 80 and 100 mev are from findlay and owens @xcite .
the high energy data for photons in the range 196 mev @xmath44 361 mev are from adams _ et al .
_ @xcite . at low energies
the calculated curves are generally close to the data , reproducing the magnitude and shapes quite well . for the higher energy data of adams
_ et al . _
the calculations tend to be close to the data points at small angles while falling below the data as the proton angle increases .
this is the behavior one expects if meson exchange processes are going to become important as the missing momentum increases . in order to remove some of the kinematic dependence from these curves
we have calculated a _ reduced cross section _ by dividing the differential cross section of eq .
( [ gp_cross ] ) by a kinematic factor @xcite : @xmath45 .
\label{reduce}\end{aligned}\ ] ] figure 7 shows the reduced cross section as a function of missing momentum for all the experimental data shown in fig .
6 , as well as additional data provided by leitch _
the curves are generated using the same ingredients as the solid curves of fig .
6 but restricted to the kinematic range covered by the data .
an interesting observation here is that the model results are close to the data for missing momentum less than about 500 mev/@xmath2 .
the vertical dotted line indicates the momentum of a free proton with kinetic energy equal to the charged pion mass .
the calculations start to fall below the data in this kinematic region , which seems to be a good indication that we are seeing the need for inclusion of pion exchange diagrams to the reaction mechanism , and provides some idea of where these diagrams become important .
figure 8 shows results for proton removal from @xmath29pb , leading to two doublets and one resolved state in @xmath46tl , for two relatively low photon energies : 45 mev and 54 mev .
the data are from bobeldijk _ et al .
these authors performed a nonrelativistic distorted - wave impulse approximation ( dwia ) analysis of the data and found that the dko contribution tends to lie up to a factor of 10 below the data .
revised recent analyses @xcite indicate that this factor may have been unrealistic .
our present analysis , on the other hand , shows that the relativistic calculations do come close to predicting the correct magnitudes of the observed cross sections . within the parameter uncertainties ,
it is evident that the dko mechanism is the leading contributor to the reaction at these energies .
in this paper we have presented relativistic calculations for the @xmath5 reaction and its inverse for a number of target nuclei .
the results for the light targets cover a wide energy range , while the results for the lead target are at low energy but for a variety of final nuclear states .
the analysis was done in a consistent manner with no free parameters . in all cases but one , @xmath3c@xmath35
, the spectroscopic factor is obtained from a parallel analysis of the corresponding @xmath47 data . in cases of transitions with simple nuclear structure ,
relativistic calculations indicate that the dko mechanism is the main contributor to the cross section for lower missing momenta . for larger
missing momenta one finds clear deviations indicating an increased role for higher order processes such as meson exchange and @xmath9isobar contributions .
nonrelativistic analyses often indicate that the contributions from the dko mechanism are small and that meson exchange effects are sometimes dominant even at lower energies .
in contrast , the present relativistic analysis suggests substantial contributions from the dko mechanism to the cross sections over a wide range of energies .
the analysis also points out that meson exchange effects are required at higher missing momenta .
in the course of this analysis we have found that in the case of the @xmath3c target for photon energies below 80 mev , the relativistic calculations appear to overestimate the cross section data by close to a factor of 2 .
this situation is puzzling and may indicate either some complications due to the structure of the @xmath3c nucleus itself or to some subtleties in the combined analysis of @xmath47 and @xmath5 for this target .
it must also be noted that these difficulties do not occur for the spherical nuclei @xmath28o or @xmath29pb , which are also considered in the present work .
it is our feeling that the differences between theory and experiment at the lower proton energies for the @xmath3c target reflect the need for a proper description of the structure of this nucleus to include the intrinsic ground state deformation .
the consistent approach based on a combined analysis of these two reactions @xcite leads , in our view , to the conclusion that the @xmath3c ground state can not be adequately described by simple single particle configurations . in the case of transitions with simple structure ( mainly single particle ) our calculations indicate that meson exchange effects will not be important until one reaches missing momentum near 500 mev/@xmath2 . with the effort to push @xmath47 reactions towards this region of missing momentum
it would be interesting to see how important mec effects will turn out to be in the relativistic model .
van der sluys _ et al .
_ @xcite have considered this question in a nonrelativistic random phase approximation ( rpa ) framework and found large contributions from mec s for larger missing momenta .
one point of interest is that the reactions discussed , @xmath41 and @xmath5 , show different sensitivities to the description of the bound state .
this is probably due to the different range of missing momenta sampled by the two reactions .
the @xmath41 reaction has been primarily concerned with low missing momenta where the bound wave function is constrained by properties such as binding energy and rms radius .
the bound state wave functions that we use show little difference in momentum space for small momenta , and so it is not surprising that the @xmath41 results are very similar in this region .
differences between the bound state wave functions do arise , however , for larger missing momenta in @xmath41 and for the inherently large missing momentum reaction @xmath5 .
this is not a surprise because this is where the nuclear wave function is poorly constrained and the region where we see differences between these bound state wave functions in momentum space .
a common criticism of the distorted - wave born approximation ( dwba ) approach , both relativistic and nonrelativistic , is the lack of orthogonality of the bound and continuum wave functions .
it is argued that this lack of orthogonality could invoke spurious contributions to the cross sections .
the distorted continuum wave function is an approximation to the many - body wave function of the nuclear system with appropriate boundary conditions .
this approximation derives its support from the fact that the wave function is constrained by proton - nucleus elastic scattering data .
nonrelativistic rpa calculations do not suffer from this lack of orthogonality , but the wave functions are not able to account for the elastic scattering data . a simple method for restoring orthogonality
has been suggested by boffi _
_ @xcite and ciofi degli atti _
these authors find the orthogonality effects to be relevant mainly at large angles .
it is likely that this feature will carry over into the relativistic calculations and hence would not substantially change the main characteristics of the present calculations .
the present results pose certain challenges for the relativistic approach .
if the dko contributions are large then the data would suggest that the mec effects are suppressed in the relativistic models , at least at the lower energies .
relativistic models must explain this suppression and in the meantime face the challenge of accounting for the observed relatively large photoneutron cross sections .
spin - dependent observables are likely to play an important role in clarifying the reaction mechanisms . it should be noted here that the cross section angular distributions for @xmath5 reactions do not have much structure in most cases .
the differences between competing models are then mainly differences in magnitudes , and hence may be related to normalization uncertainties in the models .
when we discuss spin - dependent observables these normalization uncertainties cancel out and hence a better test of the model is likely to result .
one of us ( h.s.s . ) would like to thank the members of the institute for nuclear theory at the university of washington for their warm hospitality .
we would like to thank derek branford for providing us with the data contained in rauf s thesis @xcite and for generously giving us permission to show them before publication .
we are also grateful to l.j .
de bever and e.c .
aschenauer for allowing us to use their data before publication
. 99 d.g .
ireland and g. van der steenhoven , phys .
c * 49 * 2182 ( 1994 ) .
i. bobeldijk _
et al . _ ,
b * 356 * 13 ( 1995 ) .
et al . _ ,
c * 51 * 2611 ( 1995 ) .
g. van der steenhoven and h.p .
blok , phys .
c * 42 * 2597 ( 1990 ) .
et al . _ ,
a586 * 125 ( 1995 ) .
_ , nucl . phys .
a554 * 173 ( 1993 ) . j. ryckebusch , k. heyde , l. machenil , d. ryckbosch , m. vanderhaeghen and w. waroquier , phys .
c * 46 * 829 ( 1992 ) .
johansson , h.s .
sherif and g.m .
lotz , nucl .
a605 * 517 ( 1996 ) .
lotz and h.s .
sherif , phys .
b * 210 * 45 ( 1988 ) ; and nucl .
a537 * 285 ( 1992 ) .
s. frullani and j. mougey , in _ advances in nuclear physics _ , edited by j.w .
negele and e. vogt , ( plenum press , new york , 1984 ) vol .
14 , p. 1 .
blunden and m.j .
iqbal , phys .
b * 196 * 295 ( 1987 ) ; + p.g .
blunden , in : _ relativistic nuclear many - body physics _ , edited by b.c .
clark , r.j .
perry and j.p .
vary ( world scientific , singapore , 1989 ) , p. 265 .
cooper , s. hama , b.c .
clark and r.l .
mercer , phys .
c * 47 * 297 ( 1993 ) .
m. hedayati - poor , j.i .
johansson and h.s .
sherif , phys .
c * 51 * 2044 ( 1995 ) .
udias , p. sarriguren , e. moya de guerra , and j.a .
caballero , phys .
c * 53 * 1488 ( 1996 ) .
s.v . springham _
et al . _ ,
a517 * 93 ( 1990 ) .
matthews , d.j.s .
findlay , s.n .
gardiner and r.o .
owens , nucl .
* a267 * 51 ( 1976 ) . e.c .
( to be published ) .
rauf , ph.d .
thesis , university of edinburgh , 1996 .
h. ruijter , j - o .
adler , b - e .
andersson , k. hansen , l. isaksson , b. schrder , j. ryckebusch , d. ryckbosch , l. van hoorebeke and r. van de vyver , phys . rev .
c * 54 * 3076 ( 1996 ) .
c * 51 * 1982 ( 1995 ) .
bright , b. histad , r. johansson , e. traneus and s.r .
cotanch , nucl . phys . *
a603 * 1 ( 1996 ) .
g. van der steenhoven , h.p .
blok , e. jans , m. de jong , l. lapiks , e.n.m .
quint and p.k.a .
de witt huberts , nucl . phys . * a480 * 547 ( 1988 ) .
g. van der steenhoven , nucl . phys .
* a527 * 17c ( 1991 ) .
k.i . blomqvist _
_ , z. phys .
a * 351 * 353 ( 1995 ) .
j. friedrich , mainz , germany , ( 1996 ) private communication . l.j .
de bever , ph.d .
thesis , universiteit utrecht , 1993 . d.j.s .
findlay and r.o .
owens , nucl .
a279 * 385 ( 1977 ) .
adams , e.r .
kinney , j.l .
matthews , w.w.sapp , t.soos , r.o .
owens , r.s .
turley and g. pignault , phys rev c * 38 * 2771 ( 1988 ) .
s. boffi , c. giusti and f.d .
pacati , nucl . phys .
* a359 * 91 ( 1981 ) ; m.j .
leitch , j.l .
matthews , w.w .
sapp , c.p .
sargent , s.a .
wood , d.j.s .
findlay , r.o .
owens and b.l .
roberts , phys .
c * 31 * 1633 ( 1985 ) .
aschenauer , i. bobeldijk , d.g .
ireland , l. lapiks d. van neck , b. schrder , v. van der sluys , g. van der steenhoven and r.e .
van de vyver , phys .
b * 389 * 470 ( 1996 ) .
d.j.s . findlay and r.o .
owens , nucl .
* a292 * 53 ( 1977 )
. v. van der sluys , j. ryckebusch , and w. waroquier , phys .
c * 54 * 1322 ( 1996 ) .
s. boffi , r. cenni , c. giusti and f.d .
pacati , nucl . phys .
* a420 * 38 ( 1984 ) . c. ciofi degli atti , m.m .
giannini and g. salm , il nuovo cimento a * 76 * 225 ( 1983 ) .
1 . knockout of a @xmath40 proton from a @xmath3c target leading to the @xmath31b ground state .
angular distributions for seven different photon energies ranging from 45 to 78.5 mev .
hartree bound state wave functions are used @xcite and the proton optical potentials are from ref .
the data are from refs .
@xcite , @xcite and @xcite .
curves as discussed in the text .
2 . knockout of a @xmath40 proton from a @xmath3c target leading to the @xmath31b ground state .
distributions in photon energy at four fixed proton angles : @xmath48 = 30.0@xmath34 , 60.0@xmath34 , 90.0@xmath34 and 120.0@xmath34 .
the data are from ref .
curves as discussed in the text .
3 . knockout of a @xmath40 proton from a @xmath3c target leading to the @xmath31b ground and first excited states .
left - hand column distributions in photon energy at four fixed proton angles : @xmath48 = 30.6@xmath34 , 45.8@xmath34 , 66.0@xmath34 and 91.1@xmath34 .
right - hand column angular distributions for five photon energies the data are from refs .
@xcite and @xcite .
curves as discussed in the text .
4 . differential cross section as a function of photon angle for the @xmath35 reaction on @xmath3c and @xmath31b leading to the ground state .
curves as discussed in the text .
the data are from bright _ et al .
_ @xcite .
single proton removal from the @xmath27b target leading to the ground state in @xmath42be : upper figure the @xmath41 reaction , lower figure the @xmath5 reaction .
curves as discussed in the text .
the data are from de bever @xcite .
differential cross section as a function of proton angle for the knockout of a @xmath49 proton from a @xmath28o target leading to the @xmath43n ground state .
curves as discussed in the text .
the data are from refs .
@xcite .
fig . 7 . reduced cross section as a function of missing momentum for the knockout of a @xmath49 proton from a @xmath28o target leading to the @xmath43n ground state .
data as in fig . 6 and from @xcite .
curves as discussed in the text .
differential cross section as a function of proton angle for the knockout of protons from different levels in a @xmath29pb target .
curves as discussed in the text .
the data are from bobeldijk _ et al . | results of relativistic calculations of the direct knockout ( dko ) mechanism for the photon induced removal of a proton from a target nucleus over a wide range of energies and nuclei are presented .
spectroscopic factors used in the calculations are fixed from consistent analyses of the quasifree electron scattering process @xmath0 .
the results indicate that within the uncertainties of the model , the knockout contributions are generally close to the experimental data for missing momenta below @xmath1 500 mev/@xmath2 .
this is in disagreement with nonrelativistic analyses which often find that the direct knockout contribution can be quite small compared to the data and that meson exchange corrections can be important .
the present study suggests that meson exchange current contributions may not be as large when treated in a relativistic framework .
we also point out some difficulties we encountered in analyzing the data for a @xmath3c target at photon energies below 80 mev . |
twenty five years ago aharonov , albert and vaidman published a paper entitled `` how the result of a measurement of a component of the spin of a spin-1/ 2 particle can turn out to be 100 ? '' @xcite .
the authors idea was further developed a large volume of work on the so - called weak measurements ( see , for example , @xcite-@xcite ) , culminating in a somewhat bizarre the bbc report @xcite suggesting that `` pioneering experiments have cast doubt on a founding idea of the branch of physics called quantum mechanics '' .
there seems to be room for discussion about what actually happens in a weak measurement , and this is the subject of this paper .
some of the early and more recent criticisms of the original approach used in ref .
@xcite can be found in refs .
there appear to be only two possible answers to the original question posed by the authors of ref.@xcite : ( i ) there is a new counter - intuitive aspect to quantum measurement theory , or ( ii ) the proposed measurement is flawed . in this paper
we will follow ref .
@xcite in advocating the second point of view .
the argument is subtle .
there is no error in the simple mathematics of the ref .
it is the interpretation of the result which is at stake .
below we will argue that a weak measurement attempts to answer the which way? question without destroying interference between the pathways of interest .
such an attempt must be defeated by the uncertainty principle @xcite , @xcite and the unusual weak values are just the evidence of the defeat .
a random variable @xmath0 is fully described by its probability distribution @xmath1 .
often it is sufficient to know only the typical value of @xmath0 , and the range over which the values are likely to be spread . to get an estimate for the centre and the width of the range ,
one usually evaluates the mean value of @xmath0 , @xmath2 and the standard mean deviation @xmath3 .
suppose @xmath0 can only take the values @xmath4 and @xmath5 , and its unnormalised probability distribution is @xmath6 and @xmath7 .
we , therefore , have @xmath8/[\rho(1)+\rho(2)]\approx 1.4761,\\ \nonumber \sigma % = \{[<f>^2\rho(0)+(1-<f>)^2\rho(1)]/(\rho(0)+\rho(1))\}^{1/2 } \approx 0.4994,\end{aligned}\ ] ] which reasonably well represent the centre and width of the interval @xmath9 $ ] containing the values of @xmath0 .
suppose next that , for whatever reason , the unnormalised probabilities were allowed to take negative values , e.g. @xmath10 using the same formulas , we find @xmath11 which , clearly , no longer describe the range @xmath9 $ ] , - @xmath12 is too large , and @xmath13 is purely imaginary .
the reason for obtaining such an anomalous mean value is that the denominator in eq.([n1 ] ) is small , while the numerator is not - hence the large negative expectation value in eq.([n3 ] ) . in general , the mean and the standard mean deviation of an alternating distribution do not have to represent the region of its support .
these useful properties of @xmath14 and @xmath13 are lost , once a distribution is allowed to change its sign .
to make things worse , let us assume that the unnormalised probabilities @xmath1 are also allowed to take complex values , @xmath15 while @xmath0 may take any value inside an interval @xmath16 $ ] .
as before , we will construct a normalised distribution @xmath17 , which can now be written as a sum of its real and imaginary parts , @xmath18 where @xmath19 .
now we may wonder whether the value of @xmath20 would give us an idea about the location of the interval @xmath16 $ ] . from eq.([c2 ] ) we note that if both @xmath21 and @xmath22 do not change sign , @xmath23 is a proper probability distribution , and its mean certainly lies within the region of its support .
if , on the other hand , both @xmath21 and @xmath22 alternate , the mean @xmath24 is allowed to lie anywhere , and is not obliged to tell us anything about the actual range of values of @xmath0 .
so here is how a confusion might arise : suppose one needs to evaluate the average of a variable known to take values between @xmath4 and @xmath5 indirectly , i.e. , without checking whether the distribution alternates , or is a proper probabilistic one . obtaining a result of @xmath25 may seem unusual , until it is realised that the employed distribution changes sign , and scrambles the information about the actual range values involved .
one remaining question is why was it necessary to employ such a tricky distribution in the first place ?
a chance to employ oscillatory complex valued distribution is offered by quantum mechanics , and for a good reason . consider a kind of double - slit experiment in which a quantum system , initially in a state @xmath26 , may reach a given final state @xmath27 via two pathways , the corresponding probability amplitudes being @xmath28 and @xmath29 .
there are two possibilities .
( i ) the pathways interfere , and the probability to reach @xmath27 is given by @xmath30 ( ii ) interference between the pathways has been completely destroyed by bringing the system in contact with another system , or an environment . now the probability to reach @xmath27 is @xmath31 the two cases are physically different , as are the two probabilities . in the second case
the two pathways are @xmath32 .
one can make an experiment which would confirm by multiple trials that the system travels either the first or the second route with frequencies proportional to @xmath33 and @xmath34 , respectively . in the first case the pathways remain @xmath35 .
together they form a single real pathway travelled with probability @xmath36 , and there is no way of saying , even statistically , which of the two virtual paths the system has actually travelled .
the above leads to a loose formulation of the uncertainty principle @xcite : several interfering pathways or states should be considered as a single unit .
quantum interference erases detailed information about a system .
this information can only be obtained if interference is destroyed , usually at the cost of perturbing the system s evolution , thus destroying also the very studied phenomenon , e.g. , an interference pattern in young s double - slit experiment .
let us go about the pathways in a slightly more formal way . by slicing the time interval into @xmath37 subintervals , and sending @xmath37 to infinity ,
we can write the transition amplitude for a system with a hamiltonian @xmath38 as a sum over paths traced by a variable @xmath39 , @xmath40\end{aligned}\ ] ] where @xmath41 and @xmath42 are the eigenvalues and eigenvectors of the variable of interest @xmath39 , @xmath43 . we also introduced feynman paths - functions which take the values @xmath41 from the spectrum of @xmath39 at each discrete time . in the limit @xmath44
we will denote such a path by @xmath45 .
the paths are virtual pathways , each contributing a probability amplitude @xmath46 $ ] defined in eq.([f1 ] ) . in the chosen representation they form the most detailed complete set of histories available to the quantum system
. we may be interested not in every detail of the particle s past , but only in the value of a certain variable , a functional defined for a feynman path @xmath45 as an integral @xmath47= \int_{0}^t \beta(t ' ) a(t ' ) dt',\ ] ] where @xmath48 is a known function of our choice .
we can define a less detailed set of virtual pathways by grouping together those paths for which the value of @xmath49 $ ] equals some @xmath0 .
each pathway now contributes the amplitude @xmath50 ) a^{f\leftarrow i}[path],\ ] ] where @xmath51 is the dirac delta .
the new pathways contain the most detailed information about the variable @xmath52 , while information about other variables has been lost to interference in the sum ( [ f2 ] ) .
next we can define a coarse grained amplitude distribution for @xmath52 by smearing @xmath53 with a window function @xmath54 : @xmath55 with @xmath54 chosen , for example , to be a gaussian of a width @xmath56 we are unable to distinguish the values @xmath57 and @xmath58 less than @xmath56 apart , @xmath59 , since the corresponding pathways may now interfere .
the coarse graining does , however , have a physical meaning .
consider a basis @xmath60 containing our final state @xmath27 , and construct a state @xmath61 so that @xmath62 .
it is easy to check @xcite that @xmath63 satisfies a differential equation , @xmath64 with the initial condition latexmath:[\[\label{f5 } schroedinger equation describing a system interacting with a von neumann pointer @xcite whose position is @xmath0 . with it
we have the recipe for measuring the the quantity @xmath66 $ ] : first prepare the system in the initial state @xmath26 and the pointer in the state @xmath67 . switch on the coupling , and at a time @xmath68 measure the pointer position accurately .
interference between paths with different values of @xmath66 $ ] will be destroyed , since they lead do different outcomes for the pointer .
our measurement scheme has an important parameter , the width of the window @xmath54 , @xmath56 , which determines the extent to which we can ascertain the value of @xmath66 $ ] , once the pointer has been found in @xmath0 .
this accuracy parameter also determines the perturbation a measurement exerts on the measured system .
this , in turn , can be judged by how much the probability to arrive in a final state @xmath27 with the meter on differs from that with the meter off .
the former is given by @xmath69 and , in general , is not equal to @xmath70 since @xmath71 where the last equality is obtained by integrating eq.([f2 ] ) .
the perturbation can be minimised by choosing @xmath54 to be very broad . by construction
, the value of @xmath52 typically lies within a finite interval , say , @xmath72 \le b$ ] , outside of which @xmath73 vansihes .
a very broad @xmath74 can , therefore be replaced by @xmath54 , making the l.h.s . of eq.([w2 ] ) proportional to @xmath75 .
thus , in order to study the system with the interference between the pathways intact , we must make a highly inaccurate weak measurement .
this can be achieved by introducing a high degree of uncertainty in the pointer s initial position .
the following classical example may give us some encouragement .
consider a classical system which can reach a final state by several different routes .
let us say , a ball can roll from a hole @xmath76 to a hole @xmath77 down the first groove with the probability @xmath78 , or down the second groove , with the probability @xmath79 , and so on .
it is easy to imagine a ( purely classical ) pointer which moves one unit to the right if the ball travels the first route , or two units to the right , if the second route is travelled , and so on .
the meter is imperfect : we can accurately determine its final position , while we can not be sure that it has been set exactly at zero . rather , its initial position is distributed around @xmath80 with a probability density @xmath54 of a zero mean and a known variance .
let there be just two routes .
now the final meter readings are also uncertain , with the probability to find it in @xmath0 given by @xmath81 if the meter is accurate , i.e. , if @xmath54 is very narrowly peaked around @xmath82 , we will have just two possible readings , @xmath83 , in approximately @xmath84 out of @xmath37 trials , or @xmath85 , in approximately @xmath86 out of all cases .
suppose next that the meter is highly inaccurate , and the width of @xmath87 , @xmath56 is much larger than @xmath4 .
a simple calculation shows @xcite that the first two moments of the final distribution are given by @xmath88 we have , therefore , a very broad distribution , whose mean coincides with the mean of the @xmath89 . since the second moment of @xmath87 is known , by performing a large number of trials we can extract from the data also the variance @xmath13 of @xmath89 . for instance , if the two routes are travelled with equal probabilities , @xmath90 , we have @xmath91 from this we can correctly deduce that there are just two , and not three or four , routes available to the system , and that they are travelled with roughly equal probabilities .
this simple example shows that , classically , even a highly inaccurate meter can yield limited information about the alternatives available to a stochastic system .
it is just a matter of performing a large number of trials required to gather the necessary statistics .
next we will see whether this remains true in the quantum case .
in the quantum case , employing an inaccurate meter has a practical advantage - we minimise the back action of the meter on the measured system , and may hope to learn something without destroying the interference . as discussed in sect .
vi , we can make a measurement non - invasive by giving the initial meter s position a large quantum uncertainty ( that is to say , we choose a pure meter state broad in the coordinate space ) .
we prepare the system and the pointer in a product state ( [ f5 ] ) , turn on the interaction , check the system s final state , and sample the meter s reading provided this final state is @xmath27 . from ( [ f4 ] )
the moments of the distribution of the meter s readings are given by @xmath92 as the width of the initial meter s state @xmath93 tends to infinity , assuming @xmath94 we have @xcite @xmath95 and @xmath96 where @xmath97 is a factor of order of unity , which depends only on the shape of @xmath54 @xcite .
and we have introduced the notation @xmath98 for the @xmath99-th moment of the complex valued amplitude distribution @xmath100 defined in eq.([f2 ] ) , @xmath101 it is at this point that improper averages ( [ ww5 ] ) evaluated with oscillatory distributions enter our calculation , originally set to evaluate proper probabilistic averages ( [ ww2 ] ) .
expressions similar to eq.([ww2 ] ) have been obtained earlier in @xcite for a weak von neumann measurement and in @xcite for the quantum traversal time .
they are the quantum analogues of the classical eqs.([i2 ] ) .
we see that the quantum case turned out to be different in one important aspect . where the inaccurate classical calculation of the previous section yielded the mean of the _ probability distribution _
, its quantum counterpart gives us the mean evaluated with the _ probability amplitude _ @xmath102 .
there is no _ apriori _ reason to expect that either its real or imaginary part does not change sign . as discussed in sects .
ii and iii , such averages are not obliged to tell us anything about the actual range of a random variable . thus , our attempt to answer the which way? ( which @xmath0? ) question is likely to fail , as we are not able to extract the information about the alternatives available to a quantum system .
but we have been warned : the uncertainty principle suggests that , for as long as the pathways remain interfering alternatives , the question we ask has no meaning .
to give our approach a concrete example , we return to the double slit experiment .
consider a two - level system , e.g. , a spin-1/2 precessing in a magnetic field .
the hamiltonian is given by @xmath103 where @xmath104 is the larmor frequency , and @xmath105 is the pauli matrix .
we assume that the spin is pre - selected in a state polarised along the @xmath106-axis at @xmath107 , and then post - selected in the same state at @xmath108 .
we also wish to know the state of the spin half - way through the transition , at @xmath109 .
we follow the steps outlines in sect .
v. at any given time , and in the given representation , the spin can point up or down the @xmath106-axis .
we label these two sates @xmath110 and @xmath111 , respectively .
feynman paths are , therefore , irregular curves shown in fig . 1 . the functional @xmath112 is given by eq.([f1a ] ) with @xmath113 , @xmath114 thus , we combined the feynman paths ending in the state @xmath110 at @xmath108 into two virtual pathways , one containing the paths passing at @xmath109 through the state @xmath110 , and the other - the paths passing through the state @xmath111 .
the corresponding probability amplitudes are those for evolving the spin freely from its initial state to @xmath110 or @xmath111 at @xmath109 , and then to the final state @xmath110 at @xmath108 , @xmath115 we will need a meter .
the interaction @xmath116 corresponds to a von neumann measurement @xcite of the operator @xmath117 performed at @xmath109 .
the accuracy of the measurement depends on the width @xmath56 of the initial meter s state , which we will choose to be a gaussian , @xmath118 it is easy to check that the average meter reading @xmath14 in eqs.([i2 ] ) is given by @xmath119 its dependence on @xmath56 shown in fig.2 .
this is , of course , an oversimplified version of the young s double slit experiment : the states at @xmath109 play the role of the two slits , and the states at @xmath108 - the role of the positions on the screen where an interference pattern is observed .
consider first a strong measurement of the slit number .
choose the final time such that finding the freely precessing spin in the state @xmath110 is unlikely ( our interference pattern has there a minimum , or a dark fringe ) , say @xmath120 . sending @xmath121 , for the probability distribution of the meter s readings we have [ cf .
eq.([w1 ] ) ] @xmath122 we observe that the two pathways are travelled with almost equal probability , and eq.([d3a ] ) gives us the mean slit number @xmath123 however , this is not the original spin precession we set out to study .
the interference pattern has been destroyed and the probability to arrive at the final position @xmath110 , which without a meter was latexmath:[\[\label{d5 } this is a textbook example which illustrates the uncertainty principle : converting virtual paths into real ones comes at the cost of loosing the interference pattern .
not satisfied , we try to minimise the perturbation in the hope to learn something about the route chosen by the system with the interference intact .
we send @xmath56 to infinity , and after many trials obtain the answer : the mean number of the slit used is @xmath125 which brings us back to our original question , to rephrase the title of ref.@xcite , `` how the result of measuring the number of the slit in a double slit experiment can turn out to be @xmath126 ? ''
we have tried to evaluate the mean number of the slit a particle goes through in a double slit experiment , and came up with the number @xmath126 .
the mathematics is straightforward , and we need to understand the meaning of this result before employing the weak measurements elsewhere .
there are just two slits , numbered @xmath4 and @xmath5 , so the result looks a bit strange .
has our measurement gone wrong , or is the quantum world so strange that there are slits we are not aware of ?
we opt for the first choice .
wrong measurements are common in classical physics
. they can be made and repeated , but only have meaning within the narrow context of the wrong experiment itself .
a broken speedometer will read @xmath127 m.p.h each time the car goes at @xmath128 m.p.h , and might convince the driver , but not the traffic policemen who stops him for speeding .
the slit number @xmath126 may come up in a weak measurement , but can not be used for any other purpose , such as convincing a potential user that the screen he is about to buy has more than two holes drilled in it .
there is , however , one important distinction .
classically , one can always find the right answer and correct , or re - calibrate the errant speedometer .
quantally , it is not so .
according to the uncertainty principle , there is no correct answer to the question asked .
the nearest classical analogy might be this .
suppose a ( purely classical ) charge can be transferred across one of the two lead wires , and an observer can measure , which one has been chosen .
then the wires are heated up and melted into one .
which of the two wires has the charge gone through now ?
this is what interference does , it melts the pathways through the two slits into a single one , thus depriving the which way? question of its meaning .
having started to use analogies it is difficult to stop . here is the last one : one asks a manager a question the said manager is unable or unwilling to answer properly . yet
an answer he / she must give .
the answer ( or no - answer ) given will have little to do with what one wants to know .
it will be repeated should the question be asked again .
it should nt , however , be used to draw further conclusions about the matter of interest .
the weak measurements rely on an interesting interference effect which has applications beyond measurement theory @xcite , @xcite
. they can be made , and have been made in practice @xcite .
they have useful applications in interferometry @xcite .
however , their results should not be over - interpreted .
bizarre weak values indicate the failure of a measurement procedure under the conditions where , according to the uncertainty principle , it must fail .
seen like this , the weak measurements loose much of their original appeal , and the calculation of weak values reduces to a simple exercise in first order perturbation theory . finally , throughout the paper we appealed to the uncertainty principle , seen as one of the basic axioms of quantum theory .
it is possible that the principle itself will be explained in simpler terms within a yet unknown general theory .
however , we argue , that the weak measurements have not yet given such an explanation , nor provided any deeper insight into physical reality .
i acknowledge support of the basque government ( grant no
. it-472 - 10 ) , and the ministry of science and innovation of spain ( grant no . fis2009 - 12773-c02 - 01 ) .
i am also grateful to dr .
g. gribakin for bringing the lines used in the epigraph to my attention .
references aharonov y , albert dz , vaidman l. how the result of a measurement of a component of the spin of a spin-@xmath129 particle can turn out to be 100 .
physical review letters 1988 ; 60 ( 14 ) : 13511354 .
duck i m , stevenson pm , sudarshan ecg .
the sense in which a `` weak measurement '' of a spin-@xmath129 particle s spin component yields a value 100 .
physical review d 1989 ; 40 ( 6 ) : 21122117 .
http://dx.doi.org/10.1103/physrevd.40.2112 aharonov y , vaidman l. the two - state vector formalism of quantum mechanics . in : time in quantum mechanics , muga g , mayato rs , egusquiza i ( editors ) , springer , 2002 , pp.369412 .
aharonov y , botero a , popescu s , reznik b , tollaksen j. revisiting hardys paradox : counterfactual statements , real measurements , entanglement and weak values .
physics letters a 2002 ; 301 : 130138 .
dixon pb , starling dj , jordan an , howell jc .
ultrasensitive beam deflection measurement via interferometric weak value amplification .
physical review letters 2009 ; 102 ( 17 ) : 173601 .
rozema la , darabi a , mahler dh , hayat a , soudagar y , steinberg am .
violation of heisenbergs measurement - disturbance relationship by weak measurements .
physical review letters 2012 ; 109 ( 10 ) : 100404 .
leggett aj .
comment on `` how the result of a measurement of a component of the spin of a spin-@xmath129 particle can turn out to be 100 ''
. physical review letters 1989 ; 62 ( 19 ) : 23252325 .
sokolovski d , puerto gimnez i , sala mayato r. feynman - path analysis of hardy s paradox : measurements and the uncertainty principle .
physics letters a 2008 ; 372 ( 21 ) : 37843791 .
monks pdd , xiahou c , connor jnl .
local angular momentum - local impact parameter analysis : derivation and properties of the fundamental identity , with applications to the f + h2 , h + d2 , and cl + hcl chemical reactions .
journal of chemical physics 2006 ; 125 ( 13 ) : 133504 - 133513 . | weak measurements can be seen as an attempt at answering the which way? question without destroying interference between the pathways involved .
unusual mean values obtained in such measurements represent the response of a quantum system to this forbidden question , in which the true composition of virtual pathways is hidden from the observer .
such values indicate a failure of a measurement where the uncertainty principle says it must fail , rather than provide an additional insight into physical reality . _
q. what time is it when the clock strikes 13 ? _ _
a. time to buy a new clock . _ ( a joke ) |
modified newtonian dynamics ( mond ) was originally proposed by @xcite to explain the rotation curves of disk galaxies without invoking the presence of dark matter ( dm ) and , over the years , it has been successful at reproducing the observed kinematics of several galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) . however , to some extent , the mond and dm interpretations of the kinematics of galaxies can be degenerate .
for instance , a mond rotation curve can be also described in the newtonian gravity invoking a dm halo such that the total gravitational field is the same as the mond one .
more generally , given a mond system , it is possible to construct the _ equivalent newtonian system _ ( ens ) , i.e. the newtonian system with dm in which the visible matter has the same phase - space distribution as in the mond system @xcite .
it should be noted , however , that the physical viability of the ens is not guaranteed , as for some configurations the density of the associate dm halo turns out to be negative @xcite . though a mond system and its ens are , by construction , indistinguishable from a _
point of view ( for instance , as far as the rotation curve or the velocity - dispersion profile are concerned ) , in general they are not identical from a _ dynamical _ point of view : for instance , two - body relaxation , dynamical friction and galaxy merging act differently in the two cases @xcite .
as already recognized , mond and enss might differ substantially also in terms of stability .
as one of the original motivations for invoking dm halos in disk galaxies was that purely baryonic newtonian disks are prone to bar - like instability @xcite , it is not surprising that the study of dynamical stability in mond has focused mainly on disks @xcite .
here we consider instead the so - called radial - orbit instability , which is relevant to pressure - supported stellar systems . as well known , in the context of newtonian gravity the amount of radial orbits in stellar systems is limited not only by the requirement of phase - space consistency ( i.e. positivity of the distribution function ; see , e.g. , * ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) but also by the fact that very radially anisotropic spherical systems are unstable @xcite . in newtonian gravity
dm halos tend to have a stabilizing effect against radial - orbit instability @xcite , which suggests that mond systems might be more prone to this kind of instability than their enss .
if confirmed , this could provide a discriminant between the two theories when interpreting the velocity dispersion profiles in pressure - supported stellar systems .
for these reasons , in this work we use @xmath0-body simulations to explore the stability of radially anisotropic mond spherical systems and , for comparison , of their enss and of purely baryonic newtonian systems with the same density distributions and anisotropy profiles .
the paper is organized as follows . in section [ secmod ]
the galaxy models used in the simulations are described .
their phase - space consistency is discussed in section [ seccon ] and their stability in section [ secsta ] . section [ secfin ] concludes .
we consider mond as modified gravity in the non - relativistic formulation of @xcite , in which the poisson equation @xmath1 is replaced by @xmath2 = 4\pi g \rho , \label{eqmond}\ ] ] where @xmath3 is the standard euclidean norm , @xmath4 is the mond gravitational potential , @xmath5 is a characteristic acceleration , and @xmath6 is the so - called interpolating function ( in the present work we adopt @xmath7 ; * ? ? ? * ) .
the mond gravitational acceleration is @xmath8 , just as the newtonian acceleration is @xmath9 . for a system of finite mass , @xmath10 as @xmath11 , where @xmath12 is the position vector .
as well known , from the poisson equation and equation ( [ eqmond ] ) it follows that the mond and newtonian gravitational accelerations are related by @xmath13 , where @xmath14 , and @xmath15 is a solenoidal field dependent on the specific @xmath16 considered . since in general @xmath17 , and its expression is unknown _ a priori _ , standard poisson solvers can not be used to develop mond @xmath0-body codes , in which equation ( [ eqmond ] ) must be solved at each time step ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the present work we use our original mond @xmath0-body code (
* ; * ? ? ?
* see section [ secsim ] ) .
the initial conditions of the simulations are @xmath0-body realizations of galaxy models with the stellar component described by a spherical @xmath18-model @xcite with density distribution @xmath19 where @xmath20 is the total stellar mass , @xmath21 is the scale radius , and @xmath18 is the negative of the inner logarithmic density slope ( @xmath22 ) . for simplicity
we restrict to the cases @xmath23 and @xmath24 @xcite ; we recall that for @xmath25 the newtonian potential is @xmath26}.\ ] ] for these models the mond potential , required to distribute the particles in the velocity space , is easily calculated from the newtonian one , as in spherical symmetry @xmath27 .
the ens associated with a model of stellar density @xmath28 and mond potential @xmath4 has a _ total _
density @xmath29 .
however , the resulting dm halo density ( obtained after subtraction of @xmath28 ) in principle may be negative or have a non - monotonic radial trend .
fortunately , it can be proved that the dm halos of the ens derived from the spherical @xmath23 and @xmath24 models are everywhere positive .
instead , possible non - monotonicity of the dm density distribution makes the consistency of the halo a non - trivial request , as we will discuss in section 2.1 ( see also * ? ? ?
in order to impose a tunable amount of radial anisotropy on the initial conditions , we adopt the widely used osipkov - merritt ( om ) parametrization @xcite , in which the stellar distribution function depends on energy and angular momentum per unit mass through the variable @xmath30 , where @xmath31 is the relative energy , @xmath32 is the velocity modulus , @xmath33 is the relative potential , @xmath34 is the angular momentum modulus per unit mass , and @xmath35 is the so - called anisotropy radius .
the anisotropy parameter ( e.g. * ? ? ?
* ) is @xmath36 : for @xmath37 the velocity dispersion tensor is radially anisotropic , while for @xmath38 the tensor is nearly isotropic . in the limit @xmath39 , @xmath40 and
the velocity dispersion tensor becomes globally isotropic . in the purely baryonic newtonian models , the distribution function can be written as @xmath41 where @xmath42 is the newtonian relative potential and @xmath43 is the so - called augmented density ( see , e.g. , * ? ? ?
* ) . in the mond ( and in the ens ) cases a similar formula holds , i.e. @xmath44 where now @xmath45 is the mond relative potential that , by construction , is also the total potential of the ens . note that
, at variance with equation ( 4 ) , the lower integration limit is now @xmath46 : due to the far - field logarithmic behaviour of the mond potential in isolated systems of finite mass , the distribution function must be positive for all values of @xmath47 in analogy with finite - mass newtonian system in a total isothermal potential , where stars of all energies are bound ( e.g. * ? ? ?
apart from a few exceptions ( see section [ seclive ] ) , in the simulations of the enss the halo is maintained `` frozen '' ( i.e. it acts as a fixed external potential ) , so its distribution function is not needed . as well known , in ( single- or multi - component )
om models it is possible to determine a critical value of the anisotropy radius of each density component , @xmath48 , so that for @xmath49 the models are inconsistent , i.e. , @xmath50 for some admissible value of @xmath51 .
a necessary condition for consistency of om models is @xmath52 at all radii @xcite .
as recently proved in ciotti & morganti ( 2010a , b ; see also van hese , baes & dejonghe 2011 ; an 2011 ) this condition is rigorously equivalent to the global density slope - anisotropy inequality ( gdsai ) , i.e. to the requirement that at each radius the negative of the logarithmic slope of the stellar density distribution can not be less than twice the local value of the anisotropy parameter , @xmath53 . remarkably , it is easy to show that the gdsai _ is independent of the gravity law holding the system together _ , so that it holds not only for newtonian multi - component systems ( such as the ens ) , but also in the mond case .
we applied the gdsai to our families of models , determining for each stellar density profile the value of @xmath48 ( see table 1 ) , and the obtained limits coincide with those determined in @xcite . as these are just necessary conditions for consistency , the positivity of the distribution function
must still be checked numerically for @xmath54 .
we also note that from the inequality @xmath55 it follows that in case of isotropy ( @xmath56 ) a spherical density distribution must be necessarily monotonically decreasing for increasing @xmath57 , and this imposes quite strong constraints on the dm halos of physically admissible ens .
if the halo has a central `` hole '' ( which is not unusual for enss ; see * ? ? ?
* ) , it can be physically realized only if its orbital distribution is tangentially biased , at least in the internal regions . in practice , this is not an issue in simulations with frozen dm halos , but it is an important problem when setting up initial conditions for simulations in which the dm halo is `` live '' ( i.e. modelled with particles ; see section [ seclive ] ) .
we now describe the main results of the @xmath0-body simulations . before
, we briefly illustrate the numerical code and the tests performed .
we ran @xmath0-body simulations with , our original parallel three - dimensional particle - mesh code that can be used to follow the evolution of either mond or newtonian collisionless stellar systems @xcite . in previous papers
we have already used to demonstrate significant differences in violent relaxation , merging and dynamical friction in mond and newtonian dynamics @xcite .
we refer the readers to these papers and to @xcite for a more detailed description of the code . in the present study
the spherical grid has 64 radial nodes , 32 nodes in colatitude @xmath58 and 64 nodes in azimuth @xmath59 , and the total number of particles is @xmath60 .
we verified with convergence experiments that these numbers of particles and grid points exclude important discreteness effects .
in particular we reran some of the simulations with a grid @xmath61 and @xmath62 , finding that the results are almost identical to the corresponding lower - resolution cases . in each simulation
the initial conditions consist of an @xmath0-body realization of the stellar distribution of an isolated equilibrium galaxy model , in which the particles are distributed in phase space with the standard rejection technique , using the numerically recovered distribution functions . in the ens simulations ,
even if the dm halo is frozen , the systems are truncated exponentially at a radius @xmath63 , so that the potential well is finite . as a rule ,
we adopt @xmath64 , but we ran some of the simulations also with @xmath65 and @xmath66 , finding that the stability properties of a model do not depend on the specific choice of the truncation radius .
following @xcite , we identify each mond initial condition by fixing a value of the dimensionless internal acceleration parameter @xmath67 , so @xmath20 and @xmath21 are not independent quantities . in physical units , introducing the quantity @xmath68 , @xmath69 , the time and velocity units are @xmath70 , and @xmath71 .
the simulations are evolved up to @xmath72 with a time step @xmath73 , where @xmath74 is the characteristic dynamical time of the system . in the purely baryonic newtonian @xmath18-models ,
we adopt the standard half - mass radius value @xmath75^{3/2}},\ ] ] where @xmath76 is the mean stellar density inside the half mass radius @xmath77 ( we recall that @xmath78 for @xmath23 and @xmath79 for @xmath24 ) . in the corresponding mond models ( and their enss ) , @xmath74 is given by the above expression multiplied by @xmath80 , where @xmath81 is the circular velocity at @xmath77 for the purely baryonic newtonian system and @xmath82 is the same quantity for the mond system . following @xcite , in order to determine whether a given model is unstable , we check its departures from spherical symmetry by monitoring the evolution of its intrinsic axis ratios @xmath83 and @xmath84 ( where @xmath85 , @xmath86 and @xmath87 are the longest , intermediate and shortest axes of the inertia ellipsoid of the stellar distribution ) . as preliminary tests we ran simulations in which the initial conditions are isotropic purely baryonic newtonian , mond and enss .
we found that numerical uncertainties ( due to the finite number of particles ) on @xmath83 and @xmath84 never exceed @xmath88 over @xmath72 of evolution of these stable isotropic models .
as a consequence , we define _ unstable _ the models for which the axis ratio @xmath89 after 100 @xmath74 is smaller than a fiducial threshold value @xmath90 . in the simulations
the onset of the instability is just due to numerical noise produced by discreteness effects in the initial conditions .
as expected , we found that an exact determination of the stability threshold for a given family of models is not straightforward : while for strongly anisotropic initial conditions the onset of the instability is apparent and the numerical models settle down into a final equilibrium configuration in a few dynamical times , for nearly stable initial conditions the instability can be characterized by very slow growth rates and its effects become evident even after tens of @xmath74 . for a given initial stellar density profile ( i.e. @xmath23 or @xmath24 in equation [ eqrhostar ] ) , we explored , besides the purely baryonic newtonian case ( models n@xmath91 and n@xmath92 in table 1 ) , two families of mond systems ( and the corresponding families of enss ) , characterized by the values of the acceleration ratio @xmath93 and @xmath94 ( see table 1 for a summary ) .
the cases with @xmath93 correspond to low - acceleration systems with internal accelerations everywhere much lower than @xmath95 , in the so - called deep - mond regime .
their associated enss are therefore totally dm dominated systems ( table 1 , column 3 ) . on the other hand ,
the @xmath94 cases are systems with relatively high accelerations within @xmath77 , corresponding in the newtonian context to systems dominated by baryons in the central regions and by dm only at @xmath96 ( table 1 , column 3 ) .
the four panels of fig .
[ figden ] show the baryonic and dm initial density profiles of the enss corresponding to the @xmath93 and @xmath94 mond systems , for the two explored values of @xmath18 . as can be seen
, the associated dm halos do not show a central hole and so , at least from this point of view , are not unrealistic ( see section [ seccon ] ) . for each family of models
we have a set of eight simulations , in which the initial conditions differ only in the value of the normalized anisotropy radius @xmath97 .
this is apparent in fig .
[ figcara ] , where we plot the final axis ratio @xmath89 vs. the initial value of @xmath97 for all the explored models of the ten families .
the stability properties of the models can be inferred directly from fig .
[ figcara ] : it is clear that the value of @xmath35 at which instability appears is largest for mond models , smallest for the corresponding enss , and intermediate for purely baryonic newtonian systems . because of the very nature of the radial - orbit instability ( very slow growth rates for marginally unstable systems ) , a precise determination of the value of the minimum anisotropy radius for stability @xmath98 is difficult . what can be estimated robustly with @xmath0-body simulations
is a fiducial value of @xmath98 separating apparently unstable systems from _ bona fide _ stable systems . adopting as fiducial threshold for stability @xmath99 ( horizontal dashed line in each panel of fig . [ figcara ] )
, we estimate for each family of models the minimum anisotropy radius for stability @xmath98 , which is reported ( in units of @xmath77 ) in table 1 .
as expected , the differences in the stability limit between mond and enss are more evident in systems with lower values of the acceleration ratio @xmath100 ( or , from a newtonian point of view , for more dm dominated systems ) .
for instance , if we consider @xmath23 models with @xmath93 , which are representative of low - surface density systems with flat inner stellar density profile ( such as dwarf spheroidal galaxies ) , we find @xmath101 for mond and @xmath102 for the ens ; very similar results are obtained for the more peaked hernquist @xmath24 models . unsurprisingly ,
newtonian models with the dm halo are less subject to the radial - orbit instability than the purely baryonic newtonian models . thus , from this first analysis we conclude that , when using @xmath97 as an indicator of the amount of admissible radial orbits , mond systems are more prone to radial - orbit instability than their associated enss , and also than corresponding purely baryonic newtonian models . however , as we will see in the next section , the comparison between the mond and the associated purely baryonic newtonian families is subtle : the fact that the latter typically admit smaller @xmath97 values does not imply that ( globally ) they can sustain more radial kinetic energy .
stability limitations expressed in terms of @xmath35 are particularly relevant to observational works , as @xmath35 enters directly in the jeans equations that are routinely used to fit the velocity - dispersion profiles of stellar systems . however , from a deeper point of view , the value of @xmath35 ( loosely speaking , the radius outside which orbits are mainly radial ) is not a robust measure of the fraction of kinetic energy that is stored in radial orbits , which , at least in the newtonian context , is believed to be the main indicator of the tendency to develop radial - orbit instability .
more specifically , it has been argued @xcite that a proper criterion for stability for newtonian self - gravitating systems can be expressed in terms of the global anisotropy parameter @xmath104 , where @xmath105 and @xmath106 are the radial and tangential components of the kinetic energy tensor , respectively .
global isotropy corresponds to a value of the fridman - polyachenko - shukhman parameter @xmath107 .
indications exist that there is a critical value @xmath108 such that only systems with @xmath109 are stable , and it is widely accepted that @xmath110 , relatively independent of the specific density distribution .
a priori we do not have reasons to expect that @xmath103 can be used as a stability indicator also in mond . in any case , by definition @xmath108 measures how much radial anisotropy can be supported by a stable system , so it is interesting to discuss the stability properties of our mond , purely baryonic newtonian and enss in terms of @xmath103 .
the values of @xmath108 for the families of models studied in the present work are reported in table 1 , and in fig .
[ figxi ] we show all the models in the @xmath89-@xmath103 plane to be compared with the analogous fig . [ figcara ] . for the purely baryonic newtonian models the interpretation of these numbers is straightforward :
we find @xmath111 , in agreement with the standard criterion and with previous numerical studies ( see , e.g. , * ? ? ?
for the mond models we find @xmath112 , with @xmath108 depending on both the stellar density profile and the acceleration ratio @xmath100 .
these values are substantially higher than those found for the corresponding purely baryonic newtonian models , indicating that , for given density distribution , mond systems can sustain more radial kinetic energy than newtonian systems without dm .
this is not in contrast with the findings reported in the previous section , i.e. that mond models have larger @xmath98 than purely baryonic newtonian models .
in fact , for fixed density profile and @xmath35 value , a mond model has higher @xmath103 than a purely baryonic newtonian model , because more kinetic energy is stored in the outer parts , where orbits are radially biased and the gravitational field of mond systems is stronger , have the same value of @xmath103 . ] .
apparently , this effect compensates the larger values of admissible @xmath35 . instead , the enss are again found able to sustain systematically higher values of @xmath108 than the corresponding families mond systems , and this clearly indicates that mond systems are more subject to radial - orbit instability than newtonian models with dm and identical total gravitational field . in general ,
for both mond and enss we find a substantial spread in the values of @xmath108 , supporting the expectation that a `` universal '' value of @xmath108 does not exist for these systems ( for newtonian models with dm , extended versions of the stability criterion have been proposed ; * ? ? ?
* ; * ? ? ?
the data in table 1 suggest that in mond , and even more in the enss , @xmath108 increases ( i.e. , relatively more kinetic energy can be stored in radial orbits ) for decreasing @xmath100 .
while for the enss this trend can be explained because in the limit @xmath113 the stars become just tracers , it is tempting to speculate that in mond this behaviour can be interpreted as a manifestation of the less mixing nature of mond with respect to newtonian gravity .
this interpretation is supported by the well established numerical finding that phase mixing @xcite and violent relaxation @xcite in mond systems take longer ( in units of dynamical times ) than in the newtonian case .
very qualitatively , the deep - mond force between particles behaves like @xmath114 , i.e. it is nearer to the harmonic oscillator force than the @xmath115 force , and it is easy to show that a system in which particles interact with the harmonic oscillator force , no mixing or instabilities are possible , as each particle moves independently of the others @xcite . curiously , even though mond forces in a system of particles are non - additive , a similar trend towards longer relaxation times has been also found recently in shell models interacting with additive @xmath116 forces @xcite , and the phase - space evolution of the case with @xmath114 forces is strikingly similar to the deep - mond collapses presented in @xcite .
the fact that @xmath103 in mond is strongly affected by the properties of the system at large radii suggests to consider as a possible stability indicator the quantity @xmath117 , defined as the ratio of radial to tangential kinetic energy within @xmath77 : the maximum values for stability @xmath118 for our families of models are reported in the last column of table 1 .
remarkably , all our mond and purely baryonic newtonian families have @xmath119 , while the enss have @xmath120 ( for @xmath94 ) and @xmath121 ( for @xmath93 ) , which leads us to speculate that _ a limit on the amount of the radial orbits within the half - mass radius might be used as an empirical stability criterion for mond stellar systems , valid from the newtonian to the deep - mond regime , i.e. independent of the value of the internal acceleration relative to @xmath95_. this result is reminiscent of that of @xcite , who found that an almost isotropic core can stabilize newtonian self - gravitating systems with very strong global radial anisotropy .
the results on the stability of the enss might be affected to some extent by our assumption of frozen dm halo , because it is possible that an ens with live dm halo has different stability properties ( see * ? ? ?
in order to assess to what extent our results are affected by the assumption of frozen dm halos , we reran the ens simulations of the family e@xmath122 with live dm halos .
for technical reasons we ran these live - halo simulations with our treecode ( fortran version of a fast poisson solver ; * ? ? ?
* ; * ? ? ?
* ) , which was already tested against in @xcite .
as an additional test , we reran with all the purely baryonic newtonian simulations of the family n@xmath91 , finding excellent agreement with those run with . in the live - halo simulations of the models of the family e@xmath122 we used @xmath123 particles for the stellar component and @xmath124 particles for the halo , which has mass @xmath125 for the adopted truncation radius @xmath64 .
we verified numerically that for these two - component e@xmath122 models the halo density distribution , which is monotonically decreasing ( see fig .
[ figden ] ) and thus satisfies the necessary condition for consistency ( see section [ seccon ] ) , can be generated by a positive isotropic distribution function , which we used to distribute the dm particles in phase space with the standard rejection technique .
the final axis ratios of the live - halo simulations ( hexagons in fig .
[ figlive ] ) are not significantly different from those of the corresponding frozen - halo simulations ( squares in fig .
[ figlive ] ) .
it follows that the stability properties of the e@xmath122 family of models do not depend on whether the halo is modelled with particles or it is a fixed potential : in particular the values of the minimum anisotropy radius @xmath98 ( and then of the maximum fridman - polyachenko - shukhman parameter @xmath108 ) for stability are the same in the two cases .
these results suggest that our conclusion that mond models are substantially more subject to radial - orbit instability than enss should be robust , at least if the dm halos are almost isotropic .
though the assumption of isotropy of the halos is not necessarily justified , given that the anisotropy of dm distributions is hard to constrain , it is clear that an isotropic halo can be always be invoked , in the context of newtonian gravity , to stabilize an otherwise unstable strongly radially - anisotropic stellar system , while this freedom is not allowed in the context of mond .
to some extent , the mond and dm interpretations of the kinematics of galaxies can be considered degenerate , i.e. several observational features can be satisfactorily reproduced in both paradigms .
this raises interesting questions about possible tests to discriminate between mond and newtonian gravity with dm .
fortunately , it is now well established that some important dynamical processes _ are _ different , even in systems in which the total gravitational potentials are identical .
examples are dynamical friction and two - body relaxation . in the present work we explored whether radial - orbit instability acts differently in mond and in newtonian gravity ( with and without dm ) ; in addition to the theoretical interest , this study can be useful to constrain the interpretation of the observed kinematics of stellar systems in the two cases .
in particular , we have focused on the stability of om radially - anisotropic spherical @xmath18-models ( @xmath23 and @xmath24 ) .
we compared the results obtained for mond models , enss and purely baryonic newtonian systems , all with the same stellar density distribution .
overall , we found that mond systems are more prone to radial - orbit instability than their enss , independent of the specific indicator ( @xmath35 or @xmath103 ) used to quantify the anisotropy .
compared to purely baryonic newtonian systems with the same density profile , however , mond systems have larger minimum anisotropy radius for stability @xmath98 , but nevertheless higher maximum global anisotropy parameter for stability @xmath108 , a consequence of the larger kinetic energy that can be stored in their outer regions . we speculate that @xmath108 is larger in mond systems than in newtonian systems with no dm for the same reasons that phase mixing and violent relaxation are less efficient , i.e. , because the force between particles in deep - mond decreases with distance less strongly ( qualitatively as @xmath114 ) than in newtonian gravity , so being nearer to the special case of harmonic oscillator inter - particle forces , when instabilities and energy exchanges are impossible observationally , these findings may be relevant to applications of mond to pressure - supported systems such as globular clusters , dwarf spheroidal and elliptical galaxies . for instance , the velocity - dispersion profiles of globular clusters in the outer parts of the milky way can been used to test mond . in the case of the globular cluster ngc 2419 strong om radial anisotropy ( @xmath35 close to @xmath48 )
might be required in order to bring the velocity dispersion profile predicted by mond close to the profile inferred from the observed radial velocities of the cluster stars @xcite .
a combination of the stability constraints discussed in this paper with new observations of the radial velocity of stars can make the kinematics of ngc 2419 a crucial test for mond ( ibata et al .
, in preparation ) .
another family of objects that are very interesting from the perspective of mond is that of dwarf spheroidal galaxies , because their low surface - brightness , combined with their measured velocity dispersion , lead to conclude that they must be dm dominated if newtonian gravity holds .
@xcite studied the observed line - of - sight velocity - dispersion profiles of milky way dwarf spheroidal galaxies , finding that in most cases the data can be reproduced in mond , at least with somewhat _ ad hoc _ anisotropy profiles .
this result should be reconsidered on the basis of the consistency and stability constraints discussed in the present work .
the kinematics of elliptical galaxies has also been considered as a possible test for mond ( e.g. * ? ? ?
as for globular clusters and dwarf spheroidals , when trying to reproduce the kinematics of ellipticals , the anisotropy of the velocity distribution of the stars is one of the variables in the problem and it is important to constrain it as much as possible , along the lines described in the present paper .
for instance , consistency constraints ( but not the more stringent stability constraints ) on the anisotropy of stars are considered in the recent investigation by @xcite .
some authors @xcite used the observed kinematics of planetary nebulae or of the satellites around elliptical galaxies as a test for mond , allowing for quite general anisotropy profiles for the satellite system . in this case , however , while some limits on the amount of radial anisotropy might come from the requirement of consistency , stability arguments do not necessarily lead to strong constraints as long as the considered tracers are not dynamical dominant .
most numerical simulations were performed at cineca , bologna , with cpu time assigned under the inaf - cineca agreement 2008/2010 .
l.c . and c.n .
are supported by the miur grant prin2008 .
an j. , 2011 , arxiv:1101.0006 an j.h . , evans n.w . , 2006 , apj , 642 , 752 angus g. w. , 2008 , mnras , 387 , 1481 barnes j. , hut p. , goodman j. , 1986 , apj , 300 , 112 barnes e. i. , lanzel p. a. , williams l. l. r. , 2009 , apj , 704 , 372 bekenstein j. d. , milgrom m. , 1984 , apj , 286 , 7 bertin g. , pegoraro f. , rubini f. , vesperini e. , 1994 , apj , 434 , 94 binney j. , tremaine s. , 2008 , galactic dynamics 2nd ed . , princeton university press , princeton brada r. , milgrom m. , 1999 , apj , 519 , 590 cardone v. f. , angus g. , diaferio a. , tortora c. , molinaro r. , 2010 , arxiv:1011.574 christodoulou d. m. , 1991 , apj , 372 , 471 ciotti l. , 1996 , apj , 471 , 68 ciotti l. , 1999 , apj , 520 , 574 ciotti l. , binney j. , 2004 , mnras , 351 , 285 ciotti l. , morganti l. , 2009 , mnras , 393 , 179 ciotti l. , morganti l. , 2010a , mnras , 401 , 1091 ciotti l. , morganti l. , 2010b , mnras , 408 , 1070 ciotti l. , pellegrini s. , 1992 , mnras , 255 , 561 ciotti l. , londrillo p. , nipoti c. , 2006 , apj , 640 , 741 ciotti l. , nipoti c. , londrillo p. , 2007
, in collective phenomena in macroscopic systems , ed . g. bertin et al . (
singapore : world scientific ) , p.177 ciotti l. , morganti l. , de zeeuw p. t. , 2009 , mnras , 393 , 491 dehnen , w. 1993 , mnras , 265 , 250 dejonghe h. , merritt d. , 1988 , apj , 328 , 93 di cintio p. , ciotti l. , 2011 , submitted fridman a. m. , polyachenko v. l. , 1984 , physics of gravitating systems .
springer , new york henon m. , 1973 , a&a , 24 , 229 hernquist l. , 1990 , apj , 356 , 359 hjorth j. , 1994 , apj , 424 , 106 klypin a. , prada f. , 2009 , apj , 690 , 1488 londrillo p. , nipoti c. , ciotti l. , 2003 , msais , 1 , 18 londrillo p. , nipoti c. , 2009 , msais , 13 , 89 may a. , binney j. , 1986 , mnras , 221 , 13p lynden - bell d. , lynden - bell r. m. , 1982 , proceedings : mathematical , physical and engineering sciences , vol .
1982 , p. 475
, the royal society merritt , d. 1985 , aj , 90 , 102 merritt d. , aguilar l. a. , 1985 , mnras , 217 , 787 meza a. , zamorano n. , 1997 , apj , 490 , 136 meza a. , zamorano n. , 1998 , aspc , 136 , 400 milgrom m. , 1983 , apj , 270 , 365 milgrom m. , 1986 , apj , 306 , 9 milgrom m. , 1989 , apj , 338 , 121 milgrom m. , 2001 , mnras , 326 , 1261 milgrom m. , 2010 , mnras , 403 , 886 nipoti c. , londrillo p. , ciotti l. , 2002 , mnras , 332 , 901 nipoti c. , londrillo p. , ciotti l. , 2003 , mnras , 342 , 501 nipoti c. , londrillo p. , ciotti l. , 2007a , apj , 660 , 256 nipoti c. , londrillo p. , zhao h.s . , ciotti l. , 2007b , mnras , 379 , 597 nipoti c. , londrillo p. , ciotti l. , 2007c , mnras , 381 , l104 nipoti c. , ciotti l. , binney j. , londrillo p. , 2008
, mnras , 386 , 2194 osipkov , l.p . 1979 , soviet astron . lett . , 5 , 42 ostriker j. p. , peebles p. j. e. , 1973 , apj , 186 , 467 palmer p. l. , papaloizou j. , 1987 , mnras , 224 , 1043 polyachenko v. l. , 1987 , iaus , 127 , 301 polyachenko v. l. , shukhman i. g. , 1981 , sva , 25 , 533 saha p. , 1991 ,
mnras , 248 , 494 sanders r. h. , 2000 , mnras , 313 , 767 sanders r. h. , mcgaugh s. s. , 2002 , ara&a , 40 , 263 sanders r. h. , noordermeer e. , 2007 , mnras , 379 , 702 sollima a. , nipoti c. , 2010 , mnras , 401 , 131 stiavelli m. , sparke l. s. , 1991 , apj , 382 , 466 swaters r. a. , sanders r. h. , mcgaugh s. s. , 2010 , apj , 718 , 380 tiret o. , combes f. , 2007 , a&a , 464 , 517 tiret o. , combes f. , angus g. w. , famaey b. , zhao h. s. , 2007 , a&a , 476 , l1 tremaine s. , richstone d. o. , byun y .-
dressler a. , faber s. m. , grillmair c. , kormendy j. , lauer t. r. , 1994 , aj , 107 , 634 trenti m. , bertin g. , 2006 , apj , 637 , 717 van hese e. , baes m. , dejonghe h. , 2011 , apj , 726 , 80 weinberg m. d. , 1991 , apj , 368 , 66 | the stability of radially anisotropic spherical stellar systems in modified newtonian dynamics ( mond ) is explored by means of numerical simulations performed with the @xmath0-body code .
we find that osipkov - merritt mond models require for stability larger minimum anisotropy radius than equivalent newtonian systems ( enss ) with dark matter , and also than purely baryonic newtonian models with the same density profile .
the maximum value for stability of the fridman - polyachenko - shukhman parameter in mond models is lower than in enss , but higher than in newtonian models with no dark matter .
we conclude that mond systems are substantially more prone to radial - orbit instability than enss with dark matter , while they are able to support a larger amount of kinetic energy stored in radial orbits than purely baryonic newtonian systems .
an explanation of these results is attempted , and their relevance to the mond interpretation of the observed kinematics of globular clusters , dwarf spheroidal and elliptical galaxies is briefly discussed .
dark matter galaxies : kinematics and dynamics globular clusters : general gravitation instabilities |
convergence to the poisson distribution , for the number of occurrences of possibly dependent events , naturally arises in various applications . following the work of poisson ,
there has been considerable interest in how well the poisson distribution approximates the binomial distribution . the basic idea which serves for the starting point of the so called _ chen - stein method for the poisson approximation _
is the following ( see chen ( 1975 ) ) .
let @xmath0 be independent bernoulli random variables with @xmath1 .
let @xmath2 and @xmath3 for every @xmath4 , and @xmath5 with mean @xmath6 .
it is easy to show that @xmath7 = 0 \label{eq : basic equation of the chen - stein method for poisson approximation}\ ] ] holds for an arbitrary bounded function @xmath8 where @xmath9 .
furthermore ( see , e.g. , chapter 2 in ross and pekz ( 2007 ) ) @xmath10 = \sum_{j=1}^n p_j^2 \ , { \ensuremath{\mathbb{e}}}\bigl[f(v_j+2 ) - f(v_j+1 ) \bigr ] \label{eq : first step in the derivation of the improved lower bound on total variation distance}\ ] ] which then serves to provide rigorous bounds on the difference between the distributions of @xmath11 and @xmath12 , by the chen - stein method for poisson approximations .
this method , and more generally the so called _ stein method _
, serves as a powerful tool for the derivation of rigorous bounds for various distributional approximations .
nice expositions of this method are provided by , e.g. , arratia et al .
( 1990 ) , ross and pekz ( 2007 ) and ross ( 2011 ) .
furthermore , some interesting links between the chen - stein method and information - theoretic functionals in the context of poisson and compound poisson approximations are provided by barbour et al .
( 2010 ) . throughout this letter
, the term ` distribution ' refers to a discrete probability mass function of an integer - valued random variable . in the following , we introduce some known results that are related to the presentation of the new results .
let @xmath13 and @xmath14 be two probability measures defined on a set @xmath15 .
then , the total variation distance between @xmath13 and @xmath14 is defined by @xmath16 where the supremum is taken w.r.t . all the borel subsets @xmath17 of @xmath15 .
if @xmath15 is a countable set then is simplified to @xmath18 so the total variation distance is equal to half of the @xmath19-distance between the two probability distributions .
[ definition : total variation distance ] among old and interesting results that are related to the poisson approximation , le cam s inequality ( see le cam ( 1960 ) ) provides an upper bound on the total variation distance between the distribution of the sum @xmath20 of @xmath21 independent bernoulli random variables @xmath0 , where @xmath22 , and a poisson distribution @xmath23 with mean @xmath24 .
this inequality states that @xmath25 so if , e.g. , @xmath26 for every @xmath4 ( referring to the case where @xmath11 is binomially distributed ) then this upper bound is equal to @xmath27 , decaying to zero as @xmath28 .
the following theorem combines theorems 1 and 2 of barbour and hall ( 1984 ) , and its proof relies on the chen - stein method : let @xmath20 be a sum of @xmath21 independent bernoulli random variables with @xmath1 for @xmath4 , and @xmath29 .
then , the total variation distance between the probability distribution of @xmath11 and the poisson distribution with mean @xmath30 satisfies @xmath31 where @xmath32 for every @xmath33 .
[ theorem : bounds on the total variation distance - barbour and hall 1984 ] as a consequence of theorem [ theorem : bounds on the total variation distance - barbour and hall 1984 ] , it follows that the ratio between the upper and lower bounds in is not larger than 32 , irrespectively of the values of @xmath34 .
the factor @xmath35 in the lower bound was claimed to be improvable to @xmath36 with no explicit proof ( see remark 3.2.2 in barbour et al .
( 1992 ) ) .
this shows that , for independent bernoulli random variables , these bounds are essentially tight .
furthermore , note that the upper bound in improves le cam s inequality ; for large values of @xmath30 , this improvement is by approximately a factor of @xmath37 .
this letter presents new lower bounds on the total variation distance between the distribution of a sum of independent bernoulli random variables and the poisson random variable ( with the same mean ) .
the derivation of these new bounds generalizes and improves the analysis by barbour and hall ( 1984 ) , based on the chen - stein method for the poisson approximation .
this letter concludes by outlining a use of the new lower bounds for the analysis in sason ( 2012 ) , followed by a comparison of the new bounds to previously reported bounds .
this work forms a continuation of the line of work in barbour and chen ( 2005)kontoyiannis et al .
( 2005 ) where the chen - stein method was studied in the context of the poisson and compound poisson approximations , and it was linked to an information - theoretic context by barbour et al .
( 2010 ) , kontoyiannis et al .
( 2005 ) , and sason ( 2012 ) .
in the following , we introduce an improved lower bound on the total variation distance and then provide a loosened version of this bound that is expressed in closed form . in the setting of theorem [ theorem : bounds on the total variation distance - barbour and hall 1984 ]
, the total variation distance between the probability distribution of @xmath11 and the poisson distribution with mean @xmath30 satisfies the inequality @xmath38 where @xmath39 and @xmath40 & & \hspace*{-0.5 cm } x_+ \triangleq \max\{x , 0\ } , \quad x_+^2 \triangleq \bigl(x_+)^2 , \quad \forall \ , x \in { \ensuremath{\mathbb{r}}}\\[0.1 cm ] & & \hspace*{-0.5 cm } g_{\lambda}(\alpha_1 , \alpha_2 , \theta ) \triangleq \max \left\ { \ , \left| \left(1 + \sqrt{\frac{2}{\theta \lambda e } } \cdot & & \hspace*{3.2 cm } \left .
\left| \left(2 e^{-\frac{3}{2 } } + \sqrt{\frac{2}{\theta \lambda e } } \cdot \right\ } \label{eq
: g in the lower bound on the total variation distance } \\[0.1 cm ] & & \hspace*{-0.5 cm } x(u ) \triangleq ( c_0 + c_1 u + c_2
u^2 ) \ , \exp(-u^2 ) , \quad \forall \ , u \in { \ensuremath{\mathbb{r}}}\label{eq : function x } \\[0.2 cm ] & & \hspace*{-0.5 cm } \{u_i\ } \triangleq \bigl\ { u \in { \ensuremath{\mathbb{r } } } : \ , 2 c_2 u^3 + 2 c_1 u^2 - 2(c_2 - c_0 ) u - c_1 = 0\bigr\ } \label{eq : zeros of a cubic polynomial equation } \\[0.1 cm ] & & \hspace*{-0.5 cm } c_0 \triangleq ( \alpha_2 - \alpha_1 ) ( \lambda - \alpha_2 ) \label{eq : c0 } \\ & & \hspace*{-0.5 cm } c_1 \triangleq \sqrt{\theta \lambda } \ , ( \lambda + \alpha_1 - 2 \alpha_2 ) \label{eq : c1 } \\ & & \hspace*{-0.5 cm } c_2 \triangleq -\theta \lambda .
\label{eq : c2}\end{aligned}\ ] ] [ theorem : improved lower bound on the total variation distance ] see section [ subsubsection : proof of the theorem with the improved lower bound on the total variation distance ] .
the derivation relies on the chen - stein method for the poisson approximation , and it improves ( significantly ) the constant in the lower bound of theorem 2 of barbour and hall ( 1984 ) . the upper and lower bounds on the total variation distance in scale like @xmath41 , similarly to the known bounds in theorem [ theorem : bounds on the total variation distance - barbour and hall 1984 ] ,
but they offer a significant improvement in their tightness ( see section [ section : connection of the improved lower bounds with known results ] ) .
the cardinality of the set @xmath42 in is equal to 3 ( see section [ subsubsection : proof of the theorem with the improved lower bound on the total variation distance ] ) . [
remark : the size of the set of real zeros is equal to 3 ] the optimization that is required for the computation of @xmath43 in w.r.t .
the three parameters @xmath44 and @xmath45 is performed numerically . in the following ,
we introduce a looser lower bound on the total variation distance as compared to the lower bound in theorem [ theorem : improved lower bound on the total variation distance ] , but its advantage is that it is expressed in closed form .
both lower bounds improve ( significantly ) the lower bound in theorem 2 of barbour and hall ( 1984 ) .
the following lower bound follows from theorem [ theorem : improved lower bound on the total variation distance ] via the special choice of @xmath46 that is included in the optimization set for @xmath43 on the right - hand side of . following this sub - optimal choice , the lower bound in the next corollary
is obtained by a derivation of a closed - form expression for the third free parameter @xmath45 ( in fact , this was our first step towards the derivation of an improved lower bound on the total variation distance ) . under the assumptions in theorem
[ theorem : improved lower bound on the total variation distance ] , then @xmath47 where @xmath48 & & \hspace*{-1 cm } \theta \triangleq 3 + \frac{7}{\lambda } + \frac{1}{\lambda } \cdot \sqrt{(3\lambda+7)\bigl[(3 + 2e^{-1/2 } ) \lambda + 7\bigr]}. \label{eq : optimal theta for alpha1 and alpha2 equal to lambda}\end{aligned}\ ] ] [ corollary : lower bound on the total variation distance ] see section [ subsubsection : proof of the corollary with the improved lower bound on the total variation distance ] .
we conclude our discussion in this letter by outlining a use of the new lower bounds in this work : the use of the new lower bound on the total variation distance for the poisson approximation of a sum of independent bernoulli random variables is exemplified by sason ( 2012 ) .
this work introduces new entropy bounds for discrete random variables via maximal coupling , providing bounds on the difference between the entropies of two discrete random variables in terms of the local and total variation distances between their probability mass functions .
the new lower bound on the total variation distance for the poisson approximation from this work was involved in the calculation of some improved bounds on the difference between the entropy of a sum of independent bernoulli random variables and the entropy of a poisson random variable of the same mean .
a possible application of the latter problem is related to getting bounds on the sum - rate capacity of a noiseless @xmath49-user binary adder multiple - access channel ( see sason ( 2012 ) ) .
the proof of theorem [ theorem : improved lower bound on the total variation distance ] starts similarly to the proof of theorem 2 of barbour and hall ( 1984 ) .
however , it significantly deviates from the original analysis in order to derive an improved lower bound on the total variation distance .
let @xmath0 be independent bernoulli random variables with @xmath1 .
let @xmath2 , @xmath3 for every @xmath4 , and @xmath5 with mean @xmath6 . from the basic equation of the chen - stein method , equation holds for an arbitrary bounded function @xmath8 .
furthermore , it follows from the proof of theorem 2 of barbour and hall ( 1984 ) that @xmath50 \bigr\}}{2 \ , \sup_{k \in { \ensuremath{\mathbb{n}}}_0 } \bigl| \lambda f(k+1 ) - k f(k ) \bigr| } \label{eq : fourth step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] which holds , in general , for an arbitrary bounded function @xmath8 . at this point , we deviate from the proof of theorem 2 of barbour and hall ( 1984 ) by generalizing and refining ( in a non - trivial way ) the original analysis .
the general problem with the current lower bound in is that it is not calculable in closed form for a given @xmath51 , so one needs to choose a proper function @xmath51 and derive a closed - form expression for a lower bound on the right - hand side of . to this end , let @xmath52 where @xmath44 and @xmath45 are fixed constants ( note that @xmath53 in needs to be positive for @xmath51 to be a bounded function ) . in order to derive a lower bound on the total variation distance
, we calculate a lower bound on the numerator and an upper bound on the denominator of the right - hand side of for the function @xmath51 in . referring to the numerator of the right - hand side of with @xmath51 in , for every @xmath54 , @xmath55 & & = \int_{v_j + 1 - \alpha_2}^{v_j + 2 - \alpha_2 } \frac{\mathrm{d}}{\mathrm{d}u } \left ( ( u+\alpha_2-\alpha_1 ) \ , \exp\bigl(-\frac{u^2}{\theta \lambda}\bigr ) \right ) \ , \mathrm{d}u \nonumber \\[0.2 cm ] & & = \int_{v_j + 1 - \alpha_2}^{v_j + 2 - \alpha_2 } \left(1 - \frac{2u ( u+\alpha_2-\alpha_1)}{\theta \lambda } \right ) \exp\bigl(-\frac{u^2}{\theta \lambda}\bigr ) \ ,
\mathrm{d}u \nonumber \\[0.2 cm ] & & = \int_{v_j + 1 - \alpha_2}^{v_j + 2 - \alpha_2 } \left(1 - \frac{2 u^2}{\theta \lambda}\right ) \ , \exp\bigl(-\frac{u^2}{\theta \lambda}\bigr ) \ , \mathrm{d}u - \frac{2(\alpha_2 - \alpha_1)}{\theta \lambda } \int_{v_j + 1 - \alpha_2}^{v_j + 2 - \alpha_2 }
u \ , \exp\bigl(-\frac{u^2}{\theta \lambda}\bigr ) \ , \mathrm{d}u \nonumber \\[0.2 cm ] & & = \int_{v_j + 1 - \alpha_2}^{v_j + 2 - \alpha_2 } \left(1 - \frac{2 u^2}{\theta \lambda}\right ) \ , \exp\bigl(-\frac{u^2}{\theta \lambda}\bigr ) \ , \mathrm{d}u \nonumber \\ & & \hspace*{0.4 cm } - ( \alpha_2 - \alpha_1 ) \left [ \exp\biggl(-\frac{(v_j + 2 - \alpha_2)^2}{\theta \lambda}\biggr ) - \exp\biggl(-\frac{(v_j + 1 - \alpha_2)^2}{\theta \lambda}\biggr ) \right ] .
\label{eq : fifth step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] we rely in the following on the inequality @xmath56 applying it to the integral on the right - hand side of gives that @xmath55 & & \geq \int_{v_j + 1 - \alpha_2}^{v_j + 2 - \alpha_2 } \left(1 - \frac{3 u^2}{\theta \lambda}\right ) \
, \mathrm{d}u - ( \alpha_2 - \alpha_1 ) \left [ \exp\biggl(-\frac{(v_j + 2 - \alpha_2)^2}{\theta \lambda}\biggr ) - \exp\biggl(-\frac{(v_j + 1 - \alpha_2)^2}{\theta \lambda}\biggr ) \right ] \nonumber \\[0.1 cm ] & & \geq 1 - \frac{\bigl(v_j + 2 - \alpha_2\bigr)^3 - \bigl(v_j + 1 - \alpha_2\bigr)^3}{\theta \lambda } \nonumber \\[0.1 cm ] & & \hspace*{0.4 cm } - \bigl|\alpha_2 - \alpha_1\bigr| \cdot \left| \exp\biggl(-\frac{(v_j + 2 - \alpha_2)^2}{\theta \lambda}\biggr ) - \exp\biggl(-\frac{(v_j + 1 - \alpha_2)^2}{\theta \lambda}\biggr ) \right| .
\label{eq : sixth step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] in order to proceed , note that if @xmath57 then ( on the basis of the mean - value theorem of calculus ) @xmath58 \\[0.1 cm ] & & \leq e^{-\min\{x_1 , x_2\ } } \ , |x_1 - x_2|\end{aligned}\ ] ] which , by applying it to the second term on the right - hand side of , gives that for every @xmath54 @xmath59 & & \leq \exp\left(-\frac{\min\bigl\{(v_j+2-\alpha_2)^2 , \ , ( v_j+1-\alpha_2)^2 \bigr\}}{\theta \lambda}\right ) \cdot \left(\frac{(v_j+2-\alpha_2)^2-(v_j+1-\alpha_2)^2}{\theta \lambda}\right ) \ , .
\label{eq : seventh step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] since @xmath60 then @xmath61 & & \geq \left\ { \begin{array}{cl } 0 \quad & \mbox{if $ \alpha_2 \geq 1 $ } \\[0.1
cm ] ( 1-\alpha_2)^2 \quad & \mbox{if $ \alpha_2 < 1 $ } \end{array } \right .
\nonumber \\[0.1 cm ] & & = \bigl(1-\alpha_2\bigr)_{+}^2 \label{eq : 8th step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] where @xmath62 hence , the combination of the two inequalities in gives that @xmath59 & & \leq \exp\left(-\frac{(1-\alpha_2)_+^2}{\theta \lambda } \right ) \cdot \left(\frac{\left|(v_j + 2 - \alpha_2)^2 - ( v_j + 1 - \alpha_2)^2 \right|}{\theta \lambda } \right ) \nonumber \\[0.1 cm ] & & = \exp\left(-\frac{(1-\alpha_2)_+^2}{\theta \lambda } \right ) \cdot \frac{\left|2v_j + 3 - 2 \alpha_2 \right|}{\theta \lambda } \nonumber \\[0.1 cm ] & & \leq \exp\left(-\frac{(1-\alpha_2)_+^2}{\theta \lambda } \right ) \cdot \frac{2v_j + \left|3 - 2 \alpha_2 \right|}{\theta \lambda } \label{eq : ninth step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] and therefore , a combination of the inequalities in and gives that @xmath55 & & \geq 1 - \frac{\bigl(v_j + 2 - \alpha_2\bigr)^3 - \bigl(v_j + 1 - \alpha_2\bigr)^3}{\theta \lambda } \nonumber \\[0.1 cm ] & & \hspace*{0.4 cm } - \bigl|\alpha_2 - \alpha_1\bigr| \cdot \exp\left(-\frac{(1-\alpha_2)_+^2}{\theta \lambda } \right ) \cdot \frac{2v_j + \left|3 - 2 \alpha_2 \right|}{\theta \lambda } \ ; .
\label{eq : 10th step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] let @xmath63 ; then @xmath55 & & \geq 1 - \frac{\bigl(u_j + \lambda + 2 - \alpha_2\bigr)^3 - \bigl(u_j + \lambda + 1 - \alpha_2\bigr)^3}{\theta \lambda } \nonumber \\[0.1 cm ] & & \hspace*{0.4 cm } - \bigl|\alpha_2 - \alpha_1\bigr| \cdot \exp\left(-\frac{(1-\alpha_2)_+^2}{\theta \lambda } \right ) \cdot \frac{2u_j + 2\lambda + \left|3 - 2 \alpha_2 \right|}{\theta \lambda } \nonumber \\[0.1 cm ] & & = 1 - \frac{3 u_j^2 + 3 \bigl(3 - 2\alpha_2 + 2 \lambda\bigr ) u_j + ( 2-\alpha_2+\lambda)^3 - ( 1-\alpha_2+\lambda)^3}{\theta \lambda } \nonumber \\[0.1 cm ] & & \hspace*{0.4 cm } - \bigl|\alpha_2 - \alpha_1\bigr| \cdot \exp\left(-\frac{(1-\alpha_2)_+^2}{\theta \lambda } \right ) \cdot \frac{2u_j + 2\lambda + \left|3 - 2 \alpha_2 \right|}{\theta \lambda } \ ; .
\label{eq : 11th step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] in order to derive a lower bound on the numerator of the right - hand side of , for the function @xmath51 in , we need to calculate the expected value of the right - hand side of . to this end
, the first and second moments of @xmath64 are calculated as follows : @xmath65 where equality ( a ) holds since the binary random variables @xmath0 are independent and @xmath66 . by taking expectations on both sides of , one obtains from and that @xmath67 \nonumber \\[0.2 cm ] & & \geq 1 - \frac{3 \bigl(\lambda - p_j - \sum_{i \neq j } p_i^2 + p_j^2\bigr ) + 3 \bigl(3 - 2\alpha_2 + 2 \lambda\bigr ) \bigl(-p_j \bigr ) + ( 2-\alpha_2+\lambda)^3 - ( 1-\alpha_2+\lambda)^3}{\theta \lambda } \nonumber \\[0.2 cm ] & & \hspace*{0.4 cm } - \bigl|\alpha_2 - \alpha_1\bigr| \cdot \exp\left(-\frac{(1-\alpha_2)_+^2}{\theta \lambda } \right ) \cdot \left(\frac{-2p_j + 2\lambda + \left|3 - 2 \alpha_2
\right|}{\theta \lambda } \right ) \nonumber \\[0.2 cm ] & & = 1 - \frac{3 \lambda + ( 2-\alpha_2+\lambda)^3 - ( 1-\alpha_2+\lambda)^3 - \bigl[3p_j(1-p_j ) + 3 \sum_{i \neq j } p_i^2 + 3 \bigl(3 - 2\alpha_2 + 2 \lambda\bigr ) p_j \bigr]}{\theta \lambda } \nonumber \\[0.2 cm ] & & \hspace*{0.4 cm } - \biggl(\frac{\bigl|\alpha_2 - \alpha_1\bigr| \ , \bigl(2\lambda - 2 p_j+ \left|3 - 2 \alpha_2 \right| \bigr)}{\theta \lambda } \biggr ) \cdot \exp\left(-\frac{(1-\alpha_2)_+^2}{\theta \lambda } \right ) \nonumber \\[0.2 cm ] & & \geq 1 - \frac{3 \lambda + ( 2-\alpha_2+\lambda)^3 - ( 1-\alpha_2+\lambda)^3 - \bigl(9 - 6\alpha_2 + 6 \lambda\bigr ) p_j}{\theta \lambda } \nonumber
\\[0.2 cm ] & & \hspace*{0.4 cm } - \biggl(\frac{\bigl|\alpha_2 - \alpha_1\bigr| \ , \bigl(2\lambda+ \left|3 - 2 \alpha_2 \right| \bigr)}{\theta \lambda } \biggr ) \cdot \exp\left(-\frac{(1-\alpha_2)_+^2}{\theta \lambda }
\right ) \ ; .
\label{eq : 12th step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] therefore , from , the following lower bound on the right - hand side of holds @xmath68 \bigr\ } \geq \left ( \frac{3\bigl(3 - 2\alpha_2 + 2\lambda\bigr)}{\theta \lambda } \right ) \sum_{j=1}^n p_j^3 \nonumber \\ & & \hspace*{-1.2 cm } + \left(1-\frac{3 \lambda + ( 2-\alpha_2+\lambda)^3 - ( 1-\alpha_2+\lambda)^3 + |\alpha_1 - \alpha_2| \bigl(2\lambda+|3 - 2\alpha_2|\bigr ) \exp\left(-\frac{(1-\alpha_2)_+^2}{\theta \lambda } \right)}{\theta \lambda } \right ) \sum_{j=1}^n p_j^2 \ , .
\label{eq : 12.5th step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] note that if @xmath69 , which is a condition that is involved in the maximization of , then the first term on the right - hand side of can be removed , and the resulting lower bound on the numerator of the right - hand side of takes the form @xmath70 \bigr\ } \geq \bigl(1 - h_{\lambda}(\alpha_1 , \alpha_2 , \theta ) \bigr ) \sum_{j=1}^n p_j^2 \label{eq : 13th step in the derivation of the improved lower bound on total variation distance}\ ] ] where the function @xmath71 is introduced in .
we turn now to deriving an upper bound on the denominator of the right - hand side of .
therefore , we need to derive a closed - form upper bound on @xmath72 with the function @xmath51 in . for every @xmath73 , @xmath74 + ( \lambda - k ) \ , f(k ) .
\label{eq : 14th step in the derivation of the improved lower bound on total variation distance}\ ] ] in the following , we derive bounds on each of the two terms on the right - hand side of , and we start with the first term .
let @xmath75 then @xmath76 for every @xmath73 , and by the mean - value theorem of calculus , @xmath77 \nonumber \\ & & = \left(1-\frac{2 c_k^2}{\theta \lambda}\right ) \
, \exp\left(-\frac{c_k^2}{\theta \lambda}\right ) + \left(\frac{2(\alpha_1-\alpha_2 ) c_k}{\theta \lambda}\right ) \ , \exp\left(-\frac{c_k^2}{\theta \lambda}\right ) \ , .
\label{eq : 15th step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] referring to the first term on the right - hand side of , let @xmath78 then the global maximum and minimum of @xmath79 over the non - negative real line are obtained at @xmath80 and @xmath81 , respectively , and therefore @xmath82 let @xmath83 ; then it follows that the first term on the right - hand side of satisfies the inequality @xmath84 furthermore , referring to the second term on the right - hand side of , let @xmath85 then the global maximum and minimum of @xmath86 over the real line are obtained at @xmath87 and @xmath88 , respectively , and therefore @xmath89 let this time @xmath90 ; then it follows that the second term on the right - hand side of satisfies @xmath91 hence , on combining the equality in with the two inequalities in and , it follows that the first term on the right - hand side of satisfies @xmath92 \leq \lambda + \sqrt{\frac{2\lambda}{\theta e } } \cdot |\alpha_1 - \alpha_2| \ , , \quad \forall \ , k \in { \ensuremath{\mathbb{n}}}_0 . \label{eq : 18th step in the derivation of the improved lower bound on total variation distance}\ ] ] we continue the analysis by a derivation of bounds on the second term of the right - hand side of . for the function @xmath51 in
, it is equal to @xmath93 \ , \bigl [ ( k-\alpha_2)+(\alpha_2-\alpha_1 ) \bigr ] \
, \exp\biggl(-\frac{(k-\alpha_2)^2}{\theta \lambda}\biggr ) \nonumber \\ & & = \bigl [ ( \lambda-\alpha_2 ) ( k-\alpha_2 ) + ( \alpha_2-\alpha_1 ) ( \lambda-\alpha_2 ) - ( k-\alpha_2)^2 + ( \alpha_1 - \alpha_2 ) ( k-\alpha_2 ) \bigr ] \ , \exp\biggl(-\frac{(k-\alpha_2)^2}{\theta \lambda}\biggr ) \nonumber \\[0.1 cm ] & & = \bigl [ \sqrt{\theta \lambda } \ , ( \lambda-\alpha_2 ) \ , v_k - \theta \lambda \ , v_k^2 - \sqrt{\theta \lambda } \ , ( \alpha_2 - \alpha_1 ) \ , v_k + ( \alpha_2 - \alpha_1 ) ( \lambda - \alpha_2 ) \bigr ] \ , e^{-v_k^2 } \ , , \quad v_k \triangleq \frac{k-\alpha_2}{\sqrt{\theta \lambda } } \ ; \ ; \forall \ , k \in { \ensuremath{\mathbb{n}}}_0 \nonumber \\ & & = ( c_0 + c_1 v_k + c_2 v_k^2 ) \ , e^{-v_k^2 } \label{eq : 19th step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] where the coefficients @xmath94 and @xmath95 are introduced in eqs . ,
respectively . in order to derive bounds on the left - hand side of , let us find the global maximum and minimum of the function @xmath96 in : @xmath97 note that @xmath98 and @xmath96 is differentiable over the real line , so the global maximum and minimum of @xmath96 are attained at finite points and their corresponding values are finite . by setting the derivative of @xmath96 to zero
, we have that the candidates for the global maximum and minimum of @xmath96 over the real line are the real zeros @xmath42 of the cubic polynomial equation in .
note that by their definition in , the values of @xmath42 are _ independent _ of the value of @xmath73 , and also the size of the set @xmath42 is equal to 3 ( see remark [ remark : the size of the set of real zeros is equal to 3 ] ) . hence , it follows from that @xmath99 where these bounds on the second term on the right - hand side of are independent of the value of @xmath73 . in order to get bounds on the left - hand side of ,
note that from the bounds on the first and second terms on the right - hand side of ( see and , respectively ) then for every @xmath73 @xmath100 & & \leq \lambda \ , f(k+1 ) - k \ , f(k ) \nonumber \\[0.1 cm ] & & \leq \max_{i \in \{1 , 2 , 3\ } } \{x(u_i)\ } + \lambda + \sqrt{\frac{2\lambda}{\theta e } } \cdot |\alpha_1 - \alpha_2| \label{eq : 21st step in the derivation of the improved lower bound on total variation distance}\end{aligned}\ ] ] which yields that the following inequality is satisfied : @xmath101 where the function @xmath102 is introduced in . finally , by combining the inequalities in eqs . , and , the lower bound on the total variation distance in follows .
the existing upper bound on the total variation distance in was derived in theorem 1 of barbour and hall ( 1984 ) ( see theorem [ theorem : bounds on the total variation distance - barbour and hall 1984 ] here ) .
this completes the proof of theorem [ theorem : improved lower bound on the total variation distance ] .
corollary [ corollary : lower bound on the total variation distance ] follows as a special case of theorem [ theorem : improved lower bound on the total variation distance ] when the proposed function @xmath51 in is chosen such that two of its three free parameters ( i.e. , @xmath103 and @xmath104 ) are determined sub - optimally , and its third parameter ( @xmath53 ) is determined optimally in terms of the sub - optimal selection of the two other parameters .
more explicitly , let @xmath103 and @xmath104 in be set to be equal to @xmath30 ( i.e. , @xmath46 ) . from
, this setting implies that @xmath105 and @xmath106 ( since @xmath107 ) .
the cubic polynomial equation in , which corresponds to this ( possibly sub - optimal ) setting of @xmath103 and @xmath104 , is @xmath108 whose zeros are @xmath109 .
the function @xmath96 in therefore takes the form @xmath110 so @xmath111 and @xmath112 .
this implies that @xmath113 and therefore @xmath71 and @xmath102 in and , respectively , are simplified to @xmath114 this sub - optimal setting of @xmath103 and @xmath104 in implies that the coefficient @xmath43 in is replaced with a loosened version @xmath115 let @xmath116 ; then is simplified to @xmath117 .
it therefore follows from , and that @xmath118 where @xmath119 and , in general , @xmath120 due to the above restricted constraint on @xmath53 ( see versus ) . differentiating the function inside the supremum w.r.t . @xmath53 and setting its derivative to zero , one gets the following quadratic equation in @xmath53 : @xmath121 whose positive solution is the optimized value of @xmath53 in .
furthermore , it is clear that this value of @xmath53 in is larger than , e.g. , 3 , so it satisfies the constraint in .
this completes the proof of corollary [ corollary : lower bound on the total variation distance ] .
the new lower bounds on the total variation distance in theorem [ theorem : improved lower bound on the total variation distance ] and corollary [ corollary : lower bound on the total variation distance ] scale like @xmath41 , similarly to the known upper and lower bounds in theorem [ theorem : bounds on the total variation distance - barbour and hall 1984 ] that originally appear in theorems 1 and 2 of barbour and hall ( 1984 ) .
however , the new lower bounds offer a significant improvement over the known lower bound in theorem [ theorem : bounds on the total variation distance - barbour and hall 1984 ] .
more explicitly , from theorems 1 and 2 of barbour and hall ( 1984 ) , the ratio between the upper and lower bounds on the total variation distance ( see ) is equal to 32 in the two extreme cases where @xmath122 or @xmath123 . in the following ,
we calculate the ratio of the same upper bound and the new lower bound in corollary [ corollary : lower bound on the total variation distance ] at these two extreme cases . in the limit where @xmath123 , this ratio tends to @xmath124 & & = \frac{2}{e } \ , \lim_{\lambda \rightarrow \infty } \frac{\theta \bigl(2 e^{-{1/2}}+\theta \bigr)}{\theta-\bigl(3+\frac{7}{\lambda}\bigr ) } \nonumber \\[0.1 cm ] & & = \frac{6}{e } \ , \left(1+\sqrt{1+\frac{2}{3 } \cdot e^{-1/2 } } \right)^2 \approx 10.539 \label{eq : limit of the ratio between the upper and lower bounds on the total variation distance when lambda tends to infinity}\end{aligned}\ ] ] where the last equality follows from , since @xmath125
. furthermore , the limit of this ratio when @xmath122 is equal to @xmath126 & & \stackrel{\text{(a)}}{= } \frac{28}{e } \ , \lim_{\lambda \rightarrow 0 } \left(\frac{2 e^{-{1/2}}+\theta)}{\theta-\bigl(3+\frac{7}{\lambda}\bigr)}\right ) \nonumber \\ & & \stackrel{\text{(b)}}{= } \frac{56}{e } \approx 20.601 \label{eq : limit of the ratio between the upper and lower bounds on the total variation distance when lambda tends to zero}\end{aligned}\ ] ] where equalities ( a ) and ( b ) hold since , from , it follows that @xmath127 .
this implies that corollary [ corollary : lower bound on the total variation distance ] improves the original lower bound on the total variation distance in theorem 2 of barbour and hall ( 1984 ) by a factor of @xmath128 in the limit where @xmath123 , and it also improves it by a factor of @xmath129 if @xmath122 while still having a closed - form expression for the lower bound in corollary [ corollary : lower bound on the total variation distance ] .
the only reason for this improvement is related to the optimal choice of the free parameter @xmath53 in , versus its sub - optimal choice in the proof of theorem 2 of barbour and hall ( 1984 ) . this observation has motivated to further improve the lower bound by introducing the two additional parameters @xmath44 in theorem [ theorem : improved lower bound on the total variation distance ] ; these parameters give two additional degrees of freedom in the function @xmath51 in ( according to the proof in section [ subsubsection : proof of the corollary with the improved lower bound on the total variation distance ] , these two parameters are set to be equal to @xmath30 for the derivation of the loosened and simplified bound in corollary [ corollary : lower bound on the total variation distance ] ) .
the improvement in the lower bound of theorem [ theorem : improved lower bound on the total variation distance ] ( in comparison to corollary [ corollary : lower bound on the total variation distance ] ) is especially dominant for low values of @xmath30 , as is shown in figure [ figure : ratio ot upper and lower bounds on the total variation distance ] . note , however , that no improvement is obtained for high values of @xmath30 ( e.g. , for @xmath130 , as is shown by figure [ figure : ratio ot upper and lower bounds on the total variation distance ] on noticing that the curves in this plot merge at large values of @xmath30 ) .
the lower bound on the total variation distance in theorem [ theorem : improved lower bound on the total variation distance ] implies the bound in corollary [ corollary : lower bound on the total variation distance ] ( see the proof in section [ subsubsection : proof of the corollary with the improved lower bound on the total variation distance ] ) .
corollary [ corollary : lower bound on the total variation distance ] further implies the lower bound on the total variation distance in theorem 2 of barbour and hall ( 1984 ) ( see theorem [ theorem : bounds on the total variation distance - barbour and hall 1984 ] here ) .
the latter claim follows from the fact that the lower bound in with the coefficient @xmath131 in was loosened in the proof of theorem 2 of barbour and hall ( 1984 ) by a sub - optimal selection of the parameter @xmath53 , which leads to a lower bound on @xmath131 ( the sub - optimal selection of @xmath53 in the proof of theorem 2 of barbour and hall ( 1984 ) is @xmath132 ) . on the other hand , the optimized value of @xmath53 that is used in provides an exact closed - form expression for @xmath131 in , and it leads to the derivation of the improved lower bound in corollary [ corollary : lower bound on the total variation distance ] .
theorem 1.2 of deheuvels and pfeifer ( 1986 ) provides an asymptotic result for the total variation distance between the distribution of the sum @xmath11 of @xmath21 independent bernoulli random variables with @xmath1 and the poisson distribution with mean @xmath24 .
it shows that when @xmath133 and @xmath134 as @xmath28 then @xmath135 this implies that the ratio of the upper bound on the total variation distance in theorem 1 of barbour and hall ( 1984 ) ( see theorems [ theorem : bounds on the total variation distance - barbour and hall 1984 ] here ) and this asymptotic expression is equal to @xmath136 .
therefore , the ratio between the exact asymptotic value in and the new lower bound in is equal to @xmath137 .
it therefore follows that , in the limit where @xmath122 , the new lower bound on the total variation in is smaller than the exact value by no more than 1.69 , and for @xmath138 , it is smaller than the exact asymptotic result by a factor of 2.55 . : the anonymous reviewer is acknowledged for suggestions that led to improvement of the presentation of this paper .
this research work was supported by the israeli science foundation ( isf ) , grant number 12/12 .
arratia , r. , goldstein , l. , gordon , l. , 1990 . poisson approximation and the chen - stein method . statistical science 5 , 403424 .
+ barbour , a.d . , chen , l. h. y. , 2005 .
an introduction to stein s method .
lecture notes series , institute for mathematical sciences , singapore university press and world scientific .
+ barbour , a.d . , hall , p. , 1984 .
on the rate of poisson convergence .
mathematical proceedings of the cambridge philosophical society 95 , 473480 .
+ barbour , a.d . ,
holst , l. , janson , s. , 1992 .
poisson approximation .
oxford university press .
+ barbour , a. d. , johnson , o. , kontoyiannis , i. , madiman , m. , 2010 .
compound poisson approximation via information functionals .
electronic journal of probability 15 , 13441369 .
+ chen , l.h.y .
poisson approximation for dependent trials .
annals of probability 3 , 534545 .
+ deheuvels , p. , pfeifer , d. , 1986 . a semigroup approach to poisson approximation .
annals of probability 14 , 663676 .
+ kontoyiannis , i. , harremos , p. , johnson , o. , 2005 .
entropy and the law of small numbers .
ieee trans . on information theory 51 , 466472 .
+ le cam , l. , 1960 .
an approximation theorem for the poisson binomial distribution .
pacific journal of mathematics 10 , 11811197 .
+ ross , s.m . , pekz , e.a .
2007 . a second course in probability . probability bookstore .
+ ross , n. , 2011 . fundamentals of stein s method .
probability surveys 8 , 210293 .
+ sason , i. , 2012 .
entropy bounds for discrete random variables via coupling .
[ online ] .
available : http://arxiv.org/abs/1209.5259 . | new lower bounds on the total variation distance between the distribution of a sum of independent bernoulli random variables and the poisson random variable ( with the same mean ) are derived via the chen - stein method .
the new bounds rely on a non - trivial modification of the analysis by barbour and hall ( 1984 ) which surprisingly gives a significant improvement .
a use of the new lower bounds is addressed . _
keywords : _ chen - stein method , poisson approximation , total variation distance . |
hatr 4 is known to contain a non - eclipsing photometric binary central star with a period of 1.71 days ( ( * ? ? ?
* bond 1990 ) ) . on first inspection of the original imagery presented by (
* hartl & tritton ( 1985 ) ) , hatr 4 has the appearance of a classical butterfly bipolar nebula lying in the plane of the sky with twin lobes emanating in an east - west direction .
however , deeper h@xmath0 + [ nii ] imagery acquired by d. pollacco shows faint extensions in a north - south direction indicating that the bipolarity in fact lies perpendicular to that inferred from the original imagery . this deep narrow - band imagery , in combination with high - spectral and -spatial resolution spectroscopy acquired with vlt - uves ,
have been used to derive a spatio - kinematic model of hatr 4 in order to investigate the relationship between the pn and its binary central star ( ( * ? ? ?
* ; * ? ? ?
* tyndall 2011a , 2011b ) ) . to date , only four pne have been observationally confirmed as having been shaped by their central binary stars
abell 63 ( ( * ? ? ?
mitchell 2007 ) ) , abell 41 ( ( * ? ? ?
* jones 2010 ) ) , abell 65 , ( ( * ? ? ?
* huckvale 2011 ) ) and ngc 6337 ( ( * ? ? ?
* hillwig 2010 ) , ( * ? ? ?
* garca - daz 2009 ) ) .
using the astrophysical modelling program _ shape _ ( ( * ? ? ?
* steffen 2011 ) ) , a spatio - kinematical model was developed in order to reconstruct the nebular morphology of hatr 4 . the nebular expansion velocity was assumed to follow a hubble - type flow , and the simplest best fitting model was an open - ended ovoid nebular shell waisted by a thick equatorial ring .
the model accurately reproduces both the observed velocity profiles and imagery .
the model shows that the main nebula does in fact extend in a north - south , and not east - west , direction , and that ` butterfly ' appearance of the central nebular regions is due to a projection effect .
hatr 4 possesses an elongated , axisymmetric morphology with an equatorial enhancement consistent with a nebular ring .
the ` butterfly ' appearance of the central region is shown to result from a line - of - sight inclination effect associated with the enhanced nebular waist , rather than a bipolar structure .
the modelling has revealed an age of order 8200 years at a distance of 3 kpc , and an equatorial expansion velocity of 26@xmath14 kms@xmath2 - both fairly typical for pne .
no evidence is found for extended emission or jet - like outflows associated with the pn , such as those found in ethos 1 ( ( * ? ? ?
* miszalski 2011a ) ) and the necklace ( ( * ? ? ?
* corradi 2011 ) ) .
the inclination of the nebular symmetry axis is found to be 75@xmath3@xmath15@xmath3 , consistent with the non - eclipsing nature of the binary central star .
further investigation into the central star system performed by hillwig et al .
( in preparation ) and bodman et al . ( in press ) , indicates that the inclination of the central binary plane falls within a similar range to that of the nebula as predicted by binary - induced pn shaping theories ( ( * ? ? ?
* nordhaus & blackman 2006 ) ) .
this alignment between the nebular symmetry axis and binary plane provides strong evidence that hatr 4 has been shaped by its central binary star , making it one of only five pne to have had this link observationally shown ( ( * ? ? ?
* jones 2011 ) ) . | we present the first detailed spatio - kinematical analysis of the planetary nebula hatr 4 , one of few known to contain a post - common - envelope central star system .
based on high spatial and spectral resolution spectroscopy of the [ oiii ] nebular emission line , in combination with deep , narrow - band imagery , a spatio - kinematical model was developed in order to accurately determine the three - dimensional morphology and orientation of hatr 4 .
the nebula is found to display an extended ovoid morphology with an equatorial enhancement consistent with a toroidal waist - a feature believed to be typical of central star binarity .
the nebular inclination is found to be in good agreement with that determined for the binary plane , providing strong evidence that shaping and evolution of hatr 4 has been influenced by its central binary system - making hatr 4 one of only 5 planetary nebulae to have had this observationally proven . |
many potential quantum information processing systems are controlled by means of a set of time - dependent parameters , such as quasi - static electromagnetic fields @xcite , radio - frequency , @xcite , or optical fields @xcite . for most such systems , it is relatively simple to device a set of controls that implement a given evolution in an ideal situation . often , however , a more careful choice of controls can lead to an implementation that is less sensitive to variations in unknown or uncontrollable system parameters .
examples of such robust implementations include system specific solutions such as the hot gate for ion trap quantum computing which is insensitive to vibrational excitations @xcite , as well as more general techniques such as composite pulses , a technique originating in nmr spectroscopy @xcite . in this article
we describe a numerical method for designing robust controls for systems where the evolution is adequately described by a possibly non - unitary evolution operator @xmath0 .
this form does not allow a general master equation formulation , but it is sufficient to establish worst case behavior in many quantum computing settings where the worst case effects of decoherence and loss can be adequately modeled by a schrdinger equation with a non - hermitian hamiltonian .
as an application of the method , we will consider the construction of a robust phase shift operation for the rare earth quantum computing ( reqc ) system @xcite , which is based on rare - earth ions embedded in a cryogenic crystal . in each ion ,
two metastable ground - state hyperfine levels , labeled @xmath1 and @xmath2 , serve as a qubit register which is manipulated via optical transitions from both states to an inhomogeneously shifted excited state @xmath3 .
the reqc system is an ensemble quantum computing system and macroscopic numbers of ions are manipulated in parallel , addressed by the value of the inhomogeneous shift of their @xmath3-state . to obtain a sufficient number of ions within each frequency channel ,
it is necessary to operate on all ions within a finite range of inhomogeneous shifts , and the main difficulty in operating the reqc system is to achieve the same evolution for each of these ions independent of their particular inhomogeneous shift . this article is divided into two sections : in section [ sec : design - robust - puls ] we describe the method we have used to design robust gate implementations ; these results should be applicable to a variety of quantum information processing systems . in section [ sec : optimal - control - reqc ] we present the results of applying the method to a specific problem relating to the reqc system , and show that it is indeed possible to obtain very high degrees of robustness .
we consider a collection of quantum system which evolve according to a set of time - dependent controls @xmath4 .
in addition to @xmath5 , the single system hamiltonian @xmath6 depends on a system specific set of uncontrollable or unknown parameters @xmath7 , such as field strength or quantum numbers corresponding to unused degrees of freedom .
the evolution of each system is governed by the schrdinger equation , @xmath8 where we will allow the hamiltonian @xmath9 to include non - hermitian terms describing loss and decoherence .
our goal is to choose a set of controls that lead to an evolution @xmath10 which is as close as possible to a given desired evolution @xmath11 over a range of @xmath7-values . to quantify this
, we introduce an objective functional @xmath12 which describes the performance of a set of controls @xmath5 for a given value of @xmath7 . by convention
we take a low value of @xmath13 to indicate a good performance , and the problem of finding a robust set of controls thus corresponds to minimizing @xmath14 where @xmath15 is the set of @xmath7-values for which we want the implementation to perform well . the conceptually simple approach we have taken to this problem
is to replace @xmath15 with a discrete subset @xmath16 , so that the minimization of @xmath17 has the form of a standard minimax problem , which can be solved efficiently provided that we are able to calculate @xmath18 . below
we show how to achieve this by methods from optimal control theory . in this section
we show how @xmath19 may be calculated for a quite general class of objective functionals . to keep the notation simple and avoid unnecessary restrictions
, we will consider the following generalization of the schrdinger equation , @xmath20 determining the evolution of a complex - valued , time - dependent matrix @xmath21 due to a set of real - valued , time - dependent controls , @xmath4 .
we will consider objective functionals of the form @xmath22 where @xmath23 is a real - valued function quantifying how close the final state @xmath10 is to our goal , and the real valued function @xmath24 , referred to as a penalty function , can be chosen to discourage the use of certain control values .
our goal is to calculate @xmath19 subject to the constraint that @xmath5 and @xmath21 obey eq . .
to achieve this we introduce the modified objective functional : @xmath25 + \text{h.c . }
\right ) dt,\ ] ] where the complex time - dependent adjoint state matrix @xmath26 is in effect a continuous set of lagrange multipliers leaving @xmath27 identical to @xmath13 , provided that @xmath21 and @xmath5 obey eq . .
if we require @xmath26 to obey the adjoint equations @xmath28 where @xmath21 and @xmath29 should be considered as independent with respect to the partial derivative and @xmath30 is defined as @xmath31 we find by integration by parts that the differential of @xmath27 is given by @xcite @xmath32 from which the derivatives of @xmath13 with respect to the parameters used to parametrize @xmath5 can be calculated by the chain rule .
we now return to the case of a quantum system governed by the schrdinger equation . if we assume the penalty function @xmath33 to be independent of @xmath21 , the adjoint equations in this case are @xmath34 and @xmath35
is given by with @xmath36 the role of the adjoint state and the adjoint equations is often described as back - propagating the errors in achieving the desired final state .
if @xmath9 is hermitian , the boundary value for @xmath26 can be optimized for numerical computation as shown in appendix [ sec : optimized - co - state ] . we will now discuss the choice of the function @xmath37 , quantifying how well the obtained evolution @xmath10 approximates @xmath11 .
as we are concerned with quantum information processing we will assume that all operations start out with an unknown state in the qubit subspace @xmath38 of the full system hilbert space @xmath39 , and that this subspace is invariant under the ideal evolution @xmath11 .
the function @xmath37 should not depend on the evolution of states outside @xmath38 , nor on collective phases on the states originating in @xmath38 .
a cautious choice of @xmath37 fulfilling these conditions could be based on the worst case overlap fidelity @xcite : @xmath40 which measures the least possible overlap between the obtained output state @xmath41 and the ideal output @xmath42 for initial states in @xmath38 .
this fidelity measure has the desirable quality that both population transfer from @xmath38 to @xmath43 and population transfer completely out of @xmath39 , as described by a non - unitary evolution , is counted as loss of fidelity . from the point of view of optimal control , a significant drawback of the worst case overlap fidelity , @xmath44 ,
is that it is computationally complicated @xcite .
a computationally accessible fidelity measure which share many appealing features with @xmath44 is the trace fidelity @xcite , @xmath45 where @xmath46 .
as shown in appendix [ sec : trace - fidelity ] , @xmath47 is related to @xmath44 by the strict bound @xmath48 indicating that we can safely replace @xmath44 by @xmath47 for numerical computations on a few qubits at high fidelity . for numerical calculations
it is beneficial to use @xmath49 rather than @xmath47 @xcite ; in the calculations presented in the next section we have used @xmath50 , leading to an adjoint state boundary condition of @xmath51 which can be directly computed .
one approach to minimizing @xmath13 is to directly solve the extremum condition @xmath52 for @xmath5 .
this task is significantly simplified if a penalty function proportional to @xmath53 is introduced , @xmath54 , so that the extremum condition according to reads , @xmath55 which may be used as an iterative formula for calculating @xmath5
. variations over this iterative approach give rise to the krotov @xcite and zhu - rabitz @xcite algorithms which have been shown to have excellent convergence properties , and have been successfully applied to optimal control of unitary transformations for one set of parameters by the group of r. kosloff @xcite . a unifying view of these direct methods
can be found in ref .
@xcite . in the work presented here
we have chosen to use an indirect minimization algorithm : rather than trying to solve the extremum condition directly , we have used the gradient information obtained through eq . as input for a general sequential quadratic programming procedure based on a constrained quasi - newton method @xcite .
the primary advantage of this approach is that we have total freedom to choose the parametrization of the controls , and can place arbitrary bounds on these .
this allows us to more accurately model the fact that the experimental limitations most often only distinguish between possible and impossible controls : no possible controls are significantly harder than others .
an explicit field strength limit also serves to introduce an absolute scale on which to introduce decay strengths etc .
-pulse sequence ( dashed line ) and the optimized ( solid line ) implementation of @xmath11 on far - detuned ions , as described by @xmath56 , where @xmath44 is calculated with respect to the identity , as these ions should ideally not be disturbed .
both implementations achieve worst case fidelities very close to unity , with the @xmath57-pulse sequence achieving the best results .
the curve for the optimized pulse is actually a running maximum , as the actual value of @xmath56 oscillates with a period of @xmath58 .
, width=317 ] the motivation for the work presented in this article has been the design of robust gate implementations for the reqc system mentioned in the introduction . as an example
, we will consider the construction of a robust implementation of the single qubit operation @xmath59 which could simply be implemented by a single @xmath60 pulse on the @xmath1-@xmath3 transition if we were not concerned with robustness .
our primary concern will be to make the implementation robust with respect to variations in the inhomogeneous shift @xmath61 of the @xmath3-state in order to allow the use of finite with channels .
since it is experimentally difficult to obtain a homogeneous field strength over the crystal , we would also prefer the implementation to be insensitive to variations in the relative field strength , which we will denote @xmath62 .
in addition to requiring the implementation of @xmath11 to be robust with respect to variations in @xmath61 and @xmath62 , we will add the requirement that ions outside the channels should not be affected , as this allows us to use the obtained implementation of @xmath11 as a part of a controlled phase shift operation @xcite : if the @xmath3-states of the qubit ion and a controlling ion are coupled sufficiently strongly by static dipole interaction , an excitation of the controlling ion will effectively shift the qubit ion out of the channel thus conditioning the evolution of the qubit on the state of the controlling ion . even though the simplest implementation of @xmath11 for an ideal ion with @xmath63 and @xmath64 would involve only the @xmath1 and @xmath3 levels ,
a robust implementation must also involve the @xmath2-state since the coupling of @xmath1 to @xmath3 will result in a @xmath61-dependent phase on @xmath1 which can only be compensated by introducing the same phase on the @xmath2-state , e.g. through phase compensating rotations @xcite .
a highly successful example of this approach is the @xmath57-pulse sequence suggested by roos and mlmer @xcite , which as illustrated by fig .
[ fig : fidelities](a ) yields a very robust implementation , achieving high fidelities over a wide range of parameter values .
we model the reqc system by the single ion hamiltonian @xmath65 which does not include any effects of decay or decoherence . in the notation introduced in section
[ sec : design - robust - puls ] , the system parameters are @xmath66 , the controls are @xmath67 , and the qubit subspace is @xmath68 . no penalty function is used : we use @xmath69 , and limit the field by strict bounds on @xmath70 , as this is the relevant limiting parameter in the reqc system . inspired by the success of the @xmath57-based solution and the hat - like fourier spectrum of the @xmath57-pulse , @xmath70 is parametrized in terms of a truncated fourier basis . based on trial and error
we have arrived at @xmath71 @xmath7-values to constitute @xmath72 , some within the neighborhood of @xmath63 , @xmath64 , and some at large detunings where the ions should not be disturbed .
the result of the optimization with this choice of @xmath72 is shown in fig .
[ fig : fidelities](b ) , where the circles indicate the members of @xmath72 .
it is evident from the plot that the optimization has achieved a high fidelity over an even larger range of parameters than the @xmath57-pulse sequence illustrated in fig .
[ fig : fidelities](a ) . with respect to not disturbing the detuned ions , both the optimized pulse and the @xmath57-pulses obtain fidelities within @xmath73 of unity for @xmath74 , where @xmath75 is the maximal resonant rabi frequency at @xmath64 .
as illustrated by fig .
[ fig:3levavoid_edge ] , which only shows fidelities at @xmath64 as @xmath10 is nearly independent of @xmath62 at @xmath76 , the @xmath57-pulse sequence performs better than the optimized pulse in this regime .
we have shown that it is possible to construct highly robust gate implementations for quantum information processing by a quite general method .
in particular , the method has been used to greatly enhance the performance of a gate implementation for a model reqc system by extending the range of inhomogeneous shifts and relative field strengths over which an acceptable performance is achieved .
the model reqc system used in this article ignores many performance degrading factors , the two most important being decoherence and implementation noise .
decoherence could in the present case be adequately modeled by a non - hermitian hamiltonian , for which we expect the method described in this paper to be able to find a robust implementation as in the decoherence - free case .
it is not clear , however , how the method could be extended to address the problem of robustness with respect to implementation imperfections .
the author would like to thank the people at the centre for quantum computer technology at the university of queensland for their hospitality , and klaus mlmer for valuable comments on the manuscript .
this research was funded by project esquire of the ist - fet programme of the ec .
in this section we prove the relation between the worst case overlap fidelity , @xmath44 , and the trace fidelity @xmath47 . referring to the definition ,
we note that @xmath44 is completely determined by the restriction @xmath77 of the operator @xmath78 to @xmath38 . since @xmath10 describes the evolution of a quantum system , it is possible to extend it to a unitary operation on a hilbert space containing @xmath39 , and @xmath77 is consequently the restriction of a unitary operator to @xmath38 . in the ideal case
@xmath77 will be equal to the identity on @xmath38 , perhaps with the exception of a complex phase .
@xmath44 is defined as the minimum of the overlap @xmath79 .
since the unit sphere of @xmath80 is compact , this minimum will be attained for some @xmath81 : @xmath82 .
we now extend @xmath83 to an orthonormal basis @xmath84 by the gram - schmidt process . evaluating the trace fidelity in this basis we find by the triangle inequality : @xmath85 where we have used that @xmath86 for all @xmath87 since @xmath77 is the restriction of a unitary operator . by rewriting we obtain the desired relation , @xmath88 .
we note that the established bound is strict in the sense that for any @xmath89 , the operator @xmath90 will fulfill eq . with equality for any @xmath87 .
in the case of a hermitian hamiltonian and a penalty function @xmath33 , that does not depend on the state @xmath21 , it is possible to modify the adjoint state boundary condition to reduce the required accuracy of the adjoint state propagation . in this case
, we find according to eqs . and that @xmath35 has the form @xmath91 \delta { { \ensuremath{\boldsymbol{\varepsilon}}}}(t ) \ , dt,\
] ] where @xmath92 and @xmath93 denotes the @xmath94-th columns of @xmath21 and @xmath26 respectively .
since @xmath9 and thus @xmath95 are assumed to be hermitian , @xmath35 as given by is not affected by adding to @xmath93 a component of @xmath96 for any real @xmath97 .
furthermore , since @xmath93 and @xmath92 evolve according to the same schrdinger equation , this corresponds to modifying the boundary condition for the adjoint state to read : @xmath98 for any real @xmath97 .
the obvious use of the freedom in the choice of boundary value is to minimize the norm of the adjoint state , in order to relax the requirements of the relative precision of the adjoint state propagation .
this minimum is easily calculated from , but we prefer to illustrate the physical background of the result by calculating it in a different way : the freedom in the choice of boundary value is allowed by the hermiticity of @xmath9 .
the same hermiticity ensures that @xmath92 is normalized , so that @xmath99 , where @xmath100 is the normalization function @xmath101 , is equal to @xmath102 . the gradient of @xmath103 , however , is different from that of @xmath37 .
in fact we find @xmath104\ ; { { \ensuremath{\boldsymbol{x}}}}_k,\ ] ] where @xmath92 and @xmath105 should be considered independent with respect to the derivative . comparing this expression to eq .
, it is tempting to let @xmath106 , which is indeed the answer found by minimizing @xmath107 subject to eq . . the modified boundary condition of eq
carries the same error information as that of eq . , but the required relative numerical precision when propagating the adjoint state will be significantly reduced . | quantum information processing systems are often operated through time dependent controls ; choosing these controls in a way that makes the resulting operation insensitive to variations in unknown or uncontrollable system parameters is an important prerequisite for obtaining high - fidelity gate operations . in this article
we present a numerical method for constructing such robust control sequences for a quite general class of quantum information processing systems . as an application of the method
we have designed a robust implementation of a phase - shift operation central to rare earth quantum computing , an ensemble quantum computing system proposed by ohlsson et .
al . @xcite . in this case
the method has been used to obtain a high degree of insensitivity with respect to differences between ensemble members , but it is equally well suited for quantum computing with a single physical system . |
it has been well known that many open - shell nuclei are deformed in the ground state .
the nuclear deformation generates the collective rotational motion , which is characterized by a pronounced rotational spectrum as well as strongly enhanced quadrupole transition probabilities .
theoretically , a standard way to discuss nuclear deformation is a self - consistent mean - field theory@xcite . by allowing the rotational symmetry to be broken in the mean - field potential , the mean - field theory provides an intuitive and transparent view of the nuclear deformation .
e.g. , _ ref .
@xcite for a recent systematic study of nuclear deformation based on the skyrme - hartree - fock - bogoliubov method .
the state - of - the - art mean - field approach also takes into account the effects beyond the mean - field approximation , such as the angular momentum projection and the configuration mixing @xcite .
the self - consistent mean - field method has been extensively applied also to hypernuclei @xcite ( see ref .
@xcite for a recent experimental review on @xmath0 hypernuclei ) .
these calculations have successfully reproduced the mass number dependence of @xmath0 binding energy , from a light nucleus @xmath5c to a heavy nucleus @xmath6pb .
we notice that most of these calculations have assumed spherical symmetry . only recently
, deformed calculations have been carried out in a broad mass region using the non - relativistic skyrme hartree - fock method @xcite .
the authors of ref . @xcite have reported that the hypernuclei which they studied have a similar deformation parameter to the corresponding core nuclei with the same sign . the aim of this paper is to study the deformation property of @xmath0 hypernuclei using the relativistic mean field ( rmf ) method , as an alternative choice of effective @xmath7 and @xmath8 interactions .
the rmf method has been as successful as the skyrme - hartree - fock method in describing stable nuclei as well as nuclei far from the stability line@xcite .
et al . _ have argued @xcite that the change in the nucleon spin - orbit potential due to the presence of @xmath0 particle is much more emphasized in the rmf approach as compared to the non - relativistic approach , since the spin - orbit potential in the former approach is actually given as a sum of scalar and vector potentials .
that is , even if the change in the mean - field potential ( given as a difference of scalar and vector potentials ) is small , the change in the spin - orbit potential may not necessarily be small .
therefore , a slightly different conclusion from that with the non - relativistic approach may result concerning the structure of hypernuclei .
in fact , we will demonstrate below that the shape of @xmath4c and @xmath3si nuclei are drastically changed when a @xmath0 particle is added to them .
the paper is organised as follows . in sec .
ii , we briefly summarize the rmf approach for @xmath0 hypernuclei . in sec .
iii , we apply the rmf method to ne and si isotopes , and discuss the influence of @xmath0 particle on the deformation of the hypernuclei .
we also discuss the deformation of @xmath4c and @xmath2c nuclei .
we summarize the paper in sec . iv .
in the rmf approach , nucleons and a @xmath0 particle are treated as structureless dirac particles , interacting through the exchange of virtual mesons , that is , the isoscalar scalar @xmath9 meson , the isoscalar vector @xmath10 meson , and the isovector vector @xmath11 meson .
the photon field is also taken into account to describe the coulomb interaction between protons .
the effective lagrangian for @xmath0 hypernuclei may be given as @xcite @xmath12\psi_\lambda , \label{rmf}\ ] ] where @xmath13 and @xmath14 are the dirac spinor and the mass for the @xmath0 particle , respectively .
notice that the @xmath0 particle couples only to the @xmath9 and @xmath10 mesons , as it is neutral and isoscalar . those coupling constants are denoted as @xmath15 and @xmath16 , respectively . for simplicity
, we neglect the tensor @xmath0-@xmath10 interaction .
this is justified since we consider only the ground state configuration , in which the @xmath0 particle occupies the lowest @xmath17 single - particle state@xcite , @xmath18 being the projection of the single - particle angular momentum onto the symmetry axis .
@xmath19 in eq .
( [ rmf ] ) is the standard rmf lagrangian for the nucleons .
e.g. , _ refs . @xcite for its explicit form .
we solve the rmf lagrangian ( [ rmf ] ) in the mean field approximation .
the variational principle leads to the dirac equation for the @xmath0 particle , @xmath20\psi_\lambda=\epsilon_\lambda\psi_\lambda,\ ] ] where @xmath21 is the single - particle energy for the @xmath0 particle state , and the klein - gordon equation for the mesons , @xmath22\sigma({\mbox{\boldmath $ r$}})&=&-g_\sigma\rho_s({\mbox{\boldmath $ r$ } } ) -g_2\,\sigma({\mbox{\boldmath $ r$}})^2-g_3\,\sigma({\mbox{\boldmath $ r$}})^3 \nonumber \\ & & -g_{\sigma\lambda}\psi_\lambda^\dagger({\mbox{\boldmath $ r$}})\gamma^0\psi_\lambda({\mbox{\boldmath $ r$ } } ) , \end{aligned}\ ] ] @xmath23\omega^0({\mbox{\boldmath $ r$}})=g_\omega\rho_v({\mbox{\boldmath $ r$ } } ) + g_{\omega\lambda}\psi_\lambda^\dagger({\mbox{\boldmath $ r$}})\psi_\lambda({\mbox{\boldmath $ r$}}).\ ] ] to derive these equations , we have used the time - reversal symmetry and retained only the time - like component of @xmath24 @xcite .
@xmath25 and @xmath26 are the scalar and vector densities for the nucleons , which are constructed with the spinor for the nucleons using the so called no - sea approximation , _
i.e. , _ neglecting the contribution from the antiparticles . @xmath27 and @xmath28 are the coupling constants of the nucleons to the sigma and the omega mesons , respectively , and @xmath29 and @xmath30 are the coefficients in the non - linear sigma terms in @xmath19 .
we solve these equations , together with the dirac equation for the nucleons and the klein - gordon equations for the @xmath11 meson and the photon field , iteratively until the self - consistency condition is achieved .
for this purpose , we modify the computer code rmfaxial @xcite to include the @xmath0 particle . in this code ,
the rmf equations for the nucleons are solved with the harmonic oscillator expansion method@xcite , assuming the axial symmetry .
the pairing correlation among the nucleons is also taken into account in the bcs approximation . with the self - consistent solution of the rmf equations ,
we compute the intrinsic quadrupole moment of the hypernucleus , @xmath31\,r^2y_{20}(\hat{{\mbox{\boldmath $ r$}}}).\ ] ] the quadrupole deformation parameter @xmath32 is then estimated with the intrinsic quadrupole moment as @xcite , @xmath33 where @xmath34 is the mass number of the core nucleus for the hypernucleus .
we use @xmath35 fm for the radius of the hypernucleus .
we now numerically solve the rmf equations and discuss the quadrupole deformation parameter of @xmath0 hypernuclei .
for this purpose , we use the nl3 parameter set@xcite for the rmf lagrangian for the nucleons , @xmath19 . for the @xmath0-meson
coupling constants , we follow refs .
@xcite and take @xmath36 and @xmath37 .
the value for @xmath16 was determined from the naive quark model @xcite , while the value for @xmath15 was slightly fine - tuned in order to reproduce the @xmath0 binding energy of @xmath38o @xcite . for the pairing correlation among the nucleons , we employ the constant gap approximation with the pairing gap given in ref .
@xcite , that is , @xmath39 and @xmath40 mev for the neutron and the proton pairing gaps , respectively .
( it has been known that these pairing gaps underestimate the deformation parameter of @xmath41ne nucleus when it is calculated with the nl3 parameter set@xcite .
we therefore arbitrarily switch off the pairing interaction when we calculate the @xmath41ne and @xmath42ne nuclei . )
figures 1 and 2 show the deformation parameter for the ground state of ne and si isotopes , respectively .
the dashed line is the deformation parameter for the even - even core nuclei , while the solid line is for the corresponding hypernuclei . for the ne isotopes ,
the deformation parameter is always similar between the core nucleus and the corresponding hypernucleus , although the deformation parameter for the hypernucleus is slightly smaller than that for the core nucleus .
this is consistent with the previous results with the non - relativistic skyrme - hartree - fock method @xcite . on the other hand , for the si isotopes , the deformation parameter for the @xmath43si nuclei is drastically changed when a @xmath0 particle is added , although the change for the other si isotopes is small .
that is , the @xmath43si nuclei have oblate shape in the ground state .
when a @xmath0 particle is added to these nuclei , remarkably they turn to be spherical .
ne ( the dashed line ) and @xmath44ne ( the solid line ) nuclei obtained with the constrained rmf method with the nl3 parameter set .
the energy surface for @xmath44ne is shifted by a constant amount as indicated in the inset . ]
si and @xmath45si nuclei . ]
the potential energy surfaces for the @xmath46ne and @xmath47si nuclei are shown in figs .
3 and 4 , respectively .
these are obtained with the constrained rmf method with quadrupole constraint @xcite .
the meaning of each line is the same as in figs . 1 and 2 .
in order to facilitate the comparison , we shift the energy surface for the hypernuclei by a constant amount as indicated in the inset of the figures . for the @xmath48ne nucleus
, the prolate minimum in the energy surface is relatively deep ( the energy difference between the oblate and the prolate minima is 3.04 mev , and that between the spherical and prolate configurations is 3.63 mev ) , and it is affected little by the addition of the @xmath0 particle . on the other hand ,
the energy surface for the @xmath3si nucleus shows a relatively shallow oblate minimum , with a shoulder at the spherical configuration .
the energy difference between the oblate and the spherical configurations is 0.754 mev , and could be easily inverted when a @xmath0 particle is added . in order to check the parameter set dependence of the results , we repeat the same calculation with the nlsh parameter set @xcite .
the potential energy surface for the @xmath47si nuclei obtained with the nlsh set is shown in fig .
one sees that the potential energy surface is qualitatively almost the same between the nl3 and nlsh parameter sets , although the @xmath0 binding energy is slightly different .
namely , the oblate @xmath3si nucleus becomes spherical in the presence of @xmath0 particle , again with the nlsh parameter set .
we also check the dependence of the results on the treatment of pairing correlation . for this purpose , we perform the calculations i ) without taking into account the paring correlation and ii ) with the constant force approach for the strength of the seniority pairing force . for the latter approach ,
we determine @xmath49 and @xmath50 so that they lead to @xmath39 and @xmath40 mev for the ground state of each nucleus .
( for instance , @xmath51 and @xmath52 mev for the nl3 calculation of @xmath3si nucleus . )
we confirm that our conclusion remains the same for both the treatments of the pairing correlation , due to the fact that @xmath53 or @xmath54=14 is an oblate magic number@xcite .
we therefore conclude that the @xmath0 particle significantly changes the deformation of @xmath3si nucleus , at least for the two parameter sets of the rmf lagrangian and irrespective of the treatment of pairing correlations . [ cols="^,^,^,^,^,^,^",options="header " , ] in contrast , for the @xmath55si nuclei , the dependence on the parameter set and the treatment of pairing is much stronger .
for instance , with the nlsh parameter set , the @xmath56si is slightly oblate and the deformation parameter is similar between @xmath57si and @xmath58si . without the pairing correlation , the deformation is similar between @xmath57si and @xmath58si for both nl3 and nlsh .
apparently more careful investigations will be necessary for these nuclei before we can draw a definite conclusion on their deformation parameter .
we summarize our results for @xmath43si and @xmath43si + @xmath0 in table i. c and @xmath59c nuclei . ] as another example which shows a large effect of @xmath0 particle on nuclear deformation , we next discuss the @xmath4c nucleus . for this nucleus ,
the calculation with the nl3 parameter set did not converge , due to the instability of the scalar meson field @xcite , and we here show only the results with the nlsh set . fig .
6 shows the potential energy surface obtained with the nlsh parameter set together with the constant gap approximation for the pairing correlation .
the behaviour of energy surface of @xmath4c is similar to that of @xmath3si shown in figs . 4 and 5 .
that is , the energy surface has a shallow oblate minimum and a shoulder at the spherical configuration . for this nucleus
, the energy difference between the oblate and the spherical configurations is as small as 0.13 mev . by adding a @xmath0 particle , the oblate minimum disappears and the ground state becomes spherical
this is exactly the same effect of @xmath0 particle as that in the @xmath3si nucleus .
for this light nucleus , the pairing correlation does not play an essential role , and we confirm that our conclusion remains the same even if we do not include the pairing correlation ( see table i ) .
we have used the relativistic mean field ( rmf ) theory to investigate quadrupole deformation of @xmath0 hypernuclei .
we have shown that , while an addition of @xmath0 particle does not influence much the shape of many nuclei , @xmath4c and @xmath3si make important exceptions .
that is , we have demonstrated that the @xmath0 particle makes the shape of these nuclei change from oblate to spherical . for the @xmath3si nucleus
, this conclusion was achieved both with the nl3 and nlsh parameter sets of the rmf lagrangian , although the calculation with nl3 was not converged for the @xmath4c nucleus due to the instability of sigma field .
we have also confirmed that the conclusion is independent of the treatment of pairing correlation among the nucleons .
an important next question will be how to observe experimentally the drastic structure change of the hypernuclei found in this paper .
for this purpose , a measurement of the energy of the first 4@xmath60 state , and thus a deviation from a rotational spectrum , will be extremely useful .
on the other hand , the potential energy surface for the @xmath2c and @xmath1si nuclei is somewhat soft and a large anharmonic effect of collective vibration might be expected .
one may thus need to perform _
e.g. , _ a generator coordinate method ( gcm ) calculation @xcite , on top of the mean field calculation presented in this paper , and calculate the excitation spectra before one can compare the theoretical results with experimental data .
we thank h. tamura , h. sagawa , nyein wink lwin , and khin nyan linn for useful discussions .
this work was supported by the japanese ministry of education , culture , sports , science and technology by grant - in - aid for scientific research under the program number 19740115 . | we study the deformation property of @xmath0 hypernuclei using the relativistic mean field ( rmf ) method .
we find that @xmath1si and @xmath2c hypernuclei have spherical shape as a consequence of the additional @xmath0 particle , whereas the corresponding core nuclei , @xmath3si and @xmath4c , are oblately deformed .
most of other hypernuclei have a similar deformation parameter to the core nucleus , in accordance with the previous study with the non - relativistic skyrme - hartree - fock method .
we discuss the sensitivity of our results to the choice of pairing interaction and to the parameter set of the rmf lagrangian . |
it appears that in the next decade the photon - nucleus interactions will be in the forefront of the small _ x _ qcd dynamics studies .
this is due to the possibilities of the studying coherent ( and for some channels incoherent ) photon - nucleus interactions at energies which exceed at least by a factor of 10 the energies of electron - nucleon interactions at hera .
such studies will be feasible at lhc within the program of the study of the ultra peripheral collisions ( upc ) @xcite .
this opens a challenging opportunity to get answers to a number of the fundamental questions which could be investigated in the coherent processes : how to achieve new qcd regime of strong interaction with small coupling constant , how interactions depend on the type of the projectile and how they change with an increase of the size / thickness of the target , etc .
several regimes appear possible depending on the incident energy and the target thickness . a hadronic projectile ( proton , pion , etc ) high - energy interactions with the nucleus rather rapidly approach the black body limit ( bbl ) in which the total cross section of the interaction is equal to @xmath5 ( @xmath6 ) .
another extreme limit is the interaction of small size projectiles ( or wave packages ) . in this case in a wide range of high energies the system remains almost frozen during the passage through the nucleus and the regime of color transparency is reached in which the interaction of the small size projectile with a nucleus is rather weak and proportional to a. this color transparency phenomenon in hard high energy projectile - nucleus interactions has been recently observed experimentally in the exclusive production of two jets by pions in the coherent diffraction off nuclei . at higher energies interactions of small dipole is expected to reach the regime of the perturbative color opacity . in this regime
the small size projectile still couples to the gluon field of the target via the ( skewed ) gluon density of the target , @xmath7 , like in the color transparency regime .
however , the scattering amplitude is not @xmath8 due to the leading twist ( lt ) shadowing , @xmath9 .
the onset of gluon shadowing tames somewhat the increase of @xmath10 between @xmath11 and @xmath12 and slows down the increase of the small dipole - nucleus interaction with energy .
the resulting taming is not strong enough to prevent the leading twist approximation result for the total inelastic cross section from reaching and even exceeding the bbl .
this violation of unitarity is an unambiguous signal that for sufficiently small @xmath3 the lt approximation breaks down .
an important practical issue for the studies of the small @xmath3 dynamics is whether the interactions of small dipoles with several nucleons of the target are strongly modified by the leading twist shadowing dynamics @xcite or one can neglect the lt gluon shadowing , like in the model of mclerran and venugopalan@xcite , and focus on the higher twist effects which are often modeled in the impact parameter space eikonal model @xcite .
if the leading twist shadowing was small and only higher twist effects were taming the increase of the dipole - nucleus cross section , the break down of the dglap approximation would occur at rather large @xmath3 .
the break down of dglap may result in onset of the bbl or taming of the cross sections at smaller values .
we will argue below that the relative importance of leading and higher twist interactions could be experimentally resolved using coherent onium photoproduction .
note also that if the lt gluon shadowing effects were small enough , the bbl for the interaction of @xmath13 dipoles of the size @xmath14 could be reached for central collisions with heavy nuclei already at @xmath15 that is in the kinematics where @xmath16 effects in the evolution of the parton densities are small . in any case , whatever is the limiting behavior for the interaction of the small size dipoles with heavy nuclei , it is of major theoretical interest since it represents a new regime of interactions when the leading twist approximation and therefore the whole notion of the parton distributions becomes inapplicable for the description of hard qcd processes in the small @xmath17 regime .
it is worth emphasizing that on the top of providing higher parton density targets , nuclei have another important advantage as compared to the nucleon target .
it is a weak dependence of the scattering amplitude on the impact parameter for a wide range of the impact parameters ( in fact , one can combine light and heavy nuclei to study the dependence of the amplitudes on the nuclear thickness ) . on the contrary , in the nucleon case
the scattering at large impact parameters is important even at very small @xmath3 working to mask the change of the regime of the interactions at small impact parameters .
theoretical studies of the limiting behavior of the small dipole - heavy nucleus cross sections did not lead so far to definitive results .
it is conceivable that the qcd dynamics will stop the increase of the cross section at central impact parameters at the values significantly smaller than allowed by the bbl . in the following discussion to emphasize the qualitative difference of the new regime we will use for simplicity the extreme hypothesis that the impact factor at small impact parameters reaches the value of one corresponding to the bbl pattern of the elastic and inelastic cross sections being equal .
there exist many processes where the projectile wave function is a superposition of configurations of different sizes , leading to the fluctuations of the interaction strength . in this respect ,
interactions of real and virtual photons with heavy nuclei provide unique opportunities since the photon wave function contains both the hadron - like configurations ( vector meson dominance ) and the direct photon configurations ( small @xmath13 components , heavy quark - antiquark components ) .
the important advantage of the photon is that at high energies the bbl is manifested in diffraction into a multitude of the hadronic final states ( elastic diffraction @xmath18 is negligible ) while in the hadron case only elastic diffraction survives in the bbl , and details of the dynamics responsible for this regime remain hidden .
moreover , one can _ post - select _ a small size or a large size initial state by selecting a particular final state , containing for example @xmath19 @xmath20 , or a leading light meson .
such post - selection is much easier in the photon case than in the hadron case since the distribution over the sizes of the configurations is much broader in the photon case .
spectacular manifestations of bbl in ( virtual ) photon diffraction include strong enhancement of the large mass tail of the diffractive spectrum as compared to the expectations of the triple pomeron limit , large cross section of the high @xmath21 dijet production @xcite .
we emphasize that the study of the diffractive channels will allow to distinguish between two extreme scenarios : large suppression of the cross section due to the leading twist shadowing and nonlinear regime of the bbl .
investigation of the coherent diffraction in bbl would allow to perform unique measurements of various components of the light cone wave function of the photon , providing a much more detailed information than similar measurements in the regime where leading twist dominates . in this review
we summarize our recent studies of the various regimes of the coherent photoproduction off nuclei @xcite : the onset of the bbl regime , phenomenon of color transparency and perturbative color opacity related to the leading twist nuclear gluon shadowing , and the pattern of soft qcd phenomena in the proximity to the black body limit .
in addition we consider hard leading twist diffraction off nuclei in dis .
we also outline how these effects can be studied in upc collisions and provide a comparison with the first upc data from rhic .
our consideration of the coherent photoproduction processes of light flavor and hidden charm is based on the use of the eikonal approximation that is the glauber model modified to take into account finite longitudinal momentum transfer @xcite .
the photoproduction cross section @xmath22 is given in the glauber model by the general expression @xmath23 here @xmath24 is the photon energy , @xmath25is the vector meson momentum , @xmath26 , @xmath27 is the longitudinal momentum transfer in the @xmath28 transition , and @xmath29 is the diffractive nuclear profile function .
depending on the considered vector meson production process , the glauber approach can be combined with either the traditional vector dominance model ( vdm ) ( for the detailed review see ref .
@xcite ) or with the generalized vector dominance model ( gvdm ) @xcite . more properly
this latter approximation should be called the gribov - glauber model @xcite because the space - time evolution of high energy processes is different in quantum mechanical models and in quantum field theory , and therefore theoretical foundations of the high - energy model are different . in particular , in qcd in difference from quantum mechanics a high - energy projectile interacts with all nucleons at the same impact parameter almost at the same time @xcite . due to the cancellations between diagonal and nondiagonal transitions
the gvdm allows to take into account the qcd effect of the suppression of interaction of spatially small quark - gluon wave packages with a hadron target - the so called color transparency phenomenon .
namely , cancellations occur between the amplitudes of the photon transition into a vector state @xmath30 with subsequent conversion to a state @xmath31 and direct production of the state @xmath31 , etc .
the importance of the nondiagonal transitions reveals itself in the precocious bjorken scaling for moderately small @xmath32 as due to the presence in the virtual photon of hadron - like and point - like type configurations @xcite .
these amplitudes are also crucial for ensuring a quantitative matching with perturbative qcd regime for @xmath33 few gev@xmath34 @xcite .
the amplitude of the vector meson production off a nucleon can be written within the gvdm as @xmath35 where @xmath36 are expressed through @xmath37 .
calculation of the vector meson production amplitude off nuclei within the glauber approximation requires taking into account both the nondiagonal transitions due to the transition of the photon to a different meson @xmath38 in the vertex @xmath39 and due to a change of the meson in multiple rescatterings like @xmath40 .
this physics is equivalent to inelastic shadowing phenomenon familiar from hadron - nucleus scattering @xcite .
then in the optical limit ( @xmath41 ) of the glauber multistep production theory one can introduce the eikonal functions , @xmath42 , which describe propagation of the produced objects through the medium and are related to the diffractive profile function by expression @xmath43 within the optical limit of the glauber based gvdm with accuracy @xmath44 the eikonal functions @xmath42 are determined by the solutions of the coupled channel equations @xmath45 , \label{cch}\end{aligned}\ ] ] with the initial condition @xmath46 .
the exponential factors @xmath47 $ ] are responsible for the coherent length effect , @xmath48 , @xmath49 .
the generalized glauber -based optical potentials in the short - range approximation are given by the expression @xmath50 here @xmath51 are the forward elementary amplitudes , and @xmath52 is the nuclear density normalized by the condition @xmath53 .
we calculated @xmath54 in the hartree - fock - skyrme ( hfs ) model which provided a very good(with an accuracy @xmath55 ) description of the global nuclear properties of spherical nuclei along the periodical table from carbon to uranium@xcite and the shell momentum distributions in the high energy ( p,2p)@xcite and ( e , ep)@xcite reactions .
if the nondiagonal rescattering amplitudes @xmath56 , one can easily integrate eq.[cch ] , and using the expression for the elementary amplitude @xmath57 obtain the expression for the photoproduction cross section : @xmath58 well known from early seventies ( see for example @xcite ) .
we have used the gvdm to describe coherent photoproduction of hadronic states of @xmath59 gev off nuclei and consider the onset of bbl in the soft regime . in ref .
@xcite the simplest nondiagonal model was considered with two states @xmath0 and @xmath60 .
then the gvdm comprises elementary amplitudes @xmath61 it was assumed that both @xmath0 and @xmath62 have the same diagonal amplitudes of scattering off a nucleon .
the ratio of coupling constants was fixed : @xmath63 , while the ratio of the nondiagonal and diagonal amplitudes @xmath64 , and the value @xmath65 were found from the fit to the forward @xmath66 cross sections measured at @xmath676.1 , 6.6 and 8.8 gev@xcite .
one should emphasize here that in such gvdm extension @xmath60-meson approximates the hadron production in the interval of hadron masses @xmath68 .
thus the values of the production cross section refer to the corresponding mass interval .
we refined this model in @xcite .
the dependence on the nuclear structure parameters was diminished by calculating the nuclear densities in the hartree - fock - skyrme ( hfs ) approach .
next , we used in all our calculations the parameterization of @xcite for the @xmath69 amplitude which was obtained from the fit to the experimental data on photoproduction off the proton target .
the value of @xmath70 was fixed at 0.18 to ensure the best fit of the measured differential cross section of the @xmath0-meson photoproduction off lead at @xmath71 gev and @xmath72 @xmath73 .
note that this value of @xmath74 leads to a suppression of the differential cross section of the @xmath0-photoproduction in @xmath75 by a factor of @xmath76 practically coinciding with phenomenological renormalization factor @xmath77 introduced in @xcite to achieve the best fit of the elementary @xmath0-meson photoproduction forward cross section in the vdm which neglects mixing effects . with all parameters fixed we calculated the differential cross sections of @xmath0-production off nuclei and found a good agreement with the data @xcite , see a detailed comparison in @xcite . in view of a good agreement of the model with the data on @xmath0-meson production in the low energy domain we used this model to consider the @xmath0-meson photoproduction at higher energies of photons .
the increase of the coherence length with the photon energy leads to a qualitative difference in the energy dependence of the coherent vector meson production off light and heavy nuclei and to a change of the a - dependence for the ratio of the forward @xmath60 and @xmath0-meson production cross sections between @xmath78 gev and @xmath79 gev ( fig.[rhoen ] ) .
the observed pattern reflects the difference of the coherence lengths of the @xmath0-meson and a heavier @xmath80-meson which is important for the intermediate photon energies @xmath81 gev .
the corrections due to nondiagonal transitions are relatively small ( @xmath82 15% ) for the case of @xmath0 production off a nuclei . as a result
, we find that the gvdm cross section is close to the one calculated in the vdm for heavy nuclei as well .
situation for @xmath60 production is much more interesting .
the cross section of @xmath60 production off a nucleon is strongly suppressed as compared to the case when the @xmath83 transitions are switched off .
the extra suppression factor is @xmath84 .
the non - diagonal transitions disappear in the limit of large a ( corresponding to the bbl ) due to the condition of orthogonality of hadronic wave functions in accordance with the general argument of gribov @xcite . therefore in the limit of @xmath85
we expect that the relation @xmath86 should be fulfilled for the productions of states @xmath87 of invariant masses @xmath88 at @xmath89 .
indeed we have found from calculations that in the case of the coherent photoproduction off lead the nondiagonal transitions becomes strongly suppressed with increase of the photon energy . as a result the @xmath90 ratio increases , exceeds the ratio of the @xmath91 forward cross sections calculated with accounting for @xmath92 transitions already at @xmath93 gev and becomes close to the value of @xmath94 which can be considered as the limit when one can treat the interaction with the heavy nucleus as a black one .
it is worth noting here that presence of nondiagonal transitions in terms of the formalism of the scattering eigen states @xcite corresponds to the fluctuations of the values of the interaction cross sections for the real photon .
the gvdm discussed in the paper leads to small ( @xmath95 ) color transparency effects at intermediate energies for the cross section of semiinclusive photoproduction processes .
really , this model corresponds to the propagation of states with cross sections : @xmath96 . in the case of electroproduction @xmath74
should be significantly larger : @xmath97 this equation follows from eq.([gvdm ] ) where the left hand side is put to zero because the cross section of the elastic vector meson electroproduction rapidly decreases with @xmath98 .
the presence of the ct phenomenon within the gvdm leads to a substantial modification of the pattern of the approach to bbl .
the nondiagonal transitions become more important with the increase of @xmath98 , leading to an enhancement of the effects discussed above . in particular
, the fluctuations of strengths of interaction would lead at large @xmath98 to the color transparency phenomenon .
presence in gvdm of significantly different masses of @xmath99 makes it impossible to describe all fluctuations of strengths of interaction in terms of one coherent length .
there is a qualitative difference between gvdm description of the light meson production described above and production of charmonium states . in the case of the photoproduction of light mesons the soft physics dominates both at low and high energies
hence , the gribov - glauber model should be applicable in a wide range of energies . at the same time the space - time picture of the process changes with energy . at low energies
the meson is formed at the distances smaller than the typical interaction length of a meson in the nucleus . with the increase of energy
the formation length starts to exceed the nuclear size and a photon converts to a system of @xmath13 and higher fock - components before the target , and one has to account for the interactions of this system with the media and subsequent transition of the system to a vector meson . the situation is much more involved in the case of ( charm)onium production . at high energies
both the coherence length @xmath101 and the formation length @xmath102}^{-1}$ ] ( distance on which the squeezed @xmath13 pair transforms into the ordinary meson ) are large and the color transparency phenomenon reveals itself .
in particular , it explains the fast increase of the cross section with energy observed at hera ( for reviews of small x phenomena see ref .
consequently , the value of the cross section extracted from the charmonium photoproduction characterizes the interaction of the squeezed @xmath19 pair with a nucleon rather than the charmonium - nucleon interaction . in the high energy limit the very small interquark distances in the wave function of the photon dominate , and
one has to treat the interaction of small dipoles with nuclei . in this case
the eikonal approximation gives a qualitatively wrong answer since it does not take into account the leading twist effect of the gluon shadowing(see the detailed analysis in @xcite ) , while the leading twist analysis predicts large shadowing effects @xcite , see discussion in section 3 . on the other hand , at the intermediate energies when onium states are formed inside the nucleus the nonperturbative effects at a transverse distance scale comparable to the charmonium size becomes important .
the hadronic basis description would be more relevant in this case .
however the vdm which takes into account only diagonal vector meson transitions does not account properly for the basic qcd dynamics of interaction .
in particular , slac data @xcite show that @xmath103 which within vdm corresponds to : @xmath104 .
this conclusion is in evident contradiction with the qcd expectation that the hadron interaction cross section should be scaled approximately as the transverse area occupied by color : @xmath105 .
thus in this naive qcd picture @xmath106 .
a qcd explanation of such failure of vdm is based on the observation @xcite that in photoproduction of both the @xmath107and @xmath108mesons the small relative distances @xmath109 dominate in the @xmath19 component of the photon wave function .
therefore the cross sections which enter into the ratio of @xmath107and @xmath108yields are cross sections of the interaction of the small dipoles , not genuine mesons .
the suppression of the production of @xmath108 in this picture is primarily due to a smaller leptonic decay width of @xmath108(a factor of 1/3 ) .
a significant additional suppression comes from importance of more massive @xmath19 intermediate states in the photon wave function ( @xmath110 respectively ) .
the exact value of the suppression is sensitive to the details of the onium wave functions and the dependence of the dipole - nucleon cross section on the size of the dipole @xcite .
the dominance of small @xmath19 configurations in photoproduction processes is relevant for the significant probability of nondiagonal @xmath111 diffractive transitions .
the gvdm adjusted to account for the color screening phenomenon @xcite allows to take into account qcd dynamics using a hadronic basis .
such a description is limited to the regime of small coherence lengths ( medium energies ) , where leading twist shadowing is not important .
we use the gvdm to consider the coherent photoproduction of hidden charm mesons off nuclei at moderate photon energies @xmath112 where the coherence length for the @xmath113 transition @xmath114 is still close enough to the internucleon distance in nuclei while the formation length @xmath115 is comparable to the radii of heavy nuclei . in this energy range
produced charmonium states have a noticeable probability to rescatter in sufficiently heavy nuclei .
therefore , one would be able to reveal the fluctuation of the charmonium - nucleon interaction strength as due to the diagonal @xmath116 and nondiagonal @xmath117 rescatterings for moderate energies .
the gvdm which we outlined above takes into account the coherence length effects via the glauber model approximation .
the key distinction from the @xmath118meson case is choosing the parameters of the model to account for the space - time evolution of spatially small @xmath19 pair .
it is important that the inelastic shadowing corrections related to the production of higher mass states @xcite are still insignificant .
a reasonable starting approximation to evaluate the amplitude of the charmonium - nucleon interaction is to restrict ourselves to the basis of @xmath107and @xmath108states for the photon wave function in eq.[gvdm ] .
then , similarly to the considered above case of @xmath119 production we have two equations comprising six elementary amplitudes .
the charmonium - nucleon coupling constants : @xmath120 , and @xmath121 are determined from the widths of the vector meson decays @xmath122 . since in
the photoproduction processes @xmath19 pair is produced within the spatially small configuration one can neglect for a moment by the direct photoproduction amplitude and obtain the approximative relations between rescattering amplitudes @xmath123 large values for nondiagonal amplitudes @xmath124 are a characteristic qcd property of hidden charm and beauty meson - nucleon interaction .
note that the negative sign of the nondiagonal amplitude is dictated by the qcd factorization theorem . a positive sign of the forward photoproduction @xmath125 amplitudes as well as the signs of the coupling constants @xmath126 and @xmath127 are determined by the signs of the charmonium wave functions at r=0 . from the approximative estimates above
it also follows that @xmath128 .
this is much larger than @xmath129 20 mb suggested by analyses of the data on @xmath130 absorption in nucleus - nucleus collisions @xcite .
when combined with the slac data @xcite , this corresponds to @xmath131 with large experimental and theoretical errors . to fix elementary amplitudes more accurately within gvdm ( see @xcite for details )
we parameterized the elementary photoproduction cross section in the form used by the experimentalists of hera to describe their data .
this form has no firm theoretical justification but it is convenient for the fit @xmath132 here @xmath133 is the invariant energy for the photon scattering off a free nucleon , @xmath134 is the reference point and @xmath135 is the two - gluon form factor of a nucleon .
we also used the slac results for the forward @xmath136 cross section @xcite and for the ratio @xmath137 besides , we fixed the @xmath138 cross section @xmath139 mb as measured at slac@xcite and used the reasonable assumption about the energy dependence of this cross section as the sum of soft and hard physics : @xmath140 the existence of a hard part of the @xmath141n cross section is consistent with the gvdm , because the photoproduction amplitude has a stronger energy dependence than the pomeron exchange , i.e. soft scattering amplitudes . next , we used this input , the optical theorem and the well known gribov - migdal relation @xmath142 to determine from eq.[gvdm ] all elementary amplitudes in the discussed energy range .
in particular , we found the value @xmath143 mb .
however , the input parameters of the gvdm , namely the experimental cross sections of the forward elementary photoproduction and , especially , the value of @xmath144 , are known with large uncertainties . in result
, we obtained elementary cross sections changing in the ranges : @xmath145 and @xmath146 , in the discussed energy range .
we checked how a variation of @xmath144 within the experimental errors influences our results .
= .9 = 1 .
in our papers@xcite,@xcite we analyzed the coherent photoproduction of charmonia off light(si ) and heavy(pb ) nuclei .
here we present the cross sections of the coherent photoproduction of @xmath141 and @xmath130 off ca .
the energy dependence of these cross sections is compared ( fig.[jpsipsi]a ) to that obtained in the impulse approximation where all rescatterings of the produced vector mesons are neglected , and the cross section is given by the simple formula @xmath147 the distinctive feature of the coherent charmonium photoproduction is that differential cross sections oscillate as a function of the photon energy .
the major source for such a behavior at intermediate energies is oscillating behavior of the longitudinal nuclear form factor at the relatively large values of @xmath148 in the photoproduction vertex .
the cross section of the @xmath107photoproduction in the gvdm is close to that calculated in the impulse approximation - the shapes of curves are very similar and the values of the cross sections are only slightly reduced at energies below 40 gev(fig.[jpsipsi]b ) .
this indicates that it would be very difficult to extract the total @xmath138 cross section from such a measurement . at the same time , since the the nuclear form factor is known with a high precision from the high energy elastic electron - nucleus scattering , one can use the coherent @xmath107photoproduction off the spherical nuclei to determine the elementary @xmath149 amplitude in a wide range of energies for @xmath150 .
the picture is qualitatively different for the coherent @xmath108photoproduction off nuclei . due to the higher threshold ( @xmath151 gev , the direct @xmath108production off nuclei is suppressed in the impulse approximation ( for the same incident energy ) as compared to the @xmath107production by the nuclear form factor . since @xmath152
, the minima of the impulse approximation distribution are shifted .
the contribution of the nondiagonal term @xmath153 essentially increases the @xmath108yield and produces an additional shift of the minima in the spectrum to lower photon energies ( fig .
[ jpsipsi]c ) .
this results in significant effects , especially , if one analyses the data as a function of the mass number a at different energies to extract a - dependence of the @xmath138 cross sections .
it is seen from fig.[adpsi ] that the oscillating form factor significantly influences the a - dependence that complicates extracting of the genuine @xmath138 and @xmath154 cross section from such analysis .
the oscillation of the cross section with energy can be used to measure in a new way the elementary charmonium photoproduction amplitudes as well as the charmonium - nucleon amplitudes using the coherent charmonium photoproduction off light nuclei@xcite . in particular , at the photon energies @xmath155 ( fig .
[ jpsipsi]d ) , the main contribution to the cross section originates from the nondiagonal rescattering . as a result
, one can extract from the data the nondiagonal elementary @xmath156 amplitude by measuring the ratio of the @xmath130 and @xmath141 yields at zero production angle , because other inputs to this ratio such as the elementary @xmath149 amplitude are fixed from the @xmath141 production .
moreover , since other parameters enter both in the numerator and denominator , the major uncertainties are canceled out .
on the other side , the energy dependence of this ratio should originate primarily due to the contribution of the direct @xmath130 production .
thus , one would be able to determine the @xmath157 amplitude from the measurement of the @xmath130-to-@xmath141 ratio .
we want to emphasize that the suggested procedure for extracting the nondiagonal amplitude and amplitudes of direct @xmath141 and @xmath130 photoproduction from the nuclear measurements is practically model independent .
it is much more difficult to determine the diagonal @xmath158 and @xmath159 amplitudes , which are relevant for the suppression of the charmonium yield in heavy ion collisions . a naive diagonal vdm with @xmath160 based on the slac data @xcite leads to a rather significant suppression of the @xmath141 yield : @xmath161 for light nuclei , and @xmath162 for heavy nuclei .
however , we find @xcite a strong compensation of the suppression due to the contribution of the nondiagonal transitions . as a result
we find that overall the suppression does not exceed @xmath163 for all nuclei along the periodical table .
hence , an extraction of the diagonal amplitudes from the measured cross sections would require a comparison of high precision data with very accurate theoretical calculations including the nondiagonal transitions .
another strategy is possible if the elementary @xmath164 , and @xmath157 photoproduction amplitudes , as well as the nondiagonal amplitude @xmath165 would be reliably determined from the medium energy data . in this case
it would be possible to determine the imaginary parts of the forward diagonal amplitudes from the gvdm equations ( [ gvdm ] ) @xmath166 the suggested procedure allows to determine all elementary amplitudes with a reasonable precision from the measurements of @xmath141 and @xmath130 photoproduction off the light nucleus at the medium photon energies .
the measurement of the coherent charmonium photoproduction off nuclei in this energy region are planned at slac @xcite .
the measurements of coherent production off heavy nuclei and of the quasielastic production with parameters of gvdm fixed from the analysis of light nuclei would provide a critical test of the model .
the main limitation of the suggested procedure is the restriction of the hadronic basis to the two lowest 1s , 2s charmonium states with the photon quantum numbers : @xmath141 , @xmath130 .
this is one of the key approximations in the discussed approach .
it seems quite reasonable since in the coherent production of vector meson states off nuclei at moderate energies contribution of high mass states is more suppressed by the target form factor .
however , we neglected by @xmath167 which is nearly degenerate in mass with @xmath130 : @xmath168 mev .
the properties of these two states are described well in the charmonium model @xcite . in this model
@xmath167-meson is described as a @xmath169 state with a small admixture of the @xmath170 wave , while @xmath130-meson has a small @xmath171-wave admixture .
namely @xmath172 since only the s - wave contributes to the decay of @xmath173 states into @xmath174 ( at least in the non - relativistic charmonium models ) the value @xmath175 can be determined from the data on the @xmath174 decay widths @xmath176 kev and the @xmath177 kev . due to the small difference of masses between the @xmath178 and @xmath179 mesons the produced s - wave @xmath180-state does not loose coherence while going through the media at any conceivable energies .
the soft interactions can not transform the @xmath170-state to @xmath171-state with any significant probability . in the soft qcd processes data
show that the cross sections of exclusive nondiagonal transitions are negligible for the forward angle scattering .
the same conclusion is valid in the pqcd model for the charm dipole - nucleon interactions .
thus , it is more appropriate to use the @xmath181 - @xmath182 basis for description of the propagation of @xmath19 through the nucleus .
the only resulting change is an increase of @xmath183 by a factor of @xmath184 as compared to @xmath127 which is within the uncertainties of the model .
since the @xmath182 state ultimately transforms into @xmath130 and @xmath185 we predict @xcite : ratio would be a universal number .
since @xmath185 can be easily observed via its characteristic @xmath186 decays in a number of the lepton , hadron and nucleus induced processes the study of the discussed process can provide a new way to probing dynamics of charmonium production in various reactions . ]
@xmath187 influence of the higher mass resonances is expected to be even weaker - the constants @xmath188 relevant for the transition of a photon to a charmonium state @xmath189 rapidly decrease with the resonance mass .
this is because the radius of a bound state , @xmath190 , is increasing with the mass of the resonance and therefore the probability of the small size configuration being @xmath191 is decreasing with an increase of mass ( for fixed s , l ) .
besides , the asymptotic freedom in qcd dictates decreasing of the coupling constant relevant for the behavior of the charmonium wave function at small relative distances .
experimentally one finds from the data on the leptonic decay widths that @xmath188 drops by a large factor with increasing mass .
an additional suppression arises due to the weakening of the soft exclusive nondiagonal @xmath192 amplitudes between states with the different number of nodes .
hence , it seems possible to determine the imaginary parts of diagonal rescattering amplitudes from analysis of the data using the gvdm equations .
since in the medium energy domain the energy dependence of soft rescattering amplitudes is well reproduced by a factor @xmath193 , one can determine the real parts of the amplitudes using the gribov - migdal relation ( eq .
[ gm ] ) .
the above analyses demonstrate that the relative importance of the non - diagonal transitions increases an increase of the quark mass .
it would be of interest to investigate experimentally the case of strangeness production .
it is likely to correspond to @xmath70 substantially larger than @xmath194 but much smaller than @xmath195 .
interaction of small size color singlet objects with hadrons is one of the most actively studied issues in high - energy qcd .
the qcd factorization theorem for exclusive meson electroproduction at large @xmath98 , and @xmath196 photoproduction @xcite allows to evaluate the amplitude of the production of a vector meson by a longitudinally polarized photon @xmath197 through the convolution of the wave function of the meson at the zero transverse separation , hard interaction block and the generalized ( skewed ) parton density .
production @xcite in the bfkl approximation and for the pion diffraction into two jets @xcite in the leading log @xmath98 approximation @xcite . ]
the lt approximation differs strongly from the expectations based on glauber model approaches and on the two gluon exchange models , because it accounts for the dominance of the electroproduction of spatially small quark - gluon wave package and its space - time evolution which leads to formation of a softer gluon field .
the latter effect results effectively in an increase of the size of the dipole with increase of the energy
. in perturbative qcd ( similar to qed ) the total cross section of the interaction of small systems with hadrons is proportional to the area occupied by color within projectile hadron @xcite leading to the expectation of the color transparency phenomenon for various hard processes with nuclei .
the cross sections of incoherent processes are expected to be proportional to the number of nucleons in the nuclei , while the coherent amplitude is proportional to number of nucleons times the nuclear form factor .
possibility to approximate projectile heavy quarkonium as colorless dipole of heavy quarks can be formally derived from qcd within the limit when mass of heavy quark @xmath198 but @xmath199 is fixed and not extremely small @xcite . in this
kinematics the size of heavy quarkonium is sufficiently small to justify applicability of pqcd . for practical purposes
the crucial question is at what @xmath98 squeezing becomes effective .
probably the most sensitive indicator is the @xmath200-dependence of the meson production .
the current hera data are consistent with the prediction of @xcite that the slopes of the @xmath0 and @xmath141 production amplitudes should converge to the same value .
this indicates that at small @xmath17 configurations much smaller than average configurations in light mesons ( @xmath201 ) dominate for @xmath0-meson production at @xmath202 gev@xmath34 , while the @xmath141 production is dominated by interaction in small size configurations for all @xmath98 .
therefore , one expects the regime of color transparency for @xmath203 where the gluon shadowing is very small / absent .
recently the color transparency ( ct ) phenomenon was observed at fnal by e791 experiment @xcite which studied the coherent process of dissociation of a 500 gev pion into two jets off the nuclei .
the measurement has confirmed a number of predictions of @xcite including the a - dependence , and the transverse and longitudinal momentum distributions of the jets .
previously the color transparency type behavior of the cross section was observed also in the coherent @xmath204 photoproduction at @xmath205 @xcite .
a natural question is whether the color transparency will hold for arbitrary high energies ?
two phenomena are expected to work against ct at high energies .
one is the lt gluon shadowing .
there are theoretical expectations ( see discussion below ) supported to some extent by the current analyses of the data on dis scattering off nuclei ( which do not extend deep enough into the shadowing region ) that the gluon distributions are shadowed in nuclei as compared to the nucleon : @xmath206 .
this obviously should lead to a gradual but calculable in qcd disappearance of color transparency @xcite , and to onset of a new regime , which we refer to as _ the color opacity regime _
( one can think of this regime also as a regime of generalized color transparency since a small @xmath13 dipole still couples to the gluon field of the target through a two gluon attachment and the amplitude is proportional to the generalized gluon density of the nucleus ) .
another mechanism for the violation of ct at high energies is the increase of the small dipole - nucleon cross section with energy @xmath207 . for sufficiently large energies
this cross section becomes comparable to the meson - nucleon cross sections .
one may expect that this would result in a significant suppression of the hard exclusive diffractive processes as compared to the lt approximation .
however it seems that this phenomenon is beyond the kinematics achievable for the photoproduction of @xmath141-mesons in upc of heavy ions at rhic ( @xmath208 ) but could be important at heavy ion upc at lhc . hence , a systematic study of the onium production in the coherent scattering off nuclei at collider energies will be very interesting .
one should emphasize here that with decrease of the size of the onium the eikonal ( higher twist ) contributions die out quickly ( provided @xmath3 is kept fixed ) .
in particular , for the @xmath209 case one probes nuclear gluon fields at the transverse scale of the order of 0.1 fm or @xmath210 . the @xmath141 case is closer to the border line between the perturbative and nonperturbative domains . as a result
the a nonperturbative region appears to give a significant contribution to the production amplitude @xcite .
= .50 let us discuss the photoproduction amplitude @xmath211 in more details .
we are interested here in the @xmath212 range which can be probed at lhc .
this region corresponds to rather small values of @xmath17 . in this situation interaction of @xmath13 pair , which in the final state forms a quarkonium state is still rather far from the bbl .
hence the key problem in the theoretical treatment of the process is taking into account the nuclear shadowing .
a number of mechanisms of coherent interactions with several were suggested for this process .
we focus here on the * the leading twist mechanism of shadowing * which is presented by diagrams in fig.[ltdiag ] .
there exists qualitative difference between the mechanism of interaction of a small dipole with several nucleons and the case of a similar interaction of an ordinary hadron .
let us for example consider interaction with two nucleons .
the leading twist contribution is described by the diagrams where two gluons are attached to the dipole . to ensure that nucleus remains intact in such a process
, the color singlet lines should be attached to both nucleons .
these diagrams ( especially the one of fig .
[ ltdiag]b ) are closely related to the diagrams describing the gluon diffractive parton densities ( which are measured at hera ) , and hence to the similar diagrams for the gluon nuclear shadowing @xcite .
the amplitude of high energy coherent heavy onium photoproduction is proportional to the generalized gluon density of the target , @xmath213 , which depends on the light - cone fractions @xmath214 and @xmath215 of gluons attached to the quark loop .
they satisfy a relation : @xmath216 if the quark fermi motion and binding effects were negligible , @xmath217 .
the resolution scale @xmath218 , where @xmath219 is the mass of the heavy quark .
numerical estimates for the photoproduction of @xmath141 give @xmath220@xcite reflecting a relatively small mass of c - quark and indicating that this process is on the verge between nonperturbative and perturbative regimes . on the contrary
, the mass of the beauty quark is huge on the scale of soft qcd . in this case
hard physics dominates , attachments of more than two gluons to @xmath20 are negligible and the qcd factorization theorem provides a reliable tool for description of the @xmath4 production .
this is especially true for the ratio of the cross sections of @xmath4 production from different targets since the higher twist effects due to the overlapping integral of the @xmath20 component of the photon and @xmath4 cancelled out in the ratio . as a result ,
in the leading twist shadowing approximation the cross section of the process @xmath221 is proportional to the squared nuclear gluon density distribution and can be written in the form @xmath222 ^ 2 \int \limits_{-\infty}^{t_{min } } dt { \left| \int d^2bdz e^{i{\vec q_t}\cdot { \vec b } } e^{-q_{l}z}\rho ( { \vec b},z ) \right| } ^2 . \label{phocs}\ ] ] numerical estimates using realistic potential model wave functions indicate that for @xmath141 , @xmath223 @xcite , and for @xmath4 , @xmath224 @xcite .
modeling of the generalized parton distributions at moderate @xmath98 suggests that to a good approximation @xmath225 can be approximated by the gluon density at @xmath226 @xcite . for large @xmath98 and small @xmath3 gpds
are dominated by the evolution from @xmath227 . since the evolution conserves @xmath228 , effect of skewedness
is determined primarily by the evolution from nearly diagonal distributions , see ref .
@xcite and references therein . in the case of the @xmath209 production
it increases the cross section by a factor @xmath229 @xcite and , potentially , this could even obscure the connection of the discussed effect with the shadowing of the nuclear gluon densities .
however , the analysis of @xcite shows that the ratio of gpd on a nucleus and on a nucleon at @xmath230 is a weak function of @xmath215 , slowly dropping from its diagonal value ( @xmath231 ) with the decrease of @xmath215 .
overall this observation is in a agreement with the general trend mentioned above that it is more appropriate to do comparison of diagonal and non - diagonal cases at @xmath226 . in the case of double scattering contribution
[ ltdiag]b ) there is another way to address the question of the accuracy of the substitution of the ratio of generalized gluon densities by the ratio of diagonal parton densities at the normalization scale .
we notice that the amplitude corresponding to fig .
[ ltdiag]b is expressed through the nondiagonal matrix element of the diffractive distribution function , @xmath232 , which is an analog of generalized parton distribution . in the diagonal limit of @xmath233
it coincides with the diffractive gluon distribution .
it depends on the light - cone fraction which nucleon lost in @xmath234 and @xmath235 states : @xmath236 @xmath237 respectively , @xmath238 , and @xmath239 .
if we make a natural assumption that @xmath240 we find that numerically in the kinematics we discuss the resulting skewedness effects are small as compared to the uncertainties in the input gluon diagonal diffractive pdfs . hence in the following we will approximate the ratio of generalized gluon densities in the nucleus and nucleon by the ratio of the gluon densities in nucleus and nucleon at @xmath241 . in the case of @xmath4 use of @xmath242
maybe more appropriate .
this would lead to slightly larger shadowing effect .
it was demonstrated in @xcite that one can express the quark and gluon nuclear shadowing for the interaction with two nucleons in a model independent way through the corresponding diffractive parton densities using the gribov theory of inelastic shadowing @xcite and the qcd factorization theorem for the hard diffraction @xcite .
an important discovery of hera is that hard diffraction is indeed dominated by the leading twist contribution and gluons play a very important role in the diffraction(this is loosely referred to as gluon dominance of the pomeron ) .
analysis of the hera diffractive data indicates that in the gluon induced processes probability of the diffraction is much larger than in the quark induced processes @xcite .
the recent h1 data on diffractive dijet production @xcite provide an additional confirmation of this observation .
large probability of diffraction in the gluon induced hard processes could be understood in the s - channel language as formation of color octet dipoles of rather large sizes which can diffractively scatter with a quite large cross section .
the strength of this interaction can be quantified using optical theorem and introducing @xmath243 for the hard process of scattering of a virtual photon off the gluon field of the nucleon . here
@xmath244 is the ratio of the real to imaginary parts of the elementary diffractive amplitude , @xmath98 is the momentum scale determining virtuality of the gluons , @xmath245 is the momentum fraction of the pomeron with the corresponding cut - off scale @xmath246 , and @xmath247 is the diffractive gluon density distribution of nucleon which is known from the h1 analysis of the diffractive data at the scale @xmath248 gev@xmath34 .
an important feature of this mechanism of coherent interaction is that it is practically absent for @xmath249 and may rather quickly become important with decrease of @xmath3 .
= .80 we calculated the ratio of the gluon density distributions in eq .
[ phocs ] using the leading twist shadowing model @xcite ( for the details of calculations see ref.@xcite ) . as the first step the nuclear gluon density distribution with account of the leading twist shadowing
is calculated at the starting evolution scale @xmath250 gev@xmath34 @xmath251 . \label{ga}\end{aligned}\ ] ] as an input we used the h1 parameterization of @xmath252 .
the effective cross section @xmath253 accounts for the elastic rescattering of the produced diffractive state off the nuclear nucleon and is determined by eq .
[ sigef ] .
numerically it is very large at the starting scale of the evolution , see fig .
[ figsigef ] , and corresponds to the probability of the gluon induced diffraction close to 50% .
this indicates that at the initial scale of the evolution interactions in the gluon sector are close to the bbl for @xmath254 for the nucleon case , and even more so for the nuclei , where similar regime should hold for a larger range of the impact parameters .
note that the double scattering term in eq .
[ ga ] for the nuclear parton densities satisfies qcd evolution , but the higher order terms do not . that is
if we use different starting scale of evolution we would obtain different results of @xmath255 .
the reason is that the terms @xmath256 are sensitive to degree of fluctuations in the cross sections of interaction of diffracting states .
these fluctuations increase with increase of @xmath98 .
this effect is automatically included in the qcd evolution , and it leads to violation of the glauber - like structure of the expression for the shadowing at @xmath257 .
approximation for the @xmath258 rescattering of eq .
[ ga ] corresponds to an assumption that fluctuations are small at @xmath259 scale since this scale is close enough to the scale of the soft interactions , see discussion in @xcite .
thus , at the second step of the calculation we use nlo qcd evolution equations to calculate the shadowing at larger @xmath98 using the calculation at @xmath260 as a boundary condition . in this way we also take into account the contribution of the gluon enhancement at @xmath261 which influences the shadowing at larger @xmath98 .
note here that proximity to the bbl in the gluon sector which is reflected in a large value of @xmath262 may result in corrections to the lt evolution which require further studies .
= 0.8 first we calculate the ratios of the cross sections of coherent photoproduction of @xmath141 and @xmath4 off nuclei and nucleon ( fig .
[ fshad ] ) .
such ratios do not depend on the uncertainties of the elementary cross sections and provide a sensitive test of the role of the lt shadowing effects . in a case of the @xmath141 photoproduction
the gluon virtuality scale is 3 - 4 gev@xmath34 with a significant fraction of the amplitude due to smaller virtualities @xcite .
hence we will take the gluon shadowing in the leading twist at @xmath263 .
taking a smaller value of @xmath98 would result in even larger shadowing effect . in calculations for the @xmath4
we take @xmath264 , though the result is not very sensitive to precise value of @xmath265 since the scaling violation for the gluon shadowing for these @xmath98 is rather small .
we find that in spite of a small size of @xmath4 , which essentially precludes higher twist shadowing effects up to very small x , the perturbative color opacity effect is quite appreciable .
it is worth noting that the effective cross section of the rescattering in the eikonal model is determined by the cross section of dipole -nucleon interactions at the distances @xmath266 in the @xmath141 case and @xmath267 in the @xmath4 case .
these cross sections are @xmath268 for @xmath141 and @xmath269 for @xmath4 case for @xmath12 and a factor of 1.5 - 2 smaller for @xmath270 , see fig . 13 in ref .
they are much smaller than the cross sections which enter into calculation of the gluon shadowing in the lt mechanism .
we also estimated the absolute cross section of onium photoproduction for a wide range of the photon energies .
this is of interest for the planned measurements of the onium production in the ultraperipheral heavy ion collisions at lhc .
the dependence of the momentum - integrated cross sections on the energy @xmath271 is presented in fig .
[ onics ] . in the case of @xmath141 production
the calculations are pretty straightforward as the accurate data are available from hera .
the situation is more complicated in the case of the photoproduction of @xmath4 .
so far the information about the elementary @xmath272 cross section is very limited .
there is the only zeus and h1 data on total cross section for the average energy of @xmath273 .
thus , to calculate the forward photoproduction cross section we used a simple parameterization : @xmath274 where @xmath275 gev@xmath34 , the slope parameter @xmath276 is fixed basing on the analysis of the two gluon form factor in ref .
@xcite , and the energy dependence follows from the calculations of ref .
@xcite of the photoproduction of @xmath4 in the leading @xmath277 approximation with an account for the skewedness of the partonic density distributions .
this elementary cross section is normalized so that the total cross section is in @xmath278 .
it is well known that inelastic diffraction at small @xmath200 gives information on the fluctuations of strength in the projectile - target interactions @xcite .
application of this logic to the hadron scattering off nuclei have allowed to explain the a - dependence of two measured diffractive channels in @xmath279 scattering assuming that it coincides with the a - dependence of the total cross section of the inelastic diffraction , and the absolute total cross section of pa diffraction ( the data exist for two nuclei only ) at the energies 200 - 400 gev , see @xcite for the review and references . with an increase of energy , the total cross section of @xmath280 interaction increases and fluctuations in the elementary amplitude lead to much smaller fluctuations of the absorption in the scattering off a heavy nuclei . as a result
, one can expect much weaker a - dependence of the diffractive cross section @xcite , in particular , @xmath281 at lhc energies @xcite as compared to @xmath282 at fixed target energies . for large produced masses we can also understand this suppression using the @xmath200 channel picture of the pomeron exchanges as due to the stronger screening of the triple pomeron exchange , for review and references see @xcite .
diffraction in deep inelastic scattering corresponds to the transition of the ( virtual ) photon into its hadronic components leaving the nucleus intact .
hence it is similar to elastic hadron - nucleus scattering rather than inelastic diffractive hadron - nucleus scattering .
it was demonstrated recently that in the upc collisions at lhc it would be possible to study nuclear parton densities using hard charm and beauty production in @xmath283 interactions @xcite .
naturally one can also use these and similar processes to measure diffractive parton densities of nuclei . since these quantities in the leading twist
satisfy the factorization theorem we can analyze them on the basis of the analysis of the diffraction in dis .
there is a deep connection between shadowing and phenomenon of diffractive scattering off nuclei .
the simplest way to investigate this connection is to apply the agk cutting rules @xcite .
several processes contribute to diffraction on nuclei : ( i ) coherent diffraction in which the nucleus remains intact , ( ii ) break - up of the nucleus without production of hadrons in the nucleus fragmentation region , ( iii ) rapidity gap events with hadron production in the nucleus fragmentation region . in ref .
@xcite we found that for @xmath284 gev@xmath34 , the fraction of the dis events with rapidity gaps reaches the value of about 30 - 40% for heavy nuclei , with a fraction of the events of type ( iii ) rapidly dropping with @xmath285 .
we can use the information on @xmath262 for quarks and gluons to estimate probability of diffraction for different hard triggers at the resolution scale @xmath286 .
first we consider the dependence of the fraction of the events due to coherent diffraction and due to the break - up of the nucleus on the strength of the interaction , @xmath287 , neglecting fluctuations of the interaction strength .
we find that for the realistic values of @xmath287 the probability of coherent diffraction is quite large but increases with @xmath262 very slowly and does not reach the asymptotic value of @xmath288 even for very large values of @xmath289 ( the later feature reflects presence of a significant diffuse edge even in heavy nuclei ) , see fig .
[ figcsratio ] .
thus , it is not sensitive to the fluctuations of @xmath262
. we also found that the ratio of diffraction with the nucleus break - up and with the nucleus remaining intact is small ( 10 - 20% ) in a wide range of nuclei , and slowly increasing with increase of @xmath262 , see fig .
[ figcsratio ] .
hence it would require high precision measurements to constrain the dynamics using @xmath290 ratios . comparing the values of the fraction of the diffractive events for quark and gluon induced processes off heavy nuclei and proton
, we find that the relative importance of the quark induced events is increasing .
therefore , the scaling violation at large @xmath291 for the diffractive quark distribution in nuclei will be stronger for nuclei than for a proton .
another interesting effect is that for heavy nuclei only genuine elastic components can be produced ( inelastic diffraction is zero ) .
hence , the soft contribution at @xmath260 due to triple pomeron exchange is strongly suppressed see e.g. @xcite . as a result ,
nuclear diffractive parton distributions at small @xmath291 are strongly suppressed ( by a factor @xmath292 ) at @xmath260 though this suppression will be less pronounced at large @xmath98 due to the qcd evolution .
this will lead to breakdown of the universality of the @xmath291 distributions as a function of a. though the diffractive parton densities change rather slowly with @xmath98 leading to a weak variation of the diffractive cross sections with @xmath98 ( modulus the scaling factor ) the fraction of the diffractive events at fixed @xmath3 should significantly drop with increase of @xmath98 due to a large increase of the inclusive nucleon parton densities and decrease of the nuclear shadowing .
for example , let us consider ultraperipheral collisions ( upc ) at lhc where one can measure a the process @xmath293 in the kinematics where direct photon process @xmath294 dominates . in this case
if we consider the process at say @xmath295 corresponding to @xmath296 the fraction of diffractive events will be of the order 10% .
the background from the strong interaction originates from glancing collisions in which two nucleons interact via a double diffractive process @xmath297 where @xmath298 contains jets .
probability of the hard processes with two gaps is very small at collider energies - even smaller than .the probability of the single diffractive hard processes , see e.g. @xcite .
therefore , we expect that the background conditions will be at least as good in the diffractive case as in the inclusive case considered in @xcite
. thus , it would be pretty straightforward to extract coherent diffraction by simply using anti - coincidence with the forward neutron detector , especially in the case of heavy nuclei , see discussion in @xcite . as a result it would be possible to measure in the upc the nuclear diffractive parton distributions with a high statistical accuracy .
it is important that in difference from the diffraction to a vector meson it would be possible to determine on the event by event basis the energy of the photon which induced the reaction , since the rapidity of the photon is close to the rapidity of two jets . as a result it would be possible to perform the measurements for large rapidities ( selecting the events generated by a photon of higher of two energies allowed by the kinematics of production particles in the interval of rapidities @xmath299 ) and to determine diffractive parton densities for pretty small @xmath3 .
one of the striking features of the bbl is the suppression of nondiagonal transitions in the photon interaction with heavy nuclei @xcite .
indeed , in the bbl the dominant contribution to the coherent diffraction originates from `` a shadow '' of the fully absorptive interactions at impact parameters @xmath300 , so the orthogonality argument is applicable .
we use it to derive the bbl expression for the differential cross section of the production of the invariant mass @xmath301 for scattering of ( virtual ) photons @xcite . for the real photon case : @xmath302 here @xmath303 .
comparison of the measured cross section of the diffractive production of states with certain masses with the bbl result ( eq . [ ccsb ] ) would allow to determine up to what masses in the photon wave function interaction remains black .
similar equation is valid in the bbl for the production of specific hadronic or quark - gluonic final states ( @xmath304 etc ) in the case of the coherent nuclear recoil .
this allows to measure any component of the light cone photon wave function which interacts with the bbl strength corresponding final states in the coherent processes .
the onset of bbl limit for hard processes should reveal itself also in a faster increase with energy of cross sections of photoproduction of excited states as compared to the cross section for the ground state meson
. it would be especially advantageous for these studies to use a set of nuclei - one in the medium range , like @xmath305 , and another with @xmath306 .
this would allow to remove the edge effects and use the length of about 10 fm of nuclear matter .
one especially interesting channel is exclusive diffractive dijet production by real photons .
one may expect that for the @xmath307 energies which will be available at eic or at lhc in upc the bbl in the scattering off heavy nuclei would be a good approximation for the masses @xmath308 in the photon wave function up to few gev .
this is the domain which is described by perturbative qcd for @xmath309 in the case of the proton targets , and larger @xmath3 for scattering off nuclei .
the condition of large longitudinal distances , a small longitudinal momentum transfer , will be applicable in this case up to quite large values of the produced diffractive mass . in the bbl
the dominant channel of diffraction to large masses is production of two jets with the total cross section given by eq.([ccsb ] ) and with a characteristic angular distribution @xmath310 , where @xmath311 is the c.m .
angle @xcite . on the contrary , in the perturbative qcd limit the diffractive dijet production , except the charmed dijet production ,
is strongly suppressed @xcite .
the suppression is due to the structure of coupling of the @xmath13 component of the real photon wave function to two gluons when calculated in the lowest order in @xmath312 . as a result , in the real photon case
hard diffraction involving light quarks is connected to production of @xmath313 and higher states .
distribution of diffractively produced jets over invariant mass provides an important test of the onset of bbl limit .
really , in the dglap / ct regime differential cross section of forward diffractive dijet production should be @xmath314 and be dominated by charm jet production .
this behavior is strikingly different from the bbl limit expressions of @xcite .
thus , the dijet photoproduction should be very sensitive to the onset of the bbl regime .
we want to draw attention that @xmath315 component of the photon light - cone wave function can be measured in three independent diffractive phenomena : in the bbl off the proton , in bbl off a heavy nuclei , in the ct regime where the wave function can be measured as a function of the interquark distance @xcite .
a competing process for photoproduction of dijets off heavy nuclei is production of dijets in @xmath316 collisions where the second photon is provided by the coulomb field of the nucleus .
the dijets produced in this process have positive c - parity and hence this amplitude does not interfere with the amplitude of the dijet production in the @xmath317 interaction which have negative c - parity .
our estimates indicate that this process will constitute a very small background over the wide range of energies @xcite .
ultraperipheral collisions(upc ) of relativistic heavy ions at rhic and lhc open a promising new avenue for experimental studies of the photon induced coherent and incoherent interactions with nuclei at high energies @xcite .
really , the lhc heavy ion program @xcite will allow studies of photon - proton and photon - nucleus collisions at the energies exceeding by far those available now at hera for @xmath318 scattering .
hence , we can analyze an opportunity to study the phenomena discussed above combining the theory of photo induced processes in the ultraperipheral aa collisions with our studies of the coherent photo(electro ) production of vector mesons .
we can use the standard weizsacker - williams approximation @xcite to calculate the cross section integrated over the momentum of the nucleus which emits the quasireal photons .
the cross section of the vector meson production integrated over the transverse momenta of the nucleus which emitted a photon can be written in the convoluted form : @xmath319 here y is the rapidity @xmath320 the flux of the equivalent photons @xmath321 is given by a simple expression @xcite : @xmath322.\ ] ] here @xmath323 and @xmath324 are modified bessel functions with argument @xmath325 , @xmath326 is lorentz factor and @xmath327 is the impact parameter .
the glauber profile factor @xmath328 accounts for the inelastic strong interactions of the nuclei at impact parameters @xmath329 and , hence , suppresses the corresponding contribution of the vector meson photoproduction .
recently the star collaboration released the first data on the cross section of the coherent @xmath330-meson production in gold - gold upc at @xmath331 gev @xcite .
this provides a first opportunity to check the basic features of the theoretical models and main approximations which include the weizscker - williams ( ww ) approximation for the spectrum of the equivalent photons , an approximate procedure for removing collisions at small impact parameters where nuclei interact strongly , and the model for the vector meson production in the @xmath332 interactions . in the case of the @xmath0-meson production the basic process
is understood much better than for other photoproduction processes .
hence , checking the theory for this case is especially important for proving that upc could be used for learning new information about photon - nucleus interactions .
note here that the inelastic shadowing effects which start to contribute at high energies still remain a few percent correction at energies @xmath333 100 gev relevant for the star kinematics . for lhc energy range one should account for the blackening of interaction with nuclei . in this case
cross section of inelastic diffraction in hadron - nucleus collisions should tend to 0 .
so major impact for the calculation of the process of diffractive photoproduction of @xmath0 meson would be necessity to neglect by the contribution of @xmath1 @xcite . the calculated momentum transfer distributions at the rapidity @xmath334 and the momentum transfer integrated rapidity distribution for gold - gold upc at @xmath335 gev are presented in figs .
[ figstar1]a , b @xcite .
= 0.8 = 0.8 let us briefly comment on our estimate of the incoherent @xmath330-meson production cross section .
the momentum transfer distribution ( dashed line in fig . [
figstar1]a ) is practically flat in the discussed @xmath336 range .
the total incoherent cross section obtained by integration over the wide range of @xmath336 is @xmath337 mb . to select the coherent production
the cut @xmath338 gev@xmath34 was used in the data analysis @xcite .
correspondingly , the calculated incoherent cross section for this region of @xmath336 is @xmath339 mb .
our calculations of incoherent production which are based on accounting for only the single elementary diffractive collision obviously present the lower limit
. the residual nucleus will be weakly excited and can evaporate only one - two neutrons .
the events @xmath340 were detected by the star and identified as a two - stage process - coherent @xmath330-production with the subsequent electromagnetic excitation and neutron decay of the colliding nuclei @xcite .
in particular , the cross section estimated by the star for the case when only one of the nuclei is excited and emits several neutrons is @xmath341 mb .
the momentum transfer distribution for these events is determined by the coherent production .
hence , it differs from that for incoherent events but in the region of very low @xmath336 it is hardly possible to separate them experimentally , and the measured cross section @xmath342 includes contribution of incoherent events on the level of @xmath343 .
the total rapidity - integrated cross section of coherent @xmath330-meson production calculated in the gvdm for the range of energies available at rhic is shown in fig .
[ figstar2](solid line ) .
we find @xmath344 mb at @xmath335 gev .
the value @xmath345 mb was obtained at this energy by the star from the data analysis at the low momentum transfer @xmath338 gev@xmath34 .
thus , before making a comparison we should take into account this cut .
it leads to a reduction of the cross section by @xmath346 ( the dashed line in fig .
[ figstar2 ] ) . in our calculations
we did nt account for the @xmath336-dependence of the elementary amplitudes which are rather flat in the considered range of energies and momentum transfers as compared to that for the nucleus form factor .
so , in the region of integration important for our analysis it is reasonable to neglect this slope .
nevertheless , an account of this effect would slightly reduce our estimate of the total cross section .
also we neglected a smearing due to the transverse momentum of photons and the interference of the production amplitudes from both nuclei @xcite .
.total cross sections of @xmath141 production in upc at lhc . [ cols="^,^,^",options="header " , ] this latter phenomenon results in the narrow dip in the coherent @xmath336-distribution at @xmath347 gev@xmath34 .
all these effects do not influence noticeably the value of the @xmath336-integrated cross section but can be easy treated and taken into account in a more refined analysis .
thus we find @xmath348 mb to be compared to the star value @xmath345 mb .
since our calculation does not have any free parameters , this can be considered as a reasonable agreement .
= 1.0 = 1.0 .
[ figlhc1 ] it was suggested in @xcite to look for color opacity phenomenon using @xmath204 ( photo ) electroproduction .
this however requires energies much larger than those available at the fixed target facilities and would require use of electron - nucleus colliders
. at the same time estimates of the counting rates performed within the framework of the felix study @xcite have demonstrated that the effective photon luminosities generated in peripheral heavy ion collisions at lhc would lead to significant rates of coherent photoproduction of vector mesons including @xmath4 in reaction @xmath349 as a result it would be possible to study at lhc photoproduction of vector mesons in pb - pb and ca - ca collisions at energies much higher than the range @xmath350 gev covered at the fixed target experiment at fnal @xcite .
note that even current experiments at rhic ( @xmath351 gev ) should also exceed this limit .
as it is clearly indicated by the star study the coherent photoproduction , leaving the both interacting nuclei intact , can be reliably identified by using the veto triggering from the two - side zero degree calorimeters which select the events not comprising the escaped neutrons . the additional requirement which enables to remove contribution of the incoherent events with the residual nucleus in the ground state is selection of the produced quarkonium with small transverse momentum . in fig .
[ lhcdst ] we compare the momentum transfer distributions for the coherent @xmath141 and @xmath4 photoproduction calculated in the leading twist shadowing model with the corresponding distributions for incoherent photoproduction .
note that we estimated the upper limit of incoherent cross section simply as the free elementary cross section on the nucleon target multiplied by the number of nucleons a. in fig .
[ figlhc1]a , b we present the rapidity distributions of the @xmath141 coherent production for peripheral collisions at lhc calculated including effects of gluon shadowing and in the impulse approximation . at the central rapidities we find suppression by a factor 4 for a case of ca and more strong , by a factor 6 for pb .
the total cross sections are given in table [ tcrsec1 ] .
the rapidity distributions for coherent @xmath4 production in the upc with ca and pb beams are shown in fig .
[ figlhc1]c , d and the corresponding total cross sections are given in table [ tcrsec2 ] . as it is seen from comparison of the leading twist shadowing based calculations to that performed in the impulse approximation the yield of @xmath4 is expected to be suppressed by a factor 2 at central rapidities due to the leading twist nuclear shadowing .
hence , study of the coherent photoproduction of the heavy quarkonium states at lhc opens an important avenue for investigating the nuclear gluon distributions and the shadowing effects in the kinematics which would be very hard to probe in any other experiments .
we demonstrated that coherent diffraction off nuclei provides an effective method of probing a possible onset of bbl regime in hard processes at small @xmath17 .
we predict a significant increase of the ratio of the yields of @xmath352 mesons in coherent processes off heavy nuclei due to the blackening of the soft qcd interactions in which fluctuations of the interaction strength are present .
an account of nondiagonal transitions leads to a prediction of a significant enhancement of production of heavier diffractive states especially production of high @xmath21 dijets .
study of these channels may allow to get an important information on the onset of the black body limit in the diffraction of real photons .
we argued that the fluctuations of strengths of interactions has been observed at intermediate energies in the diffractive photoproduction of vector mesons .
we discuss the opportunity to look for the transition from the nuclear color transparency to the regime of the color opacity in the ultrarelativistic peripheral ion collisions at lhc .
we thank j.bjorken , a.mueller , g.shaw , members of the upc study group , and our coauthors m.mcdermott , a.freund , v.guzey , l.gerland , for useful discussions and gif and doe for support . m.s . and
thank int for hospitality while this work was completed .
r. j. glauber , boulder lectures in theoretical physics , vol.1 , + ( interscience publ .
ny)(1959 ) ; + k. gottfried and d. r. yennie , phys . rev . *
182 * , 1595 ( 1969 ) + erratum - ibid .
d * 2 * , 2737 ( 1970 ) ; + g. v. bochmann , phys .
d6 , 1938 ( 1972 ) .
a. donnachie and g. shaw , in electromagnetic interactions of hadrons , edited by + a. donnachie and g. shaw ( plenum , new york , 1978 ) , vol .
164 - 194 ; + h. fraas , b. read , and d. schildknecht , nucl .
* b86 * , 346 ( 1975 ) ; + p. ditsas and g. shaw , nucl .
b * 113 * , 246 ( 1976 ) .
e. eichten , k. gottfried , t. kinoshita , j. b. kogut , k. d. lane and t. m. yan , + phys .
lett . * 34 * , 369 ( 1975 ) ; [ erratum - ibid . *
36 * , 1276 ( 1976 ) ] .
j. m. richard , z. phys .
c * 4 * ( 1980 ) , 211 ( 1980 ) .
l. frankfurt , l. gerland , m. strikman and m. zhalov , [ arxiv : hep - ph/0302009 ] . | we argue that study of the cross section of coherent photo(electro ) production of vector mesons off nuclear targets provides an effective method to probe the leading twist hard qcd regimes of color transparency and perturbative color opacity as well as the onset of black body limit ( bbl ) in the soft and hard qcd interactions . in the case of intermediate energies we use the generalized vector dominance model ( gvdm ) to take into account coherence effects for two distinctive limits - the soft interactions for production of @xmath0 and @xmath1-mesons and the color transparency regime for production of charmonium states .
we demonstrate that gvdm describes very well @xmath0-meson coherent photoproduction at @xmath2 gev and predict an oscillating energy dependence for the coherent charmonium production . in the limit of small @xmath3
we find that hard qcd leads to onset of the perturbative color opacity even for production of very small onium states , like @xmath4 .
the advantages of the process of coherent dijet photoproduction and hard diffractive processes in general for probing the onset of bbl and measuring the light - cone wave function of the photon in a hard scattering regime where decomposition over twists becomes inapplicable are explained .
we apply this analysis to the study of the photon induced coherent processes in ultra peripheral collisions of ions at lhc and demonstrate that the counting rates will be sufficient to study the physics of color opacity and color transparency at the energies beyond the reach of the electron - nucleon(nucleus ) colliders . |
the standard model of cosmology ( or cdm ) assumes that the dark matter(dm ) is cold and effectively collisionless .
it assumes a hierarchical framework for galaxy formation and throughout their evolution galaxies are subject to frequent collisions and interactions with nearby galaxies that leave surrounding material in the form of tidal debris , therefore studying tidal features can yield important clues on galaxy evolution .
tidal features can appear in the form of complicated structures@xcite , but it is frequent to observe structures like tidal tails around the bulge of massive galaxies@xcite , some early - type galaxies present shells@xcite which can be seen as smooth arcs of ellipses centered around its host galaxy decreasing their surface brightness for larger radii , there are also debris in the form of stellar streams@xcite .
several analyses of the spheroidal components of tidally disturbed early - type galaxies(etgs)@xcite suggest that structural properties such as the effective radius(radius at which half of the total light of the system is emitted ) and their surface brightness are quite similar from those of the other ellipticals or the spheroids of galaxies with few or no signs of tidal interaction .
other results have shown evidence of kinematically decoupled components in early - type galaxies@xcite . given the standard paradigm of cdm
it is expected that even these `` passive '' galaxies also formed by galaxy mergers , then the similarity of the bulge regions for the etgs could be due to the some minor mergers(galaxy mergers of unequal mass with mass ratio of @xmath3 ) slightly affecting the bulge but massive enough to still induce tidal debris , or by major mergers(mergers of comparable mass galaxies ) that happened a long time ago so that now the bulge has relaxed .
cosmological n - body simulations in cdm have shown@xcite that merger events are frequent when galaxies are growing , however , it must be taken into account that anticipating the final structure around a galaxy is rather difficult given the dependence on the environment and merger history in the formation of each particular galaxy@xcite .
observationally , it is not uncommon to run into difficulties identifying and studying tidal structures around galaxies , sometimes these structures are too faint compared to the central light distribution that a detailed analysis proves impossible .
nevertheless , there are observations where faint structures such as shells are clearly seen around galaxies at some particular radii from the center of their associated host@xcite . in some situations it has been possible to identify younger stellar populations in the outskirts compared to those in the inner bulge@xcite probably reminiscent of a late gas accretion .
it is worth noticing that in the cdm model the spatial location of these remnants of past interaction is not predicted and has to be tracked with numerical simulations as it depends on the history of each galaxy .
metallicity is another powerful tool to constrain galaxy evolution across cosmic epochs as it is known to be correlated to gas infall , outflows due to supernovae , and star formation rate within the galaxy .
recent studies of star formation focused on local group dwarf galaxies @xcite revealing radial gradients in metallicity @xcite , and most of them seem to have old stellar populations formed @xmath4 10 gyrs ago@xcite , followed by a more quiescent star formation likely due to loss of gas mass at the reionization epoch .
there are high redshift galaxies ( @xmath5 3 ) that also show metallicity gradients , their regions of high star formation rate(sfr ) are also of lower metallicity@xcite due to a large gas infall , they are sorrounded by a more enriched disc , in @xcite the authors concluded that these observations are consistent with accretion of metal - poor gas in massive galaxies where a high sfr can be sustained without requiring frequent mergers , this picture was also suggested by studies of gas rich disks at @xmath5 1 - 2@xcite . in order to describe the debris around galaxies and stellar population it is necessary to include the gas dynamics in simulations , even though there is an uncertainty due to our lack of understanding of the baryonic processes , their addition is also required to partially reduce some discrepancies of the cdm model , for instance , the cusp - core problem(see @xcite for a review ) , the overabundance of satellites@xcite or the too - big - to - fail@xcite issue .
however , some of these discrepancies might also be solved assuming different properties for the dark matter , leading to for instance , scalar field dark matter(sfdm ) @xcite , strongly self - interacting dm@xcite , warm dark matter@xcite .
it is of particular interest for us the sfdm alternative , here the mass of the field is assumed to be very small ( @xmath0ev/@xmath6 ) such that its de broglie wavelength is of order @xmath4 kpc , relevant for galactic scales .
the quantum behavior of the field has created much interest in the model due to its success to account for some discrepancies mentioned above with dark matter properties only , for example , the small mass keeps the central density from increasing indefinitely due to the uncertainty principle in contrast to cdm simulations where supernova feedback is required@xcite , note that the amount of feedback required to produce a constant central density may be in conflict with other observations@xcite .
an interesting property that has been noted for scalar field haloes in excited states or in a superposition of them is the appearance of `` ripples '' in their mass density profiles@xcite .
this property motivated us to investigate whether features like rings or shells in non - interacting galaxies hosted by multistate scalar field haloes could arise independently of the galaxy merger mechanism .
in the next section we give a brief overview of previous works in the scalar field dark matter model and describe a model for galaxy formation in sfdm haloes .
we modify the hydrodynamics code zeus to implement an evolving sfdm halo in which a gaseous component is present , we provide the details of our simulations in section 3 . in section 4
we discuss our results and section 5 is devoted to our conclusions .
the main idea in the scalar field dark matter model@xcite considers a self - interacting scalar field with a very small mass , typically of @xmath0ev/@xmath6 , such that the quantum mechanical uncertainty principle and the interactions prevent gravitational collapse in self - gravitating structures , thus the haloes are characterized with homogeneous densities ( usually refered as a cores ) in their centers , in general the core sizes depend on the values of the mass and the self - interacting parameters@xcite(for a review see @xcite ) .
there has been several studies of the cosmological evolution that results for a scalar field mass @xmath710@xmath8ev/@xmath6 , with this mass the cosmological density evolution of cdm can be reproduced@xcite , there is consistency with the acoustic peaks of the cosmic microwave background radiation@xcite and it implies a sharp cut - off in the mass power spectrum for halo masses below @xmath9m@xmath10 suppressing structure formation of low mass dark matter haloes@xcite . moreover , there is particular interest in finding equilibrium configurations of the system of equations that describe the field ( einstein - klein - gordon system ) and of its weak field approximation ( schrdinger - poisson(sp ) system ) , different authors have obtained solutions interpretated as boson stars or later as dark matter haloes showing agreement with rotation curves in galaxies and velocity dispersion profiles in dwarf spheroidal galaxies@xcite .
so far the large and small scales observations are well described with the small mass and thus has been taken as a prefered value but the precise values of the mass and self - interaction parameters are still uncertain , we believe tighter constraints can come from numerical simulations @xcite and comparisons with large galaxy samples which we plan to do in the future .
recently the idea of the scalar field has gained interest , given the uncertainty in the parameters the model has adopted different names in the literature depending on the regime that is under discussion , for instance , if the interactions are not present and the mass is @xmath0ev/@xmath6 this limit was called fuzzy dark matter@xcite or more recently wave dark matter@xcite , another limit is when the sf self - interactions are described with a quartic term in the scalar field potential and dominate over the mass(quadratic ) term , this was studied in @xcite and called repulsive dark matter or fluid dark matter by@xcite .
notice that for a scalar field mass of @xmath0ev/@xmath6 the critical temperature of condensation for the field is t@xmath11tev , which is very high , if the temperature of the field is below its critical temperature it can form a cosmological bose einstein condensate , if it condenses it is called bose - einstein condensed(bec ) dark matter@xcite .
@xcite mentioned that axions could also form bose - einstein condensates even though their mass is larger than the previous preferred value , notice that the result was contested in @xcite , this suggest that the condensation process should be study in more detail to confirm it can remain as bec dark matter . in @xcite , it was found that complex scalar field with @xmath12<@xmath13ev/@xmath6 that decoupled being still relativistic will always form a cosmological bose - einstein condensate described by the ground state wavefunction , this does not preclude the existence of bosons with higher energy , particulary in dark matter halos .
we see that the smallness of the boson mass is its characteristic property and cosmological condensation is a likely consequence .
the preferred mass of the scalar field dark matter lies close to @xmath0ev/@xmath6 satisfying the above constraint , although there are still uncertainties on the mass parameter , in order to avoid confusion with the known axion and help with the identification in future works , we find it useful and appropiate to name the scalar field dark matter candidate , given the above characteristics we can define it as a particle with mass @xmath12<@xmath13ev/@xmath6 , being commonly described by its wavefunction we choose to name this dm candidate _ psyon_. it is worth emphasizing that despite the variety of names given to the model the main idea mentioned above remains the same , it is the quantum properties that arise due to the small mass of the boson that characterize and distinguishes this paradigm , analoguous to the standard cosmological model represented by the cdm paradigm whose preferred dark matter candidates are the wimps(weakly interacting massive particles ) , one being the neutralino , we see that all the above regimes sfdm , repulsive dm , axion dm , or any other model assuming an ultra light bosonic particle comprise a single class of paradigm , which we call _ quantum dark matter_(qdm ) paradigm .
as pointed before , in the qdm paradigm the small mass of the dark matter boson leads to the posibility of forming cosmological condensates , even for axions which are non - thermally produced and have masses in @xmath14ev/@xmath6@xcite , from here we can obtain a characteristic property that distinguishes these dark matter candidates from wimps or neutrinos , namely , the existence of _ bosons in the condensed state _ , or simply _ bics _ , from our above discussion the axion and psyon are included in the bics .
there has been considerable work to find numerical solutions to the non - interacting sfdm in the non - relativistic regime to model spherically symmetric haloes@xcite , and also for the self - interacting sfdm @xcite , it is worth noting that as mentioned in @xcite for the weak field limit of the system that determines the evolution of a spherically symmetric scalar field , that is , the einstein and klein - gordon equations , for a complex and a real scalar field the system reduces to the schrdinger - poisson equations@xcite . from current bounds reported in @xcite obtained by imposing that the sf behaves cosmologically as pressureless matter(dust ) we obtained that the interacting parameter would be extremely small for the typical mass of @xmath410@xmath8ev/@xmath6
, therefore we expect that solutions to the sp system with no interactions would behave qualitatively similar to those when self - interactions are included , this assumption is supported by the similarity in the solutions for a small self - coupling found in other works@xcite , in this work we will study the non - interacting solutions
. one characteristic feature of stationary solutions of the form @xmath15 for the sp system is the appearance of nodes in the spatial function @xmath16 , these nodes are associated to different energy states of the sf , the zero node solution corresponds to the ground state , one node to the first excited state , and so on .
these excited states solutions fit rotation curves(rcs ) of large galaxies up to the outermost measured data and can even reproduce the wiggles seen at large radii in high - resolution observations @xcite .
however , haloes that are purely in a single excited state seem to be unstable when the number of particles is not conserved(finite perturbations ) and decay to the ground state with different decay rates@xcite , though they are stable when the number of particles is conserved(infinitesimal perturbations ) .
the ground state solution is stable under finite perturbations and infinitesimal perturbations@xcite , but has difficulties to correctly fit the rotation curves in large galaxies because its associated rc has a fast keplerian behavior shortly after reaching its maximum value unable to remain flat enough at large radii .
one way to overcome this problem was to consider that bosons are not in one state but instead coexist in different states within the halo , given our intention to describe dark matter haloes we will refer to such configurations as multistate halos(mshs)@xcite .
the size of the msh is determined by the most excited state that accurately fits the rc for large radii , excited states are distributed to larger radii than the ground state , and in contrast to the halo with single state there are mshs that are stable under finite perturbations provided the ground state in the final halo configuration has enough mass to stabilize the coexisting state@xcite . in @xcite it was shown that mshs can be studied in the classical approach as a collection of classical scalar fields coupled through gravity , one field @xmath17 for each state , this would modify the sp system such that its source of energy density would be the sum of densities in each state@xcite , where each state satisfies its respective schrdinger equation and the states are coupled through the newtonian gravitational potential @xmath18 . in spherical symmetry ( in units @xmath19=1 , c=1 ) the sf solutions , @xmath20 ,
satisfying the einstein - klein - gordon system are related to the sp wavefunctions through @xmath21 with @xmath12 the mass of the sf . for stationary solutions we can assume wavefunctions of the form @xmath22
, then the sp system reads @xmath23 @xmath24 with @xmath25 the laplacian operator in spherical coordinates , and @xmath26 the energy eigenstates .
we consider the case of a msh with only the ground and first excited states given that for this msh @xcite reported the critical value for its stability under small perturbations .
if the ground state has @xmath27 particles and there are @xmath28 in the excited state , the authors found that msh would be stable under small perturbations provided the fraction @xmath29 in all the configurations studied in which @xmath30 is larger than the maximum for stability @xmath31 , the induced instability causes the configuration to lose a few particles and a remarkable effect takes place , the excited state does a fast transition to the ground state and viceversa , that is , the populations of the states are inverted such that the final @xmath30 complies with the stability condition ( [ eta ] ) , after the inversion the msh approaches a stable configuration where the particle number remains constant for each state , after the transition the particles that were initially in the excited state are now in the ground state where they have redistributed and become more compact than in their original distribution .
this population inversion(pi ) is a characteristic feature of mshs and as we will show it can have consequences in the gas distribution and on the accretion rate during the period of galaxy formation .
in fact , given the cut - off in the mass power spectrum due to the small mass of the psyon we expect a delay in the first structures to collapse compared to the those found in cdm@xcite , this was confirmed with cosmological simulations for a mass of @xmath32@xcite , therefore , this intrinsic change of states within mshs at high redshift might have observable consequences in the gas and stellar properties of the galaxies that will form in such potential wells .
initial fluctuations that grow due to the cosmological expansion of the universe eventually separate from it and start collapsing due to its own gravity , at this time ( known as turnaround ) the halo has a number of psyons that can be in different states , their values would be determined before the collapse of the configuration , and have some dependence on the its local environment . depending on the number density of bosons populating the excited states we can have different fates for the haloes as mentioned in subsection ( 2.2 ) . from the mentioned results of the literature and based on the successful fits to different galaxies we propose the following model of galaxy formation in sfdm haloes .
the smallest and less dense systems , such as dwarf galaxies , would reside in sfdm haloes with most bosons in the ground state except possibly for just a few excited particles@xcite , this is because the potential wells of these haloes , being the lowest dense systems , would be just massive enough to collapse and allow the existence of the state of minimum energy that would form a bound configuration .
larger configurations that had initially a larger number of bosons in excited states than in the ground state can undergo a population inversion and reach a stable state , collapse to a dense ground state or become a black hole depending on how large the fraction of particles in excited states is after turnaround , given the different possible outcomes for this case we expect that most galaxies are formed this way , for instance , ellipticals can form when haloes transition to a stable configuration to form a ground state and quickly deepen the gravitational potential resulting in a dense and compact structure where the bulge can form , for high surface brightness galaxies which usually have rcs that fall slowly after its maximum , these galaxies would likely correspond to sfdm haloes whose final stable configuration has a comparable number of bosons in the excited and ground states as both states would contribute evenly in the central region increasing the potential well where more star formation can take place , and for distances larger than a given radius ( likely the first maximum peak in the rc ) the ground state would be a small contribution to the density , thus in the outer region the baryonic rc would remain below its maximum . another possible outcome for extended haloes
is when the fraction of excited particles is below its stability threshold , here gravity will slowly redistribute dm bosons in the stable msh but mostly keeping the particle number constant , considering that psyons in the excited states are subdominant but more widely distributed than the ones in the ground state of a msh , the dm gravitational potential in the center would be less dense than in the high surface brightness case so that gas accretion would also proceed more slowly resulting in rotation curves for the baryonic matter that slowly increase to large radii or with almost flat profiles , this case is quite similar to what is found in low surface brighness galaxies@xcite .
this galaxy formation scenario broadly describes the different galaxy types showing that only due to the dark matter properties in the context of sfdm it is possible to agree with the general features of several galaxies .
however , we do not expect this scenario to represent all galaxies , in fact , to get a better description of individual galaxies we require taking into account other fundamental parameters that affect their evolution , such as the environment , angular momentum , galaxy mergers etc .
, we will leave a more in dept exploration of the scenario including the full astrophysical processes for a future work . for now , we take the given description of elliptical galaxies to explore whether some substructure found in some of these galaxies can be a consequence of the properties of the background dm halo where such ellipticals are formed .
we do not discard the possibility that shells or rings result from mergers , instead , we explore a new possible mechanism in which they can arise offering an alternative explanation to the tidal debris in certain non - interacting ellipticals .
we study this issue in a multistate halo .
[ fig1 ] [ cols= " < , < " , ]
in this work we have seen that if the dark matter is a ultra light scalar field the internal structure of galaxies can reflect fossil records of their formation , in our case we study the formation of tidal structures similar to rings and shells in non - interacting galaxies .
we discussed the nature of these structures in the context of sfdm and conclude that independently of galaxy mergers they can arise as a consequence of the quantum nature of the scalar field dark matter boson(hereafter _ psyons _ ) that form the halo where galaxies are embedded in , mainly , due to the intrinsic ripples in the dark matter halo density profile .
the possibility of describing tidal substructure in scalar field dark matter halos was also suggested in @xcite but under different halo configurations and without including the evolution of a gaseous component , here we make use of spherically symmetric haloes in multistates because their stability has been studied and confirmed in previous works@xcite .
in addition , they possess desirable properties to describe a variety of galaxies .
inspired by the numerical solutions in @xcite , where they reported the maximum fraction of psyons in excited states ( eq .
[ eta ] ) to form a stable multistate halo composed of the ground state and first excited state , we propose a semi - analytical model to describe a similar halo configuration and study the distribution of a gaseous disc that is contained within such scalar field dm halo .
the halo is set to be initially unstable by choosing a fraction of excited boson above the stability limit , this induces a transition to reach a final stable configuration affecting the initial gas distribution , after the transition the population of the states is inverted and the halo stabilizes once the ground state dominates the halo mass that by gravity will be confined later in the center of the halo . at the end of the simulations ( @xmath33 gyrs ) part of
the gas keeps orbiting in the form of a ring outside the main bulge - like region , the structure of the ring is not significantly different when the halo transition happens early ( @xmath34=@xmath35 gyrs ) or late(@xmath34=@xmath36 gyrs ) but the ring broadens if the transition is slow and becomes more defined for fast transitions . due to the lack of stability parameter for msh in higher energy states we explore the first non trivial superposition of multistates , given the positive results it seems that tidal structures would also be found at larger radii for sfdm haloes that have boson in higher excited states , although they would become fainter with distance , this would agree with shells that appear to be distribited at larger radii .
the tidal features in this work have a distinct origin from the usual tidal stripping of galaxies , they emerge from the quantum properties of the psyons that compose the dark halo , the presence of the symmetric ring is due to the spherically symmetry hypothesis , deviations from this symmetry could result in incomplete rings and shells although a certain degree of symmetry should remain which distinguishes to remnants that come from recent galaxy mergers that depend on the path of the accreted satellites and does not necessarily leave tidal debris of a particular symmetry , our results suggest that outer regions of dark haloes may not be totally smooth .
exploring in more detail the regions where shells and rings form could also help to put constraints on their origin , their formation , and serve to test galaxy and halo formation scenarios such as the quantum dark matter(qdm ) paradigm .
additionally , we have provided a classification for the dark matter models that are based on the assumption that the dark matter is a scalar field of a very small mass that proves useful for future references and that helps to establish the main feature of the particle candidates of this paradigm .
we formulate a galaxy formation scenario based on known results of halo formation in the sfdm context that can explain qualitatively the general properties of the different galaxy types in the universe and applied to interpret our findings .
given the success of the sfdm model to account for the apparent discrepancies of the standard cdm model with only dark matter properties , it will be desirable to include astrophysical processes in simulations to get a fair comparison with state - of - the - art cdm simulations .
the qdm paradigm is indeed an interesting possibility that deserves further study .
this work was partially supported by conacyt mxico under grants cb-2009 - 01 , no .
132400 , cb-2011 , no . 166212 , and i0101/131/07 c-234/07 of the instituto avanzado de cosmologia ( iac ) collaboration ( http://www.iac.edu.mx/ ) .
victor h. robles and l. medina are supported by a conacyt scholarship .
99 arbey a. , lesgourgues j. , & salati p. , 2003 ,
prd , 68 , 023511 bajaja e. , van der burg g. , faber s.m . ,
et al.,1984,aa , 141 , 309 - 317 balakrishna j. , seidel e. , & suen w - m . , 1998 , prd , 58 , 104004 barnes j.e . , hernquist l. , 1992 , annual review of astronomy and astrophysics , 30 , 705 behroozi p.s . , risa h. wechsler r.h . , wu h.y
, 2013 , apj , 763 , 18 belokurov , v. , walker , m. g. , evans , n. w. et al . , 2010 , apj , 712 , l103 bendo g.j .
, buckalew b.a . ,
dale d.a . , draine b.t .
2006 , apj , 645 , 134 bernal a. , barranco j. , alic d. , & palenzuela c. , 2010 , prd , 81 , 044031 bode p. , ostriker j.p.,turok n. , 2001 , apj , 556 , 93 bhmer c.g . , & harko t. , 2007 , jcap06 , 025 boylan - kolchin , m. , bullock , j. s. , kaplinghat , m. 2011 , mnras , 415 , l40 boylan - kolchin m. , bullock j.s . , garrison - kimmel s. , 2014 , mnras , 443 , l44 bozek b. , marsh d.j.e . , silk j. , wyse r.f.g . , 2014 , arxiv:1409.3544 bray h.l . , 2012 , arxiv:1212.5745 briscese f. , 2011 , phys.lett.b , 696 , 315 bush j.w.m . , 2010 ,
, 107 , 17455 buta r.j . , sheth k. , regan m. et al . , 2010 ,
apjs , 190 , 147 chavanis p.h . , 2012 , a&a 537 , a127 colin p.,avila - reese v. , valenzuela o. , firmani c. , 2002 , apj , 581 , 777 colpi m. , shapiro s.l . , & wasserman i. , 1986 , prl , 57 , 2485 couder y. , fort e. , gautier c.h .
, & boudaoud a. , 2005 , prl , 94 , 177801 couder y. , & fort e. , 2006 , prl , 97 , 154101 cresci g. , mannucci f. , maiolino r. et al . , 2010 , nature , 467 , 811 davidson s. , & elmer m. , 2013 , jcap12 , 034 de blok w.j.g . , 2010 , advances in astronomy , article i d 789293 diez - tejedor a. , gonzalez - morales a.x . , & profumo s.,prd , 2014 , 90 , 043517 eddi a. , fort e. , moisy f. , & couder y. , 2009 , prl , 102 , 240401 eddi a. , moukhtar j. , perrard s. , fort e. , & couder y. , 2012 , prl , 108 , 264503 emsellem , e. , 1995 , aa , 303 , 673 frster schreiber n.m . , genzel r. , bouch n. et al . , 2009 , apj , 706 , 1364 fort , b. p. , prieur , j .- l . ,
carter , d. , meatheringham , s. j. , & vigroux , l. 1986 , apj , 306 , 110 fort e. , eddi a. , boudaoud a. et al . , 2010 ,
sci . , 107(41 ) , 17515 garrison - kimmel , s. , boylan - kolchin , m. , bullock , j. s. , lee , k. 2014 , mnras , 438 , 2578 garrison - kimmel , s. , boylan - kolchin , m. , bullock , j. s. , kirby , e. n. 2014 , mnras , 444 , 222 governato f. , brook c. , mayer l. et al . , 2010 , nature , 463 , 203 governato f. , zolotov a. , pontzen a. et al . , 2012 , mnras , 422 , 1231 goodman j.,2000 , new astronomy 5 , 103 guzmn f.s .
, lora - clavijo f.d .
, gonzlez - avils j.j . , & rivera - paleo f.j . , 2014 , prd , 89 , 063507 guzmn f.s . , & urea - lpez l.a . , 2004 ,
prd , 69 , 124033 guzmn , f. s. , & matos , t. 2000 , class .
grav . , 17 , l9 harko t. , 2011 , prd , 83 , 123515 hau g.k.t . , &
forbes d.a .
, 2006 , mnras , 371 , 633 hayes j.c . , norman m.l .
, fiedler r.a . , 2006 , apjs , 165 , 188 .
hu w. , barkana r. , gruzinov a. , 2000 , prl , 85 , 1158 ji s.u . , & sin s.j .
, 1994 , prd , 50 , 3655 kang m. , bieging j. h. , kulesa c.a .
, lee y. , 2009 , apj . , 701 , 454 kaup d.j . , 1968 ,
, 172 , 1331 kirby e.n . ,
cohen j.g .
, guhathakurta p. et al .
, 2013 , apj , 779 , 102 klypin a.a . , trujillo - gomez s. , primack j. , 2011 , apj , 740 , 102 klypin , a. , kravstov , a. v. , valenzuela , o. , prada , f. 1999 , apj , 522 , 82 koprolin , w. , zeilinger , w. , 2000 , a&a , 145 , 71 kuzio de naray r. , & spekkens k. , 2011 , apj , 741 , l29 kuzio de naray r. , & kaufmann t. , 2011 , mnras , 414 , 3617 kuzio de naray r. , martinez g.d . , bullock j.s .
, & kaplinghat m. et al . , 2010 , apj , 710 , l161 laurikainen e. , salo h. , buta r. , knapen j.h . , & comern s. , 2010 , mnras , 405 , 1089 lee j.w , & koh i.g . , 1996 , prd , 53 , 2236 lee j.w . , & lim s. , 2010 , jcap01 , 007 li b. , rindler - daller t. , & shapiro p.r . , 2014 , prd , 89 , 083536 lora v. , magaa j. , bernal a. , snchez - salcedo f.j . , & grebel e.k . , 2012 , jcap02 ( 2012 ) 011 lora v. , magaa j.,2014 , jcap09 , 011 macci , a. v. , paduroin , s. , anderhalden , d. et al . , 2012 mnras , 424 , 1105 madau p. , weisz d.r . ,
conroy c. , 2014 , apj , 790 , l17 magaa j. , matos t. , surez a. , snchez - salcedo f.j .
, 2012 , jcap 1210 , 003 mannucci f.,cresci g. , maiolino r. et al . , 2010 , mnras , 408 , 2115 mannucci f.,cresci g. , maiolino r. et al . , 2009 , mnras , 398 , 1915 malin , d. f. & carter , d. 1983 , apj , 274 , 534 malin , d. f. & carter , d. 1980 , nature , 285 , 643 marsh d.j.e . , & silk j. , 2014 , mnras , 437 , 2652 martnez - delgado d. , pearrubia j. , gabany r.j .
, trujillo i. et al .
2008 , apj , 689 , 184 martnez - delgado d. , pohlen m. , gabany r.j . , majewski s.r .
et al . , 2009 ,
apj , 692 , 955 martnez - delgado d. , gabany r.j . , crawford k. , zibetti1 s. et al . , 2010 , the astronomical journal , 140 , 962 martinez - medina l.a . , &
matos t. , 2014 , mnras , 444 , 185 martinez - medina l.a . , robles v.h .
, & matos t. , 2015 , prd , 91 , 023519 matos , t. , & urea@xmath37lpez , l. a. 2001 , phys rev .
d , 63 , 063506 matos t. , urea - lpez l.a . , 2007 , gen . relativ . gravit . , 39 , 1279 michard , r. & prugniel , p. 2004
, aa , 423 , 833 miyamoto m. , & nagai r. , 1975 , pasj , 27 , 533 moore , b. , ghigna , s. , governato , f. et al . , 1999 , apj , 524 , l19 ozaa a.u . ,
rosales r.r . ,
& bush j.w.m
. , 2013 , journal of fluid mechanics , 737 , 552 peebles p.j.e . , 2000 , apj , 534 , l127 , pearrubia j. , pontzen a. , walker m.g .
, koposov s.e . , 2012 ,
apj , 759 , l42 phillips j.i . , wheeler c. , boylan - kolchin m. et al . , mnras , volume 437 , 1930 pontzen a. , & governato f. , 2012 , mnras , 421 , 3464 protiere s. , boudaoud a. , & couder y. , 2006 , j. fluid mech . , 554 , 85 rindler - daller t. , shapiro p.r .
, 2012 , mnras , 422 , 135 rindler - daller t. , & shapiro p.r . , 2014 , mod .
a , 29 , 1430002 robles v.h . , &
matos t. , 2013 , apj , 763 , 19 robles v.h .
, & matos t. 2013b , phys . rev .
d 88 , 083008 robles v.h .
, & matos t. , 2012 , mnras , 422 , 282 rocha m. , peter a.h.g .
, bullock j.s .
et al . , 2013 ,
mnras , 430 , 81 rodrguez - montoya i. , magaa j. , matos t. , prez - lorenzana a. , 2010 , apj , 721 , 1509 ruffini r. , & bonazzola s. , 1969 , phys .
, 187 , 1767 scannapieco c. , wadepuhl m. , parry o.h .
et al . , 2012 , mnras , 423 , 1726 schive h.y . , chiueh t. , & broadhurst t. , 2014 , nature , 10 , 496 schive h.y . , liao m.h .
, woo t.p.,et al .
2014b , arxiv:1407.7762 schweizer , f. & seitzer , p. 1992 , aj , 104 , 1039 schweizer f. & seitzer p. , 1988 ,
apj , 328 , 88 seidel e. , & suen w - m . , 1990 ,
prd , 42 , 384 seidel e. , & suen w.m .
, 1991 , prl , 66 , 1659 sikivie p. , yang q. , 2009 , prl , 103 , 111301 simon , j. d. , & geha , m. 2007 , apj , 670 , 313 sin s.j . , 1994 , prd , 50 , 3650 slepian z. , & goodman j. , 2012 , mnras , 427 , 839 spergel d.n . , steinhardt p.j . , 2000 , prl , 84 , 3760 surez a. , robles v.h .
, matos t. , 2014 , astrophysics and space science proceedings , 38 , 107 surez a. , & matos t. , 2011 , mnras , 416 , 87 tacconi l.j . , r. genzel r. , neri r. et al . , nature , 463 , 781 taehyun k. , sheth k. , hinz j.l .
et al . , 2012 ,
apj , 753 , 43 turnbull , a. j. , bridges , t. j. , & carter , d. 1999 , mnras , 307 , 967 urea - lopez l.a .
, 2002 , class .
grav . , 19 , 2617 urea - lopez l.a . , 2009 , jcap01 , 014 urea - lpez l.a . ,
& bernal a. , 2010 , prd , 82 , 123535 urea - lopez l.a .
, valdez - alvarado s. , becerril r. , 2012 , class.quant.grav . , 29 , 065021 van dokkum , p. g. 2005 , aj , 130 , 2647 vogelsberger m.,zavala j. , loeb a. , 2012 , mnras , 423 , 3740 walker j. , 1978 , scientific american , 238 , 151 weisz d.r . , dolphin a.e .
, skillman e.d .
et al . , 2014a , apj , 789 , 147 weisz d.r . , dolphin a.e . , skillman e.d .
et al . , 2014b , apj , 789 , 148 | recent observations of tidal debris around galaxies have revealed that the structural properties of the spheroidal components of tidally disturbed galaxies are similar to those found in non - interacting early - type galaxies(etgs ) , likely due to minor merging events that do not strongly affect the bulge region or to major mergers that happened a long time ago . we show that independently of merger events , tidal features like shells or rings can also arise if the the dark matter is an ultra light scalar field of mass @xmath0ev@xmath1 . in the scalar field dark matter ( sfdm ) model the small mass precludes halo formation below @xmath2 reducing the number of small galaxies today , it produces shallow density profiles due to the uncertainty principle in contrast to the steep profiles found in the standard cold dark matter ( cdm ) paradigm , in addition to the usual soliton solution there exists dark matter haloes in multistates , characterized by ripples in their density profiles , which are stable provided that most of the halo mass resides in the ground state .
we use the hydrodynamics code zeus to track the gas evolution in a background potential given by a superposition of the ground and first excited state of the scalar field , we study this configuration when it is initially unstable ( excited state more massive than ground state ) but by a population inversion in the states it eventually becomes stable , this could happen when haloes decoupled from the expansion of the universe and collapse to reach a state of equilibrium .
we found that tidal structures like rings are formed at a particular radius as a direct consequence of the wavelike structure of the dark matter halo , thus they possess a certain degree of symmetry that can be used to distinguish them from remnants formed through the usual galaxy merging mechanism . in our simulations
most of the gas concentrates in the center of the halo forming a dense and compact structure likely enhancing star formation in this region , however , some of the gas always remains trapped between the overdensity regions of the background sfdm halo predicting surrounding features even in non - interacting galaxies . in galaxies that are not strongly interacting with their neighbors , these structures may leave traces such as radial gradients in stellar metallicity , rings of dust , or even in the form of a younger population of stars at particular radii in contrast to the old stellar population dominating the bulge region , our results offer a new mechanism in which features like rings and shells can emerge purely from the quantum properties of this ultra light scalar field .
[ firstpage ] circumstellar matter infrared : stars . |
polarization of starlight arising from aligned dust was discovered accidentally approximately 60 years ago ( hall 1949 , hiltner 1949 ) .
this gave rise to attempts to explain the alignment .
in the years that followed various interactions from paramagnetic relaxation ( davis & greenstein 1951 ) and streaming of grains ( gold 1951 ) to interaction with cosmic rays ( salpeter & wickramasinghe 1969 ) and photons ( harwit 1970 ) have been explored .
many key insights into the dynamics of grains are associated with lyman spitzer and edward purcell who addressed the problem of grain alignment on a number of occasions . while we can refer the reader interested in the history of ideas on grain alignment to a review in lazarian ( 2003 ) , in this short publication we concentrate on the modern quantitative understanding of grain alignment .
several mechanisms were proposed and elaborated to various degree ( see lazarian 2007 for a review ) , including the `` textbook solution '' , namely , the paramagnetic davis - greenstein ( 1951 ) mechanism , which matured through intensive work since its introduction ( e.g. jones & spitzer 1967 , purcell 1979 , spitzer & mcglynn 1979 , mathis 1986 , roberge et al . 1993 , lazarian 1997 , roberge & lazarian 1999 ) .
the mechanical stochastic alignment was pioneered by gold ( 1951 ) , who concluded that supersonic flows should align grains rotating thermally .
further advancement of the mechanical alignment mechanism ( e.g. lazarian 1994 , 1995a ) allowed one to extend the range of applicability of the mechanism , but left it as an auxiliary process , nevertheless .
the major problem was that even the favorite alignment mechanism , the paramagnetic one , experienced severe problems explaining observational data .
i feel that the attempts to solve the problem for spheres and spheroids , reminiscent of a theorist s favorite `` spherical cow '' , were the major stumbling block for understanding of grain alignment .
the first attempt to consider something which is not symmetric but has net helicity was a ground - breaking study by dolginov & mytrophanov ( 1976 ) .
the authors considered there a grain that has different cross - sections for the extinction of the right- and left - polarized photons and predicted that such a grain was bound to spin up and get aligned when subjected to the anisotropic external radiative field .
the study by dolginov & mytrophanov ( 1976 ) had several deficiencies , however .
first of all , it did not provide clear recipes about calculating the amplitude of radiative torques .
in addition , the functional dependences of the torques calculated there were incorrect ( hoang & lazarian 2009 ) . one way or another , the work was mostly ignored for another 20 years until it attracted attention of bruce draine , who modified his publicly available ddscat code to calculate radiative torques acting on irregular grains .
this resulted in the explosion of interest to radiative torques .
empirical studies in draine ( 1996 ) , draine & weingartner ( 1996 , 1997 ) , weingartner & draine ( 2003 ) demonstrated that the magnitude of torques is substantial for irregular shapes studied .
after that it became impossible to ignore the radiative torque alignment .
later , the spin - up of grains by radiative torques was demonstrated in laboratory conditions ( abbas et al .
2004 ) .
the initial work on radiative torque alignment did not provide quantitative predictions for the grain alignment degree . the multi - parameter space presented by grain alignment induced by radiative torques ( henceforth rats ) posed an insurmountable problem for the `` brute force '' numerical approach . at the same time , both the interpretation and modeling of polarization call for simple recipes to parameterize effects of grain alignment .
this is not feasible with numerical calculations which suggest that rats depend on grain shape , grain size , radiation spectrum , grain composition , and the angle between the radiation direction and the magnetic field .
consequently , the important empirical studies above had limited predictive powers and were used to demonstrated the radiative torque effects sometimes using one grain shape , one grain size , one wavelength of light , and one direction of the light beam with respect to the magnetic field .
the quantitative stage of radiative torque studies required theoretical models describing radiative torques . in lazarian & hoang ( 2007a )
we proposed a simple model of rats which allowed a good analytical description of the alignment .
this model was elaborated and extended in lazarian & hoang ( 2008 ) and hoang & lazarian ( 2008 , 2009ab ) .
recent reviews on grain alignment include an extended one by lazarian ( 2007 ) . however , the subject of grain alignment has been developing so rapidly , that it does not reflect all the key present - day ideas .
a short review by lazarian & hoang ( 2009 ) concentrates on the recent developments in the field .
however , both reviews are focused on how understanding of grain alignment improves magnetic field tracing . in the paper below we consider , in addition , how quantitative understanding of grain alignment can shed light onto grain composition and grain environment .
i would like to stress that grain alignment is a genetic property of astrophysical grains which applies not only to dust in interstellar gas and molecular clouds , but also to dust in accretion disks , agn environments , circumstellar regions , solar system etc .
attached to it ( @xmath1 is chosen to be @xmath2 in the standard lh07 model ) .
the `` scattering coordinate system '' which illustrates the definition of torque components : @xmath3 is directed along the maximal inertia axis of the grain ; @xmath4 is the direction of radiation .
the projections of normalized radiative torques @xmath5 , @xmath6 and @xmath7 are calculated in this reference frame . from lazarian & hoang 2007a.,scaledwidth=70.0% ] dolginov & mytrophanov ( 1976 ) were the first to propose an analytical model to describe rats .
however , their analytical predictions for the model of a twisted grain obtained using raylegh - hans approximation are in error , as they are inconsistent with the ddscat numerical simulations ( see hoang & lazarian 2009b ) .
lazarian & hoang ( 2007a , henceforth lh07 ) assumed that the most important property is grain helicity and this property should carry over to a geometric optics limit .
therefore they suggested a simple model of a helical grain that was shown by ddscat calculations to reproduce well the essential properties of rats acting on irregular grains .
the model is shown in fig .
[ amo1 ] and it consists of an ellipsoidal body with a mirror attached to its side . lh07 termed the model amo , which is an abbreviation from analytical model .
note , that grains can be both `` left - handed '' and `` right - handed '' . for our grain model to become `` right handed '' the mirror should be turned by 90 degrees .
our studies with ddscat confirmed that actual irregular grains also vary in handedness . a substantial difference in the rats acting on right and left handed irregular grains was a source of earlier confusion . to describe torques
acting of amo and irregular grains lh07 chose a system of reference with the direction of light along a vector @xmath8 , and the grain axis of maximal inertia moment @xmath3 being in the @xmath8 , @xmath9 plane ( see figure[amo1 ] ) .
the latter condition ensures that the torques calculated in the @xmath10 reference frame do not change as the grain precesses around @xmath4 .
if we recall that the equation for the changes of the angular momentum for a top is @xmath11 , it is easy to see that angle @xmath12 and angular velocity of amo depends only on the torque components @xmath5 and @xmath6 .
the third component @xmath7 induces grain precession only . in the absence of magnetic field
, this would induce the direction of the beam to serve as the alignment direction , but in most cases the precession induced by @xmath7 is subdominant to the larmor precession induced by the ambient , e.g. interstellar , magnetic field ( see more discussion in lh07 ) .
interestingly enough , the conclusion of @xmath7 is not important in terms of the rat alignment is also true for the presence of thermal fluctuation ( see hoang & lazarian 2008 ) and inefficient internal relaxation ( see hoang & lazarian 2009b ) when the alignment of angular momentum and axis @xmath3 is not enforced .
naturally , this component can not produce the rat alignment , as the helicity of an ellipsoidal grain is zero . ] .
this observation allowed lh07 to simplify the problem and consider only two torque components , namely , @xmath5 and @xmath6 instead of three rat components . for amo and irregular grains . for the latter rats
were obtained via ddscat calculations .
images of the irregular grains , namely , shapes 2 and 4 are also shown in the figure .
both irregular grains and amo have left helicity . from lazarian & hoang 2007a.,scaledwidth=60.0% ] the functional dependences of torques @xmath13 and @xmath14 , where @xmath12 is an angle between the axis @xmath15 and the radiation direction , were shown to be very similar for the analytical model in fig .
[ amo1 ] and irregular grains subject to radiation of different wavelengths . in figure [ amo2 ]
this correspondence is shown for two irregular grains ( shape 2 and shape 4 in lh07 ) and amo .
this remarkable correspondence is further quantitatively illustrated in fig .
[ chi ] using a function : @xmath16^{2 } d\theta , \label{chi_eq}\ ] ] which characterizes the deviation of the torques @xmath6 calculated numerically for irregular grains from the analytical prediction in the lh07 model . while the functional dependence of torque components @xmath13 and @xmath14 coincides for grains of various shapes , their amplitudes vary for different grains and different radiation wavelengths .
in fact , lh07 showed that the radiative torque alignment can be fully determined if the ratio @xmath17 is known . in terms of practical calculations ,
this _ enormously _ simplifies the calculations of radiative torques : instead of calculating two _ functions _ @xmath13 and
@xmath14 it is enough to calculate just two _ values _
@xmath18 and @xmath19 .
according to lh07 the maximal value of the function @xmath13 is achieved for @xmath20 of the function @xmath14 is achieved at @xmath21 . in other words , one can use a _ single number _
@xmath22 instead of _ two functions _ to characterize grain alignment .
thus , it is possible to claim that the @xmath23-ratio is as important for the alignment as the grain axis ratio for producing polarized radiation by aligned grains .
is defined by eq .
( [ chi_eq ] ) . from lazarian & hoang 2007a .
, scaledwidth=80.0% ] studying the rat alignment lh07 corrected the treatment of grain dynamics in draine & weingartner ( 1997 ) . as a result , instead of cyclic trajectories in the latter paper
, lh07 reported the situation when , instead of accelerating grains , rats were slowing grains down .
this slowing down was previously reported in weingartner & draine(2003 ) , but erroneously attributed to the effect of thermal fluctuations . on the contrary ,
lh07 had the same set up as in draine & weingartner ( 1997 ) , i.e. without any thermal fluctuations .
therefore , the effect of braking grain rotation by rats , irrespectively of any other factors , was established .
this effect happened to be very important for the alignment attractor point , in agreement with an earlier empirical study in weingartner & draine ( 2003 ) . ] .
thus , apart from high attractor points , lh07 found that for a range of @xmath23 and the angle @xmath24 between the radiation beam and magnetic field direction only low-@xmath25 attractor points exist ( see figure [ space ] ) .
a later study by hoang & lazarian ( 2008 ) established that when low-@xmath25 and high-@xmath25 attractor points coexist , the high-@xmath25 points are more stable and therefore an external stochastic driving , e.g. arising from gaseous bombardment , brings grains to high-@xmath25 attractor points .
this transfer can take several damping times which may have observational consequences for the alignment in the presence of varying sources of radiation , e.g. supernovae .
however , for the steady - state interstellar alignment for the parameter space for which a high-@xmath25 attractor point exists ( see figure [ space ] ) , one can safely assume that grains are aligned with high-@xmath25 .
as the degree alignment at high-@xmath25 attractor point is higher than at a low-@xmath25 attractor point , this is a very peculiar effect of improving alignment through random gaseous bombardment ! attractor point and both low-@xmath25 and high attractor point .
@xmath24 is the angle between the direction towards a point radiation source and magnetic field . in the situation
when the high-@xmath25 attractor point is present grains eventually get there and demonstrate perfect alignment .
in the situation when only low-@xmath25 attractor point is present , the alignment is partial . from lazarian & hoang 2007a .
, scaledwidth=80.0% ] when does the alignment happens at low-@xmath25 attractor points ?
figure [ space ] shows predictions for the existence of low-@xmath25 and high-@xmath25 attractor points for the analytical model ( amo ) for the parameter space given by @xmath23 and the angle @xmath24 .
individual horizontal lines correspond to particular grain shapes with a given @xmath23 . for the interstellar radiation field ( isrf )
the calculations of @xmath23 are performed using torques averaged over the interstellar spectrum of wavelengths ( see lh07 for more details ) .
we see that the correspondence in terms of predicting the distribution of high-@xmath25 and low-@xmath25 attractor points is also good , which , however , is not surprising due to the good correspondence between the functional dependences obtained for the amo and irregular grains depicted in figure [ amo1 ] .
further research in hoang & lazarian ( 2009b ) revealed that when the radiation arises not from a point source but the radiation field has a complex structure defining parameter spaces for low-@xmath25 and high-@xmath25 attractor points is possible if the radiation field is decomposed into multipoles , e.g. dipole and quadrupole .
that characterize the radiative torque alignment of grains depends on both grain shape and the wavelength of radiation . from lazarian & hoang 2007a .
, scaledwidth=80.0% ] the analytical description of rats in amo does substantially simplify modeling of the rat alignment .
however , modeling of the polarization arising from grains with the infinite variety of shapes and for various spectra may still present a problem . the approaches to handling this problem
are summarized below . as the torques @xmath5 and @xmath6 vary with the wavelength this results in the change of @xmath23 ( see figure [ wavelength1 ] ) .
the figure reveals systematic changes of @xmath23 that can be used in future for developing models of polarization for interstellar grains .
the magnitude of the radiative torques also changes in a systematic way as shown in figure [ wavelength2 ] .
this fact was already used to simplify calculations of grain alignment in t - tauri disks in cho & lazarian ( 2007 ) .
note that the calculations for low @xmath26 get not reliable with ddscat code that was employed .
ray tracing techniques would be much more appropriate for such calculations .
large grains are known to exist in circumstellar accretion disks and their alignment is important .
our studies of @xmath23 are limited to a handful grain shapes . to make modeling more reliable it is important to perform more ddscat studies of more grain shapes in order to reveal the statistics of @xmath23 .
naturally , obtaining the statistics of @xmath23 is a much more simple task compared to dealing with the multitude of functions that can be obtained via numerical calculations of radiative torques . .
however , the alignment of grains substantially smaller than the radiation wavelength can also be present provided that the radiation is strong enough . from lazarian & hoang 2007a .
, scaledwidth=80.0% ]
observations testify that interstellar grains tend to get aligned with long axes perpendicular to the ambient magnetic field , the fact that was frequently used to argue that the davis - greenstein ( 1951 ) mechanism is responsible for the alignment . can rats explain this observational fact ?
we note , that other mechanisms , e.g. mechanical alignment of oblate and prolate grains , i.e. gold ( 1951 ) alignment , align grains both perpendicular and parallel to magnetic field , depending on the direction of the gaseous flow in respect to magnetic field .
the direction of alignment was not an issue in the original draine & weingartner ( 1996 ) study , as there the alignment was assumed to be paramagnetic , while the radiative torques were associated only with the grain spin - up . as we discussed above , this is not what is going on and rats do produce the alignment of their own . using the amo , one can show that the rats acting on their own , without any effect of paramagnetic relaxation , tend to align grains the `` right way '' , i.e. in agreement with observations .
the exception is a narrow range of angles when the light beam direction is nearly perpendicular to the direction of magnetic field . in the latter situation
the alignment could be `` wrong '' , i.e. with long grain axes parallel to magnetic field . as the analysis testifies that the `` wrong '' alignment happens with low angular momentum , the thermal fluctuations cause wobbling of grain axes about grain angular momentum ( lazarian 1994 , lazarian & roberge 1997 ) .
the angular amplitude of this wobbling for any reasonable grain temperature , e.g. @xmath28 , exceeds the range of the beam angles for which the alignment is `` wrong '' ( lh07 , hoang & lazarian 2008 ) .
thus a remarkable fact emerges : grains get _ always _ aligned with long axes perpendicular to @xmath27 !
is it possible to understand this on a more intuitive level ?
below we try to provide an explanation why the situation when @xmath29 is aligned with magnetic field @xmath27 is special . for the sake of simplicity
, we disregard grain wobbling and assume that due to internal relaxation of energy @xmath29 is perfectly aligned with the axis of maximal inertia .
therefore , it is sufficient to follow the dynamics of angular momentum to determine grain axes alignment .
parallel ( or anti - parallel ) to @xmath27 correspond to the stationary points as at these positions the component of torques that changes the alignment angle vanishes .
as internal relaxation makes @xmath29 aligned with the axis @xmath30 of the maximal moment of grain inertia , the grain gets aligned with long axes perpendicular to @xmath27.,scaledwidth=50.0% ] as we discussed earlier , only two components of the torque , namely @xmath5 and @xmath6 are important for the processes of alignment .
the choice of scattering frame axes and the perfect alignment of @xmath29 with the axes of maximal moment of inertia ensures that @xmath29 is in the @xmath31 plane .
torques @xmath5 and @xmath6 change both the direction and the amplitude of @xmath29 .
in this situation it is convenient to present the action of these torques in the reference system with one of the axes directed along @xmath29 and the other axis being perpendicular to @xmath29 ( still in the @xmath31 plane ) . let us denote the component of torque parallel to @xmath29 by @xmath32 and perpendicular to @xmath29 by @xmath33 .
it is evident that @xmath32 torque spins the grain up or down and the torque @xmath33 aligns the grain .
figure [ alignment ] illustrates the process of alignment .
the angular momentum @xmath29 is precessing about magnetic field @xmath27 due to the magnetic moment of a grain .
the alignment torques @xmath33 are perpendicular to @xmath29 and therefore as @xmath29 gets parallel to @xmath27 the fast precession of the grain makes the torques averaged over @xmath29 precession vanish as @xmath34 .
thus the positions corresponding to @xmath29 aligned with @xmath27 are stationary points , irrespectively of the functional forms of radiative torques , i.e. of components @xmath13 and @xmath14 . in other words , grain do not experience aligning torques when @xmath35 or @xmath36 . whether these stationary points are stable , i.e. attractor points , or unstable , i.e. repellor points , it is impossible to say unless the functional form of torques is given . thus , on the intuitive level , one can understand why grains may get _ perfectly _ aligned with @xmath29 parallel to @xmath27 , but quantitative predictions of the alignment degree can be only obtained with the help of amo .
our earlier discussion was focused on ordinary paramagnetic grains , for which the effect of paramagnetic relaxation is absolutely negligible compared to the effect of rats .
we also neglected the possible action of pinwheel torques , which may arise , for instance , from h@xmath0 formation over grain catalytic sites , as it is discussed by purcell ( 1979 ) .
in addition , we assumed that internal relaxation of energy is sufficiently strong to induce the alignment of the grain axis of maximal moment of inertia with @xmath29 .
these were the simplifying assumptions of lh07 study .
below we consider special situations when one of these conditions is not true .
future observations should determine how special or generic the superparamagnetic response of the grains and the existence of strong pinwheel torques .
superparamagnetic grains , i.e. grains with enhanced paramagnetic relaxation , were invoked by jones & spitzer ( 1967 ) within the model of paramagnetic alignment ( see also mathis 1986 , martin 1995 , goodman & whittet 1994 , roberge & lazarian 1999 ) . what does happen when the dynamics of grains is determined by rats ?
we see from fig .
[ space ] that for a substantial part of the parameter space grains are driven to the low-@xmath25 states , i.e. _ subthermally _ , which is in contrast to a widely spread belief that in the presence of rats most of the interstellar grains must rotate at @xmath37 .
attractor point .
for the same set of parameters a superparamagnetic grain gets also a high-@xmath25 attractor point .
high-@xmath25 attractor points are more stable than the low-@xmath25 attractor points . as a result
, all grains eventually end up at the high-@xmath25 attractor point . from lazarian & hoang ( 2008).,scaledwidth=70.0% ] the picture above , however , is different if superparamagnetic grains dissipate ( via davis - greenstein process ) rotational energy on the time scales shorter than the gaseous damping time .
lazarian & hoang ( 2008 ) found that trajectory maps of grains with superparamagnetic inclusions _
exhibit high-@xmath25 attractor points . fig .
[ superparamagnetic ] shows that for superparamagnetic grains subject to a diffuse interstellar radiation field most grains still get to the low attractor point .
as the high-@xmath25 attractor point is more stable compared to the low-@xmath25 attractor point superparamagnetic grains get transfered by gaseous collisions from the low-@xmath25 to high-@xmath25 attractor points .
thus , superparamagnetic grains always _ rotate at high rate _ in the presence of rats and their alignment is _
perfect_. pinwheel torques were considered by purcell ( 1979 ) in the context of paramagnetic alignment .
for instance , the action of active sites forming @xmath38 molecules can be similar to the action of tiny rocket engines spinning up the grain .
how do these torques also affect the rat alignment ?
hoang & lazarian ( 2009a ) showed that the sufficiently strong pinwheel torques can create new high-@xmath25 attractor points ( see figure [ suprathermal ] ) .
therefore for strong pinwheel torques arising from h@xmath0 formation , one may observe the correlation of higher degree of polarization with the atomic hydrogen content in the media , provided that h@xmath0 torques as strong as in purcell ( 1979 ) and the subsequent papers ( see spitzer & mcglynn 1979 , lazarian 1995 , lazarian & draine 1997 ) .
the implicit assumption for observing this correlation is , however , that the grains are not superparamagnetic . for superparamagnetic grains the alignment ,
as we discussed above , is perfect anyhow .
and high-@xmath25 attractor points in the absence of pinwheel torques . in the case
when only a low-@xmath25 attractor point exists the strong pinwheel torques lift the low-@xmath25 attractor point enhancing the alignment . from hoang & lazarian 2009a.,scaledwidth=70.0% ] it may be shown ( see lazarian & hoang 2009 ) that for sufficiently large grains the internal relaxation over the time - scale of rat alignment is negligible .
large grains , however , are present e.g. in accretion disks and comets . the polarization from accretion disks can provide one with an important insight into magnetic of the objects ( see cho & lazarian 2007 ) , while understanding of the alignment for comet dust is important for explaining circular polarization observed ( see below ) .
thus , a proper description of rat alignment in the absence of internal relaxation is important . in the absence of internal relaxation grains
can get aligned not only with long axes perpendicular to the angular momentum @xmath29 , but also with the longest axis parallel to @xmath29 , i.e. the axis of minimal moment of grain inertia parallel to @xmath29 .
this complicates the analysis compared to the case of interstellar grains , for which the internal relaxation is very fast .
the corresponding problem was addressed in hoang & lazarian ( 2009b , henceforth hl09 ) .
the results of the latter study are summarized in table 1 .
llcll + & & + + * high - j * & * low - j * & * high - j*&*low - j * + & & & & + @xmath39 & @xmath40 @xmath41 or at & @xmath42&@xmath40 @xmath41 or at + & angle with @xmath43&&angle with @xmath43 + l @xmath44 & l @xmath45 or @xmath41 @xmath40 & l @xmath44&l @xmath44 + + + the study of hl09 is suggestive that grains do preferentially get aligned with long axes perpendicular to magnetic field even without internal relaxation . indeed , this is consistent with the finding there that such alignment happens for the high-@xmath25 attractor points of amo .
if this is the case , when the low-@xmath25 and high-@xmath25 attractor points coexist the steady - state alignment will happen only with high-@xmath25 attractor point ( see discussion in [ magnetic ] ) .
the `` wrong '' alignment happens only with low-@xmath25 attractor points and it is suggestive that in the presence of gaseous bombardment the grains may still spend more time in the vicinity of the high-@xmath25 repellor point , as was shown in numerical simulations by hoang & lazarian ( 2008 )
. the conclusions obtained with amo are consistent with a limited parameter study obtained in hl09 for an irregular grain .
however , it is clear that more extensive studies of the rat alignment in the absence of internal relaxation are required .
these studies are of practical interest for modeling alignment in circumstellar accretion disks , where large grains with slow rates of internal are known to be present .
the predictions that the alignment is always perfect for superparamagnetic grains opens ways to testing this hypothesis using polarimetry . for instance
, the variations of the polarization with the angle between the direction of light beam and magnetic field in the situations that rats are strong would reveal that grains are not superparamagnetic .
similarly , variations of the degree of grain alignment with fraction of atomic hydrogen in their environments would testify that the purcell s pinwheel torques arising from h@xmath0 formation strong and , simultaneously , that grains are not superparamagnetic . on the contrary , statistical studies indicating that the variations of the polarization arise only from the structure and the orientation of magnetic field in respect to the line of sight , while the grains are always perfectly aligned , would testify that grains are superparamagnetic .
the structure and orientation of magnetic field can be tested by the technique proposed in falceta - gonzalvez et al .
additional information on the orientation of 3d vector of magnetic field can be obtained using a new technique of atomic alignment polarimetry ( yan & lazarian 2006 , 2007 , 2008 ) .
note , that by increasing the sample of observations or using the observations when the relative orientation of magnetic field to the sources of orientation is known , e.g. for surcumstellar regions , it is possible to place stringent constraints on the degree of grain alignment . while the confirmation of the evidence of the strength of h@xmath0 formation torques may give insight into the intimate details of the formation of molecular hydrogen , the consequences of testing of supermagnetic nature of interstellar grains are much broader .
first of all , if grains are shown to be superparamagnetic , the modeling of polarization from them can be much simplified as assuming perfect alignment whenever the rats are strong enough will be possible ( see cho & lazarian 2005 , 2007 , pelkoen et al .
2007 , bethell et al . 2007 ) . moreover , superparamagnetic grains emit polarized magneto - dipole emission in the microwave range ( draine & lazarian 1999 ) , which can interfere with the attempts to measure cosmic microwave background polarization .
one should keep in mind that if proven that interstellar grains are not superparamagnetic this does not preclude grains in other environments to be superparamagnetic .
future research should make use of the predictions of the quantitative theory of grain alignment to study not only magnetic fields , but also the environment and the composition of grains .
abbas , m. m. et al .
2004 , apj , 614 , 781 bethell , t. j. , chepurnov , a. , lazarian , a. , & kim , j. 2007 , , 663 , 1055 cho , j. & lazarian , a. 2005 , apj , 631 , 361 cho , j. , & lazarian , a. 2007 , apj , 669 , 1085 davis , l. , greenstein j.l .
1951 , apj , 114 , 206 dolginov , a.z .
, mytrophanov , i.g .
1976 , ap&ss , 43 , 291 davis , l. & greenstein , j.l . , 1951 , apj , 114 , 206 draine , b.t . &
lazarian a. 1999 , apj , 512 , 740 draine , b.t . & weingartner , j.c .
1996 , apj , 470 , 551 draine , b. weingartner , j. 1997 , apj , 480 , 633 falceta - gonalves , d. , lazarian , a. , & kowal , g. 2008 , apj , 679 , 537 gold , t. 1951 , nature , 169 , 322 goodman , a. , & whittet , d. c. b. 1995 , apj , 455 , l181 hall , j. 1949 , science , 109 , 166 harwit , m. 1970 , nature , 226 , 61 hiltner , w.a . 1949 , apj , 109 , 471 jones , r.v .
, & spitzer , l.,jr , 1967 , apj , 147 , 943 lazarian a. 2003 , journ . of quant .
spectroscopy & radiative transfer , 79 - 80 , 881 lazarian , a. 2007 , journ . of quant .
spectroscopy & radiative transfer , 106 , 225 lazarian , a. , & draine , b.t . , 1997 ,
apj , 487 , 248 lazarian , a. , & draine , b.t.,1999a , apj , 516 , l37 lazarian , a. , & draine , b.t.,1999b , apj , 520 , l67 lazarian , a. , & hoang , t. 2007a , mnras , 311 , 903 ( lh07 ) lazarian , a. , & hoang , t. 2007b , apjl , 669 , l77 lazarian , a. , & hoang , t. 2008 , apjl , 676 , l25 lazarian , a. & hoang , t. 2009 , arxiv:0901.0146 lazarian , a. , goodman , a.a . ,
myers p.c .
1997 , apj , 490 , 273 weingartner j. , & draine b. 2003 , apj , 589 , 289 whittet , d.c.b . , gerakines , p.a . ,
hough , j.h .
& snenoy 2001 , apj , 547 , 872 whittet d.c.b , hough j.h , lazarian a. , hoang t. 2008 , apj , 674 , 304 yan , h. & lazarian , a. 2006 , apj , 653 , 1292 yan , h. & lazarian , a. 2007 , apj , 657 , 618 yan , h. & lazarian , a. 2008 , apj , 677 , 1401 | while the problem of grain alignment was posed more than 60 years ago the quantitative model of grain alignment that can account for the observed polarization arising from aligned grains has been formulated only recently .
the quantitative predictions of the radiative torque mechanism , which is currently accepted as the dominant mechanism of grain alignment , open avenues to tracing magnetic fields in various astrophysical environments , including diffuse and dense interstellar gas , molecular clouds , circumstellar environments , accretion disks , comet tails , zodiacal dust etc . at the same time
, measurements of the absolute value of polarization and its variations can , in addition , provide unique information about the dust composition and dust environment . in the review
i describe the analytical model describing well radiative torques acting on irregular grains and discuss how the alignment induced by radiative torques varies in the presence of superparamagnetic inclusions and pinwheel torques , e.g. arising from the h@xmath0 formation over grain surface .
i also describe observations that can establish whether grains are superparamagnetic and whether recoils from h@xmath0 formations are powerful enough to give rise to substantial uncompensated torques . answering to these questions
should allow for reliable modeling of astrophysical polarization with numerous important applications , from accounting for dust contribution in cosmic microwave background polarization studies to obtaining magnetic field strength using chandrasekhar - fermi technique . |
it is now well established that super - massive black holes ( smbhs ) reside in the nuclei of many galaxies .
the masses of these smbhs correlate with a range of properties of their host spheroids , including luminosity ( e.g. * ? ? ?
* ) and consequently mass ( e.g. * ? ? ?
* ; * ? ? ?
* ) ; the concentration of stellar bulge light ( e.g. * ? ? ?
* ) ; the deficit of stellar bulge light in ellipticals ; and the gravitational binding energy of the bulge . the correlation between smbh mass ( @xmath2 ) and the velocity dispersion ( @xmath1 ) of its host spheroid , referred to below as a `` @xmath3@xmath1 relation '' , has been studied by many authors .
a power - law fit to the data , @xmath4 , yielded values of @xmath5 in the range of @xmath6 ( cf . * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and is now widely interpreted as the imprint of the self - regulated growth of the smbh .
a more recent extended sample of galaxies shows that the power law index @xmath5 may be as high as @xmath7 @xcite .
furthermore , recent work has highlighted that the @xmath2@xmath1 relation one measures will depend on the nature of the galaxy sample .
for example , @xcite presents two extremely massive smbh , @xmath8 , in the centres of two giant elliptical galaxies .
combined with other previous smbh mass measurements , these authors obtain @xmath9 for elliptical and s0 galaxies .
a similar but offset to lower smbh masses result is found for spiral galaxies in the sample .
alternatively , when data for all galaxies are combined in one @xmath2@xmath1 relation , @xmath10 is found .
another important conclusion is gradually emerging from the observations : smbhs in galaxies that show flattened morphology , such as discs , pseudobulges or bars appear to be less massive at same @xmath1 than their counterparts in elliptical galaxies .
in particular , @xcite showed that pseudobulge galaxies tend to have undermassive smbhs , and @xcite showed that barred galaxies tend to have smbhs that are a factor of @xmath11 to ten smaller than elliptical galaxies at the same velocity dispersion . while @xcite , in variance with other authors , find that smbhs in pseudobulges and galaxy discs do _ not _ even follow an @xmath12@xmath1 relation , their results are consistent with that of the other authors in that most of their pseudobulge systems lie below the @xmath2@xmath1 for ellipticals .
@xcite firmed up these conclusions with a sample of narrow line seyfert 1 nuclei ( all classifiable as pseudobulge systems ) .
a useful compilation of the recent data , summarising these points , is given in the right panel of fig . 1 in @xcite .
the purpose of our paper is to show that there is a unified theoretical picture within which this observational trend can be understood , and that the observed @xmath2@xmath1 relations place very interesting constraints on the still unknown feeding mode of smbhs .
a well known difficulty in fuelling active galactic nuclei ( agn ) is that the typical angular momentum of gas in the galactic bulge is very large compared with that of the last stable orbit around a black hole ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
one expects that due to circularisation shocks , gas will end up in a planar feature , such as a disc or a ring ( cf .
simulation s30 shown in figures 1 and the left panel of figure 13 in * ? ? ?
it has been argued that large scale angular momentum transfer occurs rapidly enough through the action of stellar and gaseous bars ( e.e .
* ) or via spiral density waves @xcite to overcome this difficulty .
these galaxy - wide gas discs ( with radial scales of many kpc ) could then extend all the way to the inner galaxy and even feed the smbh .
this `` planar mode '' of feeding is schematically illustrated in the left panel of figure 1 . however , gas discs tend to become unstable to gravitational fragmentation and resulting star formation ( sf ) beyond a self - gravity radius of @xmath13 pc ( e.g. , * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in the simplest model this would result in complete consumption of gas in the disc by star formation @xcite and so there would be no way to fuel the smbh . however , stellar feedback could heat up the large scale disc to self - regulate the star formation rate in the disc to a level such that a sufficient amount of gas can reach the smbh , but there is no clear consensus on this point ( e.g. , * ? ?
* ; * ? ? ?
an alternative solution to the difficulty of fuelling agn is provided by chaotic feeding @xcite or ballistic accretion of cold streams when the gas is cold enough to form dense clouds or filaments @xcite .
large scale cosmological simulations show that cold streams may be important for supplying relatively low angular momentum gas to galaxies(e.g .
* ; * ? ? ?
* ; * ? ? ?
* ) , while higher resolution simulations that resolve the regions nearer to smbh suggest that similar processes may operate on smaller scales @xcite . in particular , the numerical experiments of @xcite show that turbulence ( driven by stellar feedback in the bulge ) in a quasi - spherical distribution of gas leads to formation of convergent flows that create high density filaments .
the latter can travel almost ballistically through the rest of the bulge , with some filaments arriving in the inner parsecs of the galaxy .
note that most of the gas still ends up in a galaxy - scale disc with a large angular momentum ; however , the smbh receives most of its fuel not from that disc but from the gas filaments arriving at its vicinity directly .
this `` chaotic cold streams '' feeding mode is schematically shown in the right panel of figure 1 . in this class of model the orientation of the innermost sub - pc scale disc
makes a random walk because of the arrival of material with a fluctuating angular momentum .
the cancellation of the angular momentum in shocks transfers gas into the smbh more rapidly than viscous torques can , and so these discs should be more resilient to star formation . furthermore , because the orientation of the inner disc is constantly changing , feedback from the agn will be quasi - isotropic .
+ the main point of our paper can be summarised as follows .
large scale gas discs are better able to withstand agn feedback than quasi - spherical inflows @xcite .
this is so even under the most optimistic assumptions about agn feedback that is , feedback acting directly on the disc without being shadowed by the inner disc , and no beaming away from the disc / galaxy plane .
we show that , under these most favourable of conditions for which agn feedback is maximal , there is a critical smbh mass at which its feedback expels the disc that feeds it .
we call this a `` disc @xmath14@xmath1 relation '' because it arises only for smbhs accreting from large scale galaxy discs . this mass is larger than the canonical @xmath14@xmath1 relation , and any dilution of agn feedback would increase it even further .
theoretically we see that if smbhs were fed by planar accretion then we could expect smbhs in pseudobulge systems to lie above the @xmath14@xmath1 relation for classical bulges .
observationally there is no strong evidence for this ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and so we conclude that most smbhs do not grow via the planar accretion mode but are much more likely to be fed by quasi - spherical inflows as in the models of @xcite and @xcite .
to avoid misunderstanding , we note that galaxy - scale gas discs and gas bars are obviously important for galaxy evolution as a whole , e.g. , as a birthplace for a significant fraction of all stars , and also by effecting a radial redistribution of gas and stars within the galaxy , but we see no clear observational evidence that these features are able to channel their gas all the way to the smbhs .
we briefly review existing ideas of agn feedback acting on quasi - spherical distributions of gas before moving on to the problem of agn feedback acting on a gas disc . @xcite assumed feedback in the form of an energy - conserving outflow and obtained a scaling @xmath15 .
subsequently , @xcite , @xcite and @xcite solved the momentum equations for momentum - conserving agn feedback acting on a spherical shell of gas in an isothermal bulge potential ( see below ) and showed that it leads to a @xmath2@xmath1 relation @xmath16 where we follow king s analysis ; here @xmath17 is the opacity , which is assumed to be dominated by the electron scattering , and @xmath18 is the baryon fraction , which is assumed to be equal to the initial cosmological value of @xmath19 ( cf . * ? ? ? * ) .
we note that all of the aforementioned theoretical approaches used a singular isothermal sphere potential ( e.g. , 4.3.3b in * ? ? ?
* ) for simplicity .
for such a potential , the one dimensional velocity dispersion is a constant independent of radius , @xmath20 , where @xmath21 is the total enclosed mass including dark matter inside radius @xmath22 ( the distance from the centre of the galaxy ) .
the enclosed gas mass , @xmath23 is proportional to @xmath22 . the gas density at radius @xmath22 for such a potential is @xmath24 for reference .
@xcite assumed that the smbh luminosity is limited by the eddington value @xmath25 , and the momentum outflow rate produced by radiation pressure @xcite is of order @xmath26 we can recover the result of @xcite using a simpler `` weight argument '' , which requires that the momentum output produced by the smbh ( @xmath27 ) should just balance the weight of the gas in the bulge , @xmath28/ r^2 $ ] , which turns out to be independent of radius : @xmath29 to order of magnitude , equation [ w ] holds for any potential if estimated at a radius which encloses most of the gas potentially available for smbh fueling , e.g. , such as the virial radius for the @xcite potential .
balancing the outward force of the outflow ( equation [ pismbh ] ) with the weight of the gas in the bulge ( equation [ w ] ) then leads naturally to equation [ msig ] .
the momentum feedback model is attractive in its physical simplicity .
the model is underpinned by observations of fast @xmath30 outflows from bright agn and quasars . studying the blue shifted ionised absorption features in the 7 to 10 kev energy range in a large sample of agn with _ xmm - newton _ , @xcite concluded that such fast agn outflows appear in @xmath31 % of the sample .
as these authors emphasise , their uniform sample overcomes `` publication bias '' and shows that fast agn outflows are widespread .
it also requires the outflows to be wide - angled ( rather than jet - like ) , which is beneficial in spreading the influence of agn feedback as broadly over the galaxy bulge as possible .
equation [ msig ] contains no free parameters , but nevertheless it is very close to the observed @xmath2@xmath1 relation .
similar reasoning helps to explain both the observed @xmath2@xmath1 relation for nuclear star clusters ( see * ? ? ?
* ) and the observation that ncs are preferentially found in low mass ( low @xmath1 ) galaxies @xcite . the model has been used recently by @xcite to explain the two @xmath3210 kpc scale symmetric lobes apparently filled with cosmic rays detected by _ fermi - lat _ telescope in the milky way @xcite . to explain the particular geometry of the bubbles ,
the only adjustment to the basic @xcite model required by @xcite has been an addition of a dense disc of molecular gas , known as the central molecular zone , found in the central @xmath32 200 pc of our galaxy @xcite .
while it is clear that the exact @xmath33 relation derived analytically is affected by the choice of the potential , we note that the main conclusion of our paper the fact that it is harder to expel a disc feeding the smbh rather than a quasi - spherical shell remains unchanged as it is based on a simple geometrical argument ( see below ) . furthermore , gravitational potentials of elliptical and lenticular galaxies appear to be relatively well approximated by a constant @xmath1 value inside their effective radii .
@xcite find that the 1d velocity dispersion of stars within the galaxies in their sample depends on radius @xmath22 ( within the galaxy ) as a very weak power - law : @xmath34 .
this implies that @xmath1 drops on average by the factor of @xmath35 when @xmath22 changes by an order of magnitude , and @xmath36 varies by the factor of @xmath37 . indeed ,
@xcite consider smbh momentum feedback for potentials other than isothermal , e.g. , the nfw potential @xcite , @xcite and @xcite potentials . in all cases momentum
driven outflow produce relations of the form @xmath38 , where @xmath39 is the maximum circular velocity in the potential .
the normalisation of the relation is slightly smaller than for the simpler isothermal potential we use here , which implies that escape from the isothermal potential is slightly more demanding than for all the others . to quote a numerical factor here ,
@xcite find a change in the critical smbh mass ( only ) by a factor of order unity only for the nfw potential , as an example , if one defines @xmath1 at the peak of the rotation curve of the potential as @xmath40 .
no analytical treatment for the complex problem at hand can be expected to be accurate within better than a factor of @xmath41 , and therefore we feel that the singular isothermal potential is an appropriate approximation to use here .
as shown by @xcite , the above argument does not quite work for a non - spherical geometry . to see this ,
consider the simplest example an axially symmetric rotating galaxy .
the gas settles in the plane of symmetry of the galaxy and forms a rotationally supported disc .
let @xmath43 be the local ( i.e. , at a given @xmath22 ) vertical disc scale - height .
let us consider a disc at radius @xmath22 .
the _ maximum _
momentum flux from the smbh outflow striking the disc is @xmath44 this is an absolute maximum given by the total momentum outflow rate ( equation 2 ) times the fraction of the solid angle the disc subtends as seen from the smbh location . in reality some of the momentum outflow from the agn could be intercepted ( shadowed ) by the disc at smaller radii ( cf .
the left panel of figure 1 ) , and also the outflow may be beamed along the direction perpendicular to the disc .
these individual effects actually strengthen the conclusion we draw because they act to lessen the effect of feedback on the disc and therefore should increase the mass of the smbh .
we first assume that the disc mass is dominated by gas , and we correct for the presence of stars * within the disc * later on .
the effective weight of the gas disc is given by @xmath45 , where @xmath46 is the disc mass , and @xmath47 is the gas disc surface density , with @xmath48 being the mid - plane disc density .
we note here that a disc in circular rotation actually has a zero weight because gravity is exactly balanced by the centrifugal force .
however , within a factor of a few the effective weight defined above still applies to infinity for a circularly rotating disc is smaller by a factor of @xmath49 than that from a static point within the same potential ] if we wish to expel the disc to infinity ( where the centrifugal force becomes negligible if the angular momentum of the material is conserved ) .
therefore , by requiring that momentum flux balances weight ( @xmath50 ) we limit the smbh mass to @xmath51 a casual look at this equation would seem to suggest that , because @xmath43 and @xmath52 may vary with radius @xmath22 and from system to system , one should not expect any robust correlation between @xmath2 and @xmath36 to arise .
what is missing in the above analysis is the possibility of star formation in the galaxy disc ( this was also not included in the numerical experiments of * ? ? ?
massive cool gas discs become gravitationally unstable when the toomre @xmath53-parameter approaches unity from above @xcite : @xmath54 where @xmath55 is the epicyclic frequency . noting that @xmath56 for the singular isothermal sphere potential , @xmath57 , and using the hydrostatic balance condition @xmath58 , we find that @xmath59 note that @xmath52 , defined just below equation [ pdisc ] , is function of @xmath22 in the above equation .
it is widely believed that self - gravitating gas discs self - regulate their star formation rates to maintain marginally stable states . as a disc becomes self - gravitating
, stars form on a local dynamical time scale ( i.e. , extremely rapidly ) and the energy and momentum released by the stars heat the disc up ; this increases @xmath53 and the star formation rate is reduced to an equilibrium rate that maintains quasi - equilibrium .
generally , the disc * first * becomes unstable when @xmath60 , and therefore we arrive at the constraint @xmath61 using this now in equation [ mmax0 ] , we arrive at the disc @xmath2@xmath1 relation " , @xmath62 where we introduce @xmath63 as the gas mass fraction in the disc to account for the fact that a realistic galaxy disc contains gas and stars as well . according to this definition ,
the fraction @xmath64 of the disc mass is in the recently formed stars or very dense massive molecular clumps on the way to collapsing and forming stars .
we see that the disk @xmath2@xmath1 relation is of exactly same form as the spherical " @xmath2@xmath1 relation ( equation [ msig ] ) derived by @xcite .
however , we note that the coefficient in front of the latter is smaller than that in the former .
the difference is difficult to quantify exactly , but would seem to be at least a factor of a few to ten .
physically , it is due to ( 1 ) the factor @xmath65 in front of the spherical relation and @xmath66 for the disk relation .
it is unlikely that @xmath66 is significantly smaller than unity during the early gas - rich epoch of galaxy formation .
furthermore , in principle a similar in meaning coefficient taking into account that part of the spherical shell is already converted into stars may be introduced in the spherical @xmath2@xmath1 relation , in which case the difference between the two relations would be exactly equal to @xmath18 .
( 2 ) when deriving the disc @xmath67 relation , we took the most optimistic assumptions about the agn feedback on the disc .
as discussed below equation [ pdisc ] , realistically , a part of the feedback can be shadowed by the inner disk ( the innermost part of which is very hard to affect by the feedback , see below ) .
beaming away from the disc plane may decrease the momentum feedback on the disc further yet .
another important point is that although our arguments were presented for a disc geometry , they are also applicable to a bar - mediated gas inflows , such as those discussed by @xcite . these authors argued that `` bars within bars '' may effectively channel the gas into the central regions of galaxies to fuel agn .
however , bars present even less of an effective solid angle than discs for agn feedback to work on : barred gas inflows are squashed geometrically not only to the galaxy midplane , as disc inflows , but also in the azimuthal direction within the galaxy plane . therefore the momentum striking
a bar would be even less than that for the disc ( equation [ pdisc ] ) , and thus we would expect even more massive smbhs if they were limited by their feedback on the gas bars only .
the derivation in this section is applicable to regions well beyond the black hole s influence radius , @xmath68 , which is typically of the order of a few to tens of pc . in the appendix
we show that a gas disc at @xmath69 is actually able to withstand the momentum feedback of the agn .
this is why the sketch in figure 1 shows small scale discs unaffected by the feedback .
@xcite found that smbhs do not correlate with circular velocities of galaxy discs .
@xcite showed that smbhs in pseudobulge galaxies are _ underweight _ by a factor of about ten with respect to their cousins in classical bulge systems .
@xcite showed that barred galaxies tend to have smbhs that are a factor of @xmath11 to ten smaller than elliptical galaxies at the same velocity dispersion .
@xcite found that observed smbhs do not correlate well with velocity dispersions of either galaxy discs or pseudobulges . in the sample of @xcite , only two smbhs in pseudobulge systems
appear to lie above the classical @xmath2@xmath1 relation for bulges , with the rest being somewhat below the relation , in a qualitative agreement with @xcite .
if these observations are confirmed in the future , then this is clearly very strongly at odds with the theory developed above : because discs are difficult to affect by agn feedback simply due to geometrical arguments ( see also simulations by * ? ? ?
* ) , a smbh fed by a disc should be able to grow more massive by at least a factor of a few ( compare equation [ new_msigma ] to equation [ msig ] ) than a smbh embedded in a bulge .
in fact , theoretically one may expect disc smbhs to be even more massive : as noted below equation [ pdisc ] , this is the absolute minimum to which the smbh can grow if fed through the disc because in general agn feedback impacting a given disc radius can be diluted at smaller radii or be beamed away from the disc ( neither of these effects is included in equation [ new_msigma ] ) .
the simplest interpretation of this result is to accept that star formation in large scale discs is too efficient a process and so significant gas flow into the sub - pc vicinity of smbh is suppressed , leading to a quenching of smbh fueling ( see the arguments by * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
therefore , we conclude that these discs do _ not _ extend to the smbh sphere of influence ( between 1 and 10 pc , depending on smbh mass ) , as suggested in the right panel of figure 1 .
@xcite proposed a star - burst disc model which appears to overcome the difficulties noted by @xcite .
specifically , they argue that the star formation rate in the disc is limited by the action of energy and momentum feedback from massive stars and supernovae , allowing some fuel to trickle down all the way to the nucleus . however , one problem with @xcite argument applied to the inner galaxy is a mismatch of time scales . gravitational collapse of the disc is expected to take place on the dynamical time scale @xcite , @xmath70 where @xmath71 is the smbh mass in units of @xmath72 , and @xmath73 is radius in units of parsec .
@xcite show that the disc cooling time is much shorter than the dynamical time in their model . on the other hand , massive stars dominating the feedback release energy on a time scale of a few million years .
therefore , for radii where @xmath74 is much shorter than the lifetime of the massive stars , the feedback might be released too slowly , i.e. , only when the disc would already have collapsed gravitationally . in conclusion
, we believe that the observations of @xcite imply that smbhs do _ not _ grow by a large scale ( e.g. tens of pc to @xmath75kpc ) discs .
surveys show little evidence for correlation between the orientation of jets from agn and the large scale structure of their host galaxies ( * ? ? ?
* ; * ? ? ?
this shows that the inner @xmath76 pc scale gas discs feeding agn are also uncorrelated with the large scale discs of host galaxies . in our own galaxy ,
the observed young stellar discs in the galactic centre @xcite are inclined at very large angles to the galactic plane .
this indicates that the disc stars originated from a deposition of one or two gas clouds with angular momentum directions very different from that of the galactic disc @xcite .
an attractive physical picture for agn feeding that explains these observations and the results of our paper is that smbhs in general are fuelled by stochastic deposition of gas clouds with randomly oriented angular momentum , as sketched in the right panel of figure 1 .
numerical simulations by @xcite demonstrated that this situation may be realised in an initially coherently rotating gas sphere if strong turbulence ( driven by supernova explosions due to star formation in the bulge , for example ) is present .
convergent turbulent gas flows lead to formation of high density regions travelling through the rest of the gas nearly ballistically
. a small fraction of such regions will be on nearly radial orbits that impact the innermost parsecs of smbh directly without going through a large scale disc .
@xcite argued that such flows create small ( pc - scale ) warped discs with a fluctuating orientation .
these discs are able to feed the smbh at much higher rates than the flat discs ( see * ? ? ?
* ) without becoming self - gravitating , effectively due to a much faster transfer of gas through the inner parsecs .
we therefore suggest that the absence of an observed @xmath2@xmath1 relation for pure disc galaxies or pseudobulges provides observational support for the chaotic / ballistic mode of smbh feeding . in our interpretation of the data ,
the lack of classical bulges in these galaxies shut down the most efficient channel by which smbh grow the cold chaotic accretion and therefore led to their smbhs growing mainly by planar ( disk or bar ) accretion which we argue is inefficient .
having said this , we note that it remains to be seen if @xmath33 relation for chaotic streams conforms to the relations obtained by @xcite .
we shall now remark on limitations of our work .
first of all , we only considered the momentum feedback here , whereas more complete treatments show that the smbh outflow switches from momentum driving to energy driving at larger radii @xcite .
however , this is not likely to affect our conclusions significantly because it is the momentum - driven part of the agn feedback process that constitutes the bottle neck for removing the gas to infinity @xcite .
furthermore , numerical simulations of zubovas and nayakshin ( 2012 ) that include both the momentum and energy feedback forms confirm the points made here on the difficulty of expelling a massive dense gas disc for the particular case of the milky way . in common with other analytical papers on the subject ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) we have also used an isothermal potential .
however the discussion in the end of section [ sec : spherical ] shows that this has little effect on momentum driven outflows studied here .
zubovas & king ( 2012 , in preparation ) have extended this result to large scale energy - driven outflows .
a more serious worry is that our theoretical model predicts @xmath77 , whereas different authors now find that @xmath1 ranges from somewhat below @xmath78 to as high as 6 ( cf .
references in the introduction ) .
one possible explanation for this within our momentum - driven feedback model is that the steeper slope for samples that _ combine different types of galaxies _ actually results from a superposition of several @xmath77 relations for different galaxies vertically offset in mass zubovas & king ( 2012 , in preparation ) .
future theoretical and observational work is needed to resolve this issue .
finally , given our smbh feeding focus , we do not attempt to relate the gas discs we consider here to exponential _ stellar _ discs observed in real galaxies .
this would require an additional multi - parameter model for transformation of gas into stars and is beyond the scope of our paper .
previous analytical treatments of the @xmath2@xmath1 relation have modelled the impact of momentum driven agn feedback on a spherical gas distribution ( e.g. * ? ? ?
* ; * ? ? ?
we have shown analytically that if smbh is fed by a large scale gas disc , there exists a `` disc @xmath2
@xmath1 relation '' .
the latter has the same shape as the `` spherical '' @xmath2 @xmath1 relation @xcite but is offset to higher masses by a factor of a few to ten , realistically .
we emphasise that the precise value of the offset is model dependent , but the general trend is undeniable even for agn feedback models not considered here , such as energy - driven models ( e.g. , * ? ? ?
this follows simply from the geometrical argument first emphasised by @xcite : a dense flat disc and even more so a bar presents a much smaller target for agn feedback to act upon than a quasi - spherical gas distribution , and therefore much more massive smbhs could be built if their masses were limited by a feedback argument
. we may further allude here to the analogy from the field of massive star formation , where radiation pressure was once thought to limit stellar growth ; instead , recent simulations by @xcite show that accretion through a disc ( which we note is stable to self - gravitational instabilities on the scales of the simulations ) is very efficient as radiation is channelled away from the disc . if a similar situation applied in galaxy formation , smbhs would grow as long as there is fuel in the massive galaxy - wide discs or bars .
one could then expect smbh becoming comparable in mass to galaxy gas discs themselves , which is clearly observationally not the case .
reconciliation of this theoretical result with observations leads to interesting and significant conclusions about the dominant mode by which smbhs grow in galaxies .
@xcite showed that pseudobulge galaxies tend to have undermassive smbhs , and @xcite showed that barred galaxies tend to have smbhs that are a factor of @xmath11 to ten smaller than elliptical galaxies at the same velocity dispersion . while @xcite argue , in variance with the above authors , that that smbhs do not even correlate with velocity dispersions of galaxy discs and pseudobulges , we note that their results also show that smbhs in pseudobulge systems are typically undermassive with respect to their classical bulge cousins ( e.g. , see fig . 2 in * ?
we suggest that these observations hint that smbhs in most galaxies ( 1 ) do not grow by accretion of gas through @xmath32 kpc gas discs , most likely because gas is turned into stars too efficiently , and ( 2 ) grow via a mode directly linked to existence of a classical quasi - spherical bulge . in this case
it is natural that the smbh in galaxies lacking the classical bulge ( pseudobulge galaxies ) are undermassive , and it is also natural that smbhs do not correlate with galaxy discs . for the smbh accretion modes physically linked to classical bulges we suggest the stochastic @xcite or ballistic @xcite accretion modes in which the bulge is the source of low and fluctuating in direction angular momentum material .
an additional indirect evidence supporting our conclusions comes from recent semi - analytical models of galaxy formation by @xcite .
these authors show that semi - analytical models which turn off smbh feeding during pseudobulge growth episodes driven by secular instabilities ( bars ) provide a much better fit to the data than models that allow a concurrent growth of smbh and pseudobulges ( see fig . 1 in * ? ? ?
we here thus provide a physical interpretation of this empirical result : `` fragmentation catastrophe '' of the outer cold and massive regions of agn discs makes them too susceptible to star formation , so that not enough gas trickles down to the smbh @xcite . to avoid confusion ,
we emphasise that we do _ not _ argue that small ( e.g. , sub - parsec ) scale accretion discs are unimportant for smbh growth . in the chaotic accretion picture , there will always be a non - vanishing amount of angular momentum associated with the accretion flow and this results in the formation of `` small '' by galactic standards discs ( @xmath79 pc , depending on the smbh mass and the accretion rate , cf .
* ) that are resilient to agn feedback ( as we show in the appendix ) .
rather , our key point is that the origin of material that feeds these small discs is distinct from the larger scale ( with respect to the smbh s sphere of influence ) disc or a galactic bar , and the orientation of these smaller - scale agn discs is likely to be chaotic because of the manner in which the quasi - spherical inflows that feeds these discs fluctuates in time .
we also note that large scale galaxy - wide gas discs and gas bars are obviously important for galaxy formation as a whole , likely accounting for a significant fraction of all stars formed and also for a radial redistribution of gas and stars within the galaxy , but we see no clear evidence that these features are able to channel their gas all the way to the smbhs .
theoretical astrophysics research at the university of leicester is supported by a stfc rolling grant .
we thank the anonymous referee whose comments helped to improve this paper significantly .
74 natexlab#1#1 , m. c. & richstone , d. o. 2007 , , 665 , 120 , j. & tremaine , s. 2008 , galactic dynamics : second edition , ed .
binney , j. & tremaine , s. ( princeton university press ) , i. a. & rice , w. k. m. 2008 , science , 321 , 1060 , m. , bacon , r. , bureau , m. , damen , m. c. , davies , r. l. , de zeeuw , p. t. , emsellem , e. , falcn - barroso , j. , krajnovi , d. , kuntschner , h. , mcdermid , r. m. , peletier , r. f. , sarzi , m. , van den bosch , r. c. e. , & van de ven , g. 2006 , , 366 , 1126 , l. , ostriker , j. p. , & proga , d. 2010 , , 717 , 708 , s. & zahn , j. 1999 , a&a , 344 , 433 , f. 2001 , in advanced lectures on the starburst - agn , ed . i. aretxaga , d. kunth , & r. mjica , 223+ , g. , capetti , a. , & marconi , a. 2006 , , 460 , 439 , w. & mclaughlin , d. e. 2005 , , 363 , 1057 , y. , pichon , c. , haehnelt , m. , kimm , t. , slyz , a. , devriendt , j. , & pogosyan , d. 2011 , arxiv e - prints , a. c. 1999 , , 308 , l39 , a. & mele , d. 2005 , international journal of modern physics d , 14 , 1861 . 2007 , international journal of modern physics d , 16 , 1261 , l. , ct , p. , jordn , a. , peng , e. w. , blakeslee , j. p. , piatek , s. , mei , s. , merritt , d. , milosavljevi , m. , tonry , j. l. , & west , m. j. 2006 , , 164 , 334 , l. & merritt , d. 2000 , , 539 , l9 , c. f. 2001 , , 553 , 174 , k. , bender , r. , bower , g. , dressler , a. , faber , s. m. , filippenko , a. v. , green , r. , grillmair , c. , ho , l. c. , kormendy , j. , lauer , t. r. , magorrian , j. , pinkney , j. , richstone , d. , & tremaine , s. 2000 , , 539 , l13 , j. 2003 , , 339 , 937 , a. w. 2004 , , 613 , l33 . 2008 , , 680 , 143 , a. w. , erwin , p. , caon , n. , & trujillo , i. 2001 , , 563 , l11 , a. w. , onken , c. a. , athanassoula , e. , & combes , f. 2011 , , 412 , 2211 , k. , richstone , d. o. , gebhardt , k. , lauer , t. r. , tremaine , s. , aller , m. c. , bender , r. , dressler , a. , faber , s. m. , filippenko , a. v. , green , r. , ho , l. c. , kormendy , j. , magorrian , j. , pinkney , j. , & siopis , c. 2009 , , 698 , 198 , n. & rix , h .- w . 2004 , , 604 , l89 , l. 1990 , , 356 , 359 , a. & nayakshin , s. 2009 , , 394 , 191 , a. , nayakshin , s. , power , c. , & king , a. 2011 , , 413 , 2633 , j. 2008 , , 386 , 2242 , s. 2004 , arxiv astrophysics e - prints , d. , katz , n. , fardal , m. , dav , r. , & weinberg , d. h. 2009 , , 395 , 160 , t. , devriendt , j. , slyz , a. , pichon , c. , kassin , s. a. , & dubois , y. 2011 , arxiv e - prints , a. 2003 , , 596 , l27 . 2005 , , 635 , l121 , a. r. & pounds , k. a. 2003 , , 345 , 657 , a. r. & pringle , j. e. 2007 , , 377 , l25 , a. r. , zubovas , k. , & power , c. 2011 , , 415 , l6 , p. i. & sunyaev , r. a. 1980 , soviet astron .
lett . , 6 , 357 , j. & bender , r. 2009 , , 691 , l142 , j. , bender , r. , & cornell , m. e. 2011 , , 469 , 374 , j. & gebhardt , k. 2001 , in american institute of physics conference series , vol . 586 , 20th texas symposium on relativistic astrophysics , ed .
j. c. wheeler & h. martel , 363381 , j. h. 1999 , active galactic nuclei : from the central black hole to the galactic environment ( active galactic nuclei : from the central black hole to the galactic environment /julian
h. krolik .
princeton , n. j. : princeton university press , c1999 . ) , m. r. , klein , r. i. , mckee , c. f. , offner , s. s. r. , & cunningham , a. j. 2009 , science , 323 , 754 , r. , gnedin , n. y. , & hamilton , a. j. s. 2010 , , 716 , 1386 , j. , tremaine , s. , richstone , d. , bender , r. , bower , g. , dressler , a. , faber , s. m. , gebhardt , k. , green , r. , grillmair , c. , kormendy , j. , & lauer , t. 1998 , , 115 , 2285 , a. & hunt , l. k. 2003 , , 589 , l21 , s. , fields , d. , peterson , b. m. , & grupe , d. 2011 , arxiv e - prints , n. j. , ma , c .- p . , gebhardt , k. , wright , s. a. , murphy , j. d. , lauer , t. r. , graham , j. r. , & richstone , d. o. 2011 , , 480 , 215 , d. e. , king , a. r. , & nayakshin , s. 2006 , , 650 , l37 , r. c. & mclaughlin , d. e. 2012 , arxiv e - prints , m. & serabyn , e. 1996 , , 34 , 645 , n. , quataert , e. , & thompson , t. a. 2005 , , 618 , 569 , j. f. , frenk , c. s. , & white , s. d. m. 1997 , , 490 , 493 , s. , cuadra , j. , & springel , v. 2007 , , 379 , 21 , s. & king , a. 2007 , arxiv e - prints , s. & power , c. 2010 , , 402 , 789 , s. , wilkinson , m. i. , & king , a. 2009 , , 398 , l54 , b. 1978 , acta astron . , 28 , 91 , t. , genzel , r. , martins , f. , nayakshin , s. , beloborodov , a. m. , levin , y. , trippe , s. , eisenhauer , f. , ott , t. , gillessen , s. , abuter , r. , cuadra , j. , alexander , t. , & sternberg , a. 2006 , , 643 , 1011 , k. a. , king , a. r. , page , k. l. , & obrien , p. t. 2003 , , 346 , 1025 , k. a. , reeves , j. n. , king , a. r. , page , k. l. , obrien , p. t. , & turner , m. j. l. 2003 , , 345 , 705 , h. r. , pringle , j. e. , clarke , c. j. , & kinney , a. l. 2002 , , 575 , 150 , f. , marulli , f. , mathur , s. , bernardi , m. , & bournaud , f. 2012 , arxiv e - prints , i. & begelman , m. c. 1989 , , 341 , 685 , i. , begelman , m. c. , & frank , j. 1990 , , 345 , 679 , j. & rees , m. j. 1998 , , 331 , l1 , e. & goodman , j. 2003 , , 341 , 501 , d. n. e. a. 2007 , , 170 , 377 , m. , slatyer , t. r. , & finkbeiner , d. p. 2010
, , 724 , 1044 , t. a. , quataert , e. , & murray , n. 2005 , , 630 , 167 , f. , cappi , m. , reeves , j. n. , palumbo , g. g. c. , yaqoob , t. , braito , v. , & dadina , m. 2010 , , 521 , a57 + , a. 1964 , , 139 , 1217 , s. , gebhardt , k. , bender , r. , bower , g. , dressler , a. , faber , s. m. , filippenko , a. v. , green , r. , grillmair , c. , ho , l. c. , kormendy , j. , lauer , t. r. , magorrian , j. , pinkney , j. , & richstone , d. 2002 , , 574 , 740 , g. a. & de zeeuw , p. t. 2005 , , 435 , 43 , k. , king , a. r. , & nayakshin , s. 2011 , , 415 , l21 zubovas , k. , & nayakshin , s. 2012 , subm . to mnras , arxiv:1203.3060
we have argued above that agn feedback can affect a large scale gas disc expelling the gas outward just as it can for a spherical distribution of gas .
we now show that there exists a critical radius within which this argument no longer holds : the disc there is resilient to agn feedback . inside the smbh s sphere of influence , marked by the radius @xmath80 ,
the gravitational potential is dominated by the smbh and so we can replace @xmath81 with @xmath2 . for a smbh on the @xmath12@xmath1 relation , the smbh s radius of influence
is given by @xmath82 as the result , the critical momentum flux striking the edge of the disc must be @xmath83 using @xmath84 for a marginally self - gravitating disc again , we arrive at the requirement that @xmath85 this can be satisfied at radii greater than @xmath86 where @xmath71 is the smbh mass in units of @xmath72 . recalling that we assumed that @xmath87 , we conclude that inside the `` feedback radius '' @xmath88 given by @xmath89 , \label{rfb}\ ] ] momentum feedback by the smbh is unable to affect a self - gravitating disc .
physically the significance of this feedback radius can be appreciated by considering a major gas mass deposition event say , the infall of a very massive giant molecular cloud or a merger with a gas - rich galaxy that dumps a significant amount of gas into the inner galaxy .
this gas will settle into a disc because it is very likely to have a non zero net specific angular momentum . at @xmath90 ,
the disc can not be expelled by the agn , whereas at @xmath90 it may as shown in the main body of the paper . somewhat counter - intuitively , gas closest to the smbh is actually hardest to expel to infinity ( although this becomes clearer if one considers that smbh s gravity is enormous in the inner @xmath32pc ) .
deposition of a small cloud at @xmath91 may be much more promising for agn feeding than having a vast amount of gas at @xmath92 as the latter is more easily affected by agn feedback and star formation . | we argue that current observations of @xmath0 relations for galaxies can be used to constrain theories of super - massive black holes ( smbh ) feeding . in particular , assuming that smbh mass is limited only by the feedback on the gas that feeds it , we show that smbhs fed via a planar galaxy scale gas flow , such as a disc or a bar , should be much more massive than their counterparts fed by quasi - spherical inflows .
this follows from the relative inefficiency of agn feedback on a flattened inflow .
we find that even under the most optimistic conditions for smbh feedback on flattened inflows , the mass at which the smbh expels the gas disc and terminates its own growth is a factor of several higher than the one established for quasi - spherical inflows .
any beaming of feedback away from the disc and any disc self - shadowing strengthens this result further .
contrary to this theoretical expectation , recent observations have shown that smbh in pseudobulge galaxies ( which are associated with barred galaxies ) are typically under- rather than over - massive when compared with their classical bulge counterparts at a fixed value of @xmath1 .
we conclude from this that smbhs are not fed by large ( 100 pc to many kpc ) scale gas discs or bars , most likely because such planar flows are turned into stars too efficiently to allow any smbh growth . based on this and other related observational evidence
, we argue that most smbhs grow by chaotic accretion of gas clouds with a small and nearly randomly distributed direction of angular momentum .
galaxies : formation galaxies : active accretion : accretion discs |
fluctuations of the x - ray background ( xrb ) provide a unique method to investigate statistical properties of the distribution of x - ray sources . at angular scales above @xmath5
the _ rosat _ all - sky survey ( rass ) revealed distinct variations of the soft xrb ( soltan et al .
subsequent analysis ( soltan et al . 1999 ) showed that clustering of active galactic nuclei ( which are the main contributor to the xrb ) potentially could explain the large amplitude of the xrb autocorrelation function ( acf ) .
however , the amplitude of x - ray source spatial clustering required to reproduce the observed acf is higher than that derived from direct investigation of the distribution of sources ( e.g. carrera et al .
thus , it is likely that fluctuations of the soft xrb are produced also by some other effects . in particular , hydrodynamical computations by cen & ostriker ( 1999 )
show that the process of accumulation of primordial gas in the potential wells created by galaxies and clusters of galaxies produced large scale concentrations of hot plasma .
this material is expected to emit thermal bremsstrahlung in the soft x - ray domain .
a clumpy distribution of emitting plasma would contribute substantially to the total fluctuations of the xrb . to investigate the importance of different mechanisms producing the xrb
anisotropy one needs to measure the amplitude of the xrb fluctuations over a wide range of angular separations and at different energies . in the present work we have determined the autocorrelation function of the soft xrb using a large number of _ rosat _ pointings . below we describe the observational material used in the analysis , the computational details and compare the present results with our earlier measurements of the acf at large separations using the _ rosat _ all - sky survey .
visual inspection of the rass x - ray maps ( snowden et al . 1997 ) shows that at low energies only selected sky regions seem to be free from thermal emission by hot galactic plasma and not affected by absorption by cold gas . to investigate fluctuations of the extragalactic component of the xrb soltan et al .
( 1996 ) used a section of the rass maps of approximately 1 sr at the northern galactic hemisphere ( @xmath6 ) apparently least contaminated by local effects . in the present analysis we have concentrated on the same region of the sky . all _ rosat _ pointed observations within this area with exposure time longer than 5000s excluding pointings at known extended sources ( supernova remnants , nearby normal galaxies , clusters of galaxies ) have been used .
total number of pointings satisfying these selection criteria is equal to 141 .
the data in the gain - corrected pulse height channels @xmath7 ( r6 band in snowden at al .
1994 ) with the energy centered at 1.15kev has been selected .
counts in each field were binned into @xmath8 pixels . for each observation the central region containing the target source
was eliminated from the data , the size of the removed area was carefully determined .
the distribution of counts within the field of view produced by the target source was calculated using the _ rosat _ telescope point spread function and the radius of the circle to be removed was obtained in such a way that the residual contribution of the target source did not exceed 5% of the local background within the remaining section of the field of view . to minimize vignetting effects
we have used the central section of the pspc field of view with radius of 135 .
contamination by particle background and by soft solar photons scattered in the earth s atmosphere was subject to scrutiny for each observation separately . in the vast majority of observations the number of non - cosmic counts was insignificant , and in few cases the observed count rates have been corrected for both these effects .
it is of fundamental importance for the fluctuation analysis to construct a sample of observations for which all the instrumental and non - cosmic effects are constant and do not vary from one observation to another . to check for the possible systematic variations of the telescope / detector system characteristics we have plotted in fig.1a and b respectively , the average count rate per pixel as a function of exposure time and date of observation .
while the count rates averaged over each field do not show any dependence on exposure time , they vary systematically with the date of the observation .
most probably this effect results from small inaccuracies in the gain corrections . to remove this weak trend ,
a straight line was fitted to the data and the average count rate in each field was corrected for the slope of the fit .
we denote the intensity of the xrb in the direction defined by the unit vector @xmath9 in the @xmath10-th pointing (= count rate in the corresponding pixel ) as @xmath11 .
the acf @xmath12 was calculated using the formula : @xmath13 ^ 2 } -1\,,\ ] ] where the sum extends over all pointings ( @xmath14 ) , the directions @xmath15 and @xmath16 are separated by the angle @xmath17 , and @xmath18 denotes the count rate in the @xmath10-th field calculated averaged over pixels used in the calculations for given @xmath17 .
the acf based on the pointings is shown in fig.2a . the strong signal at separations below @xmath19 results from `` granular '' nature of the xrb . the shape of the acf at these separations is defined by the psf of the telescope / detector system averaged over the field of view and photon energies .
variations of the acf at larger separations also result from the discrete structure of the xrb . here the major role play pairs ( or multiplets ) of strong sources in each field ; the larger the number of separate pointings used in the calculations , the smaller amplitude of the acf fluctuations .
apart from the fluctuations , the acf amplitude shows a systematic decline with increasing separation . above @xmath20
the signal becomes too weak in comparison with the fluctuations produced by the strong sources , that the acf can not be estimated reliably .
present measurements of the acf are shown together with the rass estimates ( taken from soltan et al . 1999 ) in fig . 2b ( note the logarithmic scale ) .
open circles correspond to the results shown in fig .
2a averaged in logarithmic bins .
both measurements are in satisfactory agreement taking into account large uncertainties of both measurements ( error bars of the rass result represent scatter due to statistical fluctuations ) .
the acf signal is now measured in the range @xmath21 .
it is likely that various effects contribute the xrb fluctuations at different scales . to assess what role play the bright clusters of galaxies we have repeated all the calculations without fields containing known clusters .
the acf calculated for the restricted data set ( 133 pointings ) is shown in fig .
2b with crosses .
it appears that clusters contribute to the xrb fluctuations noticeably only at separations below @xmath1 .
the acf without known clusters becomes substantially flatter than for all the data .
detailed models of the acf will be discussed in a separate paper . | the first positive detection of the x - ray background fluctuations at small angular scales is reported .
_ rosat _ pspc archive pointed observations are used to measure fluctuations at scales of @xmath0 .
the pointings have been selected from an area free from galactic contamination . at separations below @xmath1 clusters of galaxies
become a substantial source of the background fluctuations . the autocorrelation function of the fluctuations in the power law approximation
has a slope of @xmath2 for all the data but is substantially flatter ( with slope of @xmath3 ) when pointings containing bright clusters are removed . at separations @xmath4 where the acf estimates based on the rosat pointings and all - sky survey are available , both data sets give consistent results .
# 1_#1 _ # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
the research of phenomena related to light - matter interaction has been experiencing significant growth for the past decade @xcite in the domain of nano - optics . from the fundamental point of view ,
one of the most intriguing question lies in the dynamics and in the optical properties of many - body systems .
for instance , new collective modes were predicted and measured in materials composed of interacting quantum emitters ( atoms , molecules , etc . )
experiments in that regard are noticeably advanced compared to theory , as it has long been realized that treating multi - body systems is quite challenging .
even systems comprised of just a few interacting quantum dipoles are difficult to investigate @xcite .
semi - classical descriptions of light - matter interaction , in which a classical electric field interacts with a quantum system , enormously simplify modeling leading to results that both support experiments and predict new phenomena @xcite
. collective effects @xcite arise from the fact that the electric field experienced by an arbitrary quantum emitter is the sum of both the applied field and the radiation from all other atoms in the system under consideration
. when it is strong enough , the radiation from the dipoles set them coupled .
this is the source of cooperativity .
one of the important fundamental cooperative effects , the collective lamb shift which is due to the exchange of virtual photons between the particles in a dense sample of atoms was recently observed @xcite .
dependent of the cooperativeness in homogeneous and inhomogeneous systems were studied both experimentally @xcite and theoretically @xcite . at present ,
research on those cooperative effects concerns on both cold @xcite and thermal @xcite samples of atoms since they offer the observation of different interesting quantum phenomena . because of the ability of light to carry effectively quantum information , understanding the optical response and the light scattering in dense samples is also interesting for many applications such as slow and stopped light @xcite , quantum information @xcite , trapped light @xcite and optical memories @xcite . in the linear regime ,
the propagation of light through any medium can be investigated by extracting its electric susceptibility @xcite .
it was shown recently @xcite that the electric susceptibility of systems comprised of three level quantum emitters at high densities exhibit a fano - type profile that results in a new phenomenon of induced transparency called _ dipole induced electromagnetic transparency _ ( diet ) .
diet is due to the collective response of the system towards the external electromagnetic ( em ) field via dipole - dipole interactions . following this work , in this paper
, we present a detailed theory of diet and extend the model to multilevel systems .
we also apply our theory to a realistic system , namely the d1 line of @xmath0 .
the paper is organized as follows .
the next section discusses in detail our theoretical description for @xmath1-level systems and the analytic model used to analyze the results .
section [ two - levels ] provides results and a comprehensive discussion of how linear optical properties of a system comprised of interacting two - level emitters depends on various material parameters such as the particle density . in section
[ many - levels ] , we extend our discussion to systems having more than one excited states .
first we consider interacting three - level emitters .
the results provided demonstrate dipole induced electromagnetic transparency ( diet ) .
this phenomenon is explained using an analytical model and is extended to the multilevel d1 transitions of @xmath0 .
section [ applications ] discusses possible applications of our results and finally , we conclude our work in section [ conclusion ] .
our goal is to describe the linear dynamics of quantum emitters in a self - consistent manner taking into account collective effects .
the most comprehensive and thus complete description would be to implement a fully quantum model based on the jaynes - cummings hamiltonian @xcite .
on the other hand , the quasi - classical theory of light - matter interaction recently applied to hybrid materials has proven to lead to both qualitative and quantitative agreement with experiments @xcite . keeping also in mind that the majority of experiments in nano - optics , such as white light spectroscopy , for instance , are performed under conditions corresponding to high occupation numbers of photons in a given em mode , the quasi - classical description is well justified .
the idea is to separate the description of em waves from the dynamics of quantum emitters considering the spatio - temporal evolution of em radiation using classical maxwell s equations while applying the full machinery of quantum mechanics to describe the response of quantum emitters to em excitation .
this results in a system of coupled maxwell - bloch equations @xcite . if one wants to observe collective effects , it is imperative to solve the corresponding equations of motion self - consistently without any decoupling @xcite .
the dynamics of the em field components , @xmath2 and @xmath3 , is simulated using maxwell s equations [ maxwell ] @xmath4 where @xmath5 and @xmath6 are the permittivity and the permeability of free space , respectively , and @xmath7 is the macroscopic polarization of the system at position @xmath8 and time @xmath9 .
the dynamics of the latter is considered both quantum mechanically and classically . in both cases
we assume that the emitters are continuously distributed in space and we neglect static emitter - emitter interactions . under such conditions
one can express the macroscopic polarization as @xmath10 where @xmath11 is the number density of emitters .
the average dipole moment , @xmath12 , is either determined directly from classical equations of motion or quantum mechanically by evaluating @xmath13},\ ] ] where @xmath14 denotes the density matrix of the system .
each quantum emitter has @xmath15 excited states represented by @xmath16 , with @xmath17 and @xmath18 , coupled to the ground state @xmath19 via the time dependent em interaction @xmath20 .
the density matrix @xmath21 satisfies the following liouville - von neumann equation @xmath22-i\hbar\hat\gamma\hat\rho,\ ] ] where @xmath23 is the total hamiltonian and @xmath24 is the superoperator describing relaxation and dephasing processes taken in the lindblad form under markov approximation @xcite .
off - diagonal elements of @xmath24 include the pure dephasing rate @xmath25 and the diagonal elements include the radiationless decay rate @xmath26 of the excited states .
the field free hamiltonian @xmath27 is given by @xmath28 we define the bohr frequencies as @xmath29 and the interaction of a single emitter with the em fields is written as @xmath30 where @xmath31 are the instantaneous rabi frequencies associated with the interaction between the quantum system and the local em fields . in the present work
we assume that the excited states are not directly coupled to each other but only to the ground state @xmath19 .
the corresponding system of coupled differential equations reads @xcite [ opt_bloch_2 ] @xmath32\rho_{0\jmath } , \end{aligned}\ ] ] where @xmath33 eqs .
( [ maxwell ] ) and ( [ opt_bloch_1 ] ) , coupled through the evolution of the quantum polarization ( [ mean_field_approximation ] ) , form the basis of the model .
as this system of equations is propagated in space and time on a grid , one should note that each grid point is effectively a point - wise dipole with the amplitude of an individual emitter times the number density .
this essentially means that dipole - dipole interactions within a single grid point are neglected .
this approximation although valid at low densities has to be corrected for high densities . treating the dipoles exactly to include those contributions is extremely difficult and it is almost impossible to solve the system of equations by using current computational facilities @xcite .
an alternative way to include dipole - dipole interactions at a single grid point level is to introduce the well - known mean field lorentz - lorenz correction term to the local field @xcite , with @xmath34 this local electric field enters the liouville - von neumann equation ( [ opt_bloch_2 ] ) through the rabi frequencies @xmath35 , where @xmath36 denotes the transition dipole moment between states @xmath19 and @xmath37 . using the proposed model
we simulate the linear optical response of a vapor comprised of interacting quantum emitters as schematically depicted in fig .
[ system ] . to simulate a first order elastic scattering / absorption we implement the short pulse method @xcite .
a weak ultra - short incident pulse polarized along @xmath38 is launched at normal incidence ( @xmath39 direction ) on a slab of quantum emitters of finite thickness .
composed of interacting quantum emitters is exposed to linearly polarized em fields acting normal to the interface . , width=321 ] for such a geometry , maxwell s equations ( [ maxwell ] ) reduce to [ reduced_maxwell ] @xmath40 maxwell s equations are numerically integrated using a finite - difference time - domain ( fdtd ) technique @xcite with a spatial grid step , @xmath41 nm and temporal grid step , @xmath42 as .
convolutional perfectly matched layers ( cpml ) @xcite absorbing boundaries of thickness @xmath43 nm are used to avoid non - physical reflections from the boundaries of the simulation region .
concurrently , liouville - von newmann s equations ( [ opt_bloch_2 ] ) are propagated in time using the fourth order runge - kutta method , assuming that all emitters are initially in their ground state @xmath19 .
light excitation is provided by a gaussian pulse of fwhm @xmath44 , where @xmath45 is the pulse duration . to evaluate transmission , reflection and extinction
, we calculate the em energy flux @xmath46 on the input and the output sides of the slab using @xmath47 where @xmath48 , @xmath49 , @xmath50 and @xmath51 are the fourier components of the total and incident em fields .
( [ poynting2 ] ) gives the reflection spectrum @xmath52 if @xmath48 and @xmath49 are the reflected em fields and the transmission spectrum @xmath53 if @xmath48 and @xmath49 are the transmitted em fields .
both spectra carry the information about the dynamics of the system . and
finally the extinction @xmath54 is estimated as @xmath55.\ ] ] the layer of quantum emitters is placed at a distance of @xmath56 @xmath57 m from the source . the reflected field is measured at a point between the source and the layer which is at @xmath58 nm away from the source and the transmitted field is measured on the output side , @xmath59 @xmath57 m far from the source .
we verified that the results are independent of those distances as long as the pulse acts normal to the layer . in the linear regime , @xmath60 , for @xmath61 , and the homogeneous system of @xmath1-level quantum emitters
can be described by a set of @xmath15 coupled harmonic dipole oscillators .
the macroscopic polarization associated with the transition between @xmath19 and @xmath37 , @xmath62 , follows a classical equation of motion @xmath63 where @xmath64 and @xmath65 is the effective mass of the oscillating charge @xmath66 driven by the electric field @xmath67 .
taking into account eq .
( [ local_field ] ) and subsequently applying a fourier transform , we obtain @xmath68 where @xmath69 and where @xmath70 is the plasma frequency representing the collective oscillations of the dipoles in the system @xmath71 with the assumption that the maximum amplitude of dipole oscillations in the absence of a driving field is given by the quantum harmonic oscillation length , we have @xmath72 in weak electric fields where @xmath73 , eq . ( [ damped oscillator_corrected2 ] ) gives the susceptibility @xmath74 associated with the @xmath75 dipole .
summing over all dipole oscillators in the material , we obtain the total susceptibility of the system [ chi_in_coupled_n_level_systems ] @xmath76
first we consider a slab of interacting simple two - level emitters characterized by the transition energy @xmath77 ev , the transition dipole @xmath78 d , and the total decoherence rate @xmath79 thz .
we calculate the response of the system to an extremely short pulse of width @xmath80 fs using coupled maxwell - liouville - von neumann equations in different collective interaction regimes @xcite .
the results are analyzed using the semi - classical approximation described in section [ semi - classical approximation ] .
the electric susceptibility of interacting two - level emitters in the semi - classical model can be written as @xmath81 where @xmath82 is the shifted resonant frequency of the system .
the shift in transition frequency , @xmath83 , which is due to the lorentz - lorenz correction included in the formulation of the local electric field via eq .
( [ local_field ] ) can be estimated from the semi - classical model .
the first order term gives the well - known lorentz - lorenz shift ( ll shift ) @xcite .
@xmath84 the case of strong dipole - dipole interactions is characterized by @xmath85 , while @xmath86 corresponds to the weak interaction regime .
typical reflection spectra are shown in fig .
[ ref_two_level ] as a function of the reduced detuning , @xmath87 , for different values of @xmath88 for a slab of thickness @xmath89 nm . in the weak interaction regime ( @xmath90 ) , the system reflects a very small amount of the total energy ( of the order of @xmath91 ) around the transition frequency .
such a response is similar to that of a single emitter since it is characterized by a typical lorentzian profile .
as the value of @xmath88 increases , and especially for @xmath92 , the interaction between the dipoles becomes dominant and the system starts to respond to the applied field collectively . for @xmath93 ,
the reflection spectrum broadens resulting in a window of frequencies within which more than @xmath94 of incident energy is reflected .
this reflection window was explained in @xmath95 by r. j. glauber and s. prasad in ref .
@xcite by considering the exponentially decaying modes of coherent excitation of the medium that depend strongly on the frequency . for a slab of thickness @xmath96 larger than the dipole wavelength @xmath97 , the reflection at the vacuum / slab interface ( shown as interface @xmath98 in fig .
[ system ] ) dominates and the contribution from the possible multiple reflections inside the slab can be neglected .
for such a slab , the width of the reflection window can be estimated from the reflection probability at this interface .
@xmath99 where @xmath100 $ ] is the real part of the effective refractive index of the system . in the case of total reflection , @xmath101 and
the real part of the electric susceptibility is @xmath102 . equating the real part of eq .
( [ chi1 ] ) to @xmath102 leads to the range of frequencies within which the reflection probability reaches its maximum : @xmath103 and hence the frequency window with the maximum reflection shows a width of @xmath104 which agrees with the quantum simulation @xcite . for relatively thin slab of width @xmath105 , the width of the reflection window is noticeably below @xmath104 .
this is due to the fact that in this analysis we have neglected contributions from multiple reflections that occur inside the slab . for a thin system ,
contributions from the two different vacuum - medium interfaces are not negligible and obviously affect the reflection spectra . for a slab of interacting two - level emitters in different interaction regimes .
the thickness is @xmath106 nm and the transition dipole @xmath107 d. the radiationless decay rate and the pure dephasing rate are @xmath108 thz and @xmath109 thz , respectively .
reflection for small values of @xmath88 is weak and hence amplified ( for better comparison ) by the constant factors shown in the figure .
, width=321 ] fig .
[ trans_abs_two_level ] shows the extinction ( panel ( a ) ) and transmission ( panel ( b ) ) spectra at @xmath90 ( blue solid line ) and @xmath110 ( green dashed line ) . for @xmath90 , almost all the incident energy is transmitted .
clearly in the weak interaction regime , the local polarization barely modifies the local electric field experienced by a given emitter .
hence the emitters respond to the field as if they were independent . as the mutual interaction between the dipoles increases , the system becomes opaque and reflects most of the incident energy @xcite . in complement with reflection and transmission ,
the remaining energy is absorbed by the system and decays due to radiationless transitions described by @xmath24 in eq .
( [ opt_bloch_1 ] ) .
in particular at @xmath110 , the system loses its transparency , as seen in fig .
[ trans_abs_two_level](b ) , and behaves almost like a mirror around a broad range of frequencies near the transition frequency . .
the solid blue curves ( left vertical scale ) represent the results for @xmath90 and the dotted green lines ( right vertical scale ) show the data for @xmath110 .
all other parameters are the same as in the figure [ ref_two_level].,width=302 ] it should be emphasized that the reflection from strongly interacting dipoles resembles bragg mirrors ( also referred to in the literature as distributed bragg reflectors ) @xcite .
bragg mirrors are composed of alternating dielectric layers with different refractive indices . the high reflectivity is achieved by multiple reflection from layers of the mirror .
this leads to a controllable constructive interference .
spatially varied refractive index profiles determine the fringes in the spectrum .
in contrast to bragg mirrors our system is completely homogeneous .
since we perform simulations in a linear regime , the refractive index of the system is a constant throughout the layer .
this rules out the possibility of having multilayer interferences .
all the effects observed in the spectra are due to the collective response of the system .
the width and the position of the reflection window are characterized by the physical properties of the system such as the transition dipole , transition frequency , material density . the qualitative explanation for the modifications in the spectra when the system enters the strong interaction regime relies on the fact that the polarization of the system is enhanced due to the strong dipole - dipole interactions .
prior to the detailed quantitative analysis of the spectral features , we want to estimate the accuracy of the semi - classical model and compare it with results obtained via exact numerical simulations .
[ chi2 ] shows the relative errors of the electric susceptibilities for interacting two - level emitters obtained using the semi - classical model with respect to the numerical integration of the maxwell - bloch equations , eq .
( [ chi1 ] ) in the weak ( @xmath90 ) and strong ( @xmath110 ) interaction regimes .
the solid blue line shows the relative error for weakly interacting dipoles and the red dashed line , is the same for strongly interacting dipoles .
both relative errors are peaked around the corresponding transition frequencies with a maximum amplitude of about @xmath111 and we conclude that the analytical model is accurate enough to analyze the numerical results . .
the blue solid line shows the relative error for weakly interacting dipoles and the red dashed line presents the same for strongly interacting dipoles .
, width=302 ] the semi - classical model explains the spectral features such as the shift in the resonance frequency and the width of the reflection window for a system of interacting two - level emitters in the strong interaction regime .
the features that are not yet interpreted by the model are sidebands ( see fig .
[ ref_two_level ] for @xmath112 ) . those however can be interpreted as simple fabry - prot modes @xcite .
the semi - classical model does not yet contain any information about the geometry of our system , assuming homogeneous distribution of the emitters . with two or more interfaces present
, one can expect to observe interferences between reflected and transmitted fields due to the possible multiple reflections inside the slab . in this case , the system can be seen as a fabry - prot etalon consisting of two parallel partially reflecting interfaces shown as the interfaces 1 and 2 in fig .
[ system ] .
em fields at frequencies outside the transmission window are partially reflected and transmitted , resulting in non - zero optical path differences and hence giving rise to an interference pattern as seen in fig .
[ fp_modes ] for instance .
introducing the decay of the em fields inside the slab and taking into account multiple reflections with proper boundary conditions @xcite , we find the reflection @xmath113 and transmission @xmath114 coefficients for the system as [ collective ] @xmath115 e^{-ink\ell}}{[n(\omega)+1]^2 - [ n(\omega)-1]^2
e^{-i2n(\omega)k\ell } } , \\ & t(\omega ) = \frac{4n(\omega ) e^{-\kappa \ell}}{[n(\omega)+1]^2 - [ n(\omega)-1]^2 e^{-i2n(\omega)k\ell } } , \end{aligned}\ ] ] where @xmath116 is the refractive index , @xmath117 is the propagation constant of the applied field in the vacuum and @xmath118 is the imaginary part of the propagation constant inside the medium .
the associated reflection and transmission spectra predicted by the semi - classical model are given by [ ref_tran ] @xmath119 the expressions ( [ ref_tran ] ) together with ( [ collective ] ) can accurately reproduce the positions of the maxima and minima of the sidebands in the calculated spectra .
[ fp_modes ] shows the reflection spectrum calculated by integrating maxwell - bloch equations ( blue dashed line ) and the same obtained from the equation ( [ ref_tran ] a ) shown by the solid red line .
the two curves are in perfect agreement .
the dashed blue line is the reflection probability calculated by integrating the maxwell - bloch equations and the solid red line is the same obtained from the semi - classical approximation ( eq .
( [ ref_tran ] a ) ) .
other parameters are @xmath110 , @xmath78 d and @xmath79 thz .,width=302 ]
let us consider a ground state coupled to @xmath15 excited states . as in the previous case , the infinite slab of emitters is exposed to a transverse electric field . in the strong interaction regime , the lorentz - lorenz correction results in redshifts for all transitions .
we thus expect to observe @xmath15 reflection windows of width @xmath120 each .
the most interesting case is when the transitions are significantly overlapping , leading to interference effects . for three level systems , in the presence of such overlapping resonances , we observed diet which is due to the interference between two indistinguishable excitation pathways @xcite . the calculated reflection spectrum for a three level system with overlapping transitions is shown in fig .
[ three_level_ref ] as a blue solid line .
it is compared with the reflection spectrum ( red dotted line ) obtained from a slab of two types of uncoupled two - level emitters having the same physical parameters which is calculated as the product of the reflection probabilities of each dipole . for the uncoupled dipoles the reflection spectrum behaves completely
differently compared to the case when they are coupled . for uncoupled dipole systems ,
the total reflection is peaked where the two independent reflection windows corresponding to each dipole are contributing .
this region corresponds to a clear overlap of two resonances . .
the red dotted line is the reflection from the same system in the absence of coupling between the dipoles .
the excited states are separated by @xmath121 .
the reduced detuning @xmath122 , is defined with respect to the transition frequency of the first excited state of this three level system.,width=302 ] the strong interaction between the dipoles in the system broadens the transition that results in the overlapping of two closely spaced allowed transitions and hence sets up a competition between two possible excitation pathways .
we obtain a narrow window where the excitation probability becomes zero .
the presence of a hole in otherwise flat reflection spectra is a clear signature of the destructive interference between the radiation emitted by coupled dipoles and hence a signature of diet @xcite .
except for the observed hole in the reflection spectrum , all features discussed in the previous section are clearly present in the three - level system spectra . at the frequency where there is no reflection ,
we observe a strong transmission if damping processes in the system are not too efficient .
the observed transparency , similar to electromagnetically induced transparency ( eit ) @xcite , may lead to many potential applications such as slow light discussed later in this paper .
the position of the transparency window can be controlled to a certain extend by changing the material parameters @xcite .
the same type of interference effect can happen if the system has a series of closely spaced energy levels .
one of the potential actual systems for studying diet are rb atoms .
rb is used experimentally both as a thermal gas @xcite and as a cold atomic gas @xcite .
dipole - dipole interaction depends on the square of the transition dipoles and rb atoms are characterized by large @xmath123 to @xmath124 transition dipoles @xcite .
we consider the d1 transition , _
, @xmath125 to @xmath126 of @xmath127 . both @xmath125 and @xmath126
are split into two sub - levels due to the hyperfine interaction @xcite and hence there are four dipole - allowed transitions . thus the d1 transitions of @xmath0 isotope can be considered as a superposition of two three - level systems having same excited states but two different ground states . since the energy splitting between the excited states is smaller than the splitting of the ground states , it gives a possibility to observe diet between two excited states that are coupled to a specific ground state as well as between two excited states coupled to two different ground states for different atomic densities .
the susceptibility of a system consisting of two ground states can be written as @xmath128 where the index @xmath129 corresponds to the transitions from the sub - level @xmath130 and @xmath131 is for the transitions from @xmath132 .
it is important to note that eq .
( [ chi_for_mixed_dipoles ] ) can also be used to analyze systems composed of different types of atoms or molecules .
the ll shifts for d1 transitions of rb can be written as @xmath133 where @xmath134 is the ll shift defined in the absence of hyperfine splitting and @xmath135 is a measure of relative strength of the transitions @xcite .
the reflection and susceptibility of a slab of thickness @xmath136 nm comprised of @xmath0 for @xmath137 are shown in fig .
[ rb-85-weak ] .
the reduced detuning @xmath122 is defined with respect to @xmath138 thz , the d1 transition frequency in the absence of hyperfine splitting @xcite . .
the reduced detuning @xmath122 is defined with respect to the transition frequency of @xmath125 to @xmath126 in the absence of hyperfine splitting ( @xmath138 thz ) and @xmath139 mhz .
the positions of transitions @xmath140 are indicated by the red dashed lines marked with @xmath141.,width=321 ] the double peaks in the negative detuning region are due to the transitions @xmath142 and @xmath143 and those in the positive detuning region are due to the transitions @xmath144 and @xmath145 .
the separation of the double peaks is due to the hyperfine splitting of the ground state .
since the system is in the weak interaction regime , the transitions are independent and they are not overlapping .
it explains the reflection spectrum in fig .
[ rb-85-weak](b ) with a series of lorentzian profiles corresponding to different transitions . since the dipoles are relatively large for d1 transitions , relatively small increments in the gas pressure will couple the different transitions and hence lead to the diet regime .
the calculated reflection and transmission spectra and the susceptibility for such a system with @xmath146 are given in fig .
[ rb-85-strong1 ] .
panel ( a ) shows the susceptibility , panel ( b ) presents the reflection spectrum , and panel ( c ) shows the transmission . with @xmath146 .
vertical red dashed lines show the positions of destructive interference between the different transition dipoles.,width=321 ] overlapping of the transitions to @xmath147 and @xmath148 from the ground states result in two fano profiles in the susceptibility .
two sharp minima in the reflection spectrum are the signatures of destructive interferences between the transition dipoles .
concurrently we see two narrow peaks in the transmission spectrum .
these frequencies at which diet takes place are shown as red dashed lines in the figure . further increase of the dipole - dipole interaction mixes the @xmath147 and @xmath148 excited states into a single excitation state which is coupled to two ground states ( @xmath130 and @xmath132 of @xmath149 ) .
it leads to diet due to the overlapping of transitions to the excited state from two different ground states .
[ rb-85-strong2 ] shows such an effect in strong interacting samples of @xmath0 with @xmath150 .
panel ( a ) in fig .
[ rb-85-strong2 ] is the modulus of the susceptibility , panel ( b ) and ( c ) are the reflection spectra for slabs of thicknesses @xmath151 nm and @xmath152 @xmath57 m , respectively , and panels ( d ) and ( e ) are the corresponding transmissions .
atoms with @xmath150 .
reflection spectra ( panel ( b ) and c ) for thickness @xmath89 nm and @xmath153 @xmath154 and corresponding transmission spectra in panel ( d ) and ( e ) as functions of the reduced detuning .
a zoom of the susceptibility is given in the inset of panel ( a ) for better view of the faint fano profiles associated with the holes in the reflection spectra.,width=321 ] the susceptibility shows one dominant and two faint fano type profiles .
the latter clearly shown in the inset in panel ( a ) are due to the splitting of the excited states .
the dominant fano profile corresponds to the overlapping of two coupled dipoles @xmath155 and @xmath156 .
it leads to diet which is well resolved in the reflection ( fig .
[ rb-85-strong2](b ) and ( c ) ) and transmission ( fig .
[ rb-85-strong2](d ) and ( e ) ) spectra of slabs of thickness @xmath157 nm and @xmath59 @xmath57 m .
this can be understood if we assume that the d1 transitions are a mixture of two three - level systems having the same excited states .
in such a picture , the dominant diet in the strong interaction regime can be interpreted as an interference effect between the transitions to a single excited state from two different ground states .
the two small fano profiles in the susceptibility are due to the splitting of the excited states
. they also result in diet which explains two small transmission peaks on the left and right of the dominant diet signal from the @xmath157 nm slab .
these two transmission peaks are not resolved for the @xmath59 @xmath57 m slab due to damping effects .
the system , therefore , acts almost as if it was comprised of three - level emitters @xcite .
however the splitting of the excited state leaves spectral holes in the reflection spectra ( fig .
[ rb-85-strong2](c ) ) .
the destructive interference of two overlapping transitions can be used to modify the spectral envelope of an incident pulse , _
i.e. _ as a pulse shaper .
if we consider a slab of interacting multi - level quantum emitters with a reflection window wider than the fwhm , @xmath158 of the incident pulse , the collective response of the quantum emitters induce a selective reflection and transmission of the incident pulse .
this is illustrated in fig .
[ pulse_shape ] for the d1 transitions of @xmath0 at @xmath150 , with a pulse of fwhm , @xmath159 ghz .
panel ( a ) in fig .
[ pulse_shape ] compares the normalized reflected pulse envelope ( red line ) with the incident pulse shape ( blue dashed line ) of such a pulse with the central wavelength @xmath160 nm for a @xmath157 nm thick slab of @xmath0 .
panel ( b ) is the same as panel ( a ) but for a @xmath59 @xmath57 m thick slab .
the normalized transmitted pulse envelopes ( red line ) are shown in panel ( c ) for @xmath157 nm and in panel ( d ) for @xmath59 @xmath57 m .
a zoom around the central wavelength is given in the inset of panel ( d ) to see the shape of the pulse transmitted though the 2.5 @xmath57 m slab .
atoms at @xmath150 as functions of the reduced detuning @xmath122 .
the blue dashed lines in the panels are the normalized incident pulse and the red lines are the reflected and transmitted pulse envelopes .
panels ( a ) and ( c ) are for a slab of @xmath151 nm and panels ( b ) and ( d ) are for @xmath153 @xmath57 m . ,
width=321 ] from the figs .
[ pulse_shape](a ) and ( b ) it is clear that the reflection from the @xmath157 nm slab is not as efficient as the reflection from the @xmath59 @xmath57 m slab since this slab transmits a large part of the incident pulse ( fig . [ pulse_shape](c ) ) .
it is due to the mismatching between the slab thickness @xmath89 nm and the transition wavelength @xmath161 nm that creates long tails of fabry - prot modes near the reflection window which modify the transmission probability ( see fig .
[ rb-85-strong2](b ) and ( d ) ) .
but for the @xmath59 @xmath57 m slab , the condition @xmath162 is satisfied leading to the wide reflection window .
this minimizes the transmission of the pulse of width @xmath163 through the slab except at the positions of diet . if the system is comprised of emitters with many optically active excited states that can be populated by the incident laser pulse ( such as molecules with ro - vibrational levels for instance @xcite ) , each transition interacts with the others and interferes constructively or destructively at different frequencies leading to reflected pulses with more than a single frequency amplified or removed from the spectrum .
in particular , spectra of the the transmitted pulse in fig .
[ pulse_shape](c ) and ( d ) show that diet can vary dramatically depending on the system parameters . for @xmath150 .
panel ( a ) shows the refractive index : the solid blue line is the imaginary part and the dashed red line is the real part , panel ( b ) shows the group index of the layer and panel ( c ) shows the group velocity as functions of the reduced detuning @xmath122.,width=283 ] in systems comprised of interacting multilevel emitters we observe a minimum in the reflection window with a highly nonlinear dispersion ( see fig .
[ rb-85-strong1 ] and [ rb-85-strong2 ] ) .
it opens several intriguing applications , including slow light @xcite .
[ slow_light ] shows the slowing down of the transmitted pulse in a layer of @xmath164rb with @xmath150 .
panel ( a ) shows the refractive index .
the blue solid line is the imaginary part and the red dashed line is the real part of the refractive index .
panel ( b ) shows the group index @xmath165 $ ] and the group velocity @xmath166 is shown in panel ( c ) .
the system has a very large group index @xmath167 within the window of frequency where diet occurs ( see the fig .
[ slow_light](d ) and the inset of fig .
[ pulse_shape](b ) ) .
it peaks at @xmath168 similar to a recent experiment on slow light @xcite and the group velocity ( panel ( c ) ) drops below @xmath169 m / s .
but in the system we consider this lowest velocity may not be observed since the imaginary part of the refractive index ( see panel ( a ) in fig . [ slow_light ] ) is large that set the system as a lossy medium of light at this frequency .
it also explains the quick drop in the transmission pulse shown in the inset of fig .
[ pulse_shape](d ) .
the transmitted pulse peaked around the reduced detuning @xmath170 dies out at @xmath171 . within this short window of the reduced detuning the group index
@xmath167 is of the order of @xmath172 and the velocity is decreased down to @xmath173 m / s . within this window , the imaginary part of the refractive index is small enough to allow the experimental observation of slowed down transmitted pulse .
in this manuscript , we study collective effects in systems of interacting multilevel quantum emitters and their dependence on different physical parameters such as density .
such effects are studied both numerically using maxwell - bloch equations and analytically using a semi - classical model . when damping is sufficiently low , as in atomic or molecular vapors at low temperatures , for instance , narrow transmission windows are observed .
such narrow windows in an otherwise completely opaque material are due to quantum interferences between different dipoles corresponding to different induced transitions .
a clear evidence of such an interaction is the presence of fano profiles in the susceptibility of the system .
it is shown that this effect is amplified when the the dipole - dipole coupling increases .
we call the transparency of the system induced by the coupling between different kinds of dipoles _ dipole induced electromagnetic transparency _ ( diet ) .
we note that diet is similar to the well - know electromagnetic induced transparency ( eit ) @xcite the former however is fundamentally different from eit .
eit requires a strong laser pulse that couples two different quantum states , while diet is inherently internal , _
i.e. _ requires overlapping resonances induced by strong dipole - dipole couplings .
the presence of additional levels such as ro - vibrational levels in molecules for instance or multiple optically active electronic states in case of atoms can change the response of the system due to the presence of additional energetically allowed transition dipoles which leave their signatures as spectral holes in the reflection spectrum and as narrow transmission peaks in the transmission spectrum .
such windows are induced via coherent cancellation of the dipoles through a destructive quantum interference effect .
in addition , we have shown that this effect can be controlled by changing the number density of quantum emitters .
this destructive interference of two or more transition dipoles can be used for shaping laser pulses or for slow light .
the authors acknowledge support from the eu ( grant no .
itn-2010 - 264951 corinf ) and from the air force office of scientific research ( summer faculty fellowship 2013 ) | a detailed theory describing linear optics of vapors comprised of interacting multi - level quantum emitters is proposed .
it is shown both by direct integration of maxwell - bloch equations and using a simple analytical model that at large densities narrow transparency windows appear in otherwise completely opaque spectra .
the existence of such windows is attributed to overlapping resonances .
this effect , first introduced for three - level systems in [ r. puthumpally - joseph , m. sukharev , o. atabek and e. charron , phys .
rev .
lett . * 113 * , 163603 ( 2014 ) ] , is due to strongly enhanced dipole - dipole interactions at high emitters densities .
the presented theory extends this effect to the case of multilevel systems .
the theory is applied to the d1 transitions of interacting @xmath0 atoms .
it is shown that at high atomic densities , @xmath0 atoms can behave as three - level emitters exhibiting all the properties of dipole induced electromagnetic transparency .
applications including slow light and laser pulse shaping are also proposed . |
as acknowledged by nomenclature , descartes concept of coordinates depends on the notion of a cartesian product .
omnipresent in mathematics , it is already used in basic arithmetic to build number systems like the field of complex numbers or to define exterior bundles on manifolds .
the cartesian product allows to build and access higher dimensional features of geometric spaces .
many constructions in topology like suspensions , joins , fibre bundles , de rham cohomology or homotopy deformations would not work without the concept of a cartesian product .
of course , we would like to have a product in graph theory which shares the properties from the continuum .
the new graph product will achieve that .
it will allow to use coordinates " similarly as they are used in the continuum .
for two arbitrary networks @xmath39 - that is @xmath7 and @xmath11 are finite simple graphs - the coordinates of the product graph @xmath0 consists of all pairs of complete subgraphs of @xmath7 and @xmath11 .
the exterior derivatives in @xmath7 and @xmath11 will play the role of the partial derivatives " in the product and allow to build an exterior de rham derivative on the product graph @xmath0 .
the actual exterior derivative on the product graph @xmath0 operates on a much larger complex .
the later is called the whitney complex and is defined by the simplices in the product .
having a product allows to work with
rectangular boxes"s in the product space rather than with simplices .
figure ( [ poster ] ) illustrates this for a small example . to take a picture from the continuum : the curl in the plane is an infinitesimal line integral along a rectangle and uses the leibnitz rule to relate the exterior derivative of the factors with the exterior derivative @xmath40 of the product .
a simplicial point of view of cohomology integrates around infinitesimal triangles to get the curl , ignoring the product structure .
de rham establishes equivalence of the two pictures on any smooth manifold . in order to prove the knneth formula , we will need to emulate the de rham theorem combinatorially .
to do so , we explicitly construct a @xmath26-form on the whitney complex of @xmath0 from a @xmath26-form on the de rham complex defined by the two graphs @xmath39 .
the chain map is concrete as we can from this construct explicit cohomology classes of the product from the cohomology classes in @xmath7 or @xmath11 .
the relation is what one calls a chain homotopy
. the de rham picture will be useful if we work with
discrete n - manifolds " obtained by gluing together local charts of products @xmath41 of networks or when working with fibre bundles " @xmath42 obtained by gluing locally trivial charts @xmath43 of networks above a discrete manifold covered with charts @xmath44 .
the automorphism group of the fibre @xmath7 will then play the role of a gauge group on @xmath45 , as it does in the continuum .
+ when looking at graphs as geometric structures without dimension restriction , an amazing similarity with the continuum emerges .
it turns out that in the discrete , one can access the local structure of space @xmath7 directly as the complete subgraphs of @xmath7 .
these simplices can serve as fundamental entities playing the role of points " .
such insight has been promoted already in @xcite but most of the time still graphs are treated as one dimensional simplicial complexes .
an illustration on how the change of view point allows to emulate results from the continuum is the fixed point theorem of brouwer and lefschetz which looks identical to the result in the continuum @xcite .
many concepts become elementary : cohomology is part of finite dimensional linear algebra , to compute valuations , generalized volumes , one needs integral geometric tools in the continuum , while in the discrete is is just count of complete subgraphs .
the discrete hadwiger theorem @xcite is much easier than the continuum version : the numbers @xmath46 of @xmath47-dimensional simplices is a basis for the linear space of valuations .
differential forms are just functions on a simplex graph and stokes theorem in the whitney complex is a tautology ; it becomes only less obvious when stokes is considered in a de rham setup .
much intuition about higher dimensions can be obtained inductively .
there are notions of dimension , cohomology , homotopy , cobordisms , ramified covers , degree and index , spheres , geodesic lines and curvature which lead to results mirroring the results in the continuum .
this is not only true for nice geometric graphs but for general undirected networks or finite simple graphs - and much works without any exceptions .
the notion of homotopy of graphs for example immediately leads to the homotopy of a graph embedded in an other graph and so to homotopy groups for general finite simple graphs .
the product defined here will actually help to define the homotopy groups as graphs might be too small at first to have spheres embedded , so that the graph should first be refined .
but the hurewicz homomorphisms from the homotopy groups @xmath48 to the cohomology groups @xmath49 are then so explicit that one can even watch it happen : just apply the heat flow to a @xmath26-forms with support on the @xmath26-simplices on the embedded @xmath26-sphere .
it converges to a harmonic form which by hodge theory represents a cohomology classes . in the continuum
, such a proof requires de rham currents , generalized differential forms which require some functional analysis . in the graph case ,
the heat flow is just a linear ordinary differential equation of the type studied in introductory linear algebra courses .
the definition of homotopy groups which are relevant in coloring questions for graphs @xcite allows to work with spheres in graph theory in the same way as in the continuum .
+ the euclidean product is not only essential for defining fundamental objects like fibre bundles , it is also needed to construct spaces which have the same properties than classical manifolds .
examples of such properties are dimension , homotopy , cohomology or euler characteristic .
the goal of this note is to give such a product , allowing the use tools like discrete fibre bundles in graph theory .
we will see that this can be done purely algebraically : as we can glue graphs together , this gluing carries over to the product allowing to build discrete bundles . and
if the fibres carry an automorphism group @xmath50 , we get discrete analogues of principle bundles on which an enlarged gauge group acts .
+ a cartesian product for a category of a geometry needs to be dimension - additive , it needs to induce a product on the homotopy classes , it needs to be euler characteristic multiplicative , it must satisfy the knneth formula equating the tensor product of the cohomology rings with the cohomology ring of the product and it must have the property that the automorphism group contains the automorphism groups of the factors .
+ as for graphs , no previously defined product shares these properties .
the standard cartesian product @xmath51 of the cyclic graph @xmath52 with @xmath52 for example is a graph of dimension @xmath53 .
its vertices are the cartesian product of the vertices and two points @xmath54 are connected , if @xmath55 or @xmath56 .
the cohomology of the standard cartesian product has little to do with the cohomology of the factors : the betti numbers of @xmath57 for example is @xmath58 while the betti number of our product is @xmath59 , which is identical to the one of the two - dimensional torus in classical topology .
the dimension of the traditional product is @xmath53 while the dimension of our product is @xmath60 .
also other constructions like the tensor product " , the direct product " or the strong product " of graphs do not have the topological properties we want .
the reason for the shortcomings of all these products is that they do not tap into the lower dimensional building blocks of space : these are the simplices = complete subgraphs in the graph case .
what in the continuum has to be done with sheaf theoretical constructs , is already pre - wired in the graph as we can access the lower dimensional simplices as points " .
the new product has the property for example that @xmath61 is a @xmath62-dimensional torus of euler characteristic @xmath63 and betti vector @xmath64 .
it is a triangularization of @xmath65 and each unit sphere @xmath27 is a @xmath60-dimensional sphere of euler characteristic @xmath60 .
the product has practical use as it allows to construct high dimensional geometric spaces from smaller dimensional ones .
the product works for any pair of graphs and leaves geometric @xmath66-dimensional graphs invariant , graphs for which all unit spheres are @xmath67-dimensional homotopy spheres defined in @xcite .
the product will again have this property .
cohomology , dimension and homotopy properties of the product are identical to the properties in the continuum . even in the case of fractal dimension , the dimension formula matches the corresponding product formula in the continuum @xcite ( formula 7.2 ) for the hausdorff dimension of arbitrary sets in euclidean space .
there are more analogues : for hausdorff dimension , there are sets of dimension zero for which the product has dimension @xmath53 . while graphs of dimension @xmath63 are geometric , so that one gets equality , there are sequences of graphs like @xmath68 for which @xmath69 and @xmath70 , mirroring the continuum again .
an other useful feature of the product is that it allows to refine graphs : the enhanced graph @xmath71 is a barycentric refinement of @xmath7 and this can be repeated : the sequence @xmath72 produces a finer and finer mesh whose dimension converges to an integer and honor the original symmetries if @xmath38 is a subgroup of the automorphism group then @xmath50 still acts on the refinement .
the later is important when looking at discrete principal bundles . moreover , while @xmath37 is in general no more a graph , the quotient @xmath73 is , so that we can as in the continuum form covering spaces ramified over rather general subgraphs .
for example , while @xmath74 acting on @xmath75 has a quotient @xmath76 which is only a chain and no more a finite simple graph , the quotient @xmath77 is a graph . as an other example , the quotient @xmath78 of the octahedron modulo the @xmath79 action given by the antipodal reflection group @xmath80 is no more a geometric graph without boundary as it is the wheel graph @xmath81 for which we can see the octahedron as a double cover ramified over the equator ( riemann - hurwitz in the discrete is just the burnside lemma considered simultaneously for the various simplex sceletons as noted in @xcite ) , but @xmath82 is a geometric graph , a discrete projective plane . as unit spheres of geometric graphs
are spheres , the product can be used to construct new spheres . by taking refinements and then taking quotients
, one can get more general discrete graphs like the projective plane in the simplest case . the product construction can also help for other constructions , like constructing joins or building discrete hopf fibrations in arbitrary dimensions . as every unit sphere in the product has the structure @xmath83 which is the union of two solid tori glued along the torus @xmath84 . in the case
when taking the product of @xmath60-dimensional geometric graphs , we get so @xmath62-dimensional spheres which have naturally the same hopf fibration structure as in the continuum .
+ classically , the notion of a homotopy of two continuous maps @xmath85 is defined using the product : if there is a continuous map @xmath86 \to y$ ] , such that @xmath87 and @xmath88 , then @xmath89 are called homotopic .
this could be done now also for graphs : two graph homomorphisms @xmath90 are homotopic , if there exists a line graph @xmath91 and a graph homomorphism @xmath92 from @xmath93 to @xmath11 such that @xmath87 and @xmath94 .
we have suggested in @xcite an other definition : two graph homomorphisms @xmath90 are homotopic if the graph of the graph homomorphisms " are homotopic graphs .
these graphs of homomorphisms have as vertices the union @xmath95 and as edges all pairs @xmath96 and pairs @xmath97 with @xmath98 .
two graph homomorphisms are now homotopic , if the corresponding graphs are homotopic as graphs .
we believe that these two definitions are equivalent , but have not yet proven this . in any case , we have the whiteheads theorem that if there is a graph homomorphism @xmath99 which induces isomorphisms on homotopy groups @xmath100 , then @xmath39 are homotopic . and as in the continuum , having isomorphic homotopy groups does not force a homotopy equivalence . now , with a product we can take the same standard counter example @xmath101 and @xmath102 , where @xmath103 are graph implementations of the @xmath26-dimensional projective space and @xmath104 are @xmath26-spheres . since they have the same universal cover @xmath105 and the same fundamental group @xmath79 , they have the same homotopy groups but the knneth formula implies that the poincar polynomial of @xmath101 is @xmath106 while the poincar polynomial of @xmath102 is @xmath107 . having different cohomology groups
prevents the two graphs @xmath39 to be homotopic .
we see in this example , how useful it is to have a product in graph theory which shares the properties from the continuum .
the point is that one does not have to reinvent the wheel in graph theory but that one can piggy - pack on known topology .
+ from a practical point of view , the construction of the product only needs a few lines of code for a standard computer algebra system .
the full computer code is given in detail later on .
the product works for all finite simple graphs : first construct the ring elements @xmath108 which are polynomial in @xmath109 representing the vertices of @xmath7 and @xmath110 , the vertices of @xmath11 , then take the product @xmath111 and construct from this a polynomial in the variables @xmath112 , the new graph @xmath0 by connecting two monomial terms of a polynomial if one divides the other .
for example , if @xmath113 then @xmath114 and if @xmath115 then @xmath116 , we get @xmath117 which encodes a wheel graph @xmath118 with central vertex represented by @xmath119 .
we see that the product naturally extends to chains , group elements in the ring and that it corresponds to the product in the tensor product of the two rings .
every polynomial defines back a graph , but the later does in general not not agree with the original host graph .
for example , take @xmath113 and the chain element @xmath120 , which defines in turn a graph which is the disjoint union of @xmath121 ( represented by @xmath122 ) and @xmath123 represented by @xmath124 .
chains are useful constructs because natural operations escape the class of graphs : examples are forming the boundary @xmath125 or taking quotients @xmath37 with a subgroup @xmath50 of the automomorphism group of @xmath7 .
chains form a ring and have properties wanted from a geometric object : a dimension , a cohomology , a notion of homotopy , an euler characteristic , a notion of curvature and these notions match the results we expect from the familiar cases : results like gauss - bonnet @xcite , mckean singer and hodge - de - rham @xcite , poincar - hopf @xcite , brower - lefschetz @xcite , riemann - roch ( @xcite is a strong enough theory so that it can be extended to higher dimensions ) , integrable geometric evolutions @xcite , riemann - hurwitz @xcite , lusternik - schnirelmann @xcite , mirror the results in the continuum .
see also @xcite for an overview from the linear algebra point of view .
these results are more limited , if one considered graphs as one - dimensional simplicial complexes only , a common assumption taken in the 20th century . as graph theory is a discrete theory , it can surprise at first that the situation parallels the continuum so well but there are conceptual reasons for that , an example is non - standard analysis , an other is integral geometry .
some notions go over pretty smoothly , like cohomology or homotopy : others , like the notion of homeomorphisms for graphs needs more adaptations @xcite . + about the history : knneth found the formula in 1921 @xcite , where it was his dissertation under the guidance of heinrich tietze .
the paper published in 1923 @xcite is a linear algebra analysis which could probably be simplified considerably using hodge theory .
knneth attributes the cartesian product of manifolds to steinitz ( 1908 ) ( the definition is indeed given in @xcite p.44 is probably the first appearance of the simplicial product which our definition is based on ) . the notion of manifolds was initiated by poincar ( 1895 ) , weyl ( 1912 ) , veblen and alexander @xcite ( 1913 ) and whitney @xcite ( 1936 ) .
it was herbert seifert who introduced fibre bundles in 1933 @xcite .
the eilenberg - zilber theorem of 1953 @xcite .
( joseph abraham zilber was a boston born harvard graduate ( 1943 ) and phd ( 1963 ) .
interestingly , the eilenberg - zilber theorem was authored 10 years before zilber got his phd degree under the guidance of andrew gleason .
more information about the history of algebraic topology , see @xcite or the introduction to @xcite .
the classical cartesian product of graphs was introduced by whitehead and russell in principia mathematica 1912 ( it is @xcite who spotted the construction on page 384 in volume 2 of that epic work ) .
it was introduced there in the context of logical relations , not so much graph theory . this historical observation and many properties of the classical product product
are discussed in @xcite . +
as far as we know , our present paper is the first establishing a knneth formula for finite simple graphs .
there is a functorial approach to knneth for digraphs , where @xcite use path cohomology to get a functor from digraphs to cw complexes , so that one can then use the continuum result for the cw complexes .
note however that in their case , one gets knneth only indirectly by constructing a cw - complex , take the product using the cartesian embedding and then pulling the result again to graphs .
in particular , the notion of dimension is also borrowed from the continuum as cw - complexes are topological spaces .
a similar thing could be done for geometric graphs , graphs for which the unit spheres @xmath27 are homotopy @xmath19-spheres .
one can then build for each ball @xmath126 an open set in @xmath127 and use this to build a manifold @xmath128 .
the open sets form then a nice good cover and the topology of the manifold is the same than the topology of the geometric graph .
this construction is restricted however to geometric graphs .
+ the missing cartesian product bothered us for a while so that we decided to make a targeted search over several weeks ( while procrastinating from an urgent programming job still in need to be finished in geometric graph coloring ) , trying several possibilities and checking with the computer whether the construction works .
the product described here was obtained by trying random things which look beautiful " .
the current product is attractive when seen algebraically because it becomes associative on the algebraic level . until now
, we would just take the usual product and then fill out the chambers .
when working with graphs , this is cumbersome , even when staying within the discrete , as it costs programming effort to build examples , like stellated higher dimensional cubes .
the product which we propose here for graphs seems not to be known in the language of graphs .
one could of course take the product @xmath129 of the corresponding topological spaces as knneth did .
after finding the product , we looked around whether it already exists : closest to what we do here appeared as an exercise in an algebraic topology course by peter tennant johnstone at cambridge @xcite . more digging
revealed that this simplicial product has been used by eilenberg and zilber in @xcite ( page 204 ) , by de rham @xcite ( page 191 ) and earlier by steinitz in @xcite ( page 44 ) . but this product never made it to graph theory .
one could also get a product by escaping to the euclidean space .
our reluctance to use euclidean stuff is not because we feel like brouwer ( who would even refuse to accept the infinity of natural numbers ) , but simply out of pragmatism : we want to build the structures fast on a computer without having to use euclidean parts .
graphs are natural structures built into computer algebra languages and the euclidean embeddings are not needed , when doing computations ; only if we want to see the graphs visualized or modeling traditional geometric objects , the euclidean embedding is helpful . here is a self - contained full implementation of the new graph product in mathematica " , a computer algebra system which uses graphs as fundamental objects in its core language , structures which are dispatched from euclidean embeddings , unless they are drawn .
one procedures allow to translate a graph into a ring element and an other allows to get from a ring element back a graph .
the graph product is the product in the polynomial ring . ....
cliques[s_,k_]:=module[{n , t , m , u , q , v = vertexlist[s],w = edgelist[s],l } , n = length[v ] ; m = length[w ] ; u = subsets[v,{k , k } ] ; q = length[u ] ; l= { } ; w = table[{w[[j,1]],w[[j,2]]},{j , m}];if[k==1,l = table[{v[[j]]},{j , n } ] , if[k==2,l = w , do[t = subgraph[s , u[[j ] ] ] ; if[length[edgelist[t]]== binomial[k,2],l = append[l , vertexlist[t ] ] ] , { j , q}]]];l ] ; ring[s_,a_]:=module[{v , n , m , u , x},v = vertexlist[s ] ; n = length[v ] ; u = table[cliques[s , k],{k , n } ] /.
table[k->a[[k]],{k , n}];m = length[u ] ; x = sum[sum[product[u[[k , l , m ] ] , { m , length[u[[k , l]]]}],{l , length[u[[k]]]}],{k , m } ] ] ; gr[f_]:=module[{s={}},do[do[if[denominator[f[[k]]/f[[l]]]==1 & & k!=l , s = append[s , k->l]],{k , length[f]}],{l , length[f ] } ] ; undirectedgraph[graph[s ] ] ] ; graphproduct[s1_,s2_]:=module[{f , g , i , fc , tc } , fc = fromcharactercode ; tc = tocharactercode ; i[l_,n_]:=table[fc[join[tc[l],integerdigits[k]+48]],{k , n } ] ; f = ring[s1,i["a",length[vertexlist[s1 ] ] ] ] ; g = ring[s2,i["b",length[vertexlist[s2 ] ] ] ] ; gr[expand[f*g ] ] ] ; newgraph[s_]:=graphproduct[s , completegraph[1 ] ] ; example = graphproduct[completegraph[3],stargraph[4 ] ] .... the code of the above listing can be grabbed by looking at the source of the arxiv submission of this text .
+ can we write the usual product in an algebraic way ?
yes , if we have the ring element @xmath130 and take as vertices the @xmath63-dimensional simplices and as edges the @xmath53-dimensional edges , we recover the old graph .
if we take the product @xmath131 and take as vertices the pairs @xmath132 and as edges the triples @xmath133 , we regain the standard cartesian product . for example , if @xmath134 represents @xmath123 and @xmath135 represents @xmath136 , then the quadratic and cubic terms of @xmath131 are @xmath137 give the product graph which has @xmath138 vertices and @xmath139 edges .
in this section , we describe that the vertices of a graph define a ring in which every element @xmath130 can be seen as its own geometric object which carries cohomology , homotopy , euler characteristic , curvature , dimension and a dirac operator .
as we could run the wave or heat equation on such a ring element @xmath130 , the chain @xmath130 should be considered a geometric object with physical context . unlike the category of graphs , the category of these chains has an algebraic ring structure ,
is closed under boundary formation as well as taking quotients which leads to orbifolds in the continuum .
other constructs in the continuum like discrete stratifolds are already implemented as suspensions of geometric graphs and are represented by classical graphs .
we called them exotic in @xcite , where we asked whether they might lead to exotic discrete spheres ( a @xmath26-dimensional graph with geometric unit spheres which is homeomorphic to a @xmath26-dimensional homotopy sphere ) and looked at discrete varieties in @xcite .
somehow , chains form are part of a list of class of structures which resemble structures in the continuum geometric graphs @xmath140 stratifolds @xmath140 varieties @xmath140 graphs @xmath140 orbifolds @xmath140 chains .
the product goes all the way to chains and can also be used to lift the classical cartesian product to @xmath53-dimensional chains : it is obtained by taking the product and disregarding higher dimensional parts of the chain : kind of projecting the result onto curves .
+ we use multi - index notation @xmath141 . a finite simple graph @xmath7 with vertices @xmath142 defines a ring generated by all non - constant polynomial monoids @xmath143 and the chain ring .
with a given orientation , graphs are always represented by functions of the form @xmath144 with @xmath145 and @xmath146 .
the star graph for example is @xmath147 .
the ordering of the terms in the monoid allows in a convenient way define an orientation of the simplices in the graph @xmath7 which means fixing a basis for discrete differential forms @xmath148 .
+ we could add constants to get a ring with @xmath53 but we do nt yet see a use of the constants for now as we have no geometric interpretation for it .
the euler characteristic formula @xmath149 shows that the euler characteristic of @xmath53 is @xmath150 .
the @xmath53 element ca nt be interpreted as the empty graph , because the empty graph is represented by @xmath63 , and is a @xmath151-dimensional sphere with euler characteristic @xmath63 . with a @xmath53 element in the ring ,
one can produce terms like @xmath152 , an object of euler characteristic @xmath63 or @xmath153 , an object of euler characteristic @xmath154 .
+ to add an other footnote , we see that some chains represent already special differential forms , but these are forms taking values in the integers .
the triangle @xmath155 is a sum of a @xmath60-form , a @xmath53-form and @xmath63-form which are all constant @xmath53 . in this discrete setup ,
geometric objects and differential forms are already very similar , as it is custom for quantum calculus setups , where stokes is just the statement @xmath156 and where the boundary operation is truly the adjoint of the exterior derivative and both geometric objects and forms are in the same function space . for classical differential forms and geometric objects , the geometric objects are distributions ( as curves and surfaces for example are infinitely thin ) and differential forms are smooth or the dual setup used in geometric measure theory where differential forms are distributions and smooth functions are geometric objects .
its only on the level where both parts ( geometric objects and differential forms ) are represented by the same type of @xmath157 data that we have true symmetry , but then we are in a quantum setup . + every element @xmath158 defines a graph @xmath159 : the vertices of @xmath159 are the monomials of @xmath130 and two monomials of @xmath130 are connected by an edge if one is a factor of the other .
the graph of @xmath160 for example is the union of a line graph with three vertices where @xmath161 represents the middle vertex , as well as a single @xmath121 represented by @xmath122 .
the graph @xmath7 on the other hand defines the ring element @xmath162 , where @xmath163 is a simplex in @xmath7 .
the choice of the sign or permutation when writing down the polynomial monoid components corresponds to a choice of basis and is irrelevant for most considerations like for computing cohomology or when running discrete differential equations like @xcite . after fixing an orientation for each simplex , we get incidence matrices @xmath164 .
they implement the exterior derivative .
they also determine the dirac operator @xmath165 , and all such matrices are unitarily equivalent .
independent of the basis is the form laplacian @xmath166 which has a block decomposition into laplacians @xmath167 on @xmath26-forms .
the enhanced graph @xmath168 is a refinement of @xmath7 and as we will see , if @xmath7 is geometric , then the refinement is geometric again of the same dimension . for general networks
@xmath7 we will show that the dimension of @xmath9 can only increase , the reason being that higher dimensional parts will spawn off more new vertices " . for a triangle @xmath7 for example , the ring is @xmath169 $ ] and @xmath170 .
this ring element defines the graph @xmath171 for which the monomials @xmath172 are the vertices .
the divisor incidence condition leads to the wheel graph @xmath173 which has the same topological features as @xmath7 .
of course , the refinement process can be repeated , leading to larger and larger graphs with the same automorphism group .
the sequence of graphs @xmath174 define so larger and larger rings generated by more and more variables .
we will see that the dimension converges to an integer and that for large @xmath19 , the graph is close to a geometric graph as the larger simplices will overtake all others .
see figure ( [ euclideanlimit ] ) .
+ given two graphs @xmath39 , define the rings @xmath175 , take the product @xmath131 in the tensor product @xmath176 of rings and translate that product ring element back to a graph .
the dimension of @xmath7 is defined inductively as @xmath53 plus the average of the dimensions of the unit spheres and uses the foundation assumption that the empty graph has dimension @xmath150 .
a graph is geometric , if every unit sphere is a homotopy sphere .
a homotopy sphere is a geometric graph for which removing any vertex renders the graph contractible .
inductively , a graph @xmath7 is called contractible , if there exists a vertex @xmath177 such that @xmath27 and @xmath178 are both contractible , using the inductive assumption that the @xmath53-point graph is contractible .
homotopy can be defined algebraically .
inductively , a ring element @xmath31 is contractible if there exists @xmath179 such that both ring elements @xmath180 and @xmath181 are contractible .
the euler characteristic of @xmath7 is @xmath182 , where @xmath183 is the number of @xmath26-dimensional simplices @xmath184 in @xmath7 .
more generally , the euler characteristic of a ring element @xmath130 is @xmath185 .
the euler characteristic of the triangle @xmath186 for example is @xmath187 .
the euler characteristic of the chain @xmath188 is @xmath189 .
the boundary of a ring element @xmath130 is defined as @xmath190 , where @xmath191 are the usual partial derivatives and the result is projected onto functions satisfying @xmath192 . due to the orientation assumption , this gives rise to sign changes .
for example , @xmath193 if the edges basis @xmath194 was chosen .
we have @xmath195 .
the exterior derivative on @xmath196 is defined as @xmath197 . as @xmath198
, we have cohomology groups for any ring element .
let @xmath199 be the betti numbers .
the euler - poincar formula @xmath200 follows from linear algebra .
the boundary @xmath201 is no more a graph in general .
the star graph @xmath202 for example has the boundary @xmath203 which is a chain .
the space of differential forms @xmath196 is a direct sum @xmath204 , where @xmath205 is generated by functions supported on @xmath26-simplices , polynomial monoid parts in the ring of degree @xmath206 .
since @xmath148 are finite dimensional , the maps @xmath164 are represented by finite matrices .
they are called the incidence matrices and were considered by poincar already for triangulations of manifolds . the cohomology groups @xmath49 and more generally @xmath207 are independent of the chosen signs , when defining the chain ring element @xmath31 .
the matrix @xmath165 is called the dirac matrix of @xmath130 .
its square @xmath166 is the form - laplacian .
it decomposes into matrices @xmath208 . by hodge theory ,
the dimension of the kernel of @xmath167 is the @xmath26th betti number @xmath209 .
given a ring element @xmath130 and a monoid part @xmath177 in @xmath130 , its unit sphere @xmath27 is the unit sphere of @xmath177 in the graph defined by @xmath130 .
it consists of all monoids dividing @xmath177 or which are multiples of @xmath177 , without @xmath177 .
it is again a ring element .
the dimension @xmath210 is inductively defined as @xmath53 plus the dimension of the unit sphere .
the dimension of @xmath130 finally is the average of the dimensions of all monoids in @xmath130 .
the dimension of the triangle @xmath155 for example is @xmath211/7 $ ] .
we have for example @xmath212 which defines the graph @xmath213 of dimension @xmath53 and @xmath214 which is a line graph @xmath215 of dimension @xmath53 etc .
we see that the dimension of the triangle is @xmath60 . of course ,
in the graph case , the dimension can be better computed on a graph level .
the point is that the dimension extends to a nonnegative functional on the entire ring in such a way that the average of the dimensions of unit sphere @xmath27 of a monoids or generalized vertices of @xmath130 is the dimension of @xmath130 minus @xmath53 .
+ we also need an inner product @xmath216 on @xmath205 which is defined as @xmath217 @xmath218 .
it obviously satisfies all properties of an inner product and especially defines a length @xmath219 . of course , the incidence matrices @xmath220 are adjoint to each other with respect to this product . as graphs are special functions , we could use the inner product for example to define an angle between two graphs @xmath39 on the same vertex set it is the @xmath221 of the fraction @xmath222/(\sigma[g ] \sigma[h])$ ] , where @xmath223 $ ] is the number of common simplices of @xmath39 and @xmath224 $ ] is the square root of the number of simplices in @xmath7 . lets summarize the main point : \a ) every graph @xmath225 defines a ring element @xmath226 , a sum over all complete subgraphs of @xmath7 .
+ b ) every element @xmath227 defines a graph @xmath159 , where @xmath228 are the monoid entries @xmath229 in @xmath130 , with @xmath230 and where two entries are connected if one divides the other .
+ c ) if @xmath130 comes from a graph @xmath7 , then @xmath231 is a graph for which the original simplices are the points and which has the same topological features than @xmath7 and which additionally has a natural digraph structure . in other words
, there is a functor @xmath232 from the ring @xmath233 to the category of directed graphs given by the division properties ( even so we often forget about the directions ) and that there is a second functor @xmath234 from the category of undirected graphs to the ring .
they are not inverses of each other but the cohomology agrees . for other functorial relations ,
see the recent paper @xcite .
+ * remarks : * + * 1 ) * every ring element @xmath130 also defines its own geometric object which has euler characteristic , cohomology , dimension , homotopy as well as curvature . + * 2 ) * when forgetting about the anti - commutativity within the graph which is irrelevant for the graph product , the ring could be replaced with a more general integral domain .
it would allow to see the graph product @xmath0 as an element in @xmath235 $ ] where @xmath236 $ ] .
this possibility can be useful when studying fibre bundles as one can work in a ring of the fibre graph . + * 3 ) * the fact that the product of two graphs is obtained by writing the graphs algebraically using different generators and producing from it again a graph : @xmath237 is not unfamiliar to us .
if we take the cartesian product of two spaces , we use different variables for the different directions .
if we do nt take new variables and take @xmath238 in the ring , then this is in general a chain . for a triangle @xmath239 for example ,
we get ( using @xmath240 ) the chain @xmath241 .
the graph which belongs to this chain is the star graph @xmath242 as @xmath243 is the central vertex and the others the outer points ( as they divide the central point ) .
+ the imposed pauli principle imposed by anticommutativity @xmath244 is irrelevant for the euler characteristic , both for chains as well as for the graphs @xmath159 derived from the ring elements @xmath130 .
+ * 4 ) * the graph @xmath245 can be seen as a self - cobordism of @xmath9 with itself as it is a graph of one dimension more which has two copies of @xmath9 as boundary .
it is in general true that any geometric graph is self cobordant to itself as we can sandwich two copies @xmath7 with a completed dual graph @xmath246 .
but for @xmath9 we do nt have to work on a construction .
it is given .
+ * 5 ) * denote by @xmath247 the projection onto the linear subspace generated by @xmath26 until @xmath29-dimensional simplices .
as pointed out before , we can recover @xmath7 from @xmath31 by building @xmath248 and then building @xmath249 which is @xmath7 . with @xmath250 ,
then @xmath249 is the classical standard cartesian product of @xmath7 and @xmath11 .
the results for the graph product mirror results in the continuum .
first we look at some basic constructions which deal with the notion of homotopy sphere or simply @xmath26-sphere in graph theory .
the definition of a @xmath26-sphere in graph theory is recursive : a @xmath26-sphere is a @xmath26-dimensional geometric graph for which every unit sphere @xmath27 is a @xmath28-sphere and such that after removing any of its vertices , we get a graph which is contractible .
a @xmath26-ball is a @xmath26-dimensional geometric contractible graph with boundary which has a @xmath28-sphere as its boundary .
we call the interior of a ball the part of @xmath251 which is not in the boundary . a suspension @xmath252 of a graph @xmath7 is the join @xmath253 , a double pyramid construction : add two new points @xmath254 and connect the points to all the vertices of @xmath7 . a pyramid construction itself
is the join @xmath255 . from the cartesian product ,
we have construct joins by just building a product and identifying some variables in the algebraic representation of the graph .
examples are given at the end .
the join @xmath255 of a @xmath26-sphere @xmath7 is a @xmath256-ball .
the suspension @xmath257 of a @xmath26-sphere @xmath7 with a @xmath63-sphere @xmath123 is a @xmath256-sphere .
\a ) by definition , the boundary of @xmath258 is @xmath259 which is a sphere . also , the graph @xmath258 is contractible .
+ b ) removing the second point @xmath260 produces the ball @xmath258 by a ) .
* remark : * + * 1 ) * more generally , as in the continuum , and shown below , the join @xmath261 of two spheres is a sphere @xmath262 .
+ * 2 ) * also as in the continuum , the definition of the join needs the product as the join @xmath263 is a quotient of @xmath264 .
it generalizes that the 3-sphere can be written as @xmath265 , which has an interpretation of gluing two solid tori along a torus . + * examples : * + * 1 ) * the wheel graph @xmath266 is a 2-ball .
it is the join @xmath267 , where @xmath75 is the cyclic graph with @xmath19 vertices .
+ * 2 ) * the @xmath26-dimensional cross polytope is @xmath268 , where we have @xmath256 factors .
the square @xmath52 is equal to @xmath269 , the octahedron is @xmath270 etc .
assume @xmath271 are @xmath26-balls with boundaries @xmath272 and assume that @xmath273 is a @xmath28 ball with boundary @xmath274 . then @xmath275 is a @xmath26-ball with boundary @xmath276 .
this is proven by induction with respect to @xmath26 .
there are four things to show : + a ) @xmath251 is contractible .
+ b ) every unit sphere in the interior of @xmath251 is a @xmath28-sphere .
+ c ) every unit sphere in @xmath259 is a @xmath277-sphere .
+ d ) when removing a vertex from @xmath259 , we get a contractible graph .
+ for a ) take a point @xmath260 in @xmath278 .
we can retract everything in @xmath279 to @xmath260 .
+ for b ) we only have to look at a vertex @xmath177 in @xmath278 .
the unit ball @xmath126 decomposes for c ) , we only have to look at a vertex @xmath280 in @xmath242 and see whether its unit sphere in @xmath259 is a homotopy sphere . for d ) , we can retract a pointed part to @xmath242 .
the next statement is the discrete analogue of the classical statement that the boundary of the product of two balls @xmath281 is a sphere and that it can be written as as @xmath282 which is the union of two solid tori glued at a torus .
also in the discrete , we can use the intuition from the continuum : if @xmath271 are @xmath283-balls with @xmath284-spheres @xmath285 as boundary , then @xmath286 is a @xmath287-sphere provided @xmath288 .
use induction with respect to dimension .
the union is a graph of dimension @xmath289 .
a unit sphere @xmath290 is in @xmath291 or @xmath292 whose intersection is @xmath293 .
* examples : * + * 1 ) * if @xmath271 are two line graphs with boundary @xmath272 , then @xmath294 is a union of two line graphs .
similarly @xmath295 is the union of two line graphs .
the union is a square , the intersection consists of the four points @xmath296 . +
* 2 ) * if @xmath297 is a ball with @xmath60-dimensional sphere @xmath298 and @xmath299 is a @xmath53-dimensional ball with @xmath53-dimensional sphere @xmath300 , then @xmath294 is the mantle of the cylinder and @xmath295 are the top and bottom cover . + * 3 ) * if @xmath297 is @xmath53-dimensional and @xmath301 is @xmath63-dimensional , then @xmath302 is a @xmath63-sphere . +
* 4 ) * if @xmath303 are two - dimensional balls , we can see @xmath304 is the union of two solid tori glued along a @xmath60-torus .
this hopf fibration is classically given as the split of @xmath305 into two solid tori @xmath306 and @xmath307 intersecting in the @xmath60-torus @xmath308 .
among various other flavors of cohomologies , there are three equivalent cohomologies for compact @xmath19-manifolds : simplicial cohomology , de rham cohomology and ech cohomology . for simplicial cohomology , the manifold is triangulated into finitely many @xmath19-simplices leading to a differential complex . in de rham cohomology ,
one works with the complex of differential forms , partial derivatives tap into the local product structure of the manifold , for ech cohomology , the manifold is covered with a finite cover of open sets so that the nerve graph determines the cohomology .
each of the these cohomologies have advantages over the others : simplicial cohomology is the computer science or combinatorial point of view which sees space as a mesh of small simplicial building blocks , the de rham cohomology is the analysis or calculus approach , which taps into the bag of techniques used in calculus . in this flavour ,
the basic building blocks are cubes obtained from a local product structure and the different directions are accessed with partial derivatives which we think of the exterior derivatives in each factor .
the ech cohomology finally is the homotopy or topologal point of view which relies on the fact that cohomology is more robust and transcends dimension .
simplicial cohomology is straightforward and the simplest .
de rham cohomology requires local charts which are products and taps into the differential structure of the manifold , allowing for an efficient computation of cohomology .
ech cohomology finally is a flexible variant which illustrates best the homotopy invariance of cohomology .
it also can lead to considerable complexity reduction as on can for example retract a space to a much smaller dimensional set .
a solid torus for example can be retracted to a circle .
the equivalence of simplicial cohomology with de rham cohomology is due to de rham .
later proofs of this theorem use the equivalence of ech cohomology with simplicial cohomology .
+ all three cohomologies have analogue constructions in graph theory .
it is our goal to introduce the analogue of de rham cohomology and give a discrete analogue of the de rham theorem .
of course de rham cohomology emerges only in the discrete if one has a product which is compatible .
+ the simplicial cohomology of a graph @xmath7 is the clique cohomology of @xmath7 using the whitney complex of all complete subgraphs .
it is the oldest and has already been considered by poincar , even so not in the language of graphs . + the ech cohology is the cohomology of the nerve graph of a good cover , where good " in the discrete means that the nerve is homotopic to @xmath7 .
ech cohomology has first been considered for graphs in @xcite , a paper which proposes a notion of what a continuous map between graphs is , which is more tricky than one might think at first , as classical topology badly fails as the topology generated by the distance is discrete making it unsuitable .
what is important for a good notion of continuity is to merge homotopy with dimension .
the equivalence of discrete ech cohomology with discrete simplicial cohomology is there by definition just relying on the fact that homology is a homotopy invariant .
+ what was missing so far in graph theory is an analogue of de rham cohomology , where cubes " rather than simplices " play the fundamental role .
but one can not really look at a de rham cohomology for general graphs , if one does not have a cartesian product for which there is compatibility .
a de rham type cohomology has been mentioned in @xcite , where a generalized path cohomology introduced by @xcite is considered for digraphs .
as their approach uses a functor from graphs to cw complexes and pulls back results from the product of cw complexes to digraphs , there is no relation with what we do here .
other takes on discrete de rham cohomology study discrete notions for numerical purposes @xcite or @xcite .
our approach to de rham cohomology is purely combinatorial and restricted to finite constructions .
+ as in the continuum , the de rham complex for a product graph @xmath0 can use some derivatives also in the discrete , but this is merely language : while in simplicial cohomology , we write @xmath309 @xmath310 for the boundary of a simplex @xmath311 in the de rham approach , we can write @xmath312 in the algebra which is the same thing . it just uses the derivative notion in a formal way
. the de rham connection will be needed in the knneth connection , where we look at the kernel of form laplacians .
knneth will then be quite obvious . without linking de rham with simplicial cohomology
the relation is nontrivial , as the dimension of the space of differential forms on the product graph is much larger than the product of the dimensions of the space of differential forms on the factors .
already knneth had to work though such difficulties and needed dozens of pages of linear algebra reductions to tackle the issue .
our situation is also different in that we look at the product of two arbitrary networks @xmath39 which by no means have to be geometric .
+ the analytic de rham approach allows the derivative @xmath313 to be written as a sum of products @xmath314 which is leibniz formula and which reduces the exterior derivative of the product to the exterior derivative of the factors in the same way than the gradient , curl or divergence reduces the exterior derivative to partial derivatives , which are the exterior derivatives in the @xmath53-dimensional factors . to illustrate this with school calculus : infinitesimally , the curl of @xmath315 is @xmath316 . as a graph theorist we look at @xmath317 ( for fixed @xmath260 ) as the @xmath53-form restricted to the first coordinate @xmath177 which means a function on edges of the first graph and at @xmath318 ( for fixed @xmath177 ) as the @xmath53-form restricted to the second coordinate which corresponds to edges in the second graph .
the curl is so a line integral along a square .
when reducing this to simplicial cohomology , the square needs to be broken up into triangles .
our product does that very explicitly even so it needs some care as the tensor product of finite dimensional algebras has a completely different dimension in general than the cartesian product .
knnneth needed dozen of pages of rather messy linear algebra reductions to achieve this . the language of polynomials and the de rham connection will allow us to make this more clear .
+ when we start with a triangulated picture , there are no more two distinguished directions present . in graph theory
, we only can distinguish two directions if we look at the product @xmath0 of two graphs . as for manifolds
, this structure could be allowed to be present locally only ; its important however that a product structure must be present before we can even talk about discrete de rham cohomology . as mentioned before
, there is the possibility to see a graph as a triangularization of a manifold and use the euclidean product structure to emulate a discrete de rham cohomology .
notice however that this does not tickle down to the discretization .
the relation would only exist functorially and is pretty useless when working with concrete networks .
we will not leave the discrete realm and show that @xmath26-forms on the product space can be related to products of forms in the two factors . and also
, we do not only work with geometric graphs , which can be seen as discretizations of manifolds ; we work with general finite simple graphs .
+ a finite simple graph @xmath7 naturally comes with a simplicial complex , given by the set of all the complete subgraphs of @xmath7 .
this so called whitney complex can be encoded algebraically in the ring of polynomials . for a triangular graph for example
, we have the ring element @xmath319 , where the choice of the orientations of all the simplices is done arbitrarily .
the ring element @xmath320 in turn defines a new graph @xmath9 in which the polynomial monoids form the vertices and two vertices are connected if one divides the other .
it is important that it actually can be seen as a digraph , the direction is given which part is a factor of the other . in some sense , going from @xmath7 to @xmath9 frees us from having to chose an arbitrary orientation as the structure is now built in . by the way , @xmath9 appears to have other nice features like having the eulerian property allowing therefore a geodesic dynamical system @xcite .
+ in our triangular graph example @xmath7 , we get a graph @xmath9 with @xmath321 vertices because there are @xmath321 complete subgraphs of the triangle .
the graph @xmath9 is a wheel graph which shares the topological and cohomological properties with the triangle .
it is even homeomorphic to the triangle in the sense of @xcite as we can find a @xmath60-dimensional open cover whose nerve is the triangle . indeed : + the ech cohomology of @xmath9 is equivalent to the graph cohomology of @xmath7 . + the observation that @xmath7 and @xmath9 are homotopic proves that the ech cohomology of a graph with respect to a good cover in the sense of @xcite is the same than the graph cohomology . as in the continuum , the relation between simplicial and de rham cohomology is not completely obvious as we will just see .
+ now , if we take the product @xmath0 of two graphs , like for example two complete graphs @xmath322 ( illustrated in figure ( [ figurederham ] ) ) , where the ring elements @xmath323 and @xmath324 encode the graph , then the product @xmath0 is encoded by a ring element @xmath325 which by looking at division properties of the polynomial monoids produces the graph @xmath0 with @xmath139 elements .
it is the wheel graph @xmath326 , which is a discrete square with @xmath327 edges and @xmath328 triangles .
the laplacian @xmath329 with the usual exterior derivative is a block matrix decomposing into a @xmath330 block @xmath331 , a @xmath332 block @xmath333 and a @xmath334 block @xmath335 .
the algebraic representation of @xmath336 is a polynomial element with @xmath337 monoid terms . the graph @xmath338 associated with @xmath336 would have @xmath339 vertices already . from topological , algebraic or homotopical considerations ,
the graphs @xmath340 and @xmath338 are equivalent : they are homeomorphic , they are homotopic and have the same cohomology and dimension . + a boundary on the product is defined by the leibniz formula : @xmath341 which then defines an exterior derivative @xmath66 . as for ech cohomology , we will see that there is an advantage in that we can work with a smaller dimensional vector spaces : the spaces @xmath148 and @xmath342 and even the tensor product @xmath343 have in general smaller dimension than the vector spaces @xmath344 .
see figure ( [ poster ] ) . having equivalence of cohomology can be a blessing when doing computations .
+ this will be useful when looking at discrete manifolds : de rham cohomology can be used more generally also when gluing product graphs to build fibre bundle , this product structure does not have to be global .
it is the same situation as for manifolds have in general only locally neighborhoods which can be written as products . + the de rham cohomology for a product graph @xmath0 with boundary operation defined by the leibniz rule @xmath345 using the exterior derivatives @xmath66 on each factor @xmath7 and @xmath11 , is isomorphic to the cohomology on @xmath0 defined by the exterior derivative @xmath66 on the graph @xmath0 . in short ,
@xmath346 after de rham @xcite , new proofs of the de rham theorem in the continuum were given by a. weil ( 1952 ) @xcite and h. whitney @xcite .
weil had outlined a proof already in 1947 in a letter to h. cartan , a letter which initiated cartan s theory of sheaves .
argument used a staircase argument in a double complex .
the textbook proof of bredon @xcite is worked out in more detail in @xcite . see also @xcite . in the discrete , we can directly construct the chain maps @xmath347 and @xmath348 .
as we will see below , our goal will be to find a one - to - one correspondence between harmonic forms on the large simplex laplacian @xmath329 , which is a @xmath349 matrix and the de rham laplacian @xmath350 , which is a @xmath351 matrix .
this is done by a rather explicit chain homotopy .
assume @xmath39 have vertices @xmath109 and @xmath110 .
the de rham cohomology on the ring generated by functions @xmath131 is defined by using the exterior derivatives on each product using the leibniz rule .
we have to relate it with the whitney chain complex @xmath352 which features functions on the simplices of @xmath0 .
the linear spaces @xmath353 and @xmath344 have different dimensions @xmath354 and @xmath355 so that @xmath347 is not the inverse of @xmath348 ( they are chain homotopic only , which essentially means that they are equivalent modulo coboundaries on each side ) . in both cases ,
denote by @xmath356 and by @xmath357 the exterior derivatives .
we will construct two linear maps @xmath358 and check that they are chain maps : @xmath359 and @xmath360 .
this will establish the isomorphism of cohomology @xmath361 to @xmath362 . + * construction of @xmath363 : * + start with an @xmath26-form @xmath364 . as it can be written as @xmath365 in the variables
@xmath112 , it defines a function on the vertex set of @xmath0 .
given a @xmath26-simplex @xmath366 in the graph @xmath0 .
assigned to it the value of the vertex which belongs to a monomial of degree @xmath26 . if there is one , it is unique
. if there is none , assign the value @xmath63 . + * construction of @xmath367 : * + for every @xmath26-form @xmath92 in @xmath344 we build a polynomial @xmath368 as follows : the function @xmath92 defines a value @xmath369 to the vertices of @xmath0 by averaging the @xmath92 values of the @xmath26-simplices hitting @xmath177 .
we have now a function @xmath369 on the vertices of @xmath0 , written as a polynomial @xmath370 in the variables @xmath112 .
one can now add a polynomial @xmath317 of smaller degree so that @xmath371 can be factored as @xmath365 .
+ * @xmath348 is a chain map : @xmath372 * : + proof : given a @xmath28-de rham form @xmath373 , build the @xmath26-form @xmath374 using the leibniz rule , then use @xmath348 to get from this a @xmath26-form @xmath375 on @xmath0 . by linearity
, we can assume that @xmath376 with @xmath377 .
the function @xmath378 takes the value @xmath379 on each polynomial @xmath380 with @xmath381 .
now , @xmath375 is a @xmath26-form on @xmath0 which takes the value @xmath379 on those vertices @xmath380 . but @xmath382 is a @xmath383 form which assigs the value @xmath50 to every simplex having @xmath384 as a vertex . now @xmath385 is exactly @xmath50 if @xmath386 is of the form @xmath380 . +
* @xmath347 is a chain map : @xmath387 * : + proof : given a @xmath383-form @xmath92 on the product graph @xmath388 , we build the @xmath256-form @xmath374 and produce from this a polynomial @xmath389 in the de rham complex .
we check that this is the same than @xmath390 . by linearity
we can look at a @xmath383-form @xmath92 which assigns the value @xmath50 to a @xmath28 simplex @xmath366 of @xmath0 .
now take a @xmath26-simplex which has @xmath366 in its boundary , then @xmath391 .
but now , @xmath392 assigns the value @xmath50 to the vertex @xmath280 which is in @xmath366 and has degree @xmath26 .
now , @xmath393 assigns the value @xmath50 to the vertex @xmath394 which is in @xmath366 and has degree @xmath383 .
this is connected to @xmath280 so that @xmath395 is the same value .
it would be nice to have a more intuitive understanding of the function @xmath396 satisfying the chain homotopy condition @xmath397 the map @xmath398 is related to the lower degree polynomial @xmath317 added to @xmath92 so that @xmath399 .
this involves solving a system of linear equations .
the dimension of the space of scalar functions on @xmath0 is @xmath400 , where @xmath401 is the number of simplices in @xmath7 .
the space of functions @xmath365 has dimension @xmath402 .
the image of @xmath398 consists of all lower degree polynomials ( as usual satisfying @xmath403 ) which is a space of dimension @xmath404 . to solve for @xmath405 we have to solve a system of equations for @xmath19 variables and this solution gives us @xmath317 .
@xmath406^d \ar[d]^\psi & \omega_{dr}^k \ar[d]^\psi \ar[ld]_\theta \\ \omega^{k-1 } \ar[r]^d \ar@<1ex>[u]^\phi & \omega^k \ar@<1ex>[u]^\phi } \ ] ] * examples : * + * 1 ) * if @xmath7 is a forest with @xmath19 trees and @xmath11 is a forest with @xmath407 trees , then the product graph @xmath0 has @xmath408 components .
the betti vectors are @xmath409 and @xmath410 .
every harmonic form on the product space @xmath0 is of the form @xmath365 , where @xmath89 are both locally constant . because the number of vertices in @xmath0 and @xmath63-simplices in @xmath0 are the same , we do nt have to translate between simplicial and de rham @xmath63-forms . + * 2 ) * let @xmath411 , then @xmath0 is a @xmath60-dimensional ball with @xmath139 vertices @xmath412 . given a @xmath53-form @xmath92 in the product graph @xmath0 .
it is a function on the edges of @xmath0 . given a vertex which is the product of a vertex and edge like @xmath413
, we assign the sum over all edge values hitting the vertex . +
* remarks : * + * 1 ) * in the complement of the set @xmath233 of chains which are a product of graphs , the equivalence does hold any more .
take the chain @xmath414 .
its cohomology is different from the corresponding graph @xmath159 which is the @xmath62 vertex graph without edges .
+ * 2 ) * the graph @xmath159 of an abstract chain @xmath130 has as many vertices as there are simplices in @xmath130 .
the simplicial complex of @xmath159 therefore is already large .
but the linear space of all @xmath26-forms on @xmath0 is huge : if @xmath401 is the number of simplices of @xmath7 and @xmath415 the number of simplices in @xmath11 , then @xmath416 is the dimension of the set of @xmath26-forms on @xmath0 .
the de rham complex is more manageable on a computer .
if @xmath39 are geometric graphs of dimension @xmath417 then @xmath0 is a geometric graph of dimension @xmath418 .
we have only to show that @xmath290 is a homotopy sphere . depending on @xmath97
the graph is the union of two cylinders intersecting in a lower dimensional graph which is part of the sphere @xmath419 .
use the cylinder lemma .
if @xmath39 are geometric spheres of dimension @xmath420 then the join @xmath421 is a geometric sphere of dimension @xmath422 .
it is a graph of dimension @xmath423 as it is a quotient of a graph @xmath424 , where @xmath425 is a line graph with @xmath426 .
this assures that for all points which are not affected by the identification , we have a geometric unit sphere of dimension @xmath427 . for the vertices
@xmath260 which were subject of identification , the unit sphere @xmath428 is a union of two @xmath429-dimensional cylinders which by the cylinder lemma is a sphere . besides verifying that every vertex @xmath260 has a unit sphere @xmath428 which is a @xmath429-sphere , we have also to check that taking away a vertex @xmath177 from @xmath421 , leads to a remaining graph which is a contractible ball of dimension @xmath19 which has a @xmath429-dimensional boundary .
take @xmath6 . given a vertex @xmath430 .
it corresponds to a simplex in @xmath7 .
the set @xmath27 consists of all simplices in @xmath7 containing @xmath177 united with the set of all simplices inside @xmath177 .
+ * examples : * + * 1 ) * let @xmath7 be a cyclic graph @xmath75
. lets look at the sphere of the vertex @xmath177 which is now also a vertex in @xmath9 .
it becomes now the set of two vertices @xmath431 in @xmath9 corresponding to two edges in @xmath7 .
this is a @xmath63 dimensional sphere . + * 2 ) * let @xmath432 again and let @xmath433 be an edge in @xmath7 which becomes now a simplex in @xmath9 .
its sphere consists of the two simplices @xmath434 which correspond to two vertices in @xmath7 .
the sphere is @xmath435 , again a geometric graph . +
* 3 ) * in the case @xmath436 like if @xmath7 is an ecosahedron , assume that @xmath437 is the old unit sphere of a vertex @xmath177 .
the new sphere of @xmath177 in @xmath9 contains the points @xmath438 which correspond to edges in the old graph @xmath7 .
the triangles @xmath439 are the new connections in that sphere .
we have again a geometric graph . *
4 ) * now assume that @xmath433 is an edge in the old two dimensional graph @xmath7 which becomes a vertex in @xmath9 .
the sphere @xmath440 consists of the vertices @xmath254 which are the third points in the triangles @xmath441 as well as the two triangles @xmath441 .
these two original vertices and triangles form 4 points in the new graph @xmath9 which form a circular graph @xmath52 .
this is our unit sphere . + * 5 ) * now assume that @xmath442 is a triangle in the two dimensional graph @xmath7 .
it becomes a vertex in the new graph @xmath9 .
the unit sphere @xmath443 consists of all simplices @xmath444 which is a @xmath138-gon . examples 3)-5 ) together show that that @xmath9 is always geometric if @xmath7 is two dimensional .
the unit spheres always are @xmath52 and @xmath213 . in particular
, @xmath9 is eulerian . + * 6 * ) in the case @xmath445 , if @xmath177 is an original vertex , then the sphere consists of all edges and triangles in the original sphere by induction this is @xmath446 . if @xmath447 is an edge
then the sphere consist of the points @xmath254 as well as all triangles and tetrahedra containing @xmath448 .
this is a double suspension of a @xmath53-dimensional sphere and so @xmath60-dimensional .
if @xmath449 is a triangle , then the unit sphere is a suspension of the @xmath138-gon @xmath450 . +
a theorem from 1964 assures that only spheres are manifolds which are joins @xcite . since geometric graphs naturally define compact manifolds , where unit balls are filled up to become the charts , this holds also for geometric graphs .
intuitively , the reason is clear . in order that the unit spheres of the identified end points are spheres
, we better need the factors @xmath451 of the join @xmath263 to be spheres , so that @xmath7 is of the form @xmath263 where @xmath451 are spheres .
+ if @xmath7 is a @xmath19-dimensional geometric graph which is the join @xmath263 for two other graphs , then @xmath7 is a homotopy sphere .
now , we come to the main result as announced in the title : given two finite simple graphs @xmath39 .
the cohomology groups of @xmath0 are related to the cohomology groups of @xmath7 and @xmath11 by @xmath452 we know from hodge theory that the de rham laplacian @xmath453 of @xmath7 restricted to @xmath26-forms has a kernel @xmath454 of dimension is @xmath49 .
+ we have @xmath455 and @xmath456 , so that @xmath457 hodge theory also assures that that @xmath458 is equivalent to @xmath459 , so that if @xmath89 are harmonic forms in @xmath7 or @xmath11 respectively , then @xmath131 is a harmonic form in @xmath0 .
we see that we can construct the cohomology classes of the product @xmath460 from the cohomology classes of the factors @xmath39 .
it is obvious therefore that @xmath461 contains the vector space @xmath462 .
+ now assume that @xmath131 is harmonic in @xmath0 .
we want to show that @xmath463 or @xmath464 .
again we use that the kernel of @xmath465 is the intersection of the kernel of @xmath66 and the kernel of @xmath466 .
now look at the equation @xmath467 where @xmath130 is a @xmath47-form and @xmath320 is a @xmath468-form .
when we look at the terms , then @xmath469 are polynomials in @xmath179 of degree @xmath470 and @xmath471 are polynomials in @xmath472 of degree @xmath473 . but
@xmath474 consists of polynomials of degree @xmath475 in @xmath179 and of degree @xmath468 in @xmath472 and @xmath476 consists of polynomials of degree @xmath475 in the @xmath179 and of degree @xmath473 in @xmath472 .
because the degrees are different , we see that @xmath477 .
now , either @xmath478 which implies @xmath463 .
but then , @xmath479 so that also @xmath480 .
an other possibility is @xmath481 which implies @xmath464 and in turn implies @xmath463 .
a third possibility is @xmath482 .
but from @xmath483 follows then @xmath484 .
now take the inner product with @xmath131 to get @xmath485 which implies that @xmath486 , demonstrating that the third case is not possible .
we have seen that @xmath463 and @xmath464 and verified that every kernel element in @xmath487 necessarily has to have the property that both @xmath463 and @xmath464 .
+ having established that the de rham cohomology of @xmath0 satisfies the knneth formula and that the de rham theorem assures that it is equivalent to graph cohomology , we are done . *
remark : * in the continuum , the equivalence of the chain complex @xmath488 with the tensor product @xmath489 on which the exterior derivative is given by the leibniz formula is subject of the eilenberg - zilber theorem from 1953 @xcite .
they use the nerve of product coverings and so ech type ideas to verify the equivalence of the cohomologies and establish so the knneth formula more elegantly .
the above computation just reflects the fact that the tensor product of two chain complexes with exterior derivative given by the leibniz formula is again a chain complex and that we have an eilenberg - zilber formulation as in the continuum , where two chain complexes @xmath451 are chain homotopic , if there are chain maps @xmath490 such that the composition @xmath491 is the identity and @xmath492 is chain homotopic to the identity in the sense that there is a linear map @xmath493 satisfying @xmath494 .
[ eilenbergzilber ] given two finite simple graphs @xmath39 .
there is a chain map @xmath348 from @xmath495 to the chain complex @xmath352 and a chain map @xmath347 in the reverse direction such that @xmath347 and @xmath348 are chain homotopic .
* examples : * + * 1 ) * let @xmath7 be the house graph and @xmath11 the sun graph @xmath496 .
both have the betti vector @xmath497 and the product has the cohomology of the @xmath60-torus as it is homotopic to the 2-torus .
the graph @xmath9 has dimension @xmath498 , has @xmath499 vertices , @xmath500 edges and @xmath501 triangles .
the graph @xmath502 has dimension @xmath53 and @xmath503 vertices and @xmath504 edges .
the product @xmath0 is a graph of dimension @xmath505 .
it has 120 vertices and 480 edges .
the kernel of the laplacian @xmath333 on @xmath53-forms of @xmath9 is spanned by @xmath506 , + the null space of @xmath333 on @xmath53-forms of @xmath502 is spanned by @xmath507 @xmath508 .
the laplacian @xmath333 on the product graph @xmath0 is a @xmath509 matrix .
the laplacian @xmath333 on the simplicial set @xmath510 is much smaller as it works on a space whose dimension is the set of degree 3 monoids in @xmath131 which is @xmath511 .
+ * 2 ) * the cohomology of @xmath52 has the betti vector @xmath497 .
the cohomology of @xmath512 is @xmath59 .
+ * 3 ) * the cohomology @xmath513 is the same as the cohomology of @xmath7 . similarly , the cohomology of the cylinder @xmath514 obtained by taking the products of a couple of circles with the product of a couple of interval graphs is the same than of the torus @xmath515 because the cube , the product of interval graphs is contractible . + * 4 ) * the betti vector of @xmath516 with @xmath517 is the kth row of the pascal triangle @xmath518 .
easier to state , @xmath519 .
+ the reason for the following inequality is that higher dimensional simplices spawn more vertices in the graph @xmath9 .
if the dimension is not uniform , then the dimension of the enhanced graph @xmath6 will increase .
first of all , the construction of @xmath9 allows to define a local dimension for all simplices @xmath177 inside @xmath7 : it is just the usual dimension @xmath520 in the enhanced graph @xmath9 , where the simplices like @xmath177 are vertices . given a finite simple graph @xmath7 .
then @xmath521 . for @xmath123 , or any two point graph , we have equality .
for all three point graphs different from the three point graph @xmath522 , we have equality too , while for @xmath522 we have a first inequality @xmath523 and @xmath524 .
denote by @xmath525 the dimension of the sphere in @xmath9 and by @xmath526 the dimension of the vertex @xmath177 in @xmath9 .
the average of all these numbers is @xmath527 .
we can not use induction as the unit spheres @xmath27 are not of the form @xmath528 with a smaller dimensional graph in general .
however , we can do it with a chain .
we therefore prove a stronger statement for chains @xmath529 : let the degree @xmath53 monomials be the vertices and the sphere @xmath27 of a vertex @xmath177 consist of the sum @xmath320 of degree @xmath60 monomials in @xmath130 which contain @xmath177 divided by @xmath177 .
the small dimension @xmath530 of a chain @xmath130 is now recursively defined as the average of the small dimensions of the unit spheres of the vertices .
the large dimension @xmath531 of the chain is the usual dimension of the graph @xmath159 . for a chain @xmath130 which comes from a graph @xmath7 , the small dimension is the dimension of @xmath7 , the large dimension is the dimension of @xmath9 which is the average over all large dimensions of the unit spheres @xmath27 , over all vertices @xmath177 in @xmath9 .
the dimension inequality follows from the more general statement : if @xmath130 is a chain then @xmath532 . the more general statement
can now be proven with induction with respect to a lexicographic ordering which looks first at the the number of variables appearing in the chain and then at the number of monomials .
this is a total ordering so that induction works . unlike for graphs ,
the unit sphere of a vertex @xmath177 is now a chain with one variable less for which we can look at the small and large dimension .
also the unit sphere of a simplex @xmath177 is a chain for which we can look at the small and large dimension .
the unit sphere of a simplex has in general the same number of variables but less monomials .
* example . * + the chain @xmath533 has two vertices @xmath254 .
( a ) lets look at @xmath177 first : the unit sphere of @xmath177 is @xmath534 which has two vertices @xmath535 .
( i ) the vertex @xmath536 in @xmath537 has the unit sphere @xmath538 which has no vertices so that @xmath536 has dimension @xmath63 .
+ ( ii ) the vertex @xmath260 in @xmath537 has the sphere @xmath539 , a chain with one vertex @xmath280 which has a @xmath63-dimensional sphere @xmath536 so that @xmath280 has dimension @xmath53 , @xmath540 has dimension @xmath53 and @xmath260 has dimension @xmath60 in @xmath130 .
+ the dimension of the vertex @xmath177 is therefore @xmath541 . + ( b ) the unit sphere of @xmath260 in @xmath130 is @xmath542 which has no vertex so that @xmath260 has dimension @xmath63 .
( c ) the chain has small dimension @xmath543 . + the two computations ( a)-(c )
show that @xmath544 .
the large dimension of the chain is the average of the large dimensions of the monomials in @xmath130 .
it is the dimension of a graph @xmath159 with 6 vertices and 11 edges .
it has dimension @xmath545 which is @xmath531 .
+ * remarks : * + * 1 ) * the inequality looks like the known inequality between hausdorff dimension and inductive dimension in the continuum .
but we do not see any relation yet .
+ * 2 ) * when iterating the refinement construction , we get a sequence of graphs @xmath546 with a dimension which converges to an integer . lets elaborate on this remark a bit more .
the largest simplex will be the seed which grows fastest and take over .
start with a finite simple graph @xmath547 and define recursively @xmath548 , then @xmath549 , if @xmath550 is the largest clique in @xmath7 .
the dimensions can only grow or stay by the lemma . since the dimensions are bounded by @xmath66 for all @xmath19
, the dimensions must converge to some value smaller or equal than @xmath66 .
the limiting value is @xmath66 because the number @xmath551 of @xmath550 subgraphs of @xmath546 grows exponentially and the growth of that large geometric cluster @xmath552 formed by the union of the @xmath550 simplices takes over : the number @xmath553 of @xmath554 subgraphs of @xmath546 is dominated by @xmath555 of @xmath554 subgraphs of @xmath552 .
we see that when starting with an arbitrary network , we asymptotically get to a geometric graph
. this could be interesting if we look at a dynamical emergence of graphs through refinement or evolution .
it would explain , why geometric and manifold - like structures with integer dimension appear naturally : the lower dimensional structures grow slower and are overtaken by the growth of largest structures which make sure that a larger and larger part of @xmath546 will be geometric .
the geometric part grows fast like the novo - vacuum in schild s ladder where physics is described by
sarumpaet rules " @xcite .
egan s novel starts with with the words : _ in the beginning was a graph , more like diamond than graphite .
_ the dimension of @xmath71 is not necessarily equal to the dimension of @xmath7 .
for example , the dimension of the lollipop graph @xmath7 is @xmath556 , the dimension of @xmath557 is @xmath558 is slightly higher . while @xmath71 has a cover for which the nerve graph is @xmath7 , the dimensions do nt agree because higher dimensional parts of @xmath7 contribute more to @xmath9 .
+ we have looked a bit at the statistics and computed the difference @xmath559 which is related to the variance of the dimension random variable @xmath560 on a graph .
see figure ( [ dimensiondiscrepancy ] ) .
+ the preservation of dimension is an essential ingredient of a graph homeomorphisms , as it is in the continuum .
it is no surprise that in the fractal case , we have the same inequality like hausdorff dimension : @xmath561 .
note however that the equality @xmath562 holds in full generality , for any finite simple graph .
we use the notation @xmath563 where @xmath564 is the unit sphere of the vertex @xmath177 in @xmath7 .
[ inductivedimension ] given two finite simple graphs @xmath39 .
the dimension of @xmath0 is equal to the sum of the dimensions of @xmath557 and @xmath565 .
more precisely , the dimension of any vertex @xmath97 in @xmath0 the dimension is the sum of the dimensions of @xmath177 in @xmath9 and @xmath260 in @xmath502 : @xmath566 use induction with respect to the sum of dimension .
fix @xmath567 and assume @xmath568 , @xmath569 so that @xmath570 and @xmath571 .
we only have to show that the unit sphere of @xmath97 in @xmath572 has dimension @xmath573 so that @xmath97 has dimension @xmath574 .
the sphere @xmath575 is the union of two graphs @xmath576 , which overlap in @xmath419 .
the two graphs have both dimension @xmath577 and @xmath419 has dimension @xmath578 .
lemma : if two graphs @xmath451 have dimension @xmath579 and their intersection has dimension @xmath580 then @xmath581 has dimension @xmath579 . for example , if @xmath7 is the * tadpole graph * and @xmath11 is the house graph , then @xmath582 and @xmath583 .
the product graph @xmath0 has dimension @xmath584 . if @xmath39 are finite simple graphs , then @xmath585 .
we have @xmath8 and @xmath586 .
the result follows from theorem ( [ inductivedimension ] ) .
given a geometric graph @xmath7 .
a subgraph @xmath587 of @xmath7 is called convex if between any two vertices @xmath254 in @xmath587 , there exists a shortest connection between @xmath254 which is inside @xmath587 .
note that shortest connections are not necessarily unique .
the convex hull of a finite set @xmath588 of vertices in @xmath7 is the intersection of all convex subgraphs of @xmath7 which contain @xmath588 .
\a ) if @xmath7 is geometric , then @xmath6 is homeomorphic to @xmath7 in the sense of @xcite .
+ b ) if @xmath7 has no triangles , then @xmath557 is homeomorphic to @xmath7 in the classical sense of topological graph theory .
\a ) assume the graph has dimension @xmath26 .
let @xmath109 be the vertices of @xmath7 .
we have to find a cover @xmath589 of @xmath9 such that the nerve graph of the cover is the original graph @xmath7 .
as @xmath590 we look at the unit sphere @xmath591 in the original graph and build the smallest convex hull in @xmath9 containing all the vertices of this sphere .
the intersection of two @xmath592 is @xmath26-dimensional and nonempty if and only if @xmath254 were connected in the original graph .
note that according to the definition @xcite , intersections @xmath593 which are lower dimensional do not count as edges in the nerve graph .
+ b ) the graph @xmath9 is a refinement , where every edge has got an additional vertex . * examples . *
+ * 1 ) * if @xmath432 then the @xmath590 are balls of radius @xmath60 in the new graph @xmath9 centered at the original vertices . two sets @xmath592 overlap in line graphs of diameter @xmath60 if @xmath177 and @xmath260 were connected in @xmath7 .
+ if @xmath7 and @xmath11 are homeomorphic geometric graphs and @xmath594 is a third geometric graph , then @xmath388 is homeomorphic to @xmath595 .
we use that @xmath7 and @xmath9 are homeomorphic and that @xmath11 and @xmath502 are homeomorphic . if @xmath44 is a cover of @xmath9 and @xmath596 is a cover of @xmath502 such that the nerve graphs of are the same and @xmath597 is a cover of @xmath121 , then we have a cover @xmath598 and @xmath599 of @xmath388 and @xmath595 . *
remark : * we do nt yet know whether this can be proven for arbitrary finite simple graphs .
one first has to show that in full generality that @xmath7 and @xmath9 are homeomorphic .
this is not yet done .
+ let @xmath7 be a finite simple graph .
define the * curvature of a simplex @xmath177 in @xmath7 * as the usual curvature @xcite of the vertex @xmath177 in the graph @xmath71 .
the simplex curvature @xmath600 satisfies the gauss - bonnet relation @xmath601 .
the function @xmath600 is equal to the expectation @xmath602 $ ] when integrating over all colorings @xmath130 on the simplex graph .
see @xcite .
+ * remark : * we have already a natural `` curvature '' for a graph located on simplices , it is constant @xmath53 on the even dimensional simplices and @xmath150 on the odd dimensional simplices . but this has little to do with the usual curvature .
+ now , we look at symmetries : given two finite simple graphs @xmath39 .
the automorphism group @xmath603 is a subgroup of the automorphism group of @xmath0 .
each element @xmath515 of @xmath603 produces a permutation of the variables @xmath109 appearing in the algebraic description @xmath320 of @xmath7 .
if @xmath604 is the algebraic description of @xmath11 using variables @xmath110 , then @xmath50 acts on @xmath131 by permutations of the elements @xmath109 and produces a symmetry of @xmath0 .
* remarks : * + * 1 ) * the case @xmath605 with @xmath606 shows that the symmetry group can become bigger . + * 2 ) * if the vertices of @xmath557 form a group and also the vertices of @xmath565 have this property then the vertices of @xmath0 form a group , the product group .
+ next we look at homotopy : if @xmath607 and @xmath608 are two finite simple graphs and the subgraph @xmath50 of @xmath7 is homotopic to the subgraph @xmath609 of @xmath11 and @xmath251 is homotopic to @xmath610 and @xmath611 is homotopic to @xmath612 , then @xmath581 is homotopic to @xmath613 .
the graph @xmath557 is homotopic to @xmath7 .
use induction with respect to the size of the graph .
its true for @xmath121 . given a graph @xmath7 with @xmath19 vertices .
add a new vertex @xmath177 to get a larger graph @xmath11 with @xmath614 vertices .
the unit sphere @xmath27 is in @xmath7 .
now @xmath557 is a subgraph of @xmath565 .
let @xmath615 denote the simplices in @xmath27 , they are the vertices in @xmath616 .
the graph @xmath502 has new vertices @xmath617 which were not in @xmath9 .
our induction assumption is that @xmath9 is homotopic to @xmath7 and since @xmath27 is a subgraph of @xmath7 also @xmath618 is homotopic to @xmath27 .
but then the graph in @xmath11 generated by @xmath619 is homotopic to the graph in @xmath502 generated by @xmath620 .
now @xmath11 is a union of two graphs @xmath621 and @xmath7 intersecting in @xmath27 and @xmath502 is the union of two graphs @xmath620 and @xmath9 intersecting in @xmath618 .
use the lemma .
let @xmath622 be finite simple graphs .
if @xmath39 are homotopic , then @xmath388 and @xmath595 are homotopic .
show it on an algebraic level : if @xmath623 is a homotopy step , then @xmath624 is a homotopy step on each fibre .
it follows that the product defines a group operation on the homotopy classes .
this monoid defines then a grothendieck group .
+ the cartesian product also defines a direct sum on isomorphism classes of vector bundles .
+ * example . * + * 1 ) * the product of two trees is contractible . + * 2 ) * the product @xmath625 is never contractible .
+ finally , lets look at the chromatic number @xmath16 .
the graph @xmath626 satisfies @xmath627 it is minimally colorable . if @xmath628 is the largest complete subgraph of @xmath7 , then @xmath629 .
if @xmath177 is a vertex in @xmath9 , denote by @xmath630 the dimension of the corresponding simplex @xmath177 in @xmath7 .
the function @xmath631 is a coloring because two adjacent vertices have different dimension . because @xmath631 takes values from @xmath63 to @xmath19
if @xmath628 is the largest simplex in @xmath7 , the chromatic number is @xmath614 .
* remarks . *
+ * 1 ) * we see that for graphs @xmath9 we can immediatly compute the chromatic number by just looking at the largest clique . + * 2 ) * for geometric graphs @xmath7 , this means that @xmath9 is minimally colorable and therefore is eulerian in the sense of @xcite . for two dimensional graphs ,
this is equivalent to eulerian in the classical sense .
+ we can actually compute the chromatic number of any product graph @xmath0 : [ chromaticnumberofproduct ] if @xmath7 has a largest clique @xmath632 and @xmath11 has a largest clique @xmath633 , then @xmath634 .
again just look at the function @xmath631 .
it is locally injective and is so a coloring .
this result actually has the previous lemma as a corollary because @xmath121 has the largest clique @xmath121 so that @xmath635 .
+ * examples . *
+ * 1 * the chromatic number of a product of two trees is @xmath62 .
+ * 2 * the chromatic number of @xmath636 is @xmath62 .
the chromatic number of @xmath637 is @xmath638 .
* example*. + let @xmath7 be the house graph and @xmath11 the lollipop graph .
we have @xmath639 and @xmath640 , @xmath641 . the graphs @xmath642
satisfy @xmath643 and @xmath644 .
the graph @xmath0 has dimension @xmath645 .
the dimensions of the individual points are given by the dimension spectrum @xmath646\ { 1 , 1 , 1 , 7/4 , 2 , 2 , 2 , 7/4 , 1 , 2 , 1 , 2 } @xmath647 .
the dimension spectrum of @xmath0 is the cartesian product of these two lists : @xmath64815/4 , 15/4 , 15/4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 2 , 2 , 2 , 9/2 , 19/4 , 19/4 , 19/4 , 19/4 , 5 , 5 , 5 , 19/4 , 5 , 5 , 5 , 19/4 , 5 , 5 , 5 , 19/4 , 5 , 5 , 5 , 19/4 , 5 , 5 , 5 , 19/4 , 5 , 5 , 5 , 19/4 , 5 , 5 , 5 , 11/4 , 3 , 3 , 3 , 9/2 , 15/4 , 19/4 , 19/4 , 4 , 5 , 19/4 , 4 , 5 , 19/4 , 4 , 5 , 19/4 , 4 , 5 , 19/4 , 4 , 5 , 19/4 , 4 , 5 , 19/4 , 4 , 5 , 11/4 , 2 , 3 , 15/4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 2 , 19/4 , 5 , 5 , 5 , 5 , 5 , 5 , 5 , 3 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 19/4 , 5 , 5 , 5 , 19/4 , 5 , 5 , 5 , 19/4 , 5 , 5 , 5 , 19/4 , 5 , 5 , 5 , 19/4 , 4 , 5 , 19/4 , 4 , 5 , 19/4 , 4 , 5 , 19/4 , 4 , 5 , 4 , 4 , 4 , 4 , 5 , 5 , 5 , 5 , 4 , 4 , 4 , 4 , 4 , 4 , 19/4 , 5 , 5 , 5 , 19/4 , 5 , 5 , 5 , 19/4 , 4 , 5 , 19/4 , 4 , 5 , 4 , 4 , 5 , 5 , 4 , 4 , 4 , 19/4 , 5 , 5 , 5 , 19/4 , 4 , 5 , 4 , 5 , 2 , 2 , 2 , 11/4 , 3 , 3 , 3 , 11/4 , 2 , 3 , 2 , 3@xmath649 .
this list had been computed directly by computing the dimensions of each unit sphere in @xmath0 .
the knneth formula tells that @xmath0 has the betti numbers @xmath650 again for @xmath651 .
+ * example*. + * 1 ) * for an octahedron @xmath7 , the ring element is @xmath31 is @xmath652 @xmath653 .
the graph @xmath71 is a barycentric subdivision of @xmath7 of the same dimension . for a geometric graph @xmath7 ,
the graph @xmath71 is geometric again . *
2 ) * let @xmath654 be the octahedron , the product @xmath655 is a @xmath62-dimensional graph .
it is a discrete incarnation of the @xmath62-manifold @xmath656 . *
3 ) * a triangle is given by @xmath657 .
lets take the product with @xmath123 represented by @xmath658 .
it becomes a @xmath62-dimensional graph . *
4 ) * the product of @xmath136 with @xmath123 is a @xmath62-dimensional graph .
it is a @xmath62-dimensional ball of radius @xmath53 with one interior point and all other points on the boundary .
the curvature of the interior point is @xmath63 , the curvature of nine of the boundary points are @xmath659 and three are @xmath660 . by gauss - bonnet ,
the sum of the curvatures is @xmath53 , which is the euler characteristic of the ball .
* dimension * : + what is the expectation of @xmath661 on the erds - rnyi probability space .
this random variable measures in some sense the distance from a geometric graph as for geometric graphs , the value is zero . the value of @xmath662 is related to the variance of the dimension spectrum , @xmath663 which is a random variable on vertices of @xmath7 .
it would be nice to quantify this more .
+ * hausdorff dimension * : + due to the analogy of results with hausdorff dimension @xmath664 , one can ask whether there is more to it and whether for any subsets @xmath665 of euclidean space @xmath666 , there are sequences of graphs @xmath667 such that @xmath668 and @xmath669 .
+ * curvature * : + the construction @xmath670 gives a convenient natural curvature on the simplices of a graph @xmath7 .
gauss - bonnet holds as they add up to euler characteristic .
it is also true that curvature is the average over all functions @xcite + what is the relation between the curvature of @xmath39 and @xmath0 ?
the example of @xmath671 shows that the product of two zero curvature spaces can develop curvature .
the case @xmath672 shows that the curvature can become very negative . +
* spectrum * : + what is the relation between the spectrum of @xmath0 with the spectrum of @xmath7 and @xmath11 ?
we know that the laplacians live on different spaces .
the eigenvalues appear to correlated to the eigenvalues @xmath673 , where @xmath674 is an eigenvalue of @xmath675 and @xmath676 is an eigenvalue of @xmath677 .
+ * chromatic number * : + the case @xmath678 shows that the chromatic number can become smaller .
the product of two geometric graphs is eulerian in the sense @xcite which shows that if @xmath7 is a geometric graph , then @xmath557 is eulerian .
the map @xmath679 regularizes in some way as for geometric graphs , the unit spheres of @xmath9 are eulerian spheres . of course
, it would be nice to see a relation between the chromatic polynomials @xmath16 and @xmath15 of @xmath7 and @xmath9 .
it is not always true that @xmath680 divides @xmath681 but it is often the case . for the graph @xmath682 for example , @xmath683 and @xmath684 .
now @xmath685 . for @xmath686 , this chromatic fraction of the graph
@xmath7 is @xmath687 .
already for @xmath606 , the chromatic fraction is not a polynomial .
here is more random example , where the chromatic fraction is a rational function and not a polynomial is the graph @xmath7 with @xmath321 vertices and edges @xmath688 for which the chromatic fraction is only rational .
still , the chromatic number is @xmath62 for both graphs . with respect to the known relation @xmath689
, we have inequality for all graphs @xmath690 . for complete graphs
@xmath20 , both sides are @xmath614 , for trees , both sides @xmath60 . for geometric graphs , all graphs @xmath9 have the property that @xmath691 . on the list of all @xmath692 connected graphs with @xmath638 vertices
, we have equality @xmath693 , on the list of all @xmath694 connected graphs with @xmath695 vertices there are a dozen for which the chromatic number has dropped , the reason always being a @xmath695-cycle of chromatic number @xmath62 becoming @xmath60-colorable after the refinement .
+ * constructing spheres * : + what kind of spheres appear as intersections of unit spheres in discrete tori @xmath696 with @xmath697 .
we have seen that they all are unions of solid cylinders glued along a torus .
assume we have obtained a class of spheres like that .
now take the product of such spheres and take again unit spheres .
we get a larger and larger class of spheres .
can one characterize them ?
similarly , we can ask what kind of spheres can be obtained by applying the join operation to spheres , starting with a smaller class of spheres .
+ * determinant * : + can we say something about the complexity @xmath698 of the product , where @xmath699 is the pseudo determinant @xcite .
if @xmath700 , then the complexities of @xmath701 are @xmath702 and @xmath62 .
the complexity of @xmath703 is already * homotopy classes * : + if @xmath9 is homotopic to @xmath707 and @xmath502 is homotopic to @xmath708 , then @xmath703 is homotopic to @xmath709 .
the sum and product therefore produces a sum and product in the ring of homotopy classes .
the contractible ring elements form a class representing the @xmath53 element .
the empty graph is the @xmath63 element . since also @xmath710 is homotopic to @xmath711 .
homotopy extends to the ring of chains .
define contractibility in the same way by telling @xmath7 is contractible if there exists @xmath177 such that both @xmath27 and @xmath712 are contractible where @xmath712 is the non - constant part of the polynomial where @xmath177 is replaced by @xmath53 .
the unit sphere is the unit sphere of the graph @xmath159 .
homotopy is so defined algebraically as reduction steps @xmath713 with contractible @xmath27 or reverse steps .
contractibility is defined for @xmath130 in a polynomial ring as the property that there is a variable @xmath714 such that @xmath715 and @xmath716 are both contractible .
for example a line graph @xmath717 is contractible as @xmath718 and @xmath719 are both contractible .
the cycle graph @xmath720 is not contractible . also
all its unit spheres like @xmath721 or @xmath722 which represent two point graphs are not contractible .
+ * suspension and join * : + the topological notions of suspension " and join " needs the cartesian product .
the definition of a suspension of a graph @xmath7 is @xmath723 .
also in the graph case , it produces from a @xmath26-sphere a @xmath256-sphere . more generally , the join of two graphs @xmath39 can be defined as @xmath724 an obvious questions is whether results like the double suspension theorem of cannon - edwards can be proven directly in graph theory .
it certainly is true , as one can see by building corresponding manifolds from the graphs .
but it would be nice to have a direct discrete proof .
the double suspension of a graph @xmath7 is @xmath725 .
a homology @xmath19-sphere is a graph @xmath7 which is geometric of dimension @xmath19 , which has the poincar polynomial @xmath726 but which is not a @xmath19-sphere .
the discrete cannon - edwards theorem tells that if @xmath7 is a homology sphere , then @xmath727 is a sphere . to prove this within graph theory , one has to show that every unit sphere in @xmath11 is a @xmath429-sphere and that removing one vertex in @xmath11 produces a @xmath19-ball , a contractible @xmath19-dimensional graph with @xmath427 dimensional sphere boundary .
+ * higher order curvature * : + it is natural to inquire whether for @xmath60-dimensional geometric graphs @xmath557 , the higher order curvature @xmath728 produces a gauss - bonnet result .
this is not the case as for the octahedron @xmath7 , the graph @xmath9 has a sum of the second order curvatures which is @xmath63 .
the culprit are the degree 4 vertices .
+ * the sequence of graphs @xmath546 .
* + what happens with @xmath546 defined recursivly by @xmath548 is that the minimal curvature goes to @xmath729 , at least if @xmath730 is two dimensional
. it would be interesting to know how the curvature spectrum @xmath731 $ ] grows for higher dimensional graphs .
the numerics is tough as the graphs sizes explode exponentially .
l if @xmath732 , then the curvature spectra are @xmath733 $ ] , @xmath734 , \kappa(g_2)=[-1,1/3 ] , \dots$ ] . if @xmath735 , then @xmath736 , \kappa(g_1)=[0,1/6]$ ] .
+ * global de rham * : + we have constructed a chain homotopy between de rham and simplicial chains . if we define a discrete @xmath66-manifold as a graph which locally can be wrtten as a product @xmath737 of networks such that on the intersection @xmath738 there are coordinate change graph homeomorphisms , then the chain homotopy can be pushed through
. we do nt even need the graphs @xmath739 to have to be @xmath53-dimensional , nor have they be to be geometric . the combinatorial de rham theorem allows to deal with such discrete manifolds more effectively similarly as the continuum de rham theorem does .
+ * moving frames * : + while a full classical hopf - rynov theorem on graphs is impossible because of the finiteness of any reasonable tangent space and therefore , a quantum flow is needed for a reasonable notion of geodesic map , there is weaker discrete hopf - rynov theorem which tells that under some conditions , there is a global unique flow which locally minimizes length @xcite . what is needed for a geometric graph of dimension
@xmath66 is an eulerian condition which is equivalent to the graph being minimally colorable with @xmath740 colors . in @xcite
, we made a stronger assumption and required all unit spheres to be projective in order to associate to an incoming ray an outgoing ray .
there is an other approach which implements a moving frame " idea of cartan and works for all eulerian graphs and in particular with @xmath741 of @xmath0 if @xmath39 are geometric .
it in particular does not need the projective assumption .
a discrete moving frame is a pair @xmath742 , where @xmath366 is a @xmath66-dimensional simplex containing @xmath177 .
the frame propagation defines a geodesic flow in an eulerian graph without the need of a projective involution : it is described inductively with respect to dimension : use the geodesic flow on the eulerian unit sphere @xmath27 to flip the frame @xmath742 on the sphere @xmath27 to the anitpodal side @xmath743 , where either the simplex @xmath744 or the vertex @xmath745 has maximal distance ( both cases are possible as the example in figure ( [ antipodal ] ) shows ) , then reflect the vertex @xmath745 along the face @xmath746 to a vertex @xmath747 on the adjacent simplex @xmath748 .
this propagation @xmath749 defines a variant of a geodesic flow for which the simplex @xmath366 plays the role of the direction , as well ( if we keep track of the frame in the simplex ) produces a parallel transport on the graph .
the point we want to make is that for all product graphs @xmath0 and more generally for all discrete manifolds obtained by patching product graphs , where each factor is a geometric graph , there is a canonical notion of geodesic flow and parallel transport . in some sense
, there is a unique levi - civita connection on such graphs . | we define a cartesian product @xmath0 for finite simple graphs which satisfies the knneth formula @xmath1 and so @xmath2 for the poincar polynomial @xmath3 and @xmath4 for the euler characteristic @xmath5 .
the graph @xmath6 is homotopic to @xmath7 , has a digraph structure and satisfies the inequality @xmath8 and @xmath9 .
hodge theory leads to the knneth identity using the product @xmath10 of harmonic forms of @xmath7 and @xmath11 .
a discrete de rham cohomology and partial derivatives " emerge on the product graphs .
we show that de rham cohomology is equivalent to graph cohomology by constructing a chain homotopy .
the dimension relation @xmath12 holds point - wise @xmath13 and implies the inequality @xmath14 , mirroring a hausdorff dimension inequality dimension in the continuum .
the chromatic number @xmath15 of @xmath9 is smaller or equal than @xmath16 and @xmath17 .
indeed , @xmath18 is the maximal @xmath19 for which there is a @xmath20 subgraph of @xmath0 .
the automorphism group of @xmath0 contains @xmath21 .
if @xmath22 and @xmath23 are homotopic , then @xmath24 and @xmath25 are homotopic , leading to a product on homotopy classes . if @xmath7 is @xmath26-dimensional geometric meaning that all unit spheres @xmath27 in @xmath7 are @xmath28-discrete spheres , then @xmath9 is @xmath26-dimensional geometric . and if @xmath11 is @xmath29-dimensional geometric , then @xmath0 is geometric of dimension @xmath30 .
because the product writes a graph as a polynomial @xmath31 of @xmath19 variables for which the euler polynomial @xmath32 is @xmath33 and @xmath34 , the product extends to a ring of chains which unlike graphs is closed under the boundary operation @xmath35 defining the exterior derivative @xmath36 and closed under quotients @xmath37 with @xmath38 . by gluing graphs , joins or fibre bundles are defined with the same features as in the continuum , allowing to build isomorphism classes of bundles . |
in a recent calculation @xcite , two - dimensional @xmath0 theory was solved for the lowest mass eigenstates of the light - front hamiltonian .
the eigenstates were represented by truncated fock - state expansions , with momentum - space wave functions as the coefficients .
this work included estimation of the critical coupling , where the @xmath1 symmetry of the theory is broken @xcite . at this critical coupling , one would expect that the probabilities for the higher fock sectors , computed as integrals of the squares of the fock wave functions , would increase dramatically .
in particular , the probability for the one - particle sector of the lowest massive state should go to zero .
this was not observed .
the expectation that the one - particle probability would go to zero is important for the calculation of the connection between equal - time @xcite and light - front estimates @xcite of the critical coupling .
this is determined by the relationship between the different mass renormalizations in the two quantizations @xcite , which is fixed by tadpole contributions computed from the vacuum expectation value of @xmath2 .
the behavior of this vacuum expectation value is dominated by the product of the one - particle probability times the logarithm of the mass @xcite .
the mass goes to zero at the critical coupling , making zero probability a necessity for a finite result . in the previous work ,
the explanation proposed for this apparent paradox was that the calculation did not use sector - dependent bare masses .
this meant that the highest fock sector kept in the calculation used a fixed bare mass even as the eigenstate mass approached zero .
excitation of such fock states is then very unlikely .
the use of sector - dependent bare parameters , or ` sector - dependent renormalization ' as it is usually called , has a long history @xcite . a fock - space truncation forces self - energy corrections and vertex corrections to be different in different fock sectors .
this makes sector - dependent counterterms a natural choice .
in addition , the truncation causes divergences that would have been canceled by contributions from higher fock states that are now absent .
sector - dependent counterterms can take these divergences into account .
however , in at least some theories , the sector - dependent bare couplings can lead to inconsistencies in the interpretation of wave functions and fock - sector probabilities @xcite .
thus , use of sector - dependent bare masses can be a compromise .
of course , for two - dimensional @xmath0 theory , divergences are not the issue , and it is only near the critical coupling where sector - dependent masses could be a useful approximation , as already indicated by some preliminary work @xcite . the use of light - front quantization @xcite is important for its simple vacuum and well - defined fock state expansions as well as for the separation of relative and external momenta .
we define light - front coordinates @xcite as @xmath3 , with @xmath4 the light - front time .
the light - front energy is then @xmath5 , and the light - front momentum is @xmath6 . in what follows
, we will drop the superscript from @xmath7 to simplify the notation .
the inner product between momentum and position is @xmath8 , and the mass - shell condition is @xmath9 .
the light - front hamiltonian eigenvalue problem is then @xmath10 , with @xmath11 the eigenstate with mass @xmath12 and light - front momentum @xmath13 .
the eigenstate is expanded in a set of fock states , which converts the formal eigenvalue problem into a system of equations for the fock - state wave functions .
numerical approximations then transform this system into a matrix eigenvalue problem .
our chosen numerical approximation is an expansion of the wave functions in terms of symmetric multivariate polynomials @xcite .
the content of the remainder of the paper is as follows .
section [ sec : phi4theory ] provides a brief introduction to @xmath0 theory and formulates the system of equations for the fock - state wave functions .
these equations are modified in sec .
[ sec : secdepmass ] to accommodate a sector - dependent bare mass ; the results from their solution are presented and discussed . the work is summarized briefly in sec . [ sec : summary ] .
details of the numerical methods are left to an appendix .
the lagrangian for @xmath0 theory is = 12(_)^2 - 12 ^ 2 ^ 2-^4 , with @xmath14 the bare mass .
the light - front hamiltonian density is = 12 ^2 ^2+^4 .
the field @xmath15 is expanded in terms of creation and annihilation operators @xmath16 and @xmath17 as [ eq : mode ] ( x^+=0,x^-)= \ { a(p)e^-ipx^-/2 + a^(p)e^ipx^-/2}. the operators obey the commutation relation = ( p - p ) .
substitution of the mode expansion and integration of the hamiltonian density with respect to @xmath18 yields the light - front hamiltonian @xmath19 , with [ eq : pminus11 ] p^-_11&=&dp a^(p)a(p ) , + [ eq : pminus22 ] p^-_22&= & ( p_1 + p_2-p_1-p_2 ) a^(p_1 ) a^(p_2 ) a(p_1 ) a(p_2 ) , + [ eq : pminus13 ] p^-_13&= & a^(p_1+p_2+p_3)a(p_1)a(p_2)a(p_3 ) , + [ eq : pminus31 ] p^-_31&= & a^(p_1)a^(p_2)a^(p_3)a(p_1+p_2+p_3 ) .
the eigenstate of @xmath20 , with eigenvalue @xmath21 , can be expressed as an expansion [ eq : fsexpansion ] is the wave function for the fock sector with @xmath22 constituents .
the wave function depends on the momentum fractions @xmath23 , which are boost invariant , unlike the individual momenta @xmath24 .
the leading factor of @xmath25 allows the normalization of the wave functions to be independent of @xmath13 ; we require @xmath26 , which yields [ eq : normalization ] 1=_m _
i^m dy_i ( 1-_i^m y_i)|_m(y_i)|^2 .
the probability of the @xmath22th fock sector is then just @xmath27 . because the hamiltonian changes particle number by zero or two , never an odd number , the sum over fock sectors is either even or odd , depending on which state is chosen as the lowest fock state .
symmetry of the original lagrangian . ]
we will focus on the odd case .
the eigenvalue problem becomes a coupled system of equations for the wave functions : [ eq : coupledsystem ] & & + & & + ( y_1-x_1-x_2-x_3)_m+2(x_1,x_2,x_3,y_2, ,y_m ) + & & + _ m-2(y_1+y_2+y_3,y_4, ,y_m)=m^2_m(y_i ) .
this is an infinite system and requires some form of truncation before a numerical solution can be attempted .
the standard truncation is a fock - space truncation to some maximum number of constituents @xmath28 .
however , as discussed in the introduction , this causes self - energy contributions to become sector - dependent . for the sector with @xmath29
, there is no self - energy because no loop corrections are allowed ; any intermediate states would have more than @xmath28 constituents . therefore , the bare mass in the top sector is reasonably equal to the physical mass @xmath12 of the lowest state . as we step down from the top sector , the complexity of the self - energy contributions steadily increases , and the bare mass can be adjusted to compensate .
to implement a sector - dependent bare mass , we replace @xmath14 in the first term of ( [ eq : coupledsystem ] ) by @xmath30 and compute the @xmath30 for a given eigenmass @xmath12 by steadily increasing @xmath28 . for @xmath31 , we have immediately that @xmath32 and @xmath33 . for @xmath34 , we set @xmath35 and solve the following two equations for @xmath36 and @xmath37 : _ 1 ^ 2_1 + ( 1-x_1-x_2-x_3)_3(x_1,x_2,x_3)=m^2 _ 1 _ i^3 _ 3(y_1,y_2,y_3 ) & + & ( y_1+y_2-x_1-x_2 ) _
3(x_1,x_2,y_3 ) + & & + _ 1=m^2_3(y_1,y_2,y_3 ) . for @xmath38 , we set @xmath39 to the value of @xmath36 obtained for @xmath34 , set @xmath40 , and solve a system of three equations for @xmath36 , @xmath37 , and @xmath41 .
we continue in this manner until @xmath36 has converged with respect to the fock - space truncation fixed by @xmath28 .
as described in the appendix , the equations are solved numerically , with the wave functions expanded in a polynomial basis @xcite .
the principal result of the calculation is the plot of the eigenvalues versus coupling strength in fig .
[ fig : extrap ] .
these values are extrapolated in the polynomial basis size for each fock sector , and the fock - space truncation is varied from @xmath34 to 9 . with respect to the fock - space truncations ,
the results converge , in an oscillatory fashion , to within the numerical error at a given truncation .
lowest mass eigenvalue for odd numbers of constituents for different fock - space truncations to three ( circles ) , five ( triangles ) , seven ( diamonds ) , and nine ( hexagons ) constituents .
the values of @xmath42 and @xmath43 are obtained as extrapolations in the orders of basis polynomials , and the error bars estimate the range of fits for the extrapolations . , width=566 ] these results are consistent with those from the standard parameterization , with no sector dependence in the bare mass , as reported in @xcite .
this can be seen in fig .
[ fig : extrap - all ] , where the previous results are added to the plot from fig .
[ fig : extrap ] . the critical coupling , as indicated by the point where @xmath44 reaches zero , is again estimated to be 2.1 .
the sector - dependent results do converge more slowly ; they require @xmath45 compared to the @xmath38 required for the standard parameterization .
this is to be expected , even desired , because we expect that higher fock states should become more important as the critical coupling is approached .
same as fig .
[ fig : extrap ] , but with the standard parameterization results ( open circles , triangles , and diamonds ) of @xcite , which include up to seven constituents , and the light - front coupled cluster results ( open hexagons ) of @xcite added for comparison . without sector - dependent masses , the fock state expansion converges more quickly , and the five and seven - constituent results are nearly identical , between themselves and with the nine - constituent sector - dependent results . , width=566 ] to understand what might be happening in the structure of the eigenstate , we plot the relative fock - sector probabilities in fig . [
fig : relprob ] .
these are computed as @xmath46 , for both the sector - dependent and standard parameterizations . in the former case , @xmath45 and in the latter , @xmath47 . for the sector - dependent calculations , results do not extend beyond the critical coupling ,
because negative @xmath44 is ill - defined for sector - dependent renormalization ; the bare mass in the highest fock sector would then be imaginary .
the relative probabilities are essentially the same in the three - body fock sector , indicating full convergence with respect to the fock - space truncation . in fock sectors with five and seven constituents , the relative probability for the sector - dependent case rises above the probability in the standard case as the critical coupling is approached .
the greater probability is expected ; however , the full expectation was that these probabilities would have a much more striking increase .
in fact , as relative probabilities , they will tend to infinity as the one - body probability @xmath48 goes to zero , and obviously this is not happening , and the original hypothesis , that sector - dependent bare masses would resolve the paradox , must be incorrect .
relative fock - sector probabilities for the three , five , seven and nine - constituent sectors when the truncation is @xmath45 .
, width=566 ] the finite one - body probability also prevents any improvement in the calculation of the difference in mass renormalization between equal - time and light - front quantization , as attempted in @xcite .
therefore , no new calculation is attempted here .
contrary to expectations , we have found that a sector - dependent bare mass does not provide any significant improvement in the solution of @xmath49 theory .
the sector dependence does allow higher fock states to have a larger contribution , but the contribution to the lowest massive state is not large in an absolute sense , and the one - body contribution remains dominant , even at the critical coupling .
convergence of the bare mass in the lowest sectors , as the fock - space truncation is relaxed , is quite rapid , as can be seen in fig .
[ fig : mu2vsnmax ] .
there is , however , a hint as to what might be needed in earlier work @xcite that explored the light - front coupled - cluster ( lfcc ) method @xcite . in this method ,
all of the higher fock states are kept . to keep the calculation finite in size ,
the relationship between fock wave functions is truncated , so that wave functions of the higher fock states are related to those of the lower states in a simple way .
the wave functions are then determined by a nonlinear equation that sums over contributions from all fock states . in this calculation
, a relative probability shows a rapid increase , in fig . 5 of @xcite , although at an unexpected value of the coupling
. may be due to the simplicity of the particular lfcc approximation ; a higher - order lfcc approximation should be investigated . ]
the hint is that coherent effects across all of fock space are important , something that ordinary fock - space truncation would not be able to reproduce .
that this would happen at the phase transition to broken symmetry is actually not surprising .
this work was supported in part by the minnesota supercomputing institute through allocations of computing resources .
the coupled system ( [ eq : coupledsystem ] ) , modified to use sector - dependent masses , is solved numerically , with each wave function expanded in a basis of symmetric multivariate polynomials @xcite @xmath50 . here
@xmath51 is the order and @xmath22 the number of momentum fractions ; the index @xmath52 differentiates between linearly independent polynomials of the same order .
the expansion for a wave function is [ eq : expansion ] _ m(y_1,
,y_m)=_ni c_ni^(m ) p_ni^(m)(y_1,
,y_m ) projection of the system of equations onto the basis functions then yields a system of matrix equations [ eq : matrixequations ] _
m^2_nic^(m)_ni + _ ni v^(m , m+2)_ni , ni c^(m+2)_ni + _
ni v^(m , m-2)_ni , ni c^(m-2)_ni = ^2_nib^(m)_ni , nic_ni^(m ) , with @xmath53 , @xmath54 , and matrices defined as given in the appendix of @xcite .
the matrices @xmath55 come from the overlap between basis functions in each fock sector .
if the basis was orthonormal , @xmath55 would be the identity matrix ; however , due to round - off errors that would be associated with the construction and use of orthonormal combinations , the basis functions are not chosen to be orthonormal .
instead , we implicitly orthonormalize the basis by using a singular - value decomposition @xmath56 .
the columns of the matrix @xmath57 are the eigenvectors of @xmath55 .
the matrix @xmath58 is a diagonal matrix of the eigenvalues of @xmath55 .
we then define new vectors of coefficients @xmath59 and new matrices , such as @xmath60 , with the @xmath61 matrices defined analogously .
the equations now become [ eq : reducedequations ] _ m^2_nic^(m)_ni + _ ni v^(m , m+2)_ni , ni c^(m+2)_ni + _
ni v^(m , m-2)_ni , ni c^(m-2)_ni = ^2 c_ni^(m ) , in exact arithmetic , this transformation is well defined .
the overlap matrix @xmath55 is a symmetric positive - definite matrix , and the eigenvalues must be positive , making @xmath62 real . in practice , though , round - off error can produce small negative eigenvalues . also , at high orders , some of the original polynomials are nearly linearly dependent , which is signaled by small positive eigenvalues .
in some sense , the basis is too large and not fully independent .
a robust linear independence is restored by reducing the basis size , keeping in @xmath57 only those columns associated with eigenvalues above some positive threshold @xcite .
the transformation is then implicitly a projection onto a smaller basis . for the results presented here , the threshold was @xmath63 , because the need was driven by round - off errors in double - precision arithmetic . in order to solve for @xmath64
, we define a set of matrices @xmath65 recursively , from @xmath29 down to 3 , as g^(m)=^-1 , with the initial form given by g^(n_max)=^-1 and @xmath66 the identity matrix in the @xmath22th sector .
the bare mass in the lowest sector is then simply _ 1 ^ 2= , where @xmath67 is a @xmath68 matrix and therefore just a number .
the coefficients for the wave - function expansions are constructed recursively from @xmath69 up to @xmath28 by ^(m)/c^(1)=g^(m)v^(m , m-2)c^(m-2)/c^(1 ) , with the value of @xmath70 set last by the normalization ( [ eq : normalization ] ) , which becomes 1=_m=1 c^(m ) b^(m)c^(m ) = |c^(1)|^2+_m=3 |c^(m)|^2 .
the values of intermediate @xmath30 are set by values of @xmath64 obtained in calculations with smaller @xmath28 , again recursively . as @xmath28
is increased , the @xmath71 converge to the dimensionless bare mass @xmath72 obtained in the standard parameterization , where the bare mass is not sector dependent .
this is just the reciprocal of the dimensionless coupling @xmath43 used in @xcite .
this allows us to estimate @xmath73 as @xmath74 .
the ratio @xmath42 is then obtained as @xmath75 .
the convergence of @xmath36 is illustrated in fig .
[ fig : mu2vsnmax ] for representative values of the mass scale @xmath76 . at @xmath31
, the points correspond to @xmath77 ; most of the change as self - energy contributions become active occurs immediately at @xmath34 .
plot of the scaled bare mass @xmath78 versus the fock - space truncation at @xmath28 constituents , for three values of @xmath76 : 1.0 ( circles ) , 0.5 ( triangles ) , and 0.05 ( diamonds ) .
the error bars represent uncertainties in extrapolations .
, width=566 ] | as an extension of recent work on two - dimensional light - front @xmath0 theory , we implement fock - sector dependence for the bare mass .
such dependence should have important consequences for the convergence of nonperturbative calculations with respect to the level of fock - space truncation .
the truncation forces the self - energy corrections to be sector - dependent ; in particular , the highest sector has no self - energy correction .
thus , the bare mass can be considered sector dependent as well .
we find that , although higher fock sectors have a larger probability , the mass of the lightest state and the value of the critical coupling are not significantly affected .
this implies that coherent states or the light - front coupled - cluster method may be required to properly represent critical behavior . |
nuclear matter present in compact and hot astrophysical objects generated in a gravitational collapse suffers from high thermal excursions.in the accessible range of temperature and densities , nuclear matter presents phase transitions characterized by density fluctuations . because of the presence of the long range coulomb interaction , those charge fluctuations give rise to so called `` frustration effects '' @xcite . under these conditions ,
the system organises in inhomogeneous phases , named `` pasta phases '' due to their unusual topologies @xcite .
the evolution of these structures with density is expected to be connected to the properties of the inner neutron - star crust.the survival of pasta - phases at finite temperature has been inferred from recent molecular - dynamics simulations @xcite . as these complex structures are related to coherent neutrino scattering @xcite , their evolution with density and temperature or , more generally , the underlying equation of state , becomes a key quest for exploring several interconnected astrophysical phenomena , like the dynamics of supernova explosion and the thermodynamics of proto - neutron - stars cooling .
a tempting connection can be searched with the physics of atomic nuclei , which are experimentally accessible .
comparable conditions of temperature and density can be probed in rather violent processes like nuclear multifragmentation . however , compact - star matter differs from normal nuclear matter due to the presence of electrons .
they constitute an incompressible degenerated gas , which establishes a condition of global charge neutrality at large scale .
to understand the phenomenology of phase transitions in compact stars , we consider an ising model .
indeed widely used mean - field approximations @xcite are known to fail to describe critical phenomena in 3d .
the ising model is specifically suited for the analysis of charge fluctuations by exact calculations , since phase transitions are universal phenomena .
moreover , it was largely employed in the study of ferromagnetic systems subjected to frustration @xcite as well as for nuclear matter @xcite where , in the form of the lattice - gas model , magnetic spins are replaced by site occupations .
we construct a cubic periodic lattice composed of @xmath0 sites , each characterised by a position @xmath1 and an occupation number @xmath2 or @xmath3 , to indicate the presence or the absence of positive charge , respectively .
this distribution of positive charge simulates the distribution of nuclear - matter charge ( neutron and protons without distinction are considered as nucleons with an effective charge z / a ) .
the strongly incompressible gas of electrons is represented by a uniform distribution of negative charge @xcite imposing a strict condition of neutrality . in any site
we find therefore an effective charge @xmath4 , where @xmath5 is the charge per site of the background of negative charge .
the schematic hamiltonian @xmath6 with @xmath7 is introduced to study the interplay of nuclear - like ( @xmath8 ) and coulomb - like forces ( @xmath9 ) .
@xmath10 is a sum extended over closest neighbors , and @xmath11 is the distance between sites @xmath1 and @xmath12 . the short - range and long - range interactions are characterized by the coupling constants @xmath13 and @xmath14 respectively , where @xmath15 is the nuclear saturation density , and @xmath16 is the proton fraction .
the ratio of the two coupling constants @xmath17 measures the strength of frustration . in the sequel ,
@xmath16 is fixed to @xmath18 , a value expected for proto - neutron stars .
@xmath9 can be rewritten in terms of occupation products times the geometric constants @xmath19 as @xmath20 translational invariance , which is ensured by imposing the distance @xmath21 to be the shortest between @xmath1 and @xmath12 in the periodic space , imposes @xmath22 .
hence , @xmath19 can be calculated by adopting periodic boundary conditions in the form @xmath23 where @xmath24 if @xmath25 , and @xmath26 otherwise ; @xmath27 is calculated with respect to any position @xmath28 .
@xmath9 preserves all the characteristic symmetries of the ising model and the coulomb energy is therefore invariant for lattice - occupation reversal .
we should also point out that , despite some tempting similarities , the correspondent of eq.([eq : hamiltonian ] ) in spin variable @xmath29 and @xmath30 with the magnetization @xmath31 is not identical to the hamiltonian of the frustrated ising ferromagnet @xcite , defined by @xmath32 indeed , @xmath33 and @xmath34 are related by@xmath35 this connection includes a term in @xmath36 , scaling with @xmath37 .
such term deeply modifies the thermodynamics in a finite system , and imposes @xmath38 at the thermodynamic limit for @xmath33 .
the strict constraint @xmath38 on the order parameter @xmath39 leads to a substantial modification of the thermodynamics of the system@xcite .
the phase diagram of the frustrated ferromagnet @xcite , where , as in nuclei , the coulomb interaction reduces the phase - coexistence region ( i.e. decreases the limiting temperature ) , can not be directly exported to our system @xmath34 .
example of a metropolis calculation ( @xmath40 ) .
density distribution for a fixed value of @xmath41 and a series of temperatures , each one associated to one spectrum .
the ensemble of explored temperatures and densities defines the phase diagram .
the ridge of the distribution defines three branches joining at the limiting temperature .
the phase - coexistence region is delimited by the gas ( low density ) and liquid ( high density ) branches .
, scaledwidth=35.0% ] for the numerical simulation of the @xmath34 system , in order to take into account the long - range coulomb interactions in a rapidly converging form , the finite lattice is repeated in all three directions of space a large number @xmath42 of times .
each site @xmath1 has @xmath42 replicas of itself , each one displaced from @xmath1 of a vector @xmath43 , where @xmath44 has integer components and @xmath45{v}$ ] . to treat a frustrated system , described by the interplay of several interactions , the multi-(grand ) canonical ensemble@xcite
is specifically suited .
it consists in associating a lagrange multiplier to each component of the hamiltonian that we want to treat as a separate observable .
in particular , it makes possible to construct one single phase diagram for both neutral and charged matter by associating one lagrange multiplier , @xmath46 , to the nuclear energy ( related to @xmath8 ) and another , @xmath41 , to the coulomb energy @xmath47 ( related to @xmath9 ) .
the multi - canonical partition sum reads : @xmath48 where @xmath49 is the density of states . for any finite value of @xmath41
, we treat the system as equilibrated at the temperature @xmath50 and described by an equivalent effective charge @xmath51 .
when @xmath52 , the coulomb energy @xmath47 does no more influence the partition sum and the system behaves as uncharged , reducing to the standard ising model .
the generalised grand potential is defined as @xmath53 where the fugacity is linked to the chemical potential by @xmath54 . when @xmath55 the ensemble coincides with the conventional grand - canonical form .
we sampled the density distribution from eq.([macro ] ) for a chemical potential value @xmath56 , by employing a standard metropolis technique @xcite .
an example is presented in fig .
[ fig : fig1 ] where , for a fixed value of @xmath41 , several density distributions are calculated for different temperatures @xmath57 .
all bimodal distributions present a peak for the gas phase ( low - density ) and one for the liquid ( high - density ) phase .
their height is equal because of the choice of @xmath40 .
all distributions are composed together to draw a phase diagram : the ridge of the overall distribution of density and temperature delimits with its gas branch and liquid branch the phase coexistence .
the two branches join at the limiting temperature .
for two cases , either @xmath58 ( corresponding to the standard ising model ) , or @xmath59 , and for a series of different lattice sizes @xmath60 , we collected a series of calculations of the phase diagram .
the results are schematized in fig .
[ fig : fig2 ] .
the phase - coexistence region in the system @xmath58 shrinks for progressively larger lattice sizes according to the characteristic finite - size scaling behaviour of the ising system .
in particular , since the asymptotic value of the limiting temperature @xmath61 corresponds to a critical point , finite - size scaling requires that @xmath62 evolves as @xmath63 @xcite .
this is comfirmed by the calculated values of the critical exponent @xmath64 and the critical point @xmath65 , which correspond to the ising values .
* left panel*. calculations of phase diagrams for the systems @xmath58 ( ising - like ) and @xmath59 ( frustrated ) .
calculations for a series of different lattice sizes @xmath60 are associated to different limiting temperatures for the system @xmath58 ; they overlap for the system @xmath59 . *
right panel*. for the system @xmath58 the test of the finite - size scaling law @xmath63 gives ising values for the critical exponent @xmath64 and the critical temperature @xmath66 .
no scaling is observed for the system @xmath59 : this is a first indication of the quench of criticality in presence of a coulomb field . ] at variance with the ising behaviour , the phase diagram of the system @xmath59 does however not vary with the lattice size .
this first sign of incompatibility of the frustrated system with the ising model , is also a first indication of quenching of the critical behaviour .
@xmath64 rules in fact the divergence of the correlation length at the critical point , which evolves as @xmath67 with @xmath68 .
if a critical point existed , it would correspond to a very large value for @xmath64 , which would then be hardly compatible with a diverging correlation length .
the major result we infer from comparing the two systems and the corresponding asymptotic limiting temperature , is that the coexistence region expands under the action of a coulomb field . *
( a ) * increase of the limiting point and expansion of the phase - coexistence region with increasing frustration ( i.e. for higher @xmath69 ratios ) .
four points determine different thermodynamical situations : points ( 1 ) and ( 3 ) belong to the coexistence region , points ( 2 ) and ( 4 ) are at the limiting temperature ; while points ( 1 ) and ( 2 ) refer to the system @xmath70 ( ising - like ) , points ( 3 ) and ( 4 ) refer to the system @xmath71 ( frustrated ) .
the calculation refers to @xmath72 ; * ( b ) * in correspondence with the points marked in panel ( a ) , the probability distribution is evaluated for the two hamiltonian components , @xmath73 , which corresponds to the nuclear - energy density @xmath74 , and @xmath75 , representing the coulomb - energy density @xmath76 in the case of the frustrated system .
contour plots refer to @xmath72 ; logarithmic cluster plots refer to @xmath77 .
points ( 1 ) and ( 3 ) manifest bimodal patterns .
the critical scaling effect exhibited at point ( 2 ) quenches at point ( 4 ) . ]
the connection between the increase of the limiting point and charge fluctuations can be searched in the evolution of the event distribution with temperature and coulomb - field strength .
[ fig : fig3]a illustrates the evolution of the limiting temperature with the strength of the coulomb field and the phase coexistence is indicated for the uncharged system @xmath70 and the frustrated system @xmath71 , in correspondence with the phase diagrams shown in fig .
[ fig : fig2 ] .
we analyzed the event distribution along the path 1 - 2 - 3 - 4 , determined by four different thermodynamic situations . in fig .
[ fig : fig3]b the distributions are shown , as a function of the two hamiltonian components , @xmath73 and @xmath75 .
point ( 1 ) belongs to the region of phase coexistence sampled for the uncharged system @xmath70 and presents a bimodal pattern with respect to the abscissa , which corresponds to the nuclear - energy density @xmath74 , so that pure liquid or gas partitions are preferred to mixed events .
point ( 2 ) corresponds to the limiting temperature of the uncharged system @xmath70 , where mixed partitions are favoured . to pass from point ( 2 ) to point ( 3 ) the temperature is left unchanged while @xmath41 is increased from @xmath3 till it equals @xmath46 .
once introduced the coulomb field , @xmath75 represents the coulomb - energy density @xmath76 , which should be minimized in the frustrated system . in order to satisfy this constraint ,
the positive - charge distribution searches uniform patterns , which are better compensated by the uniform background of negative charges . as a result
, the system moves to the phase - coexistence region to search pure partitions , and bimodality appears . in the frustrated system , the temperature should be further increased with respect to the limiting temperature of the system @xmath70 , point ( 3 ) , in order to find the limiting temperature , as indicated at point ( 4 ) .
[ fig : fig3]b presents the event distributions sampled for two different lattice sizes , in order to test scaling effects which could indicate the presence of critical points . at a critical point
the correlation length @xmath78 diverges and determines the exponential decay of the correlation function @xmath79 according to the expression @xmath80 , where @xmath81 is the space dimension and @xmath82 is a critical exponent .
on the basis of the properties of the @xmath19 matrix , @xmath83 is related to the mean coulomb - energy density by @xmath84 it is therefore typical of a critical point to exhibit a divergence of the quantity @xmath85 for progressively larger lattice sizes .
this indicates that , at point ( 2 ) , the limiting point is compatible with a critical point , as we already expected from the finite - size scaling law tested in fig .
[ fig : fig2 ] for the critical exponent @xmath64 . on the contrary ,
no scaling is observed at the point ( 4 ) .
indeed , as indicated by eq.([eq : energydens ] ) , when the coulomb field is introduced in the system , the diverging quantity would be the coulomb - energy density @xmath76 .
to avoid such a singularity , the correlation length keeps a finite value at the limiting temperature and the critical character of the limiting temperature is lost @xcite .
a more formal discussion on the suppression of criticality on the basis of finite - size - scaling analysis is presented in ref .
by employing a specific model for studying the thermodynamical features of a frustrated system , we could define general properties of a neutral system in presence of charge fluctuations , in analogy to compact - star matter .
the study of the phase diagram and the phase - transition phenomenology indicated that the introduction of a coulomb field has the effect of increasing the limiting temperature without preserving any critical character .
this result is the opposite as found for many other physical systems subject to coulomb frustration in the absence of global charge neutrality , from hot atomic nuclei to frustrated ferromagnets @xcite .
this discrepancy with the phase - transition phenomenology characteristic of atomic nuclei , indicates that a connection between compact - star matter and nuclei is not trivial .
conversely a widening of the density range connected to the pasta phases was found @xcite within the rmf model , when the coulomb field is included under the constraint of global charge neutrality over the wigner - seitz cells .
such a finding is directly compatible with the thermodynamic phenomenology we are drawing .
it was also discussed @xcite that neutrino propagation can be highly concerned by the presence of pasta phases and characterizes the process of neutron - star formation and cooling . with respect to homogeneous phases , the inhomogeneous phases ( like pasta phases ) present in the coexistence region
are characterized by a larger opalescence to neutrino propagation . in the presence of a critical point
the opalescence would diverge as a direct consequence of the divergence of the correlation length , and neutrinos would be trapped . on the contrary , we conclude that the medium stays `` grey '' at any temperature .
this result seems also connected to the observation of small opacity for the transport of neutrinos discussed in ref .
. 9 j.w .
negele and d. vautherin , nucl .
phys . a * 207 * , 298 ( 1973 ) .
ravenhall , c.j .
pethick , and j.r .
wilson , phys .
. lett . * 50 * , 2066 ( 1983 ) ; m. lassaut _
et al . _ ,
. astrophys . * 183 * l3 ( 1987 ) ; c.j .
pethick and d.g .
ravenhall , annu .
nucl . part .
sci . * 45 * , 429 ( 1995 ) ; g. watanabe _
et al . _ , nucl.phys .
a * 676 * , 455 ( 2000 ) ; n.k .
glendenning , phys .
rep . * 342 * , 393 ( 2001 ) ; j.m .
lattimer and m. prakash , phys .
rep . * 333 * , 121 ( 2000 ) . c. horowitz _ et al .
_ , phys . rev .
c * 70 * , 065806 ( 2004 ) .
g. watanabe _
et al . _ ,
* 94 * , 031101 ( 2005 ) ; c.j .
et al . _ ,
c * 69 * , 045804 ( 2004 ) .
a * 432 * , 646 ( 1985 ) .
f. douchin , p. haensel and j. meyer , nucl .
a * 665 * , 419 ( 2000 ) . c. ducoin _ et al . _ , to be published .
p. bonche , s. levit and d. vautherin , nucl .
a * 436 * , 265 ( 1985 ) .
f. gulminelli _
et al . _ ,
lett . * 91 * , 202701 ( 2003 ) .
m. grousson , g. tarjus and p. viot , phys . rev .
e * 62 * , 7781 ( 2000 ) ; phys .
e * 64 * , 036109 ( 2001 ) .
t. maruyama _
et al . _ , phys .
c * 72 * , 015802 ( 2005 ) .
j. barr , d. mukamel and s. ruffo , phys .
rev . lett.*87 * 030601 ( 2001 ) .
f. gulminelli _
et al . _ ,
e * 68 * , 026120 ( 2003 ) .
chomaz , f. gulminelli , in _
`` dynamics and thermodynamics of systems with long range interactions '' _ , lecture notes in physics vol.*602 * , springer ( 2002 ) . k. binder , _ `` monte carlo methods in statistical mechanics '' _ , 2nd ed .
berlin : springer - verlag ( 1986 ) .
_ , arxiv : astro - ph/0507633 ( 2005 ) .
p. napolitani _ et al .
_ , in preparation .
lee , a.z .
mekjian , phys .
c * 63 * , 044605 ( 2001 ) .
raduta , ad.r .
raduta , nucl .
a * 703 * , 876 ( 2002 ) .
j. margueron , j. navarro and p. blottiau , phys .
c * 70 * , 28801 ( 2004 ) .
g. watanabe , h. sonoda , aip conf .
proc . * 791 * , 101 ( 2005 ) .
horowitz , and j. piekarewicz , phys .
c * 72 * , 035801 ( 2005 ) . | in the formation and evolution of compact stars , nuclear matter explores high thermal excursions and is the site of intense neutrino emission .
neutrino transport as well as structural properties of this matter depend on the presence of inhomogeneous phases ( named `` pasta '' phases ) , which are the result of coulomb frustration of the liquid - gas phase transition .
we take into account charge fluctuations by employing a frustrated lattice - gas model to which we impose a neutrality constraint by the addition of an homogeneous background of charge , representing delocalised electrons . within this schematic model
we highlight a generaic feature of the phase - transition phenomenology : the temperature interval where pasta phases are formed is enhanced by coulomb - frustration effects .
this result is at variance with the behaviour of frustrated ferromagnetic systems as well as hot nuclei and mean - field approaches .
moreover , the region of phase coexistence is not found to end upon a critical point , indicating that no critical opalescence can occur in compact - star matter . |
a long held expectation is that the singularities predicted from general relativity signal a breakdown of the classical theory requiring a more proper accounting of the quantum effects of gravity .
the situation is analogous to hydrogen atom in ordinary mechanics , where classically the electron is expected to in - spiral towards the proton leading to a singularity .
this behavior is cured when quantum mechanics is taken into account leading to a stable non - singular hydrogen atom .
in general relativity , two of the most relevant forms of singularities are the big - bang cosmological singularity , as well as black - hole singularities .
a central question to be answered by any quantum theory of gravity is whether these singularities are regulated by incorporating quantum effects .
furthermore , if a quantum theory of gravity were to resolve the classical singularities , it would be of paramount interest as to what replaces them dynamically . a leading candidate theory of quantum gravity is known as loop quantum gravity ( for reviews see @xcite ) .
the application of loop quantum gravity techniques to cosmological models is known as loop quantum cosmology @xcite which has led to a resolution of the big - bang singularity , replacing it with a big - bounce for homogeneous and isotropic models @xcite .
these results provide tantalizing first hints that loop quantum gravity may indeed provide the singularity resolution hoped for in a quantum theory of gravity .
thus the next step is to consider loop quantum gravity for the black hole scenario .
the simplest first step in considering loop quantum black holes consists of examining the interior of a schwarzschild black hole .
there , the temporal and radial coordinates flip roles , and the interior becomes a kantowski - sachs cosmological model whereby the metric components are homogeneous and only time dependent .
the interior therefore can be quantized in a similar fashion as loop quantum cosmology leading to the possibility that the singularity is resolved as in the cosmological case .
the loop quantization of the schwarzschild interior has been initially developed in @xcite and @xcite .
there , the quantization indicates that the quantum einstein equations are non - singular in a similar way loop quantum cosmology was originally shown to be non - singular @xcite .
however , the question of what replaces the black hole singularity is not answered .
therefore , we attempt to provide an answer as to what dynamics are predicted from the loop quantization of the schwarzschild interior .
initial attempts in this direction are given in @xcite which is based on the quantum einstein equations of the original quantization of @xcite . however , this quantization is the direct analog of the original formulations of loop quantum cosmology which have been shown to not have a good semi - classical limit @xcite . in particular ,
a crucial parameter in the construction of the hamiltonian constraint operator is taken to be a constant in the early constructions and it is this assumption that can spoil the semi - classical limit in loop quantum cosmology .
an improved quantization of loop quantum cosmology whereby the parameter is assumed to explicitly depend on the scale factor has been applied in @xcite and leads to the correct semi - classical behavior as well as more physically sensible quantum corrections to the standard cosmological dynamics . in this paper , we consider the improved quantization scheme of loop quantum cosmology applied to the loop quantization of the schwarzschild interior .
we will consider the resulting modifications to the classical interior by examining effective semi - classical equations that incorporate quantum effects . while we do not have a complete handling of the physical sector of the quantum theory owing to its complexity
, the effective semi - classical description is intended to be an approximate shortcut to the physical predictions of the quantum theory . in the framework of the effective description
, we will show that the singularity is resolved and discuss the implications for an in - falling test particle as well as discuss the resulting conformal diagram for the non - singular interior region .
the interior of the schwarzschild geometry can be described by a kantowski - sachs homogeneous model whereby the line element is given by [ metric ] ds^2 = -n(t)^2 dt^2 + g_xx(t ) dx^2 + g_(t ) d^2 with @xmath0 being the freely specifiable lapse function and @xmath1 representing the unit two - sphere metric given in polar coordinates d^2 = d^2 + ^2d^2 .
the topology of the spatial slices is @xmath2 hence the coordinates range from @xmath3,\ , \phi \in [ 0 , 2\pi]$ ] . in the standard schwarzschild solution ,
the metric components are given by [ swcl ] n(t)^2 = ( -1)^-1 , g_xx(t ) = ( -1 ) , g_(t ) = t^2 for @xmath4 in the range @xmath5 $ ] , where @xmath6 is the mass of the black hole .
loop quantum gravity is based on a connection - triad hamiltonian formulation of general relativity .
considering connections and triads that are symmetric under the kantowski - sachs symmetry group @xmath7 , the connection @xmath8 and triad @xmath9 after gauge fixing of the gauss constraint are of the form @xcite a_a^i _
i dx^a & = & _ 3 dx + _ 2 d - _ 1 d+ _ 3 d + e^a_i _ i _ a & = & _ c _ 3 _ x + _ b _ 2 _ - _ b _ 1 _ where the dynamical phase space variables @xmath10 are all only functions of time , and @xmath11 are @xmath12 generators satisfying @xmath13 = \epsilon_{ij}^k \tau_k$ ] .
the relationship between the triad variables @xmath14 and the metric variables @xmath15 is given through the general relation e^a_i e^b_i = ( q ) q^ab which explictly implies g_xx & = & + g _ & = & |_c| .
we thus see that the triad component @xmath16 directly determines the physical radius of the two sphere which is proportional to @xmath17 .
we will use this fact when interpreting the quantum dynamics .
the classical hamiltonian consists of a single constraint known as the hamiltonian constraint .
the explicit form of the hamiltonian is given by h= dx d. performing the spatial integrations we run into a problem as the integral over @xmath18 diverges .
following @xcite , if we restrict the integration over @xmath18 to a finite interval @xmath19 , the hamiltonian becomes [ hcl ] h= written in terms of rescaled variables [ untild ] b = , c = l_0 , p_b = l_0 _ b , p_c = _ cthe poisson structure of these phase space variables is given by the only non - vanishing poisson brackets \{b , p_b } & = & g + \{c , p_c } & = & 2 g written in terms of these variables , the metric is given by [ triadmetric ] ds^2 = - n^2 dt^2 + + |p_c| d^2 the equations of motion are given as in an ordinary hamiltonian dynamical system based on the hamiltonian .
the dynamical behavior of the phase space variables is governed by hamilton s equations which give for instance _
b = \{p_b , h } = - g and so forth .
additionally , the lapse @xmath20 appears in the hamiltonian as a lagrange multiplier and this implies that the hamiltonian must vanish leading to an additional constraint on the dynamics h= 0 .
we will next consider two classes of solutions that will be relevant to the discussion .
the first consists of the usual schwarzschild interior we are most interested in .
the second corresponds to the nariai solution which will be of relevance when we discuss the modified quantum dynamics .
to arrive at the schwarzschild interior solution let us first choose for simplicity the lapse ( for @xmath21 ) to be given by n = [ eq : lapse1 ] whence the hamiltonian becomes h= .
the equations of motion are given by & = & - ( b+^2/b ) + & = & -2c + _ b & = & ( 1-^2/b^2 ) + _ c & = & 2 p_c + h&= & 0 2c = -(b + ^2/b ) .
the solution to these equations of motion is given by p_b(t ) & = & p_b^(0 ) e^t [ pbcl ] + p_c(t ) & = & e^2 t [ pccl ] + b(t ) & = & [ bcl ] + c(t ) & = & -m p_b^(0 ) e^-2 t [ ccl ] .
with @xmath22 a constant of motion . transforming to a new time coordinate @xmath23 , we have @xmath24 and the metric is given by ds^2 = - dt^2 + ( ) ^2 ( -1 ) dx^2 + t^2 d^2 if we then identify @xmath25 , we arrive at the schwarzschild interior solution . in these coordinates ,
the singularity is located at @xmath26 , while the horizon is at @xmath27 .
we can identify the singularity and horizon in a coordinate invariant fashion by noting that at the singularity both triad components @xmath28 and @xmath29 vanish .
the horizon then occurs where @xmath30 and @xmath31 . before moving to the second class of solutions
let us discuss certain gauge freedoms of the schwarzschild solution . classically there are two gauge freedoms corresponding to coordinate changes in either @xmath4 or @xmath18 variables .
a change in the time coordinate @xmath4 leads to a change in the lapse leaving all physics unchanged .
on the other hand if we consider a coordinate rescaling of @xmath18 , @xmath19 is correspondingly rescaled such that the ratio @xmath32 is left invariant .
this implies that a rescaling of @xmath18 does not rescale the value of @xmath29 .
note that this point was mistaken in @xcite where it was mentioned that a rescaling of @xmath22 can be accommodated by a rescaling in @xmath18 .
in addition to the classical coordinate gauge freedoms , there is an additional gauge in the loop setup pertaining to the choice in restriction of the spatial integration leading to the @xmath19 parameter . from the definitions of @xmath29 and @xmath33 in , it is easy to see that if we choose a different size of @xmath19 then this will rescale the value of @xmath29 and @xmath33 and hence in terms of the schwarzschild solution , @xmath22 will rescale accordingly .
the loop quantization is based on the phase - space variables @xmath34 all of which are invariant under a coordinate rescaling of @xmath18 .
the quantization is thus manifestly invariant under the first gauge transformation .
however the quantum effects in principle can depend on the second gauge freedom which does not appear classically .
this will be manifested in the quantum dynamics by a dependence of the physical results on the parameter @xmath22 .
any dependence of the quantum physics on this gauge freedom implies the need for additional input into the theory ( namely the specification of the value of @xmath22 ) which must be done by hand .
we will show that the improved quantization we focus on in section [ gooddelta ] will be independent of @xmath22 and hence will not require additional input of its value . an additional classical solution that will be of relevance to the discussion of the effective dynamics is the nariai solution @xcite .
it differs from the schwarzschild solution in that the two - sphere radius determined from @xmath28 is assumed to be constant and the model is sourced by a cosmological constant with the possible inclusion of charge .
the assumption that @xmath28 is a constant requires that @xmath35 and hence the lapse used in the schwarzschild case becomes degenerate .
we must therefore choose another form of the lapse to generate the nariai solution . choosing the lapse to be equal to one and adding in a cosmological constant @xmath36 leads to the hamiltonian @xmath37 + \frac{\lambda}{2 g } p_b \sqrt{p_c}. \label{lam1}\end{aligned}\ ] ] we are interested in a solution where the radius of the two - sphere is constant implying @xmath38 , such that @xmath39 .
the resulting equations of motion imply @xmath35 .
putting @xmath35 into the equation for @xmath40 yields the relation between the constant radius of the sphere and the cosmological constant @xmath41 .
the remaining two equations of motion simplify to @xmath42 which can be written as a single second order differential equation for @xmath29 given by @xmath43 the general solution is given by @xmath44 where @xmath45 and @xmath46 are two constants of integration . to obtain the global cosmological metric of the nariai universe we set @xmath47 and @xmath48 to get @xmath49 hence , we obtained the following solution of the classical field equations @xmath50 compare for example @xcite , which corresponds to the uncharged nariai universe .
geometrically nariai type spacetimes can be understood as four dimensional submanifolds of a flat six dimensional lorentzian manifold being the product of two ( three dimensional ) spaces of constant curvature .
if the respective length scales are denoted by @xmath51 and @xmath52 say , the generic nariai type metric takes the form @xmath53 for @xmath54 , this corresponds to the uncharged nariai universe , given by ( [ lam11 ] ) upon a rescaling of the time @xmath55 .
if @xmath56 , the spacetime corresponds to the charged nariai universe .
it should be noted that the knowledge of the metric alone , does not suffice to distinguish between the two types of possible charges , electric or magnetic . in order to interpret the quantum dynamics
, we will also consider the trajectory of a radially in - falling test particle .
the geodesics can be derived from the line element @xmath57 since we are mainly interested in radial geodesics we henceforth assume @xmath58 .
the geodesic equations are most easily read off from the lagrangian @xmath59 where @xmath60 for a massless particle and @xmath61 for a massive particle , and @xmath62 refers to the proper time for a massive particle .
since the lagrangian ( [ r2 ] ) is independent of the variable @xmath18 , its conjugate momentum @xmath63 is conserved @xmath64 where @xmath65 represents the total energy including the gravitational potential energy for a static observer at infinity .
since @xmath66 the lagrangian can now be written as @xmath67 hence , we obtain @xmath68 in the classical schwarzschild case , we have @xmath69 and hence the geodesic equation can be understood as a point particle with energy @xmath65 moving in an effective potential given by @xmath70 .
moreover , the geodesic equation is simple to interpret because of the simple relation between the coordinate parameter @xmath4 and the two - sphere radius determined from @xmath28 since we have @xmath71 .
however , when we come to the quantum dynamics , the relationship between the coordinate parameter @xmath4 and @xmath28 will become complicated and we will no longer be interested in the behavior of the coordinate parameter @xmath4 as a function of proper time
. we will therefore be most interested in @xmath72 which can be derived from [ geodesic ] = where @xmath73 is to be determined from the equations of motion derived from the hamiltonian , and @xmath74 is given by .
classically , for any given energy @xmath65 , the in - falling particle reaches the singularity at @xmath75 in finite proper time
. we will be interested in whether the quantum dynamics implies that the singularity is never reached by the in - falling particle .
we will see that the loop quantum dynamics imply a minimum value of the two - sphere radius @xmath28 and hence the solutions are non - singular with an in - falling particle never reaching the classical singularity .
in attempting to ascertain the quantum corrections to the classical behavior implied by the loop quantization , we will work in an effective semi - classical description that leads to modifications of the classical hamiltonian constraint .
the effective description is motivated from the construction of the quantum hamiltonian operator , although a more rigorous understanding of the quantum dynamics would require more sophisticated machinery .
our aim here is to describe the expected dominant corrections arising from the quantum theory and as such , there is room for additional corrections that could modify some of the results presented here .
evidence that the effective theory we describe provides an accurate description of the quantum dynamics is given by homogeneous and isotropic loop quantum cosmology with a massless scalar field with or without a cosmological constant @xcite .
the main quantum correction arises from the fact that in the loop quantization no direct operators exist corresponding to the connection components @xmath52 and @xmath33 .
instead , the quantization proceeds as in loop quantum gravity by considering holonomies of the connection .
the holonomies roughly consist of exponential terms like @xmath76 with @xmath77 corresponding to the _ edge length _ of the holonomy .
the end result in the effective hamiltonian is to replace the classical @xmath52 terms in the hamiltonian , with @xmath78 terms for instance .
thus the @xmath79 factors play a crucial role in determining the effects of the quantum corrections .
the classical limit is recovered by taking the @xmath80 limits .
the @xmath79 parameters , arising from the holonomy edge lengths , are a measure of the quantum discreteness arising from the loop quantization . in the original loop quantization of the schwarzschild interior @xcite
, the @xmath79 terms were assumed to be a constant value labeled by @xmath81 analogous to the original quantization of loop quantum cosmology where similar terms appear in the hamiltonian operator . in @xcite
, it was shown that because of the assumption that @xmath81 is a constant ( in loop quantum cosmology the parameter is labeled @xmath82 ) , loop quantum cosmology does not have the correct semi - classical limit .
the solution to this issue presented in that article is to relax the assumption that the @xmath81 factors are constant and instead consider them to be functions of the triad variables .
the precise prescription given leads to the non - constant factor labeled @xmath83 which varies inversely with the scale factor of the universe and with this behavior the correct semi - classical limit is achieved @xcite .
thus our goal will be to consider the effective dynamics with a quantization scheme arising from the improved prescription for loop quantum cosmology given in @xcite .
in particular , the @xmath79 factors will depend explicitly on the triad components @xmath29 and @xmath28 .
we will start however , with the case of constant @xmath81 since the effective dynamics are exactly solvable and allows to contrast the modifications from those arising from the new non - constant @xmath81 quantization .
the results for the constant @xmath81 case have been previously derived in @xcite .
we merely repeat the results that are relevant to our discussion . with the considerations of the previous discussion
the quantum modifications can be understood as leading to an effective hamiltonian with the holonomy parameter @xmath81 playing a crucial role in determining the magnitude of the corrections .
the effective hamiltonian constraint reads @xcite @xmath84 .
\label{a9}\end{aligned}\ ] ] it is easy to see that in the limit @xmath85 , the classical hamiltonian is recovered . in analogy to the classical case , we choose @xmath86 and the equations for the pairs @xmath87 and @xmath88 decouple to two independent pairs of differential equations @xmath89 the equation of motion for @xmath33 can be integrated to give @xmath90 where the constant of integration has been fixed so that the limit of the effective @xmath91 as @xmath85 coincides with the classical expression ( [ ccl ] ) .
next , the equations for @xmath28 can be solved @xmath92 also the equation for @xmath52 can be integrated and yields @xmath93 .
\label{a16}\end{aligned}\ ] ] finally , @xmath94 can be obtained from the vanishing of the hamiltonian which gives p_b = - .
> from the form of @xmath95 , it is evident that the evolution ends at a minimum value @xmath96 where the absolute value of the argument of the @xmath97 function equals one .
moreover , from the form of @xmath98 it is evident that the group orbits also have a minimal value at @xmath99 , with @xmath28 reaching its maximum value at the classical horizon and at @xmath96 .
this minimal two - sphere radius is given from @xmath100 , which depends on the mass of the black hole and also on @xmath22 .
figure [ fig : const1 ] shows the plots of @xmath101 for various values of @xmath6 and @xmath22 . the singularity is thus avoided through a bounce in the two - sphere radius and the quantum dynamics matches two black holes together though in a generically non - symmetric value .
a symmetric solution can be constructed by fine tuning the value of @xmath22 . as we have seen in the figures , the mass parameter @xmath6 controls the mass of the first black hole whereas @xmath22 controls the mass of the second black hole .
this suggests that in the effective quantum theory , @xmath22 would not be regarded as a meaningless constant of integration or a certain choice of gauge . for the constant @xmath81 quantization of the schwarzschild interior spacetime this parameter acquires an important physical meaning , and two different choices
correspond to different effective quantum theories .
any physical results derived from the constant @xmath81 quantization must include a prescription for specifying the value of @xmath22 which has to be input by hand .
we now wish to relax the assumption that the @xmath81 parameters are constant . in accordance with the results from loop quantum cosmology @xcite
, we will assume that @xmath79 depend on the triad variables @xmath28 and @xmath29 .
the method to constrain the exact dependence relies on the fact that the holonomies used to construct the quantum hamiltonian operator form closed loops .
the behavior of @xmath77 and @xmath102 can then be constrained if we demand that the physical area of the closed loop be equal to the minimum area gap predicted by loop quantum gravity a_min = = 2 l_^2 .
this procedure can be executed in different possible ways in the kantowski - sachs model considered here , whereas homogeneous isotropic models lead to a unique choice .
we discuss in this section what we feel to be the most geometrically natural scheme .
we will show in the section , that this scheme has the advantage that the resulting dynamics is independent of the @xmath22 parameter .
we will discuss an alternative scheme in appendix [ alt ] .
the procedure considered in this section amounts to evaluating the classical area of the holonomy loops and constraining the area based on this .
for instance if we consider the holonomy loop in the @xmath103 plane , the classical physical area is given by a_x = _ b _ c p_b which is evident from the form of the metric ( [ triadmetric ] ) .
for the holonomies on the two - sphere , the loop does not close as discussed in @xcite , but we can assign an effective area given by a _ =
_ b^2 p_c .
if we constrain these areas to be equal to @xmath104 we have the following behavior [ deltagood ] _ b & = & + _ c & = & . with this , the effective hamiltonian constraint reads @xmath105 .
\label{d1}\end{aligned}\ ] ] once again , in the limit @xmath106 the effective hamiltonian ( [ d1 ] ) reduces to the classical hamiltonian ( [ hcl ] ) .
we can now derive the equations of motion as in the previous case .
choosing the lapse function to be @xmath107 , the resulting equations of motion are given by & = & - - -c c _ c + & & + b b _ b ( 1- ) + [ d3 ] + & = & 2 p_c c _ c [ d4 ] + & = & - ( + ) - + c c _ c [ d5 ] + & = & b _ b ( 1 - ) p_b .
[ d6 ] before considering numerical solutions to these equations , we can get an idea on where the quantum modifications should be appreciable .
it is evident from the effective hamiltonian , the the modifications arise when either @xmath108 or @xmath109 are of the order of one . near the classical singularity we have @xmath110 and @xmath111 in such a way that @xmath109 and @xmath108 blow up
, hence we expect the quantum corrections to become important near the singularity .
if we also consider the horizon , we have @xmath112 and @xmath113 such that @xmath109 is small , but @xmath108 diverges
. thus we can expect quantum corrections also as the classical horizon is approached .
therefore , the picture that arises is a region in the interior where classical behavior is recovered with quantum effects near the classical singularity and also near the classical horizon .
this behavior is borne out in the numerical simulations which we now discuss .
we focus in this article on the effects near the singularity and will comment later on the horizon effects . accordingly , in the numerical simulations we will specify initial conditions starting in the interior region where the quantum effects are negligible and evolve towards the singularity .
to solve the equations of motion numerically , we start at some initial point near the horizon with initial conditions @xmath114 that match the classical solutions ( [ pccl],[pbcl],[bcl ] ) for some initial @xmath115 and the constraint @xmath116 is then solved to get @xmath117 .
the coupled differential equations of motion are then solved for the approach towards the classical singularity ( @xmath118 ) .
the results of the simulations for @xmath29 as a function of @xmath28 appear in figure [ fig : deltawell1 ] for various initial conditions corresponding to different values of the classical black hole mass @xmath6 as well as different values of @xmath22 .
the results all indicate that the classical singularity is resolved with the two - sphere radius @xmath28 being bounded from below .
the second figure of [ fig : deltawell1 ] indicates that choosing different values of @xmath22 only rescale @xmath94 as in the classical case , and thus the effective dynamics predicted is insensitive to this classical gauge choice as opposed to the constant @xmath81 case . unlike the constant @xmath81 case of section [ constantd ]
, the solutions do not match up to a separate black hole solution .
instead , the solutions asymptote to a nariai type solution for @xmath119 .
the solution is characterized by @xmath120 where the barred quantities and @xmath121 are constants .
figure [ fig : deltawell2 ] shows this behavior with plots of @xmath122 and the ratio of @xmath123 as functions of time .
it is clear that @xmath28 undergoes damped oscillations settling in to a fixed finite value as @xmath124 goes to minus infinity .
similarly the ratio of @xmath123 settles into a finite value given by @xmath125 as expected from a nariai type solution .
> from the equations of motion we can determine the asymptotic values of the constants @xmath126 and @xmath125 . assuming @xmath127 is a constant , eq .
( [ d4 ] ) implies @xmath128 @xmath129 which also yield @xmath130 . assuming furthermore that @xmath131 is constant , the resulting equation of motion for @xmath40 , namely eq .
( [ d5 ] ) yields @xmath132 the same equation is obtained by evaluating the vanishing of the hamiltonian constraint ( [ d1 ] ) .
next , eq . ( [ d6 ] ) implies that @xmath133 and fixes the constant @xmath121 to be @xmath134 finally , the above equations eqs .
( [ d9])([d11 ] ) can be inserted in the equation of motion for @xmath33 leading to @xmath135 for given value of @xmath136 this equation can be solved for the value of the ratio @xmath137 . a find root algorithm is necessary since an analytical solution can not be obtained .
the ratio @xmath137 then specifies the constant in the exponent @xmath121 by eq .
( [ d11 ] ) .
moreover , with @xmath138 , @xmath136 and @xmath137 given , the relation ( [ d10 ] ) can be used to extract the value of @xmath139 which in turn yields the value of @xmath140 . finally , eq . ( [ d9 ] )
fixes the ratio @xmath125 .
however , there is a remaining scaling freedom in @xmath141 and @xmath142 since only their ratio is specified .
note that this implies that rescalings in @xmath142 have to be compensated by simultaneous rescalings in @xmath141 .
> from the previous equations it is clear that the asymptotic value of @xmath139 depends only on the constants @xmath136 and @xmath138 as |p_c = ^2 g ( ) where @xmath143 is some function determined from the solutions of equations and . for the natural choice
@xmath144 the following asymptotic values are obtained @xmath145 where @xmath146 is the asymptotic value of the lapse which also behaves as a constant .
these values agree with the asymptotic region of the numerical solution that we studied , see in particular figure [ fig : deltawell2 ] .
the fact that @xmath28 asymptotes to a fixed value on the order of the planck length squared implies that the radius of the two - sphere asymptotes to a value on the order of the planck length . furthermore this value is independent of the mass of the black hole .
the interpretation of the effective dynamics becomes clear if we consider the radially in - falling test particle using equation .
a plot of the numerical integration of @xmath72 is shown in figure [ fig : geodesici ] .
an in - falling particle classically would reach the singularity at @xmath147 in finite proper time as evidenced by the solid black line , however as a result of the quantum effects , the particle oscillates about the fixed value of @xmath139 .
the particle thus settles in at a fixed planckian radius .
the geodesic reaches this asymptotic value in infinite proper time and hence we can say that the singularity is truly absent in the quantum modified spacetime .
the fate of an in - falling test particle is also of interest in the context of the information loss paradox .
the results are suggestive that the black hole can store information infinitely long , and strictly speaking no information can ever be lost . however , once the thermal black hole radiation is taken into account , things are still more involved .
in this respect , however , we would like to point out that the improved quantization scheme also leads to quantum modification of the horizon . in the context of euclidean quantum gravity
, it has been argued in @xcite that indeed horizon modification might resolve the paradox .
we have also considered the behavior of the kretschmann invariant , the square of the riemann tensor @xmath148 . for the metric ( [ triadmetric ] ) , written out explicitly , this yield a rather complicated expression .
however , plotting this quantity for our numerical solution also shows that the geometry of the spacetime changes significantly , from the classical schwarzschild like behavior where @xmath149 ( which diverges at @xmath150 ) towards a space of constant curvature , as expected for the nariai type metrics whose form is given by k = 4 ( + ) .
plugging in the theoretical values in gives an expected value k 124.36 m_^4 which is in good agreement with the numerical results .
furthermore this asymptotic value is independent of the black hole mass .
we find that the quantum effects become appreciable when the kretschmann curvature invariant approaches the planck scale .
figure [ fig : geodesici ] shows a plot indicating this behavior .
the quantum effects removing the singularity can best be visualized by carter - penrose diagrams , see e.g. @xcite .
the classical diagrams for both the schwarzschild and nariai spacetimes are given in figure [ figa ] .
the quantum effects remove the classical schwarzschild singularity shown as the wavy line at the top of region ii in the carter - penrose diagram and glue the diagram to the shaded region of the nariai diagram .
the resulting diagram of the interior is shown in figure [ figc ] .
at the level of the effective description we have presented , we have been able to answer the questions posed at the beginning , namely what replaces the singularity in the schwarzschild interior . in all of the quantization schemes
we have discussed , singularity avoidance arises naturally thus providing further evidence that loop quantum gravity successfully resolves the classical black hole singularity .
we have seen two possible replacements for the singularity , one in which quantum effects connect two separate ( non - identical ) black holes resembling a wormhole - like solution . the other solution predicting repulsive gravitational effects leading to an equilibrium two - sphere radius to which an in - falling particle would asymptote .
the results are suggestive but can not be considered as a proof that the lqg resolves the black hole singularity .
the effective equations considered here are not rigorously derived from the quantum theory and thus can only be considered approximations of the quantum dynamics . since understanding the quantum dynamics is difficult , we do not have a complete handle on when the approximation might break down .
as was mentioned however , the effective equations have approximated the quantum dynamics in the models where both can be developed completely , namely the homogeneous and isotropic models of lqc sourced with a massless scalar field . some initial attempts at constructing semi - classical states for the constant @xmath81 quantization have be made in @xcite and @xcite and it is worthwhile to compare our results with the previous work . in the works cited , numerical instabilities of the difference equation are found which bring into question whether any effective picture is valid near the singularity . in @xcite
it is argued that no proper semi - classical states exist unless the immirzi parameter @xmath138 is less the the accepted value in the lqg literature raising a further question . in this article , our aim has been to probe the improved , non - constant @xmath81 quantization and thus we do not see a direct contradiction .
in particular , the bound on the immirzi parameter depends explicitly on the constant value of @xmath81 and so it is not clear if a bound exists in the improved quantization .
in addition , both papers evolve gaussian semi - classical states with the quantum difference equation which seem to imply that semi - classicality breaks down near the singularity . the numerical results are based on using the triad component @xmath28 as an internal clock , however , our results suggest that @xmath28 bounces and hence would not play the role of a good clock . thus to reconcile our results with those of @xcite
, the numerical evolutions would probably require different initial conditions to incorporate both the expanding and contracting parts of the wave - packets at the initial instance in `` time '' .
this issue is solved in the isotropic case by including a scalar field that plays the role of a global internal clock .
we thus feel that our results are not in conflict with @xcite , and that all of our works requires further development of the quantum dynamics including the specification of a good internal clock , physical inner product , and dirac observables .
additionally , the validity of the kantowski - sachs description of the interior could be questioned .
boundary conditions must be provided for the numerical evolution at the horizon which are prescribed using classical solutions . however , this could be modified when considering a true quantization of the full spacetime .
in particular , the improved quantization scheme described in section [ gooddelta ] leads to modifications at the horizon which may require input from the full quantization . additionally , by considering the homogeneous interior , we can not probe a realistic collapse scenario .
therefore , an interesting next step would be to apply the techniques described here to the inhomogeneous spherically symmetric quantization .
the results described in @xcite provide the initial steps at an amenable quantization , though the results presented there are only valid in the exterior . if the results can be extended to cover the whole spacetime , then an effective dynamics analysis could be extended . with this model
, one could analyze a genuine collapse scenario .
we thank abhay ashtekar and roy maartens for valuable discussion .
the work of cgb was supported by research grant bo 2530/1 - 1 of the german research foundation ( dfg ) .
kv is supported by the marie curie incoming international grant m1f1-ct-2006 - 022239 .
here we consider an alternate behavior for the @xmath81 factors that are more heuristically motivated . in the language of
@xcite this corresponds to a lattice refinement model whereby the number of vertices is proportional to the transverse area .
in addition this scheme has been considered in anisotropic bianchi i models in @xcite .
a stability analysis indicates that this scheme leads to an unstable difference equation @xcite thus it may not represent a good quantization scheme .
an advantage of this scheme is that the quantum difference equation is much simpler than in the previous one and more amenable to a proper analysis of the true quantum dynamics although the instability of the difference equation could spoil this . in this scheme
we have the @xmath81 factors behaving as _ b & = & + _ c & = & . again the effective hamiltonian is given by @xmath151 , \label{d1a}\end{aligned}\ ] ] leading to equations of motion with lapse @xmath107 & = & c c _ c - 3 + _ c & = & 2 p_c c _ c + & = & - + - - + _ b & = & p_b b _ b - again we can examine where the quantum corrections are appreciable by determining where @xmath152 and @xmath153 are on the order of one . from the classical behavior ,
both @xmath152 and @xmath153 blow up at the singularity and thus quantum effects are expected there . at the horizon ,
@xmath152 vanishes and @xmath153 is small provided @xmath136 is small .
hence with this quantization scheme , the effects near the horizon are minimal .
the resulting numerical solutions are qualitatively very similar to those obtained for the constant @xmath81 quantization scheme .
as discussed above , the parameter @xmath22 controls the mass of the second black hole in the effective quantum theory which can be tuned to give a symmetric spacetime .
we note that similarly to the other quantization schemes , the singularity is resolved dynamically .
h. nariai . on some static solutions of einstein s gravitational field equations in a spherically symmetric case .
31:951961 , 1999 . originally published in _
the science reports of the tohoku university _ series i , vol .
* xxxiv * , no . 3 ( 1950 ) ,
p. 160167 .
h. nariai . on a new cosmological solution of einstein s field equations of gravitation .
, 31:963971 , 1999 .
originally published in _
the science reports of the tohoku university _ series i , vol .
* xxxv * , no . 1 ( 1951 ) , | we examine the schwarzschild interior of a black hole , incorporating quantum gravitational modifications due to loop quantum gravity .
we consider an improved loop quantization using techniques that have proven successful in loop quantum cosmology .
the central schwarzschild singularity is resolved and the implications for the fate of an in - falling test particle in the interior region is discussed .
the singularity is replaced by a nariai type universe .
we discuss the resulting conformal diagram , providing a clear geometrical interpretation of the quantum effects . |
arising from the light deflection by the inhomogeneous mass distribution of the universe , gravitational weak - lensing effects have been applied to map out , individually or statistically , the mass distribution of galaxies , clusters of galaxies , and large - scale structures ( e.g. , bartelmann & schneider 2001 ; mandelbaum et al .
2008 ; li et al . 2009 ; clowe et al . 2006 ; corless et al . 2009 ; fu et al .
deep observations also reveal the time evolution of the structure formation ( massey et al .
cosmic shear analyses have been used to constrain different cosmological parameters ( e.g. , benjamin et al .
2007 ; kilbinger 2009 ; li et al . 2009 ) .
future wide and deep weak - lensing surveys will give rise to observational data with greatly improved quantity and quality .
therefore the derived cosmological information is expected to be much more accurate than that of today , which will make the weak - lensing effect one of the most important probes in cosmology , notably , in understanding the nature of dark energy ( e.g. , albrecht et al .
however , it has been realized that weak - lensing effects are not as clean as they appear to be .
being very weak , they can be affected and contaminated easily by different observational and physical effects .
considerable efforts have been paid to understand these effects and their influences on cosmological studies ( e.g. , tang & fan 2005 ; heymans et al .
2006 ; ma et al . 2006
; huterer et al . 2006 ; bridle & king 2007 ; fan 2007 ; amara & refregier 2008 ; sun et al .
2009 ) . among others , intrinsic ellipticities of source galaxies are known to dominate over their weak - lensing induced shear signals by a large factor .
the environmental dependence of galaxy formation can lead to intrinsic - intrinsic and shear - intrinsic correlations , which can contaminate directly the two - point weak - lensing shear correlations at percent to tens of percent level ( e.g. , hirata & seljak 2004 ; mandelbaum et al .
2006 ; fan 2007 ; zhang 2008 ; joachimi & schneider 2009 ) .
moreover , in reconstructing the foreground mass distribution from shear measurements of background galaxies , their intrinsic ellipticities leave considerable noise even if they are completely random ( e.g. , van waerbeke 2000 ; hamana et al .
2004 ; white et al . 2002 ) .
understanding fully the statistical properties of the noise field and the mutual influence between noise and the true lensing effects are therefore crucially important in studies aiming to reveal the mass distribution of large - scale structures .
particularly , chance alignments of intrinsic ellipticities can form false peaks in a lensing convergence map , and mimic dark clumps , namely structures with large mass to light ratios ( e.g. , erben et al . 2000 ; linden et al .
2006 ; gavazzi & soucail 2007 ; schirmer et al . 2007 ; fan 2007 ) .
furthermore , noise can affect true lensing signals , causing large scatters and possibly systematic biases to them .
these can lead to considerable uncertainties in analyzing the mass distribution of clusters of galaxies , in investigating the possible existence of dark clumps , and in deriving constraints on cosmological parameters from weak - lensing cluster abundances .
different statistical methods have been developed to suppress such noise in weak - lensing mass reconstructions , and the smoothing is one of the simplest ways for doing so .
given the rms of intrinsic ellipticities , the amplitude of the smoothed noise depends on the smoothing scale and on the surface number density of source galaxies thus the observational depth ( e.g. , van waerbeke 2000 ) . because of the central limit theorem , it is expected that the statistics of the noise field after smoothing should be close to gaussian if a large enough number of galaxies are included within the smoothing window . by employing a gaussian smoothing function , van waerbeke ( 2000 )
shows that the noise field of the convergence is indeed approximately gaussian .
he further analyzes its properties , such as the number and the shape distribution of the noise peaks .
fan ( 2007 ) extends the study by including the intrinsic alignment of background galaxies in the analyses , and shows that the number of false peaks are enhanced depending on the strength of the intrinsic alignment .
schneider ( 1996 ) consider the effects of noise on the height of the peaks associated with true mass concentrations in aperture - mass maps by adding a gaussian scatter to the peak height .
from numerical simulations , hamana et al .
( 2004 ) analyze noise effects on true peaks , and present a phenomenological correction to the abundance of weak - lensing detected peaks . in this paper
, we put forward a theoretical model to study the mutual influence of noise and true peaks .
specifically , we consider the lensing effects of a cluster of galaxies taking into account noise from intrinsic ellipticities of source galaxies .
we investigate the effect of the density profile of the cluster on the distribution of noise peaks , and conversely , how noise can change the true peak signal of the cluster .
based on these analyses and the mass function of dark matter halos , we develop a model to compute statistically weak - lensing convergence peak abundances over a large field , including both the peaks associated with clusters of galaxies and the false ones from noise .
therefore we can use convergence peaks directly as cosmological probes without the need to sort out true and false peaks .
moreover , the distribution of noise peaks depends on the density profile of true clusters , and thus contains additional cosmological information , which is naturally included in our model .
the rest of the paper is organized as follows . in 2 , we present our theoretical framework .
the results are shown in 3 with 3.1 for individual cluster cases and 3.2 for statistical results on a large field .
4 contains summaries and discussions .
with the intrinsic ones taken into account , the observed ellipticities of source galaxies can be written in the weak lensing regime as ( e.g. , bartelmann & schneider 2001 ) @xmath20 where @xmath21 is the complex lensing shear due to the foreground mass distributions , and @xmath22 and @xmath23 represent the observed and the intrinsic complex ellipticities of source galaxies , respectively .
the corresponding smoothed quantities are @xmath24 where @xmath25 and @xmath26 are the smoothed @xmath22 and @xmath27 , respectively , @xmath28 is the smoothing function , and @xmath29 and @xmath30 are , respectively , the surface number density and the total number of source galaxies in the field . the lensing convergence @xmath31 can be obtained from the shear through their following relation in the fourier space @xmath32 where the summation over @xmath33 is implied , and @xmath34 $ ] with @xmath35 ( kaiser & squires 1993 ) .
then the real - space convergence reconstructed from the observed ellipticities is @xmath36 where @xmath31 and @xmath37 are the lensing convergence and the noise from the intrinsic ellipticities , respectively . after smoothing ,
we have @xmath38 with @xmath39 and @xmath40 the corresponding smoothed @xmath31 and @xmath37 .
it is therefore clear that in order to extract the true @xmath39 from the noisy @xmath41 , we must understand the influence of @xmath40 . in this paper
, we mainly consider foreground clusters of galaxies as lens objects .
considering an individual cluster , we take @xmath42 to be the smoothed convergence of the cluster , which is determined by its assumed known density profile . the noise @xmath40 is regarded as a random field .
the total @xmath43 is therefore also a random field with its statistics determined by the noise field @xmath44 and modulated by the cluster @xmath42 . concerning peaks in the noisy convergence @xmath43 around an individual cluster
, there should be a dominant one corresponding to the main cluster .
there are also false peaks arising from @xmath40 .
it is expected while the properties of the dominant peak in @xmath43 should be closely related to those of true @xmath42 , they are inevitably affected by the existence of @xmath40 . on the other hand , the noise peaks in @xmath43
should also be modulated by the cluster @xmath42 in comparison with those in pure noise field @xmath40 .
it should be noted that the peak statistics depend not only on the value of @xmath41 , but also on its profile through the first and second derivatives of @xmath41 .
thus @xmath39 and @xmath40 need to @xmath45collaborate@xmath45 to give rise to peaks in @xmath41 .
therefore the peak properties in @xmath41 differ from those in @xmath40 or in @xmath39 in ways that are more complex than just changing the peak height by a value of @xmath39 or @xmath40 .
our main focus in the paper is to understand such mutual influences between @xmath39 and @xmath40 in terms of the peak properties in @xmath41 , and further their effects on relevant cosmological studies .
the formulations presented in the following are for noisy @xmath43 around a cluster with a known density distribution .
these set the theoretical framework for analyses presented in 3.2 on large - scale statistics beyond individual clusters incorporating the mass function of dark matter halos .
it has been shown that the smoothed noise field @xmath40 is approximately gaussian because of the central limit theorem ( van waerbeke 2000 ; fan 2007 ) .
for an individual cluster , its convergence @xmath39 is regarded as a known quantity .
thus the statistics of the field @xmath41 is also gaussian and follows that of @xmath40 with the modifications due to the presence of @xmath39 from the cluster . for the random gaussian field @xmath41 ,
its peak statistics involves random variables @xmath41 , @xmath46 ( denoted as @xmath47 ) and @xmath48 ( denoted as @xmath49 ) ( @xmath50 ) .
their joint probability distribution function can be readily written out as follows ( e.g. , bardeen et al .
1986 ; bond & efstathiou 1987 ) @xmath51^{1/2}}\exp\bigg \{-{\{(k_n - k)/\sigma_0+\gamma_n [ ( k^{11}_n - k^{11 } ) + ( k^{22}_n - k^{22})]/\sigma_2\}^2\over 2(1-\gamma_n^2)}\bigg \}\times \nonumber \\ & & { 1\over 2\pi\sigma_2 ^ 2}\exp\bigg \{-{[(k^{11}_n - k^{11})-(k^{22}_n - k^{22})]^2\over 2 \sigma_2 ^ 2 } -{(k^{11}_n - k^{11})^2\over \sigma_2 ^ 2}-{(k^{22}_n - k^{22})^2\over \sigma_2 ^ 2}\bigg \}\times \nonumber \\ & & { 8\over ( 2\pi)^{1/2}\sigma_2}\exp\bigg \{-{4(k^{12}_n - k^{12})^2\over \sigma_2 ^ 2}\bigg \ } \times \nonumber \\ & & { 1\over \pi\sigma_1 ^ 2}\exp\bigg [ -{(k^1_n - k^1)^2\over \sigma_1 ^ 2}-{(k^2_n - k^2)^2\over \sigma_1 ^ 2}\bigg ] \hbox{\ } dk_n\hbox{\ } dk^{11}_{n}\hbox{\ } dk^{22}_{n}\hbox { } dk^{12}_{n}\hbox { } dk^1_{n}\hbox { } dk^2_{n},\end{aligned}\ ] ] where all the quantities related to @xmath41 are attached a subscript n , while those without the subscript refer to the ones from @xmath39 .
the @xmath52 , @xmath53 and @xmath54 are the moments of the noise field @xmath40 , which are given by , with @xmath55 being the fourier transform of @xmath40 ( e.g. , van waerbeke 2000 ) , @xmath56 the quantity @xmath57 is defined as @xmath58 . by diagonalizing the second derivative tensor @xmath59 to obtain the eigen values of @xmath60 and @xmath61 and
by defining @xmath62 we have @xmath63 where @xmath64 is the rotation angle in the range of @xmath65 $ ] .
we further define @xmath66 , then eq . ( 6 ) can be simplified to @xmath67\times \nonumber \\ & & \bigg \{{d\nu_n\over [ 2\pi(1-\gamma_n^2)]^{1/2}}\exp\bigg \{-{[(\nu_n - k/\sigma_0)-\gamma_n x_n -\gamma_n(k^{11}+k^{22})/\sigma_2]^2\over 2(1-\gamma_n^2)}\bigg \}\times \nonumber \\ & & { dx_n\over ( 2\pi)^{1/2}}\exp\bigg ( -{x_n^2\over 2}\bigg ) { 8\over \pi}x_n^2e_n\hbox { } de_n\hbox { } d\theta_n \exp(-4x_n^2e_n^2)\exp\bigg [ -x_n\bigg ( { k^{11}+k^{22}\over \sigma_2}\bigg ) \bigg ] \times \nonumber \\ & & \exp\bigg [ -4x_ne_n\cos2\theta_n\bigg ( { k^{11}-k^{22}\over \sigma_2}\bigg ) \bigg ] \exp\bigg ( -8x_ne_n\sin2\theta_n{k^{12}\over \sigma_2}\bigg ) \times \nonumber \\ & & { 1\over \pi\sigma_1 ^ 2}\exp\bigg [ -{(k^1_n - k^1)^2\over \sigma_1 ^ 2}-{(k^2_n - k^2)^2\over \sigma_1 ^ 2 } \bigg ] \hbox { } dk^1_{n}\hbox { }
dk^2_{n}\bigg \}.\end{aligned}\ ] ] considering peak statistics , the average number density of maxima given @xmath68 , and @xmath69 is ( bond & efstathiou 1987 ) @xmath70 where the average is calculated with the probability function given by eq .
( 10 ) , and the step functions @xmath71 and @xmath72 are required to satisfy the conditions of maxima with @xmath73 . thus the number density of peaks given @xmath74 and @xmath75 is @xmath76\exp \bigg [ -{(k^1)^2+(k^2)^2\over \sigma_1 ^ 2}\bigg ] \times \nonumber \\ & & \bigg \ { { 1\over 2\pi\theta_*^2}{1\over [ 2\pi(1-\gamma_n^2)]^{1/2}}\bigg\ } \exp\bigg \{-{[(\nu_0-k/\sigma_0)-\gamma_n x_0 -\gamma_n(k^{11}+k^{22})/\sigma_2]^2\over
2(1-\gamma_n^2)}\bigg \}\times \nonumber \\ & & { 1\over ( 2\pi)^{1/2}}\exp\bigg [ -{x_0 ^ 2\over 2}-x_0\bigg ( { k^{11}+k^{22}\over \sigma_2}\bigg ) \bigg ] \int_0^{1/2}de_n \hbox { } { 8\over \pi}(x_0 ^ 2e_n)x_0 ^ 2(1 - 4e_n^2 ) \exp(-4x_0 ^ 2e_n^2 ) \times \nonumber \\ & & \int_0^{\pi } d\theta_n \hbox { } \exp\bigg [ -4x_0e_n\cos2\theta_n\bigg ( { k^{11}-k^{22}\over \sigma_2}\bigg ) \bigg ] \exp\bigg ( -8x_0e_n\sin2\theta_n{k^{12}\over \sigma_2}\bigg ) , \nonumber \\\end{aligned}\ ] ] where @xmath77 .
we reorganize the above equation and obtain the following @xmath78 \bigg \ { { 1\over 2\pi\theta_*^2}{1\over ( 2\pi)^{1/2}}\bigg\ } \exp\bigg [ -{1\over 2}\bigg ( \nu_0-{k\over \sigma_0}\bigg ) ^2\bigg ] \times \nonumber \\ & & { 1\over [ 2\pi(1-\gamma_n^2)]^{1/2}}\exp\bigg \{- { [ { x_0+(k^{11}+k^{22})/ \sigma_2 -\gamma_n(\nu_0-k/\sigma_0)}]^2\over 2(1-\gamma_n^2)}\bigg \ } \times f(x_0 ) , \nonumber \\ \end{aligned}\ ] ] where @xmath79 \times \nonumber \\ & & \int_0^{1/2}de_n \hbox { } 8(x_0 ^ 2e_n)x^2(1 - 4e_n^2 ) \exp(-4x_0 ^ 2e_n^2 ) \times \nonumber \\ & & \int_0^{\pi } { d\theta_n\over \pi } \hbox { } \exp\bigg [ -4x_0e_n\cos2\theta_n\bigg ( { k^{11}-k^{22}\over \sigma_2}\bigg ) \bigg ] \times \nonumber \\ & & \exp\bigg ( -8x_0e_n\sin2\theta_n{k^{12}\over \sigma_2}\bigg ) .\end{aligned}\ ] ] further we have @xmath80 \bigg \ { { 1\over 2\pi\theta_*^2}{1\over ( 2\pi)^{1/2}}\bigg\ } \exp\bigg
[ -{1\over 2}\bigg ( \nu_0-{k\over \sigma_0}\bigg ) ^2\bigg ] \times \nonumber \\ & & \int { dx_n\over [ 2\pi(1-\gamma_n^2)]^{1/2}}\exp\bigg \{- { [ { x_n+(k^{11}+k^{22})/ \sigma_2 -\gamma_n(\nu_0-k/\sigma_0)}]^2\over 2(1-\gamma_n^2)}\bigg \ } \times f(x_n ) .
\nonumber \\\end{aligned}\ ] ] it is seen clearly that the number of peaks near a cluster of galaxies depends on the density profile of the cluster through @xmath39 , @xmath81 , and @xmath82 @xmath83 . assuming a spherical mass distribution for a cluster of galaxies , we can calculate @xmath84 at different radii along a specific direction where @xmath85 in our chosen coordinate system .
then we have @xmath86 \times \nonumber \\ & & \int_0^{1/2}de_n \hbox { } 8(x_n^2e_n)x_n^2(1 - 4e_n^2 ) \exp(-4x_n^2e_n^2 ) \times \nonumber \\ & & \int_0^{\pi } { d\theta_n\over \pi } \hbox { } \exp\bigg [ -4x_ne_n\cos 2\theta_n{(k^{11}-k^{22})\over \sigma_2}\bigg ] .
\nonumber \\\end{aligned}\ ] ] without the foreground cluster , the function @xmath87 can be written out analytically as @xmath88 ( bond & efstathiou 1987 ) . with a non - zero @xmath89
, we find a fitting formula for @xmath87 with @xmath90 - 1\},\ ] ] where @xmath91 and @xmath92 is a fitted value depending on @xmath93 . in figure 1
, we show two sets of results for @xmath94 $ ] with @xmath95 ( upper set ) and @xmath96 ( lower set ) , respectively .
for each set , both the numerical result ( solid line ) and the one calculated from the fitting formula ( dashed line ) are shown . for @xmath95 and @xmath96 , we have @xmath97 and @xmath98 , respectively .
it is seen that eq.(17 ) provides an excellent approximation to @xmath87 , which will be adopted for further calculations .
given a spherical cluster profile , we can obtain its convergence @xmath31 , and further the smoothed @xmath39 with a smoothing function @xmath99 through @xmath100 the relevant derivatives are @xmath101 and @xmath102 in our analyese , we use the gaussian smoothing function with @xmath103 where @xmath104 is the smoothing scale .
we calculate @xmath39 , @xmath81 and @xmath82 along the @xmath105 ( i.e , @xmath106 ) direction .
then the relevant quantities become @xmath107\kappa(x_1^ { ' } , x_2^{'}),\ ] ] @xmath108\kappa(x_1^ { ' } , x_2^{'}),\ ] ] @xmath109|_{x_2=0}= \int \hbox { } dx_1^{'}dx_2^ { ' } \hbox { } { 1\over \pi \theta_g^2 } { ( -4)\over \theta_g^2 } \exp\bigg [ -{(x_1^{'}-x_1)^2+(x_2^{'})^2\over \theta_g^2}\bigg ] \kappa(x_1^ { ' } , x_2^ { ' } ) \nonumber \\ & & + \int \hbox { } dx_1^{'}dx_2^ { ' } \hbox { } { 1\over \pi \theta_g^2 } { 4\over \theta_g^2}{(x_1^{'}-x_1)^2+(x_2^{'})^2\over \theta_g^2 } \exp\bigg [ -{(x_1^{'}-x_1)^2+(x_2^{'})^2\over \theta_g^2}\bigg ] \kappa(x_1^{'},x_2^ { ' } ) , \end{aligned}\ ] ] @xmath110|_{x_2=0}=\int dx_1^{'}dx_2^ { ' } & & { 1\over \pi \theta_g^2 } { 4\over \theta_g^2}{(x_1^{'}-x_1)^2-(x_2^{'})^2\over \theta_g^2}\nonumber \\ & & \times \exp\bigg [ -{(x_1^{'}-x_1)^2+(x_2^{'})^2\over \theta_g^2}\bigg ] \kappa(x_1^{'},x_2^ { ' } ) , \nonumber \\\end{aligned}\ ] ] and @xmath111 and @xmath85 .
we consider two types of most commonly used cluster profiles , namely , the isothermal profile and the nfw profile ( navarro , frenk & white 1996 ) .
for the isothermal one , we adopt the non - singular isothermal sphere model with its lensing potential having the form ( van waerbeke 2000 ) @xmath112 where @xmath1 is the core radius , @xmath113 is the central position of the cluster , and @xmath114 here @xmath115 and @xmath116 are the angular diameter distances between the observer and the lens , between the lens and the source , and between the observer and the source , respectively , and @xmath117 represents the velocity dispersion of the isothermal cluster .
we choose @xmath118 , then the convergence @xmath31 of the cluster is @xmath119^{3/2}},\ ] ] where @xmath120 is related to @xmath121 and @xmath1 . for the nfw profile , we have ( navarro , frenk & white 1996 ) @xmath122 where @xmath123 and @xmath124 are the characteristic mass density and scale of a dark matter halo .
the corresponding convergence is @xmath125 with @xmath126}+{1\over \bigg [ 1-{(\tilde r^s)}^2\bigg]^{3/2}}\hbox { arccosh } ( 1/ { \tilde r^s})\quad \quad \hbox{for } \tilde r^s < 1 \nonumber \\ & & = { 1\over 3 } \qquad \qquad \qquad \qquad \qquad \qquad \quad \quad \qquad \qquad \qquad \quad \hbox{for } \tilde r^s=1 \nonumber \\ & & = -{1\over \bigg [ 1-{(\tilde r^s)}^2\bigg ] } -{1\over \bigg [ { ( \tilde r^s)}^2 - 1\bigg ] ^{3/2}}\hbox { arccos}(1/ { \tilde r^s})\quad \quad \hbox{for } \tilde r^s > 1\end{aligned}\ ] ] where @xmath127 .
given the redshift of the lens , @xmath128 can also be expressed in angular coordinates with @xmath129 , so for the isothermal case of eq .
( 28 ) . in figure 2
, we show the smoothed quantities @xmath130 , @xmath131 , @xmath132 , and @xmath133 for the two profiles , respectively .
the specific parameters are @xmath0 , @xmath134 and @xmath135 for the isothermal model ( solid lines ) . for the nfw model ( dashed lines ) , we adjust @xmath136 and the central amplitude of @xmath137 so that it has the same mass and the same smoothed central @xmath39 as those of the isothermal cluster . in this case ,
note that the horizontal axes in the plots are @xmath139 .
it is seen that the two profiles are different , especially in the inner region .
thus we expect to see differences of the peak statistics in the two cases .
this section contains two parts . in
3.1 , we show the results around individual clusters .
the isothermal and the nfw clusters are analyzed .
the field extends to @xmath140 for all the considered cases .
the analyses shown here are relevant to weak lensing studies targeted at individual clusters . in 3.2 , the statistical results over a large field are shown . specifically , we present a theoretical model based on the mass function and the nfw density profile of dark matter halos to predict the numbers of high peaks in the weak lensing convergence map over a large field .
our model takes into account the effects of noise on the peak heights of true clusters and the enhancement of noise peaks near clusters of galaxies .
given a cosmology , our model can then predict the number distribution of peaks with different heights .
these cosmology - dependent quantities can be used as very promising and efficient probes to constrain cosmologies .
the advantages of these probes are two folds .
firstly , because we concern the total number of peaks , we do not need to differentiate true peaks and false peaks , which often requires extensive follow - up observations .
secondly , the redshift information of the peaks is not necessary since we use the peak height distribution rather than the redshift distribution of peaks as cosmological probes .
here we discuss the peak statistics around individual clusters of galaxies .
two aspects are addressed .
one is how the distribution of noise peaks is influenced by the density profile of a cluster .
the other is how the properties of the peak corresponding to the true cluster , such as its height and position , are affected by the presence of noise . in our calculations , we assume that a considered cluster is located at the center , and the field extends to @xmath140 .
our main results are presented for the two clusters shown in figure 2 .
in addition , we also analyze different nfw clusters of the same mass but with different concentrations to further demonstrate the dependence of the peak statistics on cluster profiles . besides theoretical analyses , we perform monte carlo simulations with different realizations of the noise field added to the considered cluster @xmath31-map .
the theoretical results are then compared with the results averaged over @xmath141 simulations for each of the clusters . for a visual impression
, we first present in figure 3 a typical set of maps from our monte carlo simulations for the two clusters shown in figure 2 .
figure 3a and figure 3b are for the isothermal cluster , and the nfw cluster , respectively .
it is seen that in the regions outside @xmath1 , there is almost a one - to - one correspondence between the peaks in the two maps .
this is because in these outer regions , the density profiles of the two clusters are relatively flat with small values of @xmath131 , @xmath142 and @xmath143 .
then the peaks are essentially the ones existed in the pure noise field but with their heights modified by the term @xmath130 .
however , it should be noted that @xmath130 depends on the radius @xmath144 , and thus the peak statistics over a range of @xmath144 can not be described by an overall constant shift in peak height from that of @xmath40 . within @xmath1 , the peaks look very different in the two maps , which clearly shows the effects of the density profile ( @xmath131 , @xmath142 and @xmath143 ) in addition to the term @xmath130 . for the nfw cluster , there is a single high peak which corresponds clearly to the peak signal of the main cluster . for the isothermal cluster ,
however , two prominent peaks with comparable heights occur .
the positions of the peaks are also shifted from the center .
such peak configurations due to noise can cause complications in weak lensing studies of the structure of clusters . to be quantitative
, we present in figure 4 the radial distributions of the number of peaks above a certain threshold around the two clusters .
in each panel , the lines are the results from our theoretical calculations , and the symbols are the average results from the @xmath141 monte carlo simulations for each case . the dash - dotted line and the + symbols are for the nfw cluster . the solid line and the diamonds are for the isothermal cluster .
the vertical axes are for the number of peaks within rings of width @xmath145 , where @xmath139 .
different scales of the horizontal and vertical axes in different panels should be noted .
it is seen that the results from the theoretical model and from the simulations agree very well .
the peak distributions for the two cluster profiles are distinctively different , especially in the inner region and for high peaks with @xmath4 . for the nfw cluster
, high peaks are mainly found near the central region of the cluster , and the total number of high peaks is smaller than that of isothermal cluster . in the region with @xmath146 , the average numbers of peaks for @xmath147 and @xmath148 are approximately @xmath149 , @xmath150 , @xmath151 and @xmath152 , respectively , for the nfw case . for the isothermal case ,
the corresponding numbers are @xmath153 , @xmath154 , @xmath155 and @xmath156 .
therefore , for the nfw cluster , the inner region is relatively clean and often contains a single and well defined high peak as seen in figure 3b . on the other hand , for the isothermal cluster
, there are on average @xmath157 peaks with @xmath158 in the inner region , of which , one corresponds to the real cluster and the other is generated due to noise ( figure 3a ) .
we further consider the inner ( @xmath2 ) and outer ( @xmath159 ) regions separately . in the outer regions ,
we analyze the occurrence probability of high noise peaks , which is relevant to the dark clump problem .
we emphasize , with respect to the case without clusters , the enhancement due to the cluster density distribution .
figure 5a shows the distribution of the cumulative number of peaks above the threshold @xmath160 in the region @xmath161 ( @xmath139 ) and
figure 5b presents the corresponding enhancement factor @xmath162 , where @xmath163 and @xmath164 denote , respectively , the cumulative numbers of peaks in the cases with and without a cluster in the center . in figure 5a , the solid , dash - dotted and dashed lines are for the isothermal cluster , the nfw cluster , and the case without a cluster , respectively . the diamond symbols and the plus symbols
are for the results averaged over @xmath141 simulations of the isothermal and nfw clusters , respectively . in figure 5b , the solid and dash - dotted lines are for the corresponding enhancement factors of the two clusters .
it is seen that the results are similar for the two clusters because their density profiles are similar in the outer regions as shown in figure 2 .
however , differences can still be seen for high peaks . for @xmath165 ,
the enhancement factor is larger for the isothermal cluster than that for the nfw cluster .
this is due to the fact that in the outer regions , these high peaks are mainly found in @xmath166 , where @xmath130 is higher for the isothermal cluster ( figure 2a ) . for @xmath167 , @xmath168 , @xmath169 , @xmath170 , @xmath171 , @xmath172 , @xmath173 and @xmath174 for the isothermal cluster .
the corresponding @xmath11 for the nfw cluster are @xmath175 , @xmath172 , @xmath176 , and @xmath177 . therefore depending on the density profile of a cluster
, the occurrence probability of high noise peaks around it can be greatly boosted .
the effects are larger for higher peaks .
as discussed earlier , in the outer region , the effect of the density profile of a cluster on noise peak statistics comes in mainly through the term @xmath130 .
it increases the heights of the peaks originally with @xmath178 to be above the threshold @xmath179 . because of the approximately exponential dependence of @xmath180 on @xmath7 , the enhancement at @xmath181 changes sensitively with the values of @xmath130 and @xmath179 .
this enhancement in outer regions can be analogous to the peak - background split scenario in explaining the biased dark matter halo distributions with respect to the dark matter mass density distribution ( bardeen et al .
1986 ; mo & white 1996 ) . it is noted , however , that @xmath130 changes with radius .
thus the local enhancement depends on the radius , and the overall enhancement factor @xmath11 over the outer region with @xmath161 can not be modeled by a constant shift of @xmath179 with respect to that of @xmath40 . when the observational field is comparable to the size of a cluster , we should also be aware that the mass - sheet degeneracy in lensing analyses brings some arbitrariness in the reconstructed mass distribution , which in turn affects the noise peak statistics within the field . in the weak lensing regime , this arbitrariness is mainly reflected by an additive constant to @xmath31 . in real studies , different people may choose different constants .
it can be chosen so that the averaged @xmath182 in the observed field ( e.g. , erben et al .
2000 ) , or to make the @xmath31 averaged over the outer - most ring of the observed field be zero ( e.g. , kaiser & squires 2003 ) .
it is also attempted to recover the true mass distribution of a cluster by adjusting the constant to make the reconstructed mass distribution be consistent with the distribution determined by some other observations ( e.g. , umetsu & futamase 2000 ) . in figure 6a and figure 6b
, we show @xmath11 for the isothermal and nfw clusters , respectively , by applying an additive constant to the @xmath31 field to have @xmath182 over the area of @xmath183 ( dashed ) , to have @xmath182 at @xmath184 ( dash - dotted ) .
the solid lines are for the true mass distribution .
it is seen that different choice of the constant results significantly different enhancements .
however , even in the case of @xmath182 over the field within @xmath184 , which gives the lowest enhancement factor , @xmath185 and @xmath186 at @xmath187 for the isothermal and nfw clusters , respectively .
the strong association of noise peaks with their nearby clusters can have significant effects on cosmological studies with weak lensing effects .
it has been argued that the @xmath45dark clump@xmath45 found near cluster a1942 ( e.g. , erben et al .
2000 ) is unlikely to be a noise peak resulting from the chance alignment of source galaxies .
this is based on the peak statistics without considering the cluster environment , which gives the average number of noise peaks with @xmath188 in the considered area of @xmath189 to be about @xmath190 for @xmath0 . from our results shown here
, the number of noise peaks can be increased by a factor of @xmath185 if the density profile of the cluster is approximately isothermal .
this leads to an order of magnitude increase in the probability that the seen @xmath45dark clump@xmath45 near a1942 is a false one without corresponding to real mass concentrations .
thus in order to properly quantify the statistical significance of a @xmath45dark clump@xmath45 near a cluster , the cluster mass distribution has to be taken into account .
specifically , in observational weak lensing studies , such as the one for a1942 , one can first make an approximate modeling of the mass distribution of the main cluster from the weak lensing measurements .
then our model can be applied to study the occurrence probability of noise peaks in its surrounding area .
such a probability is more appropriate than the one from the blind estimate without considering the existence of the cluster itself .
it should also be emphasized that this probability depends sensitively on the density profile of the cluster . moreover , because of its strong dependence on the density profile of the cluster , the spatial distribution of noise peaks can mimic , to a certain extent , the true substructure distribution within the cluster .
thus the noise peaks can contaminate the cosmological information inferred from weak lensing substructure analyses .
we now turn to the inner region with @xmath2 or @xmath191 .
the highest peak within this region is assumed to correspond to the signal from the cluster itself .
we then analyze how the height and position of this main - cluster peak are affected by the existence of noise . in the case without noise
, the expected weak lensing peak heights should be the same for a sample of clusters with the same mass distribution and at the same redshift . with noise
included , however , their measured peak heights do not have the same value anymore . instead
, they follow a probability distribution . based on our analyses shown in the previous section
, the probability can be written as @xmath192},\ ] ] where @xmath193 is given by eq.(15 ) and the central values of @xmath194 and @xmath195 at @xmath196 are used in calculating the probability . in figure 7 , we show @xmath197 for the isothermal cluster ( upper panel ) and the nfw cluster ( lower panel ) , respectively . in each panel ,
the solid line is for the theoretical result from eq .
( 31 ) , and the histograms are for the result from the @xmath141 monte carlo simulations . for each simulation
, we identify the highest peak within the region @xmath191 to be the one corresponding to the main cluster .
the true cluster mass distribution is considered without additional treatments for the mass - sheet degeneracy .
firstly , it is seen that the solid line is in very good agreement with the results from simulations , showing that eq.(31 ) can properly model the effects of noise on the measured height of the main - cluster peak .
secondly , it is noted that the average value of the measured peak height is shifted toward larger @xmath7 with respect to the true height of the cluster ( indicated by the vertical dotted line in each panel ) .
the specific value of the shift depends sensitively on the density profile of the cluster . for the isothermal cluster ,
the shift is about @xmath198 from @xmath6 of the true peak height to @xmath199 , whereas for the nfw cluster , the shift is smaller with @xmath200 .
this systematic shift due to the existence of noise is closely related to the fact that the differential peak number distribution of a pure gaussian noise field ( considering only maxima peaks ) is peaked at @xmath201 , rather than at @xmath202 .
such a positive shift can increase the number of weak - lensing cluster detections , and thus affect the corresponding cosmological applications considerably . in the next subsection , we present a detailed model to calculate quantitatively the expected number of weak - lensing convergence peaks , true or false , from a large - scale weak lensing survey , taking into account such a shift and the enhancement of the noise peaks around clusters of galaxies .
this model allows us to be able to use directly the high peaks detected in a large - scale weak - lensing convergence map as cosmological probes without the need of expensive follow - up observations to differentiate true and false peaks .
the existence of noise also results a spatial offset for the identified main - cluster - peak from the true central position of the cluster .
this can be understood as follows . without noise ,
the main - cluster - peak in the weak lensing convergence map locates exactly at the cluster center @xmath203 in our considered cases of spherical clusters . at this spatial location , the conditions @xmath204 @xmath205 are satisfied . with noise , however , the location of a peak should satisfy necessarily the conditions @xmath206 @xmath205 , rather than @xmath204 @xmath205 . given a realization of a noise field n and considering a small offset @xmath207 @xmath205
, the original central peak now shifts to a location which can be estimated by @xmath208 @xmath209 , where we have used @xmath210 .
thus the offset distribution depends on the distribution of @xmath211 , which , in the considered gaussian case , is characterized by @xmath53 defined in eq .
it also depends sensitively on the density profile of the main cluster through @xmath212 .
specifically , the offset distribution can be approximately modeled by @xmath213},\ ] ] where the peak number density @xmath214 can be obtained from eq .
( 15 ) by taylor expansion of the quantities of @xmath215 and @xmath82 with respect to @xmath216 . in our considered case , @xmath217 and @xmath218 ,
then to the linear order of @xmath216 , we have @xmath219 ^ 2\ } \over \int_0^{r_c } d\delta r\hbox { } \delta r \exp\{-[k^{11}(0)\delta r/\sigma_1]^2\}]}.\ ] ] in terms of @xmath220 , eq.(33 ) can be written as @xmath221\over \{0.5{\tilde \sigma}^2[1-\exp(-{\tilde\sigma}^{-2})]\}}\ ] ] where @xmath222(\sigma_2/\sigma_1)(r_c)$ ] . in our considerations with the gaussian smoothing scale @xmath0 and @xmath223 , we have @xmath224\approx 0.25 $ ] and @xmath225 , and the corresponding @xmath226 and @xmath227 for the isothermal and nfw cases , respectively .
we then expect a wider distribution of the offset for the isothermal case than that of the nfw case .
figure 8 shows the results of @xmath228 .
the symbols are the simulation results with diamond and plus for the isothermal and nfw cases , respectively .
the solid and the dash - dotted lines are the corresponding results calculated from eq .
( 34 ) . because we use the linear expansion in @xmath220 in eq .
( 34 ) , we expect some inaccuracies of eq .
( 34 ) when @xmath220 is relatively large . on the other hand
, we see that eq .
( 34 ) still gives a reasonable description about the offset distribution , especially in terms of the location of the maximum probability .
for the isothermal case , the offset distribution is peaked around @xmath229 , and for the nfw case , it is peaked at @xmath230 .
the relatively flat density distribution of the cored isothermal cluster can therefore lead to an offset of order of @xmath1 for the main - cluster - peak identified from weak - lensing analyses .
this large offset may have significant effects on the cluster mass and the mass profile estimations from weak - lensing observations .
for the nfw cluster , the offset is smaller .
the above analyses are mainly done for the two clusters shown in figure 2 , in which , the nfw cluster has the same central @xmath39 and the same mass within @xmath231 as those of the isothermal cluster .
this leads to @xmath138 , or a concentration parameter @xmath232 for the nfw cluster . to further demonstrate the dependence of the peak statistics on the density profile of a cluster , we also carry out detailed analyses for two more nfw clusters with different concentrations and peak heights
the parameters are @xmath233 , and the central @xmath234 for one cluster , and @xmath235 , and the central @xmath236 for the other .
these parameters are chosen so that the clusters have about the same mass as the nfw cluster with @xmath138 considered above . in figure 9
, we show the results for the three nfw clusters .
figure 9a presents the profiles of @xmath130 .
figure 9b and figure 9c are for the height distributions and spatial offset distributions of the main - cluster peak . in each panel , the solid , dashed , and dash - dotted lines are for the cases with @xmath235 , @xmath233 and @xmath138 , respectively .
the histograms in figure 9b and symbols in figure 9c are for the results from monte carlo simulations .
the vertical lines in figure 9b indicate the true cluster peak height ( dotted ) and the shifted peak height corresponding to the maximum of the distribution ( dash - dotted ) in each case . the dependence of the results on the cluster profile is clearly seen . with
the same mass , the more centrally concentrated the cluster density distribution is , the less the main - cluster - peak properties are affected by noise .
the peak height shift due to noise ( figure 9b ) is @xmath237 , @xmath238 , and @xmath152 for the @xmath235 , @xmath233 and @xmath138 , respectively .
the corresponding spatial offset distribution is peaked at @xmath239 , @xmath240 and @xmath241 for the three cases .
it is known from eq .
( 31 ) and ( 34 ) that @xmath242 and @xmath216 are mainly determined by the central value of @xmath243 of a cluster .
a more centrally concentrated density profile leads to a larger value of @xmath243 , and thus smaller @xmath242 and @xmath216 .
it is possible to incorporate our model on the height and the position of the main - cluster peak [ eq . ( 31 ) and eq .
( 34 ) ] into observational weak - lensing analyses of cluster density profiles so that the noise effects can be properly taken into account .
the detailed implementations will be explored in our future studies .
weak - lensing cluster search and subsequently statistical studies are among the important cosmological applications of weak - lensing observations .
it is seen from our discussions in 3.1 that noise due to intrinsic ellipticities of background galaxies results false peaks in the weak - lensing convergence map and the noise peak distributions near clusters of galaxies are boosted depending on the density profile of the clusters .
it is also shown that the peak heights corresponding to clusters of galaxies are affected by the existence of noise that generates large scatters as well as systematic shifts .
these effects can contaminate the cosmological applications of weak - lensing cluster statistics considerably if they are not taken into account carefully . in this subsection , we present a model to calculate the number distribution of peaks @xmath193 over a large area , where @xmath193 includes both true peaks corresponding to clusters of galaxies and the false peaks from intrinsic ellipticities of background galaxies .
it should be noted that the false peak distribution near clusters of galaxies depends on the density profile of clusters , and thus also carries important cosmological information .
the model then allows us to use directly the peaks detected in the large - scale reconstructed convergence map from weak - lensing observations as cosmological probes without the need to differentiate true or false peaks with follow - up observations . in our model ,
the surface number density of convergence peaks is written as @xmath244 here @xmath245 calculates peaks within virial radii @xmath246 of dark matter halos , which includes the peaks corresponding to halos themselves taking into account the modified peak heights due to noise , and the noise peaks with enhanced number distributions near clusters .
the term @xmath247 is for the noise peaks in the field area away from dark matter halos .
specifically , we have @xmath248 where @xmath249 is the cosmological volume element at redshift @xmath250 , @xmath251 is the solid angle element , @xmath252 is the mass function of dark matter halos given the mass @xmath253 and redshift @xmath250 , and @xmath254 with @xmath255 given by eq . ( 15 ) . the quantities related to the density profile of dark matter halos @xmath256 , and @xmath82 in eq . ( 15 ) are computed assuming a nfw profile for a halo with its characteristic density @xmath123 and scale @xmath124 ( or the concentration @xmath257 ) determined by @xmath253 and @xmath250 ( navarro , frenk & white 1996 ) . for @xmath247 , it can be written as @xmath258\bigg \},\ ] ] where @xmath259 is the differential surface number density of noise peaks without foreground clusters .
we should emphasize that eq .
( 35 ) does not correspond to a simple and direct summation of the number of peaks corresponding to true dark matter halos and the peaks from pure noise field .
rather , the two terms in the right - hand side of eq . ( 35 ) represent peaks from different environments .
the considered area ( @xmath260 here ) is divided into regions occupied by dark matter halos and the field region away from halos .
the first term counts for total number of peaks in the regions occupied by dark matter halos , which includes both the peaks of the halos themselves and the noise peaks within the virial radii of the halos .
the halos are assumed to be randomly distributed spatially with their average abundance given by the mass function @xmath252 .
the peak abundance within the virial radius of a dark matter halo is given by @xmath11 calculated through eq .
( 37 ) and eq .
( 15 ) , which takes into account the effect of the halo mass distribution on noise peak statistics and the effect of noise on the peak height of the halo itself . to a certain extent ,
the treatment of the first term is similar to the model of halo occupation distribution for galaxies , where the peaks corresponding to halos can be analogous to central galaxies and the noise peaks can be analogous to satellite galaxies .
the second term @xmath261 calculates the number of peaks in the field region away from dark matter halos , which is the product of of surface number density of pure noise peaks and the spatial area not occupied by dark matter halos [ cf .
( 38 ) ] .
it should also be noted that to apply eq .
( 35 ) , we implicitly assume that the true convergence peaks corresponding to real mass concentrations are linked to individual dark matter halos .
this is a very reasonable assumption concerning peaks with @xmath262 . on the other hand , for lower peaks
, they can arise from the projection effects of large - scale structures without being associated with certain primary halos . to analyze these low peaks
, it may be more appropriate to model the large - scale density field as a gaussian random field @xmath263 .
then the noise effect on the statistics of these low peaks can be investigated by analyzing peaks in the total field @xmath264 , which is also a gaussian random field .
in fact , it is possible to separate the mass distribution in the universe into highly nonlinear part @xmath39 which can be modeled by dark matter halos and the part associated with large - scale structures that can approximately be treated as a gaussian random field @xmath263 . by studying the peaks in the total field @xmath265
, one can possibly obtain the peak statistics over a wide range of peak height , from low to high .
we will pursue along this line in our future investigations . in this paper
, we primarily concern high peaks with @xmath266 , and thus eq .
( 35 ) is used in our analyses . to validate our modeling of eq .
( 35 ) , we also analyze the convergence peak statistics from the ray tracing simulations by white & vale ( 2004 ) .
the simulations we use are for the flat @xmath267cdm model with @xmath268 , and @xmath269 , where @xmath270 and @xmath37 are the dimensionless matter density of the universe , the present hubble constant in units of @xmath271 , the rms of the extrapolated linear density fluctuation over @xmath272 and the power index of the initial power spectrum of density fluctuations , respectively . in total , 16 simulated convergence maps , each with @xmath273 pixels corresponding to a field of view of @xmath274 , are used in our analyses .
the source redshift is @xmath275 .
the details of the simulations can be found in white & vale ( 2004 ) .
for each simulated convergence map , we add in noise from intrinsic ellipticities following hamana et al .
specifically , a random gaussian noise is added to each pixel with the variance given by @xmath276 where @xmath277 is the rms of the intrinsic ellipticity of source galaxies , @xmath29 is the surface number density of source galaxies , and @xmath278 is the pixel size of the simulated map .
we take @xmath279 , @xmath280 , and @xmath281 .
we then apply a gaussian smoothing to each noisy convergence map . here
we use the smoothing scale @xmath282 , which is presumably the most optimal smoothing scale in weak - lensing cluster detections ( e.g. , hamana et al .
2004 ) . in our model calculations
, we use the same set of parameters as described above . the sheth - tormen mass function ( 1999 ) and the nfw density profile for dark matter halos
are applied . in figure 10
, we show the surface number density @xmath283 from our theoretical calculations ( solid line ) in comparison with those from numerical simulations ( upper histogram ) .
the simulation results are the results averaged over @xmath284 of @xmath285 convergence maps , i.e. , we count the peaks over the total area of @xmath286 , and then divide the total peak number by @xmath287 to obtain the average surface number density of the peaks .
we see that our model calculations agree with the simulation results very well . in the plot
, we also show @xmath193 without considering the noise ( dotted line ) , which is given by @xmath288,\ ] ] where @xmath289 $ ] is a @xmath290 function representing the fact that without noise , a dark matter halo with given @xmath253 and @xmath250 has a determined peak height .
the lower histograms are for the results from numerical simulations without adding noise in the convergence maps .
it is seen that this set of histograms is systematically higher than that of the dotted line , indicating significant projection effects from large - scale structures . furthermore , to demonstrate the effects of noise on the main - cluster - peak heights , we also show ( long - dashed line ) the results of @xmath193 for peaks associated with true dark matter halos but taking into account the change of the main - cluster - peak heights due to noise ( cf .
figure 7 and figure 9b ) .
this is given by @xmath291 where @xmath292 is given by eq .
( 31 ) with @xmath293 , and @xmath82 determined by @xmath253 and @xmath250 . comparing the long - dashed line with the dotted line
, we see that at @xmath262 , the number of peaks @xmath193 associated with real dark matter halos increases by a factor of @xmath294 .
this is clearly due to the positive shift of the peak height of a dark matter halo from the effects of noise ( see figure 7 and figure 9b ) , which boosts certain peaks of relatively small dark matter halos initially with @xmath295 in the case without noise to @xmath262 with the presence of noise .
this is highly relevant to cosmological studies with the number of true weak - lensing peaks , i.e. , confirmed peaks by follow - up observations . without properly taking into account the noise effects on the peak heights , eq .
( 40 ) would normally be used to predict the abundance of weak - lensing peaks above a certain threshold . by fitting such a prediction with the observed abundance of true weak - lensing peaks to constrain cosmological parameters ,
we would obtain a set of biased cosmological parameters .
the detailed analyses of such bias effects on cosmological parameters will be presented in our forthcoming paper .
it should also be pointed out that the differences between the long - dashed and solid lines represent the number of noise peaks .
these differences are much larger than those calculated from @xmath259 without considering the enhancement of noise peaks near dark matter halos , indicating the importance in taking into account such an enhancement in peak number calculations . the results shown in figure 10 are directly comparable with the ones shown in figure b1 of hamana et al .
( 2004 ) . by analyzing the results from numerical simulations , they present a fitting correction to the number of peaks taking into account the noise effects ( the solid line in their figure b1 ) . while in qualitative agreements , our results match the simulation results better .
more importantly , our model [ eq .
( 35)-(38 ) ] is based on statistical analyses of peaks near cluster of galaxies , but not a fitting model to the simulations .
the excellent agreement between our model prediction ( the solid line in figure 10 ) and that from simulations ( the upper histogram ) shows that our model can indeed give a reliable prediction about the number of convergence peaks in weak - lensing analyses .
thus we can use directly the distribution of the total number of peaks , true and false ones , to probe cosmologies .
this advantage would greatly strengthen the cosmological applications of weak - lensing peak statistics .
in this paper , we analyze the peak statistics in weak - lensing convergence maps taking into account the mutual influences between the mass distribution of clusters of galaxies and noise from intrinsic ellipticities of background galaxies .
concerning the main convergence peak corresponding to a cluster , the existence of noise affects both its height and its location , resulting scatters as well as a systematically positive shift for its height and an offset distribution for its position .
the effects are sensitive to the density profile of the cluster itself . in the two considered cases of the cored isothermal and the nfw clusters with the same mass and the same central @xmath39
, the flat density distribution of the isothermal cluster leads to a relatively large shift of the peak height with @xmath16 and a wide offset distribution of the peak location with @xmath216 extending to the core radius @xmath1 . for the nfw cluster ,
its inner density profile changes more rapidly and has a relatively large @xmath243 at the center .
this leads to smaller @xmath242 and @xmath216 , with @xmath200 and @xmath216 peaked at @xmath296 .
we further analyze different nfw clusters of the same mass but with different concentrations .
it is shown that the weak - lensing convergence peak signal of a dark matter halo is affected less as its concentration gets higher .
we have @xmath297 and @xmath152 , and @xmath298 and @xmath241 , for @xmath235 , @xmath233 and @xmath138 , respectively . for noise peaks ,
their occurrence probability is boosted near cluster of galaxies . in the isothermal case ,
there are on average about @xmath155 peaks with @xmath299 ( note that we have @xmath6 for the considered cluster without noise ) in the central region with @xmath2 . while the higher one is usually defined to be associated with the cluster itself , the other is a noise peak without any related real structures .
this noise peak can be mistakenly identified as a large substructure and therefore can complicate the interpretation of the cluster mass distribution and further the formation of clusters .
for the nfw cluster with @xmath138 , the central region is cleaner than that of the isothermal case , and contains on average @xmath8 peaks with @xmath299 . in outside regions around a cluster with @xmath300 ,
high noise peaks can be confused with @xmath45dark clumps@xmath45 , and thus lead to incorrect conclusions regarding the structure formation .
we find that depending on the density profile of a cluster , the average number of high noise peaks in the regions @xmath301 can be significantly enhanced .
this is locally due to the quantity @xmath302 mainly .
but it is important to realize that the @xmath144-dependence , i.e. , the profile of @xmath302 , plays key roles resulting a net overall enhancement in the considered outside region , especially when the mass - sheet degeneracy is taken into consideration . for the cored isothermal cluster , the enhancement factor @xmath11 ranges from @xmath12 to @xmath13 for the number of peaks with @xmath10 , depending on the treatment of the mass - sheet degeneracy . for @xmath4 ,
@xmath11 is in the range of @xmath303 . for the nfw cluster ,
the enhancement is weaker than that of the isothermal cluster for high peaks , with @xmath304 for @xmath10 , and @xmath305 for @xmath4 .
this enhancement is important in analyzing the statistical significance of @xmath45dark clump@xmath45 occurrence near clusters of galaxies .
taking into account the effects of noise on the main - cluster - peak heights and the enhancement of the number of noise peaks near dark matter halos , we propose a model incorporating the mass function of dark matter halos to calculate the statistical abundance of convergence peaks over large scales .
the model prediction is in good agreement with that from numerical simulations .
it should be pointed out that because of the mutual effects of the mass distribution of dark matter halos and noise , the noise peak abundance also carries important cosmological information , especially the information related to the density profile of dark matter halos .
our model then provides a theoretical framework to use statistically the total number of convergence peaks , including the false ones from noise , as cosmological probes .
this facilitates the cosmological applications of weak - lensing peak statistics in saving the expensive follow - up observations to distinguish true and false peaks .
furthermore , the useful cosmological information contained in the noise peak distribution is also included . concerning high peaks with @xmath306
, our model can predict the total convergence peaks very well because most of the high peaks are associated with massive virialized dark matter halos ( and the noise peaks around them ) .
for relatively low peaks , the projection effects of large - scale structures can be important , which can change the peak height of clusters as well as produce their own convergence peaks without being associated with virialized clusters ( e.g. , tang & fan 2005 ; marian et al .
2009 , 2010 ; maturi et al .
2009 ) . to a certain extent
, the total weak - lensing convergence can be written as @xmath307 , where @xmath39 and @xmath40 are the respective contributions from virialized dark matter halos and the noise as discussed in this paper , and @xmath263 denotes the large - scale structure contribution beyond dark matter halos .
on very large scales where density perturbations are approximately linear , the effects of large - scale structures can be described by a gaussian random field reasonably well ( maturi et al .
2009 ) . under this gaussian assumption , our current model on peak distribution near clusters of galaxies can be extended to include the effects of large - scale structures by analyzing the sum of the two gaussian random fields @xmath308 , which is also gaussian , while treating the cluster @xmath39 as a background
. then eq .
( 35 ) to eq .
( 38 ) can be used to study large scale peak statistics , in which the gaussian field @xmath40 is replaced with the total gaussian field @xmath308 .
it should be pointed out that physical correlations between @xmath39 and @xmath263 can exist , which may complicate the analyses considerably .
the above approach essentially divides the structures into two parts , virialized dark matter halos , and linear large - scale perturbations described by a gaussian random field . on scales in between where nonlinear effects are significant and the gaussian approximation of @xmath263 is not appropriate , the modeling on @xmath263 deserves careful investigations .
a possible way to study them is to extend the mass function of dark matter halos to include such nonlinear but non - virialized structures .
detailed investigations on the projection effects of large - scale structures , both linear and nonlinear , will be performed as one of our main future tasks .
weak lensing effects hold great potentials in cosmological studies . with thorough investigations on different systematics
, we expect that future weak - lensing observations would bring us a greatly improved understanding about the dark side of the universe .
we sincerely thank the referee for the comments and suggestions , which help to improve the paper considerably .
this research is supported in part by the nsfc of china under grants 10373001 , 10533010 and 10773001 , and the 973 program no.2007cb815401 .
zuhui fan is very grateful for the hospitality of the institute of astronomy , university of cambridge where part of the paper is done .
amara , a. , & refregier , a. 2008 , , 391 , 228 albrecht , a. et al . , 2006 , astro - ph/0609591 bardeen , j. m. , bond , j. r. , kaiser , n. , & szalay , a. s. 1986 , , 304 , 15 bartelmann , m. , & schneider , p. 2001 , physics reports , 340 , 291 benjamin , j. et al .
2007 , , 381 , 702 bond , j. r , & efstathiou , g. 1987 , , 226 , 655 bridle , s. , & king , l. j. 2007 , new j. phys . , 9 , 444 clowe , d. , bradac , m. , gonzalez , a. h. , markevitch , m. , randall , s. w. , jones , c. , & zaritsky , d. 2006 , apjl , 648 , 109 corless , v. l. , king , l. j. , & clowe , d. 2009 , , 393 , 1235 fan , z. h. 2007 , , 669 , 10 fu , l. et al .
2008 , a&a , 479 , 9 gavazzi , r. , & soucail , g. 2007 , a & a , 462 , 259 hamana , t. , takada , m. , & yoshida , n. 2004 , , 350 , 893 heymans , c. et al . 2006 , , 368 , 1323 hirata , c. m. , & seljak , u. 2004 , phys .
d. , 70 , 063526 huterer , d. , takada , m. , bernstein , g. , & jain , b. 2006 , , 366 , 101 joachimi , b. , & schneider , p. 2009 , a & a , 507 , 105 kaiser , n. , & squires , g. 1993 , , 404 441 kilbinger , m. et al .
2008 , a&a , 497 , 677 kratochivil , j. m. , haiman , z. , & may , m. 2010 , phys .
d , 81 , 043519 li , h. , liu , j. , xia , j. q. , sun , l. , fan , z. h. , tao , c. , tilquin , a. , & zhang , x. m. 2009 , phys .
b , 675 , 164 li , r. , mo , h. j. , fan , z. h. , cacciato , m. , van den bosch , f. c. , yang , x. h. , & more , s. 2009 , , 394 , 1016 ma , z. , hu , w. , & huterer , d. 2006 , , 636 , 21 mandelbaum , r. , hirata , c. m. , ishak , m. , seljak , u. , & brinkmann , j. 2006 , , 367 , 611 mandelbaum , r. , hirata , c. m. , & seljak , u. 2008 , jcap , 8 , 6 marian , l. , smith , r. e. , & bernstein , g. m. 2009 , , 698l , 33 marian , l. , smith , r. e. , & bernstein , g. m. 2010 , , 709 , 286 massey , r. et al .
2007 , nature , 445 , 286 maturi , m. , angrick , c. , bartelmann , m. , & pace , f. 2009 , arxiv : 0907.1849 mo , h.j . , & white , s. d. m. 1996 , , 282 , 347 navarro , j. , frenk , c. & white , s. d. m. 1996 , , 462 , 563 schirmer , m. , erben , t. , hetterscheidt , m. , & schneider , p. 2007
, a & a , 462 , 875 schneider , p. 1996
, , 283 , 837 sheth , r. k. , & tormen , g. 1999 , , 308 , 119 sun , l. , fan , z. h. , tao , c. , kneib , j .-
jouvel , s. , & tilquin , a. 2009 , , 699 , 958 tang , j. y. & fan , z. h. 2005 , 635 , 60 umetsu , k. , & futamase , t. 2000 , , l5 van waerbeke , l. 2000 , , 313 , 524 wang , s. & haiman , z. , & may , m. 2009 , , 691 , 547 white , m. , van waerbeke , l. , & mackey , j. 2002 , , 575 , 640 white , m. , & vale , c. 2004 , astroparticle physics , 22 , 19 zhang , p. j. 2008 , arxiv : 0811.0613 | taking into account noise from intrinsic ellipticities of source galaxies , in this paper , we study the peak statistics in weak - lensing convergence maps around clusters of galaxies and beyond .
we emphasize how the noise peak statistics is affected by the density distribution of nearby clusters , and also how cluster - peak signals are changed by the existence of noise .
these are the important aspects to be understood thoroughly in weak - lensing analyses for individual clusters as well as in cosmological applications of weak - lensing cluster statistics .
we adopt gaussian smoothing with the smoothing scale @xmath0 in our analyses .
it is found that the noise peak distribution near a cluster of galaxies depends sensitively on the density profile of the cluster . for a cored isothermal cluster with the core radius @xmath1
, the inner region with @xmath2 appears noisy containing on average @xmath3 peaks with @xmath4 for @xmath5 and the true peak height of the cluster @xmath6 , where @xmath7 denotes the convergence signal to noise ratio . for a nfw cluster of the same mass and the same central @xmath7 ,
the average number of peaks with @xmath4 within @xmath2 is @xmath8 .
thus a high peak corresponding to the main cluster can be identified more cleanly in the nfw case . in the outer region with @xmath9 , the number of high noise peaks is considerably enhanced in comparison with that of the pure noise case without the nearby cluster . for @xmath10 , depending on the treatment of the mass - sheet degeneracy in weak - lensing analyses , the enhancement factor
@xmath11 is in the range of @xmath12 to @xmath13 for both clusters as their outer density profiles are similar .
the properties of the main - cluster - peak identified in convergence maps are also affected significantly by the presence of noise .
scatters as well as a systematic shift for the peak height show up .
the height distribution is peaked at @xmath14 , rather than at @xmath15 , corresponding to a shift of @xmath16 , for the isothermal cluster . for the nfw cluster , @xmath17 .
the existence of noise also causes a location offset for the weak - lensing identified main - cluster - peak with respect to the true center of the cluster .
the offset distribution is very broad and extends to @xmath18 for the isothermal case . for the nfw cluster
, it is relatively narrow and peaked at @xmath19 .
we also analyze nfw clusters of different concentrations .
it is found that the more centrally concentrated the mass distribution of a cluster is , the less its weak - lensing signal is affected by noise . incorporating these important effects and the mass function of nfw dark matter halos , we further present a model calculating the statistical abundances of total convergence peaks , true and false ones , over a large field beyond individual clusters .
the results are in good agreement with those from numerical simulations .
the model then allows us to probe cosmologies with the convergence peaks directly without the need of expensive follow - up observations to differentiate true and false peaks . |
in recent papers the tilted bianchi models @xcite have been studied using dynamical systems methods @xcite . in this paper
we will study the general tilted was studied in @xcite . ] perfect fluid bianchi model of type vi@xmath0 , with a constant linear equation of state parameter @xmath1 satisfying the causality conditions ( i.e. , no superluminal speed of sound ) , which was not studied in detail in @xcite .
the tilted models have shown a wide variety of phenomena including , for example , the existence of closed curves @xcite and incompleteness of the fluid congruence @xcite .
the general irrotational perfect fluid type vi@xmath0 model is the most general of the irrotational bianchi models .
in fact , the irrotational type vi@xmath0 model has a 6-dimensional state space ; just one dimension less than that of of the most general tilted type vi@xmath0 model .
we will show that , regardless of whether we consider the general type vi@xmath0 model with or without vorticity , the important late time asymptotes are the non - tilted collins type vi@xmath0 solution ( for @xmath2 ) , @xmath3 where @xmath4 , or the collinson - french ( or robinson - trautman ) vacuum solution ( for @xmath5 ) , @xmath6 the tilt of the fluid can be asymptotically non - tilted , intermediate or extreme depending on the value of @xmath1 .
the collinson - french solution is a petrov type iii vacuum solution , and hence it has the peculiar feature that all the curvature invariants of zeroth order vanish . ] .
the spatial hypersurfaces in a bianchi type vi@xmath0 cosmology are defined as the orbit of the action of the bianchi type vi@xmath0 lie group .
this lie group can be equipped with the left - invariant one - forms : @xmath7 where @xmath8 is a constant .
the exceptional feature of this model does not relate to the group itself ( for which there is nothing extraordinary ) but arises from the einstein field equations . for this particular model one of the einstein constraint equations vanishes exactly and hence the vacuum case allows for an extra shear degree of freedom .
this extra shear is present in the collinson - french vacuum solution and therefore this metric has no analogue in the other bianchi vacuum cosmologies .
the aim of this paper is to fill one of the gaps in the analysis of the general tilted type vi@xmath9 models presented in @xcite .
the @xmath10 case is special ( as explained above ) and requires a separate analysis , which has only been done in part previously . in particular
, we will show that closed periodic orbits exist for the general vortic model .
however , for the irrotational models , no such closed curve exists .
moreover , we will also complete a centre manifold analysis appearing in one 6-dimensional invariant subspace describing vortic bianchi type vi@xmath0 models .
all other attractors will be included for completeness .
the line - element of a bianchi cosmology can be written s^2=-t^2+_ab^a^b , where @xmath11 is the co - moving cosmological time .
the one - forms @xmath12 are left - invariant one - forms on the hypersurfaces spanned by the group orbits and can be related to the left - invariant one - forms , @xmath13 , given above by @xmath14 .
the geometric ( or normal ) congruence , @xmath15 , is given by @xmath16 .
it is also useful to define and the hubble scalar and the shear associated with the congruence @xmath17 : h13n^ _ ; , _ n_;-hh_. the matter variables are chosen to be the energy density , @xmath18 , and the tilt - velocity , @xmath19 , which is defined as the 3-velocity of the fluid with respect to the geometric ( or normal ) congruence , @xmath15 .
the equations of motion can now be written down in terms of the hubble scalar , @xmath20 the shear , @xmath21 ; the curvature variables @xmath22 and @xmath23 ; and the matter variables @xmath18 and @xmath19 .
expansion - normalised variables are introduced ( by dividing the variables with the appropriate powers of @xmath24 ) .
the papers @xcite contain all the details regarding the determination of the evolution equations for the tilted cosmological models under consideration . here
, we shall adopt the so - called @xmath25-gauge in which the function @xmath26 is purely imaginary ; this is realised by the choice @xmath27 , where @xmath28 is defined by @xmath29 . the evolution equation for @xmath30 can then be replaced by an evolution equation for @xmath28 , which ensures a closed system of equations .
we will also adopt the dimensionless time parameter @xmath31 , which is related to the cosmological time @xmath11 via @xmath32 , where @xmath24 is the hubble scalar . using expansion - normalised variables , the equations of motion are ( see @xcite for the complete derivation of the equations ) : _
+ &= & ( q-2)_++3(_12 ^ 2+^2_13)-2n^2 + ( -2v_1 ^ 2+v_2 ^ 2+v_3 ^ 2 ) + _ -&=&(q-2 - 2_23)_-+(_12 ^ 2-_13 ^
2 ) + 2an + ( v_2 ^ 2-v_3 ^ 2 ) + _12&= & ( q-2 - 3_+-_-)_12 -(_23+_-)_13 + v_1v_2 + _13&=&(q-2 - 3_++_-)_13-(_23-_-)_12+v_1v_3 + _23&=&(q-2)_23 - 2n^2 + 2_-^2 + 2_12_13 + v_2v_3 + n&= & ( q+2_++2_23)n + & = & 2_23(1-^2 ) + a&= & ( q+2_+)a .
the equations for the fluid are : &= & \{2q-(3 - 2)+2av_1 + v^2 } + v_1 & = & ( t+2_+)v_1 - 2_13v_3 - 2_12v_2-a(v_2 ^ 2+v_3 ^ 2)-n(v_2 ^ 2-v_3 ^ 2 ) + v_2&= & ( t-_+-_-)v_2-(_23+_-)v_3+v_1v_3+(a+n)v_1v_2 + v_3&= & ( t-_++_-)v_3-(_23-_-)v_2-v_1v_2+(a - n)v_1v_3 + v&= & , where q&= & 2 ^ 2 + 12 + ^2 & = & _ + ^2+_-^2+_12 ^ 2 + _ 13 ^ 2+_23 ^ 2 + & = & _ abc^ac^b , c^ac_a=1 , v^a = vc^a , + v^2 & = & v_1 ^ 2+v_2 ^ 2+v_3 ^ 2 , + t&= & + g_+&=&1+(-1)v^2 .
these variables are subject to the constraints 1&= & ^2+a^2+n^2+[const : h ] + 0 & = & 2_+a+2_-n+ [ const : v1 ] + 0 & = & -+ [ const : v2 ] + 0 & = & + [ const : v3 ] + 0&= & 3a^2-(1-^2)n^2.[const : group ] the parameter @xmath1 will be assumed to satisfy @xmath33 .
the generalized friedmann equation ( [ const : h ] ) yields an expression which effectively defines the energy density @xmath34 .
we will assume that this energy density is non - negative : @xmath35 .
therefore , the state vector can be considered @xmath36 $ ] modulo the constraint equations ( [ const : v1])-([const : group ] ) .
thus the dimension of the physical state space is seven ( and hence of equal generality to the other models of type vi@xmath9 for a given value of @xmath37 ) .
additional details are presented in @xcite .
the dynamical system is invariant under the following discrete symmetries : @xmath38 [ cols= " < " , ] + the various invariant subspaces can also be categorised in terms of the ( @xmath39-normalised , where @xmath40 ) fluid vorticity , @xmath41 .
the vorticity of the fluid for the type vi@xmath0 models is given by : w_a=(n_ab+_abca^c)v^b , w_0=0 , [ eq : vorticity]where @xmath42\sqrt{1-v^2}}.\ ] ] for the invariant sets : 1 .
@xmath43 : @xmath44 , most general vortic type vi@xmath0 .
@xmath45 : @xmath44 .
@xmath46 : @xmath47 , non - vortic .
@xmath48 : @xmath49 .
@xmath50 : @xmath47 , non - vortic .
@xmath51 : @xmath47 , non - vortic
@xmath52 : @xmath47 , non - tilted and non - vortic .
we note that the most general non - vortic model is of dimension 6 . hence , since the non - vortic type vi@xmath9 and vii@xmath9 models regarding @xmath37 as fixed are of dimension 5 , the type vi@xmath0 model is the most general non - vortic model of all bianchi models .
in general we can also solve for the vorticity component @xmath53 : @xmath54 this follows from the constraint equations and eq.([eq : vorticity ] ) .
there are a number of monotone functions in the state space of interest . for @xmath55
, there exists a monotonically increasing function @xmath56 defined by z_1 & & ^1- , = , + z_1&=&z_1 .
this function can be used to show @xcite : [ thm : nontilted ] for @xmath55 , all tilted bianchi models ( with @xmath57 , @xmath58 ) of solvable type are asymptotically non - tilted at late times . for @xmath57 , @xmath58 , and
@xmath59 we have that @xmath60 [ no - hair ] moreover , the following function is a monotone function in @xmath43 : z_2&= & , + z_2&=&(5 - 6)(3 - 2av_1)z_2 , where @xmath61 is defined in eq.([dhat ] ) .
this function is monotonically decreasing for @xmath62 and monotonically increasing for @xmath63 .
we note that in the subspace @xmath46 we have the monotone function : z_3&= & , + z_3&=&-(2-)(3 - 2av_1)z_3 .
this function is monotonically decreasing in @xmath46 .
the monotonic function @xmath64 immediately implies : for @xmath65 , @xmath57 , @xmath66 , @xmath67 , @xmath68 we have that :
@xmath69 this implies that all irrotational type vi@xmath0 universes are either asymptotically vacuum or non - tilted at late times .
1 . @xmath71 : @xmath72 and @xmath73 . here
, @xmath74 and is an unphysical parameter .
this represents the flat friedman - lematre model .
the remaining equilibrium points are all in @xmath75 .
all of the tilted equilibrium points come in pairs . these represent identical solutions ( they differ by a frame rotation ) ; however , since their embeddings in the full state space are inequivalent , two of their eigenvalues are different .
all of these equilibrium points have an unstable direction with eigenvalue @xmath77 corresponding to the variable @xmath78 .
these equilibrium points are given in @xcite .
1 . @xmath79 : collins perfect fluid solution , @xmath80 + @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath85 , @xmath86 .
this equilibrium point is in @xmath52 .
@xmath87 : the apostolopoulos @xmath10 solution @xcite , @xmath88 + this is a vortic solution lying in the invariant subspace @xmath89 .
the solution is given in terms of the expansion - normalised variables in @xcite with @xmath10 , @xmath90 .
bianchi type vi@xmath0 vacuum plane waves .
all of these solutions have @xmath91 it is avantageous to introduce @xmath92 , which implies that we can write @xmath93 we will also define @xmath94 by @xmath95 + the equilibrium points are then determined by the tilt velocities : 1 .
@xmath96 : @xmath97 .
these represent non - tilted plane waves and lie in @xmath52 .
@xmath98 : @xmath99 , @xmath68 , @xmath100 .
these represent intermediately tilted plane waves and lie in @xmath51 .
@xmath101 : @xmath102 , @xmath68 .
these represent extremely tilted plane waves and lie in @xmath51 .
[ deff ] @xmath103 : here , @xmath104\leq \gamma\leq 3(3+r^2)/(9-r^2)$ ] and v_1&=&-,v_2 ^ 2-v_3 ^ 2= r + & = & these represent intermediately tilted plane waves and lie in @xmath45 ( for @xmath105 they lie in @xmath48 ) . 5 .
@xmath106 , @xmath107 : v_1&=&-,v_2 ^ 2-v_3 ^ 2=r + & = & 1-v_1 ^ 2= these represent extremely tilted plane waves and lie in @xmath45 .
for @xmath105 these equilibrium points lie in @xmath48 , and due to the special importance for the late - time behaviour , we will denote @xmath108 .
therefore , @xmath109 4 .
@xmath110 : the collinson - french ( robinson - trautmann ) solution is given by : & & _ + = -13,_-= , _
13=,n= , a= , + & & _ 12=_23===0 .
there are the following equilibrium points associated with the collinson - french solution : 1 .
@xmath111 : @xmath112 , @xmath107 .
2 . @xmath113 : @xmath114 , @xmath115 , @xmath116 .
@xmath117 : @xmath118 , @xmath119 , @xmath120 .
@xmath121 : @xmath102 , @xmath119 , @xmath107 .
+ these equilibrium points , and their stability , were studied in @xcite .
@xmath122 : wainwright @xmath123 solution : & & _ + = -13,_-=,0<_13<,n= , a = n,=59 - 3_13 ^ 2 , + & & _ 12=_23==v=0 .
the late - time behaviour of models with @xmath59 is determined by ( the cosmic no - hair ) corollary 3.2 . here , we have the following late - time attractors : * @xmath124 : the collins solution , @xmath79 .
* @xmath88 : the apostolopoulos solution , @xmath87 . *
@xmath125 : `` extremely tilted '' vacuum plane waves , @xmath126 the stability of these points for @xmath127 follows from the eigenvalues of the linearised matrix . for @xmath125
several zero - eigenvalues occur and a centre manifold analysis is needed to determine the late - time behaviour .
let us present the centre manifold analysis of the equilibrium point @xmath126 in some detail .
the centre manifold in this case is a 2-dimensional submanifold of the 5-dimensional extremely tilted invariant subspace @xmath129 . to find the centre manifold
, we will therefore set @xmath130 .
let us choose variables ( _ + , _
23,n,,v_2)=(-34+x_1,x_2,+x_3,x_4,3+x_5 ) , let @xmath131 , @xmath132 , @xmath133 and @xmath34 be determined from the constraint equations , and @xmath134 .
let us define the column vector @xmath135^t$ ] .
we can now expand the equations of motion to 2nd order in @xmath136 : =ax+c(x , x)+(x^3 ) , where @xmath137 is a bilinear vector - valued function .
it is convenient to align the vector @xmath136 with the jordan canonical form of @xmath138 .
this can be accomplished by defining = 75 & -25 & 0 & 0 & 0 + 0 & 1 & 0 & 0 & 1 + 5 & -5 & 0 & 0 & 0 + 4 & -4 & 0 & 4 & -4 + - & & & 0 & 0 .
now , defining @xmath139 , the equation for @xmath140 becomes @xmath141 where @xmath142 is the jordan block matrix @xmath143 the centre manifold correspond to the zero - eigenvalues of @xmath142 .
we can therefore parameterise the centre manifold using the variables @xmath144 .
the next step is to expand the variables @xmath145 , @xmath146 , in terms of @xmath144 on the centre manifold , to second order .
we therefore define the quadratic forms @xmath147 , @xmath146 , such that @xmath148 is ( to second order ) an invariant subspace . on the centre manifold
we then have : y_i = y_i(y_1,y_4)+(y^3 ) . by substituting these into the @xmath149 and @xmath150 evolution equations ,
we finally get , on the centre manifold : y_1&=&-16y_1 ^ 2+(y^3 ) , + y_4&=&-16y_1y_4+(y^3 ) .
to lowest order , these equations can be solved to give @xmath151 , @xmath152 , where a constant of integration has been eliminated by a translation of time . note
that , in principle , there are essentially two kinds of behaviour : for @xmath153 the variables diverge , while for @xmath154 the variables decay . however
, requiring @xmath57 leaves only the decaying mode as physically acceptable .
these decay rates will therefore be the dominant ones since the centre manifold will dominate the behaviour at late times . in terms of the original variables
, we therefore get the decay rates in @xmath45 ( only dominant decay rates are included ) : _ + & & -34(1- ) , + _ - & & 4(1 + ) , + _ 12 & & , + _ 13 & & - , + _ 23 & & - , + n & & 4(1 + ) , + & & , + & & , + v_1 & & -3(1 + ) , + v_2 & & 3(1- ) , + & & ( ) _ 0 .
this confirms that the extremely tilted vacuum plane wave @xmath155 is the attractor for @xmath45 , @xmath156 .
note , however , that this solution is unstable in the fully tilted space @xmath43 , which can be verified by calculating , for example , @xmath157 close to @xmath155 . * @xmath2 : the collins solution , @xmath79 . * @xmath123 : the wainwright solution , @xmath122 . *
@xmath5 : the collinson - french solution , @xmath158 .
we have the following refinement for the asymptotic tilt ( which follows from an analysis of the eigenvalues ) : * * @xmath159 : the tilt is asymptotically zero ( @xmath160 ) . * * @xmath161 : the tilt is either asymptotically zero ( @xmath160 ) or extreme ( @xmath162 ) . * * @xmath163 : the tilt is either asymptotically intermediate ( @xmath164 ) or extreme ( @xmath162 ) . * * @xmath165 : the tilt is asymptotically extreme ( @xmath162 ) . for the fully tilted model , the late - time behaviour can be summarised as follows : * @xmath2 : the collins solution , @xmath79 .
* @xmath123 : the wainwright solution , @xmath122 .
* @xmath5 : the collinson - french solution , @xmath158 .
we have the following refinement for the asymptotic tilt : * * @xmath159 : the tilt is asymptotically zero ( @xmath160 ) . * * @xmath166 : the tilt is either asymptotically zero ( @xmath160 ) or extreme ( @xmath162 ) . *
* @xmath167 : the tilt is either asymptotically intermediate ( @xmath168 ) or extreme ( @xmath162 ) . * * @xmath169 : the tilt is either asymptotically oscillatory ( closed curve ) or extreme ( @xmath162 ) . *
* @xmath170 : the tilt is asymptotically extreme ( @xmath162 ) . most of these results comes from an analysis of the eigenvalues of the various equilibrium points .
however , for @xmath171 the equilibrium point @xmath168 undergoes a hopf - bifurcation and a stable closed curve results .
the analysis of this hopf - bifurcation is given below .
interestingly , this closed curve co - exists with an extremely tilted attractor .
our main aim for this section is to show the following .
there exists a @xmath172 such that for @xmath173 there exists a closed orbit , @xmath174 , acting as the future attractor for a set of non - zero measure of tilted bianchi type vi@xmath0 models . to prove this theorem
we will first show the existence of a closed period orbit acting as an attractor in a particular subset .
we consider the invariant subset given by the collinson - french solution with 2 tilts .
we introduce @xmath175 : x&=&(t-23)x - y , + y&=&2(t+3x)y .
[ 2dsys ] we set @xmath176 , @xmath177 , and perform the transformation @xmath178 , @xmath179 with respect to the equilibrium point @xmath168 . the normal form of a hopf - bifurcation can be written z=(+b|z|^2)z , where @xmath8 is some complex number and @xmath180 is a parameter . if @xmath181 for @xmath182 , there exists a stable closed orbit for @xmath183 sufficently small .
we will therefore set @xmath184 and expand to cubic terms in @xmath185 and @xmath186 .
it is also convenient to introduce a complex variable @xmath187 chosen such that it aligns with the jordan form of the linearised matrix .
this can be achieved by setting : @xmath188 where @xmath189 is a real number chosen such that the linear term of @xmath190 is _ zf(0,0)= , _
|zf(0,0)=0 . by a transformation @xmath191
we can choose the coefficients @xmath192 and @xmath193 so that the equation for @xmath194 takes the form z=(+b|z|^2)z + ( |z|^4 ) , where @xmath195 hence , there exists a @xmath172 such that there exists a closed stable orbit for @xmath173 .
the next step is to show that this orbit is also stable in the fully tilted type vi@xmath0 models .
we can show this as follows : for a function @xmath196 , we introduce _ the average _
, @xmath197 , with respect to the closed orbit @xmath174 , defined by @xmath198 we can use this average , using similar manipulations as for the closed curves in the type iv , vi@xmath9 and vii@xmath9 models @xcite to show that : assume that there exists a closed periodic orbit @xmath174 for the dynamical system ( [ 2dsys ] ) .
then @xmath199 from the @xmath200 equation , we get @xmath201 .
a manipulation of the @xmath202 equation yields @xmath203 , and @xmath204 .
these can now be solved to yield the desired value for @xmath205 .
a similar manipulation of the @xmath34 equation yields @xmath206 .
this implies that the vacuum solution is stable when the closed curve is perturbed by @xmath34 .
hence , since the collinson - french solution is stable with respect to vacuum perturbations , this closed curve is stable for the fully tilted type vi@xmath0 models whenever @xmath173 .
it remains to determine the maximal value for @xmath172 ?
it seems that this limiting value is related to the existence of a heteroclinic orbit originating and ending at the saddle points @xmath160 and @xmath164 , respectively . for the limiting value of @xmath1 , which we will call @xmath207
, there exists such a heteroclinic orbit , while for values @xmath208 no such heteroclinic orbit exists connecting these two equilibrium points .
we can use this to numerically estimate the value for @xmath207 .
our estimate gives : @xmath209 since @xmath210 this means that the region in which the closed orbit exists is extremely small .
models also appears to be extremely small . ]
note also that this implies that these closed orbits co - exist with the extremely tilted attractor @xmath162 , which makes this closed orbit even more difficult to detect .
in addition , it appears as if the curves asymptote to the closed curve relatively slowly , which implies that the numerics have to run for a relatively long time in order to see the late - time asymptote .
a numerical plot of some generally tilted type vi@xmath0 models approaching this closed orbit is shown in fig.[figcc ] .
@xmath211{sixesninthed2.eps}\ ] ]
in this paper we have examined the tilted bianchi type vi@xmath0 model in some detail .
this model is a special case ( due to the vanishing of one of the constraint equations ) and necessitates a separate analysis from the general type vi@xmath9 models .
we showed that in these models there exists a tiny region of parameter space where there exists a closed curve acting as an attractor .
this closed curve co - exists with an extremely tilted attractor . in the case of the most general irrotational models this closed curve , which appears in terms of the tilt velocities ,
is absent .
we have also confirmed the analytical results with an extensive numerical investigation .
coley and s. hervik , _ class .
* 22 * ( 2005 ) 579 .
s. hervik , r.j .
van den hoogen and a.a .
coley , _ class .
* 22 * ( 2005 ) 607 .
s. hervik , r.j .
van den hoogen , w.c .
lim and a.a .
coley , _ class .
* 23 * ( 2006 ) 845 .
s. hervik and w.c .
lim , _ class .
quantum grav . _
* 23 * ( 2006 ) 3017 .
coley , s. hervik and w.c .
lim , _ phys .
b _ * 638 * ( 2006 ) 310 - 313 ; a.a .
coley , s. hervik and w.c .
lim , _ class .
* 23 * ( 2006 ) 3573 - 3591 ; a.a .
coley , s. hervik and w.c .
lim , _ int .
phys . _ * d15 * ( 2006 ) 2187 - 2190 ; w.c .
lim , a.a .
coley and s. hervik , , _ class .
* 24 * ( 2007 ) 595 - 604 | we study tilted perfect fluid cosmological models with a constant equation of state parameter in spatially homogeneous models of bianchi type vi@xmath0 using dynamical systems methods and numerical simulations .
we study models with and without vorticity , with an emphasis on their future asymptotic evolution .
we show that for models with vorticity there exists , in a small region of parameter space , a closed curve acting as the attractor . |
biological and ecological research has investigated cell migration . to model cell migration , studies have been composed to include migration , diffusion , haptotaxis , and chemotaxis , @xcite . in this paper , the focus is chemotaxis .
chemotaxis describes the movement of an organism and/or groups of cells that either move toward or away from a chemical or sensory stimulus . in the early work by keller and segel @xcite , chemotactic responses of amoebae to bacteria
is studied in a cellular slime mold .
bacterial chemotaxis , which describes the ability of bacteria to move toward increased or decreased concentrations of attractants is analyzed at the macroscopic level through a microscopic model of individual cells , @xcite .
it was first observed by engelmann @xcite in 1881 .
for example , if _ salmonella typhimurium _ , a strain of salmonella associated with meat and poultry products , is introduced to a petri dish filled with a nutrient , the bacteria will migrate outward , consuming the nutrient .
as they consume the nutrient , they secrete a chemoattractant .
after several days , the bacteria will have clustered in the areas of high chemical concentration .
a structure of concentric rings is usually observed experimentally .
chet and mitchell s work @xcite describes patterns formed from _
e. coli _ movement toward amino acids .
allweis et .
al @xcite investigate _ vibrio cholerae _ which are inhibited by a pepsin digest that reduces the possibility of the vibrios attaching to the intestinal wall .
other authors @xcite have addressed chemotaxis in immune cell motility which when combined with tumor morphology is hoped to provide new avenues of treatment strategies .
in addition , authors have analyzed chemotactic responses in ecology @xcite and have investigated mathematical issues for the existence of global solutions in multiple dimensions , @xcite .
chemotaxis also arises in a variety of medical applications .
in particular , it has been studied in connection with myxobacteria @xcite , leukocyte mobility in tissue inflammation @xcite , the migration of tumor cells towards bone @xcite , and other issues in morphogenesis @xcite .
another interesting problem involves the study of vascular tumors through angiogenesis .
angiogenesis involves the formation of capillary networks of blood vessels that are vital for the growth of tumors .
mathematical modeling of angiogenesis @xcite has given new insight into tumor structure .
normal tissue , lymphocytes , and other types of cells grow at the tumor site or are recruited through chemotaxis .
the need to identify the nature of this recruitment is at the heart of this paper .
the identification of a chemotactic term falls under the umbrella of an inverse problem . in principle
, we can measure certain characteristics of the tumor concentration and use mathematical techniques to recover the chemotactic term , in particular the chemotactic sensitivity , that is driving the tumor growth . to our knowledge , this _ inverse problem _ approach has only been used in the analysis of chemotaxis models by dolak - stru and kgler @xcite under the assumption that the chemical concentration is explicitly known .
since there are many applications in which chemotaxis arises , there are also different models of the chemotactic effect .
there have been many different expressions proposed that model chemotactic velocity , see keller and segel @xcite , lapidus and schiller @xcite , ford and lauffenburger @xcite , and tyson et al .
. this velocity is used in a bacterial conservation equation in the formulation of a system of partial differential equations that governs the particular application .
the chemotactic sensitivity determines the velocity .
our goal is to develop a technique whereby the appropriate chemotactic sensitivity model , and hence chemotactic velocity , can be determined from available data .
in particular , we consider a system of partial differential equations that was developed by oster and murray @xcite to model the pattern formation of cartilage condensation in a vertebrate limb bud .
a similar system was studied by myerscough et al . , @xcite . the numerical solution of similar systems were recently studied by tyson et al .
@xcite and by nakaguchi and yagi @xcite .
work by fister and mccarthy @xcite has shown that the system of partial differential equations can in fact be controlled theoretically through the introduction of a mechanism controlling the number of cells being generated .
simulations provide optimal drug treatment programs for patients to facilitate the rebuilding of cartilage or the reduction of cancerous tumors .
the chemotactic sensitivity in @xcite was known and the control parameter was a harvesting term .
our goal in this work is to _ identify _ the chemotactic sensitivity .
the paper is organized into six sections . in section two ,
the existence of the forward problem is proven . in section three , identifiability of the chemotactic sensitivity
is established using the weak formulation of the state problem . in section four , tikhonov regularization
is used to approximate the solution through the use of minimization arguments .
the rate of convergence of the approximate minimizer of the chemotactic sensitivity to the true parameter follows next . in section five ,
numerical experiments provide graphical depictions of the accuracy of the recovery of the parameter . in section six
, conclusion remarks are made .
in this model , @xmath0 and @xmath1 represent the concentration of the cells and the chemoattractant , respectively .
the cells and the chemoattractant are governed by a convection - diffusion equation and a reaction - diffusion equation as @xmath2 where @xmath3 is the outward unit normal . @xmath4 and
@xmath5 represent the diffusion coefficients of the cells and the chemoattractant .
the michaelis - menten term , @xmath6 , represents a response of the chemoattractant to a maximum carrying capacity or saturation rate , assuming @xmath7 .
we incorporate a decay term where @xmath8 denotes the degradation rate .
we assume that there is no flux of the concentrations across the boundary , and that the initial concentrations for the cells and chemoattractant are @xmath9 and @xmath10 respectively . here
, @xmath11 is the chemotactic sensitivity which monitors the chemical gradient attraction of the cells .
it is this term that we seek to identify . in @xcite
the term @xmath11 is simply a constant .
more generally , @xmath11 is a linear function of @xmath12 in @xcite , while in @xcite it is a nonlinear function of @xmath13 we assume henceforth that the chemotactic sensitivity has the form @xmath14 and is a bounded function .
we restrict our analysis to the dimensionless system @xmath15 we will establish a technique for the identification of @xmath16 where @xmath17 observe that , with the available data , we can only expect to recover @xmath18 on the interval @xmath19.$ ] the lipschitz condition on the derivative of the chemotactic parameter is quite reasonable , since the chemotactic parameter has a rate of change that is bounded for bacteria growth , @xcite . in order to prove identifiability and to establish the rate of convergence ot our method , we will need to establish existence of a solution of ( [ forward ] ) . using the standard notation @xmath20 to represent the sobelev space @xmath21
let @xmath22 denote the intermediate space between @xmath20 and @xmath23 for any @xmath24 let @xmath5 be an interval in @xmath25 the space @xmath26 is the @xmath27 space of measurable functions in @xmath5 with values in the banach space @xmath28 the space @xmath29 is the space of @xmath30times continuously differentiable functions in @xmath5 with values in @xmath31 while the space @xmath32 is the space of hlder - continuous functions in @xmath5 with values in @xmath28 [ existencet ] if @xmath33 for @xmath34 , and @xmath35 , @xmath36 on @xmath37 , then a real unique local solution @xmath38 of ( [ forward ] ) exists on an interval @xmath39 $ ] such that @xmath40 with @xmath41 and @xmath42 .
the solution satisfies the lower bounds @xmath43.}\ ] ] * proof : * let @xmath44 and @xmath45 . the system ( [ forward ] )
can be formulated as an abstract quasilinear equation @xmath46 on the banach space @xmath28 let @xmath47 clearly @xmath48 .
we define @xmath49 to be the linear operator in @xmath50
such that @xmath51 with domain @xmath52 let the vector @xmath53 be the function @xmath54 since @xmath53 is lipschitz , application of yagi s work ( * ? ? ?
* thm 2.1 and 3.4 ) yields our result .
( see appendix a for statements of yagi s results . )
in this section , we begin by establishing the identifiability of the parameter @xmath18 from the available data @xmath0 and @xmath1 almost everywhere in @xmath56 . note that , in order for chemotaxis to be observed biologically , cells must be present and
a chemical gradient must exist .
this means that @xmath0 and @xmath57 must be nonzero for a measurable subset of @xmath58 we denote by @xmath59 and @xmath60 the solution pairs of ( [ forward ] ) with chemotactic sensitivities @xmath18 and @xmath61 respectively .
[ identification ] let @xmath59 and @xmath62 both be solutions in @xmath63 of the direct problem ( [ forward ] ) corresponding to @xmath64 and @xmath65 .
if @xmath66 and @xmath67 almost everywhere in @xmath68 $ ] , then @xmath69 .
* proof : * we consider the weak form of the first equation of the direct problem ( [ forward ] ) for @xmath70 and @xmath60 and subtract them .
@xmath71 \nabla{\phi } \
dx\ , dt \\\end{aligned}\ ] ] since @xmath72 and @xmath73 a.e .
, this reduces to @xmath74 u_a\nabla { c_a } \nabla{\phi } \ dx\ , dt = 0.\ ] ] by definition , @xmath75 is in @xmath76 , then we can choose @xmath77 .
hence , @xmath74
u_a \left ( \nabla { c_a } \right)^2 \ dx\ , dt = 0.\ ] ] our existence result , theorem [ existencet ] , says that @xmath78 we also employ our biological assumptions that @xmath79 and @xmath80 on a measurable subset of @xmath58 thus @xmath81 almost everywhere .
we wish to identify a function @xmath82 from noisy measurements @xmath83 of @xmath84 recall that @xmath85 where @xmath19.$ ] we define @xmath86 with @xmath87 in the presence of perfect data @xmath83 , we would solve the non - linear ill - posed problem @xmath88 where @xmath89 is the solution of the direct problem with @xmath90 to do this using tikhonov regularization would involve approximating the solution by minimizing @xmath91 where @xmath92 is a small parameter and @xmath93 is an _ a priori _ guess of the true solution @xmath94 in real applications , measurement errors mean that exact data is not available .
noisy data is assumed to have an error level @xmath95 which means that @xmath96 we assume attainability of a true solution , i.e. if @xmath97 there exists @xmath98 such that @xmath99 in the presence of noisy data @xmath100 the minimizer @xmath101 of ( [ minimization_problem ] ) minimizes @xmath102 for appropriate choices of @xmath103 and @xmath104 we begin by establishing the weak - closedness of the map @xmath105 if @xmath106 then @xmath107 and @xmath108 in @xmath109 * proof : * here , we give the outline of the proof and refer the reader to @xcite for details . using that the solution to the state system ( [ forward ] ) is unique , one can define @xmath110 and @xmath111 .
a transformation involving @xmath112 times each component of the solution pair is made with @xmath113 to be chosen in order to obtain the boundedness of the solution in @xmath114 .
the weak definition of the solution associated with the transformed @xmath115 and @xmath116 in equation ( [ weakdiff ] ) is analyzed via cauchy s inequality and the boundedness of the coefficients . using the boundedness ( independent of @xmath117 ) of the solution pairs
, subsequences are extracted that converge weakly to @xmath118 and @xmath119 .
lastly , comparison results are used so that one can pass to the limit in the weak formulation of the solution to show that @xmath120 and @xmath121 .
@xmath55 existence of a minimizer @xmath122 now follows from the lower semi - continuity of @xmath123 for any data @xmath124 a minimizer @xmath122 of ( [ tikobj ] ) exists .
continuous dependence on the data @xmath125 for fixed @xmath126 and the convergence of @xmath122 toward the true parameter @xmath127 as the noise level @xmath128 and the regularization parameter @xmath129 go to zero also follow from standard results @xcite . for fixed @xmath126
the minimizers depend continuously on the data @xmath130 if @xmath131 satisfies @xmath132 then @xmath133 although we have noted ( without proof ) the convergence of the minimizer @xmath122 to the true parameter @xmath134 the rate of convergence may be arbitrarily slow . we wish to determine a source condition that will guarantee a certain rate of convergence .
even when our regularization parameter @xmath129 is comparable to our noise level @xmath95 such a source condition will require assumptions involving @xmath12 and @xmath135 recall that we seek to solve the nonlinear problem ( [ nonlinear - eqn ] ) , @xmath136 where @xmath137 the true solution is @xmath134 and @xmath93 is an _ a priori _ guess . in order to apply the theory of engl , hanke , kunisch and neubauer @xcite
, we must establish the following : * @xmath138 is frechet differentiable , * @xmath139 is lipschitz with @xmath140 * there exists @xmath141 satisfying the source condition @xmath142 * @xmath143 in practice , although computing @xmath139 and @xmath144 is not difficult , it can be quite tricky to establish the lipschitz condition on @xmath139 with our system of coupled nonlinear partial differential equations .
instead , our approach involves developing a _ source condition _ without imposing differentiability constraints on @xmath145 thus we establish @xmath146 convergence .
this technique is also found in the work of engl and kgler , @xcite .
suppose that there exists a function @xmath147 satisfying @xmath148 such that for any @xmath149 @xmath150 if @xmath151 then @xmath152 and @xmath153 [ convergence - theorem ] * proof : * for clarity , we briefly describe the techniques used in this proof .
using that a minimizer to @xmath154 exists , we obtain an upper bound in terms of the error level @xmath128 and the norm of the difference in the minimizer and optimal @xmath155 values .
we then use our source condition with the weak formulation of the cell and chemical differential equations to obtain a representation of the inner product of the appropriate differences of the approximating minimizers .
this allows us to bound @xmath154 .
specifically , we use triangle and young s inequalities to bound the time and spatial derivatives of the differences in the state variables . integration by parts and hlder s inequality enable us to successfully bound the spatial derivatives of the states in terms of the states themselves . using the assumptions from @xmath156 and choosing @xmath157 sufficiently small , we can obtain the error of order @xmath158 with @xmath151 . since @xmath122 is a minimizer of @xmath159 we have @xmath160 using our definition of noise level ( [ noise - defn ] )
, we find that @xmath161 adding @xmath162 to both sides of the inequality and using inner product properties yields @xmath163 observe that our source condition @xmath164 with @xmath165 together with the weak forms of the cell equation in the forward problem ( [ forward ] ) for @xmath127 and @xmath122 is @xmath166 \cdot \nabla w\ dx \ dt\end{aligned}\ ] ] and ( [ source - ineq2 ] ) becomes @xmath167 \cdot \nabla w\ dx \ dt .
\label{source - ineq3}\end{aligned}\ ] ] we bound each integral in ( [ source - ineq3 ] ) separately using triangle and young s inequalities . for the estimates of @xmath168 and @xmath169 we refer the reader to the appendix for some of the details of the bounds used .
we find that @xmath170 and @xmath171 where @xmath172 is an arbitrary parameter resulting from the use of young s inequality .
we utilize the assumptions @xmath173 green s theorem , the boundary conditions , and hlder s inequality to obtain the estimate,@xmath174 \cdot \nabla w\ dx \ dt } \\ & \leq & \varepsilon \left [ \hat{\eta}_1 ^ 2 \int_{0}^t \norm{u_{a_\alpha^\delta}-z_u^\delta}^2_{l^2(\omega ) } \
dt + \norm{u}_{l^\infty(\omega)}^2 k^2 \int_{0}^t \norm{c_{a_\alpha^\delta}-z_c^\delta}^2_{l^2(\omega ) } \ dt \right ] \\ & & + \frac{\alpha^2}{\varepsilon } \int_{0}^t \norm{\delta w}^2_{l^2(\omega ) } \ dt + \varepsilon \hat{\eta}_1 ^ 2\delta^2 + \varepsilon \norm{u}_{l^\infty(\omega)}^2k^2\delta^2 \\ & & + \varepsilon \norm{\nabla u}_{l^\infty(\omega)}^2 k^2 \int_{0}^t \norm{c_{a_\alpha^\delta}-z_c^\delta}^2_{l^2(\omega ) } \ dt + \varepsilon \hat{\eta}_1 ^ 2 \mu^2 \int_{0}^t \norm{u_{a_\alpha^\delta}-z_u^\delta}^2_{l^2(\omega ) } \ dt \\ & & + \frac{\alpha^2}{\varepsilon } \int_{0}^t \norm{\nabla w}^2_{l^2 } \ dt + \varepsilon \norm{\nabla u}_{l^\infty(\omega)}^2 k^2 \delta^2 + \varepsilon \hat{\eta}_1 ^ 2\mu^2\delta^2.\end{aligned}\ ] ] grouping terms and relabeling constants , we have @xmath175 if @xmath172 is chosen to be sufficiently small , then for the choices @xmath151 we obtain @xmath176 and @xmath177 @xmath55
in order to demonstrate the effectiveness of tikhonov regularization for this application , we consider several examples .
all computations were carried out in matlab .
the tikhonov functional @xmath178 was minimized using lsqnonlin , a matlab implementation of the levenberg - marquardt method with line search @xcite .
although it was not tractable to do so in the convergence analysis , a gradient based algorithm is appropriate here because computing the gradient and its adjoint is straightforward .
we restrict our discussion to @xmath179.$ ] recall that @xmath180 and @xmath181 represent noisy data and @xmath93 represents an _ a priori _ guess of the chemotactic sensitivity @xmath182 cell and chemoattractant concentration data on @xmath179 $ ] was generated using pdepe with high accuracy . during the computation of @xmath159 cell and chemoattractant concentrations @xmath0 and @xmath1 associated with a particular @xmath155 were computed using pdepe with moderate accuracy over coarser space and time meshes than those used to simulate data . since lsqnonlin requires objective functions of the form @xmath183 we approximated the first two terms of @xmath184 by @xmath185 ^ 2 + \sum_{i=1}^n \left[c(x_i , t_j)-z_c(x_i , t_j)\right]^2 \right\ } \left ( \delta x\right ) \left(\delta t\right ) \\\end{aligned}\ ] ] where @xmath186 for @xmath187 with @xmath188 and @xmath189 for @xmath190 with @xmath191 we approximate @xmath18 by @xmath192 where @xmath193
are the usual piecewise linear hat functions defined over a partition of @xmath194.$ ] note that any function @xmath18 can be represented by its corresponding vector @xmath195 since the values of @xmath196 and @xmath197 may vary considerably for each @xmath155 used during the optimization , we choose instead an interval that is sufficiently large to include the range of @xmath198 for each @xmath155 considered by the algorithm . in practice , this means making a guess , and expanding the interval when @xmath198 leaves our chosen interval .
the penalty term @xmath199 can be replaced by @xmath200 where the components of the matrix @xmath201 are given by @xmath202 various strategies for the choice of regularization parameters are discussed in @xcite . in each of the following examples
, we used an l - curve method to choose an optimal regularization parameter @xmath104 recall that our chemotaxis system is @xmath203 a similar system was used by myerscough et al .
@xcite in their numerical simulations of chemotaxis in limb - bud development with parameters @xmath204\ ] ] where @xmath205 $ ] and @xmath206 was a bounded perturbation function . in examples 1 - 3 , we used the myerscough parameters with @xmath207 [ [ example-1 ] ] example 1 + + + + + + + + + consider the chemotactic coefficient used by myerscough et al .
@xcite @xmath208 the cell and chemoattractant concentrations associated with this @xmath155 are shown over the time interval @xmath209 $ ] in figure [ concentrations_example1 ] .
an initial guess of @xmath210 was used .
the _ a priori _ guess was also chosen to be @xmath211 the parameter @xmath155 was recovered to within @xmath212 of the true value at @xmath213 commented these next two examples out ! [ [ example-2-keller - segel - model ] ] example 2 : keller - segel model + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we consider the nonlinear chemotactic coefficient @xmath214 proposed in the original keller - segel model for chemotaxis @xcite .
the cell and chemoattractant concentrations associated with this @xmath155 are shown at steady state in figure [ concentrations_example4 ] . from the data
, we find that @xmath194 = [ 0.1794 , 0.6398]$ ] when @xmath215 since the optimization algorithm will use approximations of other chemotactic functions , we attempt to reconstruct @xmath18 over a larger interval .
we found the interval @xmath216 $ ] to be sufficiently large to include the range of @xmath198 for each @xmath155 considered by the algorithm .
an initial guess of @xmath217 was used .
the _ a priori _ guess was also chosen to be @xmath218 figure [ a_example4 ] shows the chemotactic function @xmath155 and its recovery of @xmath219 with and without regularization .
a regularization parameter of @xmath220 was chosen by an l - curve method .
notice that the regularized recovery is quite reasonable over the interval @xmath194 = [ 0.1794 , 0.6398]$ ] and that its quality degrades , as expected , outside this interval .
[ [ comments ] ] comments + + + + + + + + the number of iterations used by the algorithm is quite sensitive to the choice of initial function @xmath221 and the number of piecewise linear basis functions used in equation ( [ a - approximation ] ) . for experimental data , we must acknowledge that the quality of the recovery degrades with increased noise in the data . in certain applications such as pattern formation in _ escherichia coli _ or _ salmonella typhimurium _ , see tyson et al .
@xcite , the size of the interval @xmath194 $ ] is sometimes too small to give adequate information for the recovery of the chemotactic coefficient .
this can be avoided by taking a larger time interval @xmath39.$ ] in numerical simulations , this requires a careful choice of numerical method for the solution of the chemotaxis system , see @xcite .
an alternative approach is to restrict our measurements to a particular time , rather than an interval of time .
the efficacy of this approach will be discussed in a future paper .
in this work , we have explored a particular mathematical aspect of the chemotactic sensitivity within the gradient .
the identification of a chemotactic sensitivity with functional dependence has been determined .
the interesting aspect of this work is that , to our understanding , no one has been able to capture the chemotactic sensitivity information from limited data with dependence on the chemical in a system .
we have proven the existence of the state solutions in specific sobolev spaces and formulated an inverse problem .
we have employed tikhonov regularization to recover the chemotactic sensitivity from noisy measurements . in doing so ,
a minimization problem is formed and the necessary convergence results for an approximating minimizer to the true parameter are discussed .
another significant result is that we have established a source condition that guarantees a particular rate of convergence by imposing a lipschitz condition on the derivative of the chemotactic sensitivity . in practice , this is biologically reasonable , since the chemotactic sensitivity has a rate of change that is bounded for bacterial growth , @xcite .
numerically , we have utilized models from myerscough et .
al , @xcite and keller and segel , @xcite for the studies of the comparison of our proposed work to the actual scenarios . with the use of tikhonov regularization ,
we have been able to recover the chemotactic sensitivity with reasonable accuracy .
a biological benefit of this knowledge is the ability for one to understand the growth associated with chemotaxis within tumor studies , leukocyte dynamics , and bacterial patterns based on the specific gradient information that can be recovered from imperfect data .
this work was partially supported by the howard hughes medical insitute as part of its undergraduate biological sciences education program award to murray state university , and by the national science foundation awards dms-0209562 , dms-0414011 and due-0531865 .
any opinions , findings , and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the howard hughes medical insitute or the national science foundation .
[ existencelocalyagi ] let @xmath223 for @xmath224 , and @xmath35 , @xmath225 on @xmath37 . assume a real local solution @xmath226 to ( [ yagiprob ] ) exists on the interval @xmath227 $ ] such that @xmath228 for some @xmath229 in addition , assume that @xmath230 satisfies @xmath231 on @xmath232 $ ] and an estimate@xmath233 for some @xmath234 and constant @xmath235
. then @xmath236$},\ ] ] where @xmath237 denotes a positive function defined as the global solution to an ordinary differential equation,@xmath238 [ uniqueyagi]let @xmath223 for @xmath224 , and @xmath35 , @xmath239 on @xmath37 and let @xmath240 .
then , in the function space , @xmath241 , the problem ( [ yagiprob ] ) possess a unique local solution@xmath242with the lower bounds@xmath243$},\ ] ] where @xmath244 is a positive function defined by equations in ( [ peqn ] ) .
the interval @xmath227 $ ] on which the solution exists at least is determined by the norms @xmath245 , @xmath246 and by the initial lower bound @xmath247 .
it is to be noted that by this theorem from yagi s work that a maximal solution to ( [ yagiprob ] ) can be uniquely defined in the space @xmath248 for @xmath240 for each @xmath249 such that @xmath223 for @xmath224 , and @xmath35 , @xmath250 on @xmath37 .
ford , r. , lauffenburger , d. , 1991 .
analysis of chemotactic bacterial distributions in population migration assays using a mathematical model applicable to steep or shallow attractant gradients .
, 53(5):721729 .
hillen , t. , levine , h. , 2003 .
blow - up and pattern formation in hyperbolic models for chemotaxis in 1-d .
, 54:839868 .
hillen , t. , painter , k. , 2001.global existence for a parabolic chemotaxis model with prevention of overcrowding . , 26:4 , 280 - 301 .
nakaguchi , n. , yagi , , a. , 2001 .
full discrete approximation by galerkin method for chemotaxis - growth model .
, 47:60976107 .
painter , k. , hillen , t. , 2003 .
volume - filling and quorum - sensing in models for chemosensitive movement .
, 10:4,501 - 543 .
tyson , r . , stern , l.g . ,
leveque , r.j .
fractional step methods applied to a chemotaxis model . , 41:455475 .
velazquez , j.j.l .
point dynamics for a singular limit of the keller - segel model , i. motion of the concentration regions .
, 64:4 , 1198 - 1223 .
velazquez , j.j.l .
point dynamics for a singular limit of the keller - segel model , ii .
formation of the concentration regions .
, 64:4 , 1224 - 1248 . | chemotaxis is the process by which cells behave in a way that follows the chemical gradient .
applications to bacteria growth , tissue inflammation , and vascular tumors provide a focus on optimization strategies .
experiments can characterize the form of possible chemotactic sensitivities .
this paper addresses the recovery of the chemotactic sensitivity from these experiments while allowing for nonlinear dependence of the parameter on the state variables .
the existence of solutions to the forward problem is analyzed .
the identification of a chemotactic parameter is determined by inverse problem techniques .
tikhonov regularization is investigated and appropriate convergence results are obtained .
numerical results of concentration dependent chemotactic terms are explored . * keywords : * inverse problem , chemotaxis , tikhonov regularization |
finite size systems have a great deal of interest to our comprehension of several phenomena , particularly those involving critical systems @xcite .
the main question is : how far can one reduce the size of a system without disturbing its critical universal properties ?
consider three - dimensional infinite ( bulk ) systems in a cubic lattice .
if one dimension perpendicular to the planes of the critical material slab is reduced , one has a geometry of thin films , the critical temperature is lower in comparison with that from the bulk and the system still has divergences in its thermodynamic potentials .
the thermodynamic limit can be safely taken in these thin films provoking the physical divergences , in stark contrast with thin films which are finite in all directions . in the latter , the thermodynamic limit _
can not be taken _ and instead of the cusp - like divergences , the maximum corresponding to the critical temperature is no longer discontinous but it is `` rounded '' @xcite .
we can also consider systems with two reduced dimensions and one infinite dimension .
their classical phase transitions effects appear , for instance in a cylindrical geometry regarding @xmath2 nanowires @xcite , in the study of potts model in long cylinders @xcite , etc . from the experimental point of view
, the fabrication of high - quality thin films is commonly based on epitaxial growth over a substrate @xcite .
in general , ferromagnetic properties of epitaxially grown thin films depend upon the substrate . if the substrate forms a heterostruture with the original film , for instance , in @xmath3 @xcite , ferrolectricity is induced in the heterostructure due to epitaxial strain in the @xmath4 planes ( which is originally bulk ferromagnetic ) .
depending on the method chosen and the substrate where the film is epitaxially grown , the critical properties of the bulk can be very different from that of the film itself : there can be effects of interfaces @xcite ( and its applications in the electroceramics industry @xcite ) , surfaces @xcite , etc .
the parallel plate geometry is certainly important in the construction of microscopic devices as a stack of several layers of material in manufacturing , for example , high performance disks for memory storage @xcite . in particular , multiferroic materials show ferromagnetism and ferroelectricity and are ideal for that purpose @xcite , such that the comprehension of those effects in finite - systems would be worthwhile . for
a recent overview with a broader perspective and a mix of applications , see for instance @xcite .
however , the epitaxial growth method of thin films named metal - organic chemical vapor deposition ( @xmath5 ) is rather efficient to produce high - quality films with bulk properties , since the substrate can be `` sheared '' in the final process .
this technique was employed in single crystal films of @xmath2 @xcite and @xmath6 @xcite .
the post - growth analysis exhibited normal bulk magnetic properties .
( there are also examples of materials whose molecular structure is composed by different atomic species in which this method of producing thin films show similar magnetic properties with those from bulk samples , albeit not so simply ; see ref .
hence , we learn that thin films produced in this manner yields a firm basis to investigate finite size effects realized experimentally in a simplified way .
the field - theoretical description of critical properties for this geometry utilizing green s functions in momentum space was initiated by nemirovsky and freed @xcite utilizing a massive method involving only bulk fields with several boundary conditions on the limiting surfaces .
recently , the renormalized one - particle irreducible ( @xmath7 ) framework was devised and extended to massless fields as well , subject to periodic ( @xmath8 ) and antiperiodic ( @xmath9 ) boundary conditions @xcite .
after that , the @xmath7 vertex part formalism was generalized to include dirichlet ( @xmath10 ) and neumann ( @xmath11 ) boundary conditions in a non - trivial unified fashion by renormalizing the primitively divergent vertex parts at nonzero external quasi - momenta @xcite .
the simplest situation within this picture is realized when each plate has linear dimensions of infinite extent .
these thin films satisfy the thermodynamic limit and were shown to have critical exponents identical to bulk systems and independent of boundary conditions , using massive or massless regimes for the renormalized volumetric order parameters ( fields ) , provided the distance between the limiting surfaces is not too small . in this work
we propose the renormalized one - particle ( @xmath7 ) irreducible formalism at null external quasi - momenta in the treatment of a critical system defined in a parallel plate geometry , whose order parameter is subject to neuman boundary conditions ( @xmath11 ) at the limiting surfaces constituted by @xmath12-dimensional hyperplanes located at @xmath13 , where @xmath14 is the space dimension perpendicular to the plates .
we see finite size corrections to the bulk behavior which are @xmath0-dependent . in the limit @xmath1
we show that the bulk behavior is recovered and all corrections tend to zero .
contrarily , in the limit @xmath15 we demonstrate how the `` dimensional crossover '' @xcite shows up , overcoming the bulk behavior through non - trivial surface left - over , entirely out of bulk fields . differently from previous approaches ,
the appearance of surface effects is due neither to the presence of interfaces between two materials ( or the same material in different structural phases ) nor the presence of external surface fields .
interestingly , surface effects emerge through dimensional crossover for @xmath8 and @xmath11 but are absent ( or are much weaker , so to speak ) in @xmath10 and @xmath9 .
in contrast with the situation for @xmath8 and @xmath9 which preserve translation invariance @xcite , we will show that in the present framework @xmath11 do break the translation invariance of the theory in line with the result from the unified treatment @xcite .
moreover , the present method is by far simpler than the latter in the treatment of @xmath11 .
if the system is kept away from the dimensional crossover region , we prove that the critical exponents are the same of those from the bulk system , which is consistent with the results from the above previous @xmath7 vertex part and independent of boundary conditions .
the ideas and results are presented as follows .
we write down the bare lagrangian density out of solely bulk fields and offer a quick review of the decomposition of the fields in terms of its fourier components and basis functions in section ii .
section iii will be the starting point with the basis functions from the lagrangian density in the exponential representation .
we write down the tensors corresponding to @xmath7 vertex parts ( with and without composite operators ) . utilizing the zero mode basis function as well , due to our choice of normalization conditions
, we find later the complete set of feynman graphs up to three - loop order . in section
iv we tie up the formulation of massive fields with a quick analysis of the one - loop diagram for the coupling constant and prove that the finite size corrections do not require normal ordering at zero external quasi - momenta .
we show that the limit of validity of the @xmath16-expansion can be interpreted consistently with that from the unified technique and compute the critical exponents . in appendix
a we summarize the results of all massive integrals . in section
v we introduce the massless framework in parallel with the massive case .
thereby , the critical exponents are shown to be identical in both massive and massless formulations .
they agree exactly with those obtained from the infinite ( bulk ) system .
the results of all massless diagrams are presented in appendix b. the conclusions are the subject of section vi .
the bare lagrangian density involving only volumetric ( bulk ) fields in the parallel plate geometry is given by : @xmath17 , @xmath18 and @xmath19 are the bare order parameter , mass ( @xmath20 is the bare reduced temperature proportional to @xmath21 ) and coupling constant , respectively .
note that although @xmath22 denotes the critical temperature of the infinite system ( `` bulk '' ) and is in general different of the shifted critical temperature of the finite system @xmath23 , in principle the difference is not sufficient to provoke a functional variation in the critical exponents at the critical region . obviously , proving this with quantitative computation of the critical exponents is appropriate , but it is reasonable to note that the fluctuations there do not distinguish between these two types of temperatures @xcite , provided the plates are of infinite extent in each linear dimension . as discussed above ,
if the plates are not infinite , instead of a spike - like divergence of the critical quantity under consideration with its associated critical exponent , the finite size of the system provokes a `` rounding '' in the divergence . on aforementioned experimental grounds ,
the temperature @xmath23 yields a qualitative picture which resembles that from the bulk system .
we shall use henceforth the bulk critical temperature in what follows .
we refer the reader to ref .
@xcite in order to fix the notation , which is quite similar to that we shall employ hereafter .
the field is a vector of @xmath24 components @xmath25 , whose internal indices of the @xmath26 symmetry are omitted .
the coordinate vector @xmath27 belonging to the @xmath12-dimensional plates along with the perpendicular @xmath14 axis constitutes the collective vector @xmath28 .
there are parallel plates separated by a lattice constant space along the @xmath14 direction and filling in the region between @xmath29 and @xmath30 .
the field satisfies @xmath31 for @xmath11 .
the order parameter is related to its fourier modes in momentum space through @xmath32 , where @xmath33 is the momentum vector characterizing the @xmath12-dimensional space .
the basis functions @xmath34 specify the ( hyper)plate and have a discrete index , with eigenvalues @xmath35 , where @xmath36 @xmath37 is the quasi - momentum along the @xmath14-direction .
the free bare massive ( @xmath38 ) propagator in momentum space is given by the expression @xmath39 .
the propagator can be represented graphically by a line with two extremities : the left one characterized by the index @xmath40 and the right labeled by @xmath41 , which implies the presence of the overall factor @xmath42 , which will be ignored in what follows without loss of generality .
the propagator for the massless case is obtained by setting @xmath43 . from our construction of a generic feynman diagram previously in the unified formalism
, we just highlight that each momentum line ( propagator ) must be multiplied by @xmath44 and the @xmath45 vertices are multipled by the tensor @xmath46 . recall that this is the way in which the finite size effect is implemented as an internal symmetry , consisting of the direct product of these tensors with the usual ones from the @xmath26 symmetry .
the basis functions for @xmath11 have the nonzero modes @xmath47 ( @xmath48 ) as well as the zero mode @xmath49 .
the explicit choice of vanishing external quasi - momenta in the renormalized theory implies that the zero mode basis function participates in the feynman rules in a simpler but different form in comparison with the unified approach involving @xmath10 and @xmath11 renormalized at nonzero external quasi - momenta .
let us start the construction representing mathematically the several feynamn diagrams required to our computation by writing down the tensors @xmath52 , @xmath53 and @xmath54 corresponding to the composite field .
we multiply the tensors with four indices by @xmath55 in order to construct the diagrams , namely @xmath56 .
we are using explicitly the exponential representation for the nonzero mode basis functions ; compare the tensors with those from ref .
@xcite when @xmath57 .
the explicit form of the tensors required using the above eigenfunctions are @xmath58 : @xmath59,\\ & & \tilde{s}_{0 i_{1 } i_{2 } i_{3}}= \sqrt{2 } \tilde{\sigma } [ \delta(i_{1 } + i_{2 } + i_{3 } ) + \delta(i_{1 } + i_{2 } - i_{3 } ) + \delta(i_{1 } - i_{2 } + i_{3 } ) + \delta(i_{1 } - i_{2 } - i_{3})],\\ & & \tilde{s}_{0 0 i_{1 } i_{2 } } = 2 \tilde{\sigma } [ \delta(i_{1 } - i_{2 } ) + \delta(i_{1 } + i_{2 } ) ] , \;\;\tilde{s}_{0 0 0 i_{1 } } = 0,\;\ ; \tilde{s}_{0 0 0 0 } = 2 \tilde{\sigma},\\ & & \hat{s}_{i_{1 } i_{2 } i_{3}}=\frac{1}{2}[\delta(i_{1 } + i_{2 } + i_{3 } ) + \delta(i_{1 } + i_{2 } - i_{3 } ) + \delta(i_{1 } - i_{2 } + i_{3 } ) + \delta(i_{1 } - i_{2 } - i_{3})],\\ & & \hat{s}_{0 i_{1 } i_{2}}= \frac{1}{\sqrt{2}}[\delta(i_{1 } - i_{2 } ) + \delta(i_{1 } + i_{2 } ) ] , \;\ ; \hat{s}_{00i}= \delta(i ) .
\end{aligned}\ ] ] note that we separated the zero mode in all tensors above ( the other indices appearing in the above expressions are explicitly nonzero ) .
there is an argument presented in the unified picture which simplifies enormously our task in evaluating feynman graphs out of these tensors .
we shall outline it very briefly here and take advantage of it in our present discussion . in order to restrict our discussion to a minimal number of diagrams
we start with a tree level bare mass parameter @xmath18 , perform the diagrammatic expansion of the @xmath60 and identify it with the three - loop bare mass @xmath61 . after the inversion in order to obtain @xmath62 and expressing all integrals in terms of @xmath61 , this reparameterization in conjunction with the non - trivial tadpole cancelation eliminates all tadpole diagrams and tadpole mass insertions in all primitively divergent diagrams , _ i. e. _ , @xmath50 and @xmath51 @xcite .
therefore , we just have to analyze the `` sunset '' ( two - loop ) and the nontrivial three - loop diagram of @xmath63 . in addition , we are left only with the evaluation of the diagrams from @xmath64 @xmath65 and @xmath66 up to two - loop order without any mass insertions . before analyzing any loop diagram
let us determine the value of all tree - level primitively divergent bare vertex parts at zero external quasi - momenta .
first , the bare tree diagram of @xmath63 is just equal to the inverse propagator , i.e. , @xmath67 and represented by the graph @xmath68{fig9.eps}}^{-1}$ ] . at this point
is convenient switching to a simpler notation for the momenta @xmath69 .
the bare vertex @xmath70 starts with the zero - loop contribution @xmath71{fig29.eps}}=2 \lambda \tilde{\sigma}$ ] , whereas the composite vertex operator @xmath72 has tree - level value @xmath73{fig28.eps}}\qquad=1 $ ] @xcite . therefore , differently from the unified approach , here
each @xmath74-loop diagram from @xmath70 receives a coefficient @xmath75 , such that this coefficient factors out in the diagrammatic expansion of this vertex part .
we shall omit this trivial global factor in the construction of arbitrary loop diagrams from @xmath70 , but will recuperate it during the discussion of the normalization conditions later on .
we also choose not to write explicitly the coupling constant in front of each diagram ; rather we use their different powers as coefficients of the graphs ( see below ) . the other vertex parts @xmath76 and @xmath77
do not receive global factors in arbitrary loop order and have pretty much the same form as in the usual @xmath45 theory describing bulk systems .
although not depicted in the diagrams , we will employ the convention that all external quasi - momenta of all diagrams ( obviously including those from tree level contributions ) are implicitly set to zero .
we analyze the minimal set of diagrams from @xmath78 to begin with .
the preliminary expression for its two - loop diagram is given by : @xmath79{fig6.eps } } \quad = \bigl(\frac{n+2}{3}\bigr ) \overset{\infty}{\underset{j_{1},j_{2},j_{3}=0}{\sum } } \tilde{s}_{0 j_{1 } j_{2 } j_{3 } } \tilde{s}_{j_{1 } j_{2 } j_{3 } 0 } \int d^{d-1}q_{1 } d^{d-1}q_{2 } g_{0}(q_{1}+q_{2}+k , j_{3 } ) g_{0}(q_{2 } , j_{2 } ) \nonumber\\ & & \qquad \times \;\ ; g_{0}(q_{1 } , j_{1 } ) .
\end{aligned}\ ] ] after explicitly separating the zero mode components from the nonzero ones and perform some manipulations one can show that this expresssion reduces to @xmath80{fig6.eps } } \quad = \left(\frac{n+2}{3}\right)\bigl[i_{3}(k;0;\tilde{\sigma},\mu ) + 3 \tilde{i}_{3}(k;0,0;\tilde{\sigma},\mu)\bigr]\end{aligned}\ ] ] where , [ 5 ] @xmath81[(q_{1}+q_{2}+k)^{2 } + \tilde{\sigma}^{2}(j_{1}+j_{2})^{2}+\mu^{2}]}\nonumber\\ & & \;\;\;\quad \times \frac{1}{[q_{1}^{2}+\tilde{\sigma}^{2}j_{1}^{2}+\mu^{2}]},\\ & & \tilde{i}_{3}(k;0,0;\tilde{\sigma},\mu)= \tilde{\sigma}^{2 } \overset{\infty}{\underset{l=-\infty}{\sum } } \int \frac{d^{d-1}q_{1 } d^{d-1}q_{2 } } { [ q_{1}^{2}+\tilde{\sigma}^{2}l^{2}+\mu^{2}][q_{2}^{2}+\mu^{2 } ] } \frac{1}{[(q_{1}+q_{2}+k)^{2 } + \tilde{\sigma}^{2 } l^{2}+\mu^{2}]}.\end{aligned}\ ] ] interestingly , the first integral is just identical to the similar contribution coming from @xmath8 at zero external quasi - momenta @xcite , whereas the second one correspond to the `` non - diagonal '' term which breaks explicitly the translational invariance at zero external quasi - momenta .
the latter just shows up for @xmath11 and @xmath10 and was discussed previously in the unified picture with nonvanishing external quasi - momenta .
however , the present form is much simpler than in the situation when the external quasi - momenta are not chosen equal to zero .
let us analyze now the three - loop contribution of this vertex part at zero external quasi - momenta .
in terms of the finite size tensors , it reads @xmath82{fig7.eps } } \quad = \frac{(n+2)(n+8)}{27 } \overset{\infty}{\underset{j_{1},j_{2},j_{3 } , j_{4 } , j_{5}=0}{\sum } } \tilde{s}_{0 j_{1 } j_{2 } j_{3 } } \tilde{s}_{j_{2 } j_{3 } j_{4 } j_{5 } } \tilde{s}_{j_{4 } j_{5 } j_{1 } 0 } \int d^{d-1}q_{1 } d^{d-1}q_{2 } d^{d-1}q_{3 } g_{0}(q_{1 } , j_{1 } ) \nonumber\\ & & \qquad \qquad \times \;\ ; g_{0}(q_{1}+q_{2}+k , j_{2 } ) g_{0}(q_{2 } , j_{3 } ) g_{0}(q_{1}+q_{3}+k , j_{4 } ) g_{0}(q_{3 } , j_{5}).\end{aligned}\ ] ] the manipulation here involving the separation of zero modes and nonzero modes components of each index in the summation of the finite size tensors is really lengthy and we let the details to the reader figure them out .
fortunately , after a formidable large number of cancelations , the outcome turns out to be simple and we get to @xmath83{fig7.eps } } \quad = \frac{(n+2)(n+8)}{27 } [ i_{5}(k,0,\tilde{\sigma},\mu ) + \tilde{i}_{5}(k,0,0,\tilde{\sigma},\mu ) + 4 \hat{i}_{5}(k,0,0,0,\tilde{\sigma},\mu)\nonumber\\ & & \;\;+\;\ ; 2 \bar{i}_{5}(k,0,0,0,\tilde{\sigma},\mu)].\end{aligned}\ ] ] the integrals in last equation are defined by : [ 8 ] @xmath84[q_{2}^{2}+\tilde{\sigma}^{2}j_{2}^{2}+\mu^{2}][q_{3}^{2}+\tilde{\sigma}^{2}j_{3}^{2}+\mu^{2}]}\nonumber\\ & & \times \frac{1}{[(q_{1}+q_{2}+k)^{2 } + \tilde{\sigma}^{2}(j_{1}+j_{2})^{2}+\mu^{2}][(q_{1}+q_{3}+k)^{2 } + \tilde{\sigma}^{2}(j_{1}+j_{3})^{2}+\mu^{2}]},\\ & & \tilde{i}_{5}(k;0,0;\tilde{\sigma},\mu)= \tilde{\sigma}^{3 } \overset{\infty}{\underset{j_{1},j_{2}=-\infty}{\sum } } \int \frac{d^{d-1}q_{1 } d^{d-1}q_{2}d^{d-1}q_{3 } } { [ q_{1}^{2}+\mu^{2}][q_{2}^{2}+\tilde{\sigma}^{2}j_{1}^{2}+\mu^{2}][q_{3}^{2}+\tilde{\sigma}^{2}j_{2}^{2}+\mu^{2}]}\nonumber\\ & & \times \frac{1}{[(q_{1}+q_{2}+k)^{2 } + \tilde{\sigma}^{2 } j_{1}^{2}+\mu^{2}][(q_{1}+q_{3}+k)^{2 } + \tilde{\sigma}^{2 } j_{2}^{2}+\mu^{2}]},\\ & & \hat{i}_{5}(k;0,0,0 ; \tilde{\sigma } , \mu)=\tilde{\sigma } \overset{\infty}{\underset{j=-\infty}{\sum } } \int \frac{d^{d-1}q \tilde{i}_{2}(q+k;0,j;\tilde{\sigma } , \mu ) i_{2}(q+k;j;\tilde{\sigma } , \mu)}{q^{2 } + \sigma^{2}j^{2 } + \mu^{2}},\\ & & \bar{i}_{5}(k;0,0,0;\tilde{\sigma},\mu)= \tilde{\sigma } \overset{\infty}{\underset{j_{1},j_{2}=-\infty}{\sum } } \int \frac{d^{d-1}q \tilde{i}_{2}(q+k;j_{2};j_{2}+j_{1 } ; \tilde{\sigma } , \mu)}{q^{2 } + \sigma^{2}j_{1}^{2 } + \mu^{2 } } \nonumber\\ & & \;\ ; \times \;\ ; \tilde{i}_{2}(q+k;j_{1}+j_{2},j_{2 } ; \tilde{\sigma } , \mu).\end{aligned}\ ] ] it is important to emphasize that the integrals @xmath85 and @xmath86 correspond to one - loop subdiagrams of the vertex part @xmath87 and are defined by @xmath88[q^{2 } + \tilde{\sigma}^{2 } j^{2 } + \mu^{2}]},\label{9a}\\ & & \tilde{i}_{2}(k;0,0;\tilde{\sigma},\mu ) = \tilde{\sigma } \int \frac{d^{d-1}q}{[(q+k)^{2 } + \mu^{2}][q^{2 } + \mu^{2}]}.\label{9b}\end{aligned}\ ] ] the pattern of @xmath8 ( integral @xmath89 ) plus `` non - diagonal '' terms happens in a way consistent with the unified framework at nonzero external quasi - momenta for @xmath11 .
that is why we included an extra entry of quasi - momenta in the the integral @xmath86 .
this is a generic feature of the non - diagonal terms showing up in arbitrary loop diagrams : if the diagonal contribution ( integral ) has @xmath90 entries in its argument for the external quasi - momenta , the nondiagonal contribution will have @xmath91 entries .
this fact is trivial when addressing all the contributions at zero external momenta , but should be kept in mind in order to make contact with the more general unified formalism .
we now discuss the vertex part @xmath87 .
the one - loop diagram in terms of the finite size tensors and propagators can be written as @xmath92{fig10.eps}}\quad = \frac{(n+8)}{9 } \overset{\infty}{\underset{l_{1},l_{2 } = 0}{\sum}}\tilde{s}_{00 l_{1 } l_{2 } } \tilde{s}_{l_{1 } l_{2}00 } \int d^{d-1}q g_{0}(q+k , l_{1 } ) g_{0}(q , l_{2}).\end{aligned}\ ] ] we employ the standard notation that in this diagram the external momentum @xmath93 can be connected with the actual external momenta @xmath94 through the possibilities @xmath95 , @xmath96 and @xmath97 .
( we could also have used @xmath98 due to the overall external momenta conservation . )
all of them should be included in the computation of the contribution of this diagram .
the same happens in the trivial two - loop diagram to be discussed below .
however , for the two - loop non - trivial diagram of this vertex part , there are twice more terms in the complete computation due to these diagrams in the diagrammatic expansion . for the sake of simplicity
we adopt the notation @xmath93 to designate a generic external momenta ( and adopt a similar convention for the diagrams involving composite fields ; see below ) .
since we are going to be interested in particular configurations of the external momenta ( `` symmetry points '' ) in the renormalization scheme , we shall not worry about these details .
the reader is invited to consult the book by amit and martin - mayor @xcite at this point .
we can handle this expression in order to get the expressions @xmath99{fig10.eps}}\quad
= \frac{(n+8)}{9 } \bigl(i_{2}(k;0 ; \tilde{\sigma } , \mu ) + \tilde{i}_{2}(k;0,0;\tilde{\sigma } , \mu)\bigr),\end{aligned}\ ] ] where these integrals were defined in eqs .
( [ 9a ] ) and ( [ 9b ] ) . by the same token ,
the trivial two - loop diagram computed at zero external quasi - momenta can be written as @xmath100{fig11.eps}}\qquad = \frac{(n^{2}+6n+20)}{27 } \overset{\infty}{\underset{j_{1},j_{2 } , j_{3 } , j_{4 } = 0}{\sum } } \tilde{s}_{00 j_{1 } j_{2 } } \tilde{s}_{j_{1 } j_{2 } j_{3 } j_{4}}\tilde{s}_{j_{3 } j_{4 } 00 } \int d^{d-1}q_{1 } d^{d-1}q_{2 } g_{0}(q_{1}+k , j_{1 } ) \nonumber\\ & & \;\ ; g_{0}(q_{1},j_{2 } ) g_{0}(q_{2}+k , j_{3 } ) g_{0}(q_{1},j_{4}).\end{aligned}\ ] ] when the same systematics is carried out by replacing the values of the tensors we are led to @xmath101{fig11.eps}}\qquad = \frac{(n^{2}+6n+20)}{27}\bigl [ i_{2}^{2}(k;0;\tilde{\sigma } , \mu ) + 2 i_{2}(k;0;\tilde{\sigma } , \mu ) \tilde{i}_{2}(k;0,0;\tilde{\sigma } , \mu ) \nonumber\\ & & \;\ ; + \;\ ; \overset{\infty}{\underset{j=-\infty}{\sum}}\tilde{i}_{2}^{2}(k;j , j;\tilde{\sigma } , \mu)\bigr].\end{aligned}\ ] ] in order to complete our task of computing the two - loop contributions of the four - point vertex part , consider the non - trivial two - loop diagram @xmath102{fig12.eps}}\qquad = \frac{(5n+22)}{27 } \overset{\infty}{\underset{j_{1},j_{2 } , j_{3 } , j_{4 } = 0}{\sum } } \tilde{s}_{00 j_{1 } j_{2 } } \tilde{s}_{j_{1 } j_{3 } j_{4 } 0}\tilde{s}_{j_{2 } j_{3 } j_{4 } 0 } \int d^{d-1}q_{1 } d^{d-1}q_{2 } g_{0}(q_{1}-k , j_{1 } ) \nonumber\\ & & \;\ ; g_{0}(q_{1},j_{2 } ) g_{0}(q_{1}-q_{2}+k_{3},j_{3 } ) g_{0}(q_{2},j_{4}).\end{aligned}\ ] ] after working out some set of calculations we find the simple result @xmath103{fig12.eps}}\qquad = \frac{(5n+22)}{27 } \bigl[i_{4}(k , k_{3};0,0;\tilde{\sigma},\mu ) + \tilde{i}_{4}(k , k_{3};0,0,0;\tilde{\sigma},\mu ) \nonumber\\ & & \;\;+\;\ ; 2 \hat{i}_{4}(k , k_{3};0,0,0;\tilde{\sigma},\mu)\bigr],\end{aligned}\ ] ] where @xmath95 and the above integrals read [ 16 ] @xmath104[(q_{1}-k)^{2}+\tilde{\sigma}^{2 } l^{2}+\mu^{2 } ] [ q_{2}^{2}+\tilde{\sigma}^{2}m^{2}+\mu^{2}]}\nonumber\\ & & \;\;\times\;\ ; \frac{1}{[(q_{1}-q_{2}+k')^{2}+\tilde{\sigma}^{2}(l - m)^{2}+\mu^{2}]},\\ & & \tilde{i}_{4}(k , k';0,0,0 ; \tilde{\sigma } , \mu)=\tilde{\sigma}^{2}\overset{\infty}{\underset{m=-\infty}{\sum}}\int \frac{d^{d-1}q_{1}d^{d-1}q_{2 } } { [ q_{1}^{2}+\mu^{2}][(q_{1}-k)^{2}+\mu^{2 } ] [ q_{2}^{2}+\tilde{\sigma}^{2}m^{2}+\mu^{2}]}\nonumber\\ & & \;\;\times\;\ ; \frac{1}{[(q_{1}-q_{2}+k')^{2}+\tilde{\sigma}^{2}m^{2}+\mu^{2}]},\\ & & \hat{i}_{4}(k , k';0,0,0 ; \tilde{\sigma } , \mu)=\tilde{\sigma}^{2}\overset{\infty}{\underset{m=-\infty}{\sum}}\int \frac{d^{d-1}q_{1}d^{d-1}q_{2 } } { [ q_{1}^{2}+\tilde{\sigma}^{2}m^{2}+\mu^{2}][(q_{1}-k)^{2}+\tilde{\sigma}^{2}m^{2}+\mu^{2 } ] [ q_{2}^{2}+\mu^{2}]}\nonumber\\ & & \;\;\times\;\ ; \frac{1}{[(q_{1}-q_{2}+k')^{2}+\tilde{\sigma}^{2}m^{2}+\mu^{2}]}.\end{aligned}\ ] ] it is worthy noting that all integrals above are straightforward particular cases from their counterparts in the unified framework with nonvanishing external quasi - momenta @xcite . from this viewpoint , it is really rewarding to employ the present formalism to study @xmath11 .
let us examine the diagrams of the composite field .
they have the same integral structure of the four - point integral just discussed .
it will simplify the subsequent description , for we restrict ourselves to the explanation of the results in terms of the above integrals .
we start with the tensors @xmath54 whose contracted products produce the diagram with null external quasi - momenta .
the one - loop graph @xmath105{fig14.eps}}\qquad = \frac{(n+2)}{6 } \overset{\infty}{\underset{j_{1},j_{2 } = 0}{\sum}}\tilde{s}_{00 j_{1 } j_2 } \hat{s}_{j_{1 } j_{2 } j } \int d^{d-1}q g_{0}(q+k , j_{1 } ) g_{0}(q , j_{2}),\end{aligned}\ ] ] can be easily shown to be given by @xmath106{fig14.eps}}\qquad = \frac{(n+2)}{6 } \bigl[i_{2}(k;0 ; \tilde{\sigma } , \mu ) + \tilde{i}_{2}(k;0,0;\tilde{\sigma } , \mu)\bigr],\end{aligned}\ ] ] in order to make a connection with the unified framework , we treat both two - loop contributions for the composite field simultaneously .
their expressions with respect to the finite size tensors are @xmath107{fig16.eps}}\qquad = \frac{(n+2)^{2}}{108 } \overset{\infty}{\underset{j_{1},j_{2 } , j_{3 } , j_{4 } = 0}{\sum } } \tilde{s}_{00 j_{1 } j_{2 } } \tilde{s}_{j_{1 } j_{2 } j_{3 } j_{4 } } \hat{s}_{j_{3 } j_{4 } j } \int d^{d-1}q_{1 } d^{d-1}q_{2 } g_{0}(q_{1}+k , j_{1 } ) \nonumber\\ & & g_{0}(q_{1},j_{2})g_{0}(q_{2}+k , j_{3 } ) g_{0}(q_{2},j_{4}),\\ & & \parbox{13mm}{\includegraphics[scale=1.0]{fig17.eps}}\qquad = \frac{(n+2)}{36 } \overset{\infty}{\underset{j_{1},j_{2 } , j_{3 } , j_{4 } = 0}{\sum } } \tilde{s}_{0 j_{1 } j_{3 } j_{4 } } \tilde{s}_{0 j_{2 } j_{3 } j_{4 } } \hat{s}_{j_{1 } j_{2 } j } \int d^{d-1}q_{1 } d^{d-1}q_{2 } g_{0}(q_{1}-k , j_{1 } ) \nonumber\\ & & g_{0}(q_{1},j_{2})g_{0}(q_{1}-q_{2}+k_{3},j_{3 } ) g_{0}(q_{2},j_{4}).\end{aligned}\ ] ] working out the details encoded in the finite size tensors , the reader can check that [ 20 ] @xmath108{fig16.eps } } \qquad = \frac{(n+2)^{2}}{108 } \bigl[i_{2}^{2}(k;0 ; \tilde{\sigma } , \mu ) + 2 i_{2}(k;0 ; \tilde{\sigma } , \mu ) \tilde{i}_{2}(k;0,0 ; \tilde{\sigma } , \mu ) ) \nonumber\\ & & + \overset{\infty}{\underset{l=-\infty}{\sum } } \tilde{i}_{2}^{2}(k;l , l ; \tilde{\sigma } , \mu)\bigr ] , \\ & & \parbox{10mm}{\includegraphics[scale=1.0]{fig17.eps } } \qquad = \frac{(n+2)}{36 } \bigl\{i_{4}(k , k_{3};0,0 ; \tilde{\sigma } , \mu ) + \tilde{i}_{4}(k , k_{3};0,0,0;\tilde{\sigma},\mu ) \nonumber\\ & & + 2 \hat{i}_{4}(k , k_{3};0,0,0 ; \tilde{\sigma } , \mu ) \bigr\}.\end{aligned}\ ] ] this concludes our discussion to getting all integrals required in the computation of critical exponents .
the main difference with respect with other momentum space approaches for @xmath11 are the multiplicities of some @xmath7 primitively divergent vertex parts @xcite .
we turn now our attention to the renormalization of the vertex parts in the massive and massless regimes .
our succint exposition of these topics herein should be complemented with the detailed account given in the unified approach involving @xmath11 and @xmath10 .
let us focus on the integrals belonging to the one - loop diagram of the four - point function at zero external momenta .
they will determine under what conditions the perturbative expansion is well - defined .
generically , two types of potential divergences are expected : the traditional ones realized as dimensional poles in @xmath109 and those directly connected with the lenght @xmath0 .
the latter appear as finite size corrections .
taking certain limits to @xmath0 the corrections either are well - behaved , thus validating perturbation theory , or their contribution in the integral blow up leading to a breakdown of the @xmath16-expansion .
their solutions are given in appendix a. the combination appearing in the diagram yields : [ 21 ] @xmath110 , \label{21a}\\ & & \tilde{\zeta}_{0}(\tilde{r } ) = \frac{1}{2 } \bigl[f_{\frac{1}{2 } } ( 0 , \tilde{r}^{-1 } ) + \tilde{r}\bigr ] , \label{21b}\\ & & \tilde{r}=\frac{\tilde{\sigma}}{\mu},\label{21c}\\ & & f_{\alpha}(a , b)= 4\overset{\infty}{\underset{m=1}{\sum } } cos(2\pi ma ) \bigl(\frac{\pi m}{b}\bigr)^{\alpha -\frac{1}{2}}k_{\alpha -\frac{1}{2}}(2 \pi mb),\label{21d}\end{aligned}\ ] ] where @xmath111 is the modified bessel function of second kind . here
@xmath112 corresponds to the finite size correction .
the limit @xmath1 of this function was already studied in @xcite in the cases of @xmath8 and @xmath11 . since @xmath113 in this limit ,
we collect together both outcomes to determine that @xmath114 and correctly reduces to the bulk result .
consider the @xmath15 limit .
the functional form of the finite size correction is exceedingly simpler here : at zero external quasi - momenta we do not need to devise any `` normal ordering '' and we obtain @xmath115 which is divergent , since @xmath116 , therefore invalidating @xmath16-expansion perturbative results .
the dominant divergence is that linear in @xmath117 and is responsible for the appearance of surface effects .
the origin of this term comes not only from the translation invariant symmetry breaking piece from the integral @xmath118 , but also from the `` diagonal '' integral @xmath119 @xcite .
they have the same sign and add up to produce this behavior . for the sake of comparison , for @xmath8 the coefficient of the linear term in @xmath116
is half the value of that from @xmath11 , whereas in @xmath10 the diagonal contribution linear in @xmath117 coming from @xmath119 is exactly canceled by the contribution of the non - diagonal translational symmetry breaking integral @xmath118 .
that is why @xmath10 has the same logarithmic behavior in the limit @xmath120 as in @xmath9 . from this simple analysis
we see the reason why @xmath11 rules the bulk - surface transition : the coefficient of the linear divergence when @xmath15 is twice as bigger compared with the one from @xmath8 , preventing the latter to take over in the aforementioned structural phase transition .
nevertheless , this behavior is in exact agreement with the unified approach for @xmath11 when normal ordering of the finite size correction is taken into account .
this procedure had nothing to do with physical reasons .
as pointed out before , had we treated @xmath8 and @xmath9 renormalized with nonvanishing external quasi - momenta , we would have gotten to the same regularization .
the physical results are equivalent independently of our choice to define the theory either with zero or nonzero external quasi - momenta . in the normalization conditions with null external quasi - momenta
we will use the minimal set of diagrams with the three - loop level bare mass .
the conventions utilized in ref .
@xcite will be useful to our purposes .
the primitively divergent vertex parts are sufficient to renormalize all those which are renormalizable multiplicatively . a generic vertex part of @xmath121 `` external legs '' and @xmath24 insertions of composite operators with all external quasi - momenta equal to zero ,
are renormalized multiplicatively according to the rule ( @xmath122 ) : @xmath123 in the above equation @xmath124 and @xmath125 are the normalization functions , @xmath126 , where @xmath127 is the `` lattice constant '' ) is the `` cutoff '' that shall be suppressed from now on : we are going to use dimensional regularization in the remainder of the discussion .
although not obvious , the cases under consideration are restricted to @xmath128 , @xmath129 and our condensed notation actually means _ that all external null quasi - momenta are included in the argument of the vertex parts _ , namely @xmath130 .
the notation @xmath131 will be adopted henceforth .
the normalization conditions for the primitively divergent bare vertex parts are defined by : [ 24 ] @xmath132 where @xmath133 is the renormalized mass and @xmath134 is the dimensionful coupling constant .
it is convenient to write dimensionful coupling constants in terms of dimensionless ones .
the dimensionful bare coupling constant is written as @xmath135 , whereas the renormalized one is defined by @xmath136 , where @xmath137 and @xmath138 are the dimensionless bare and renormalized coupling constants , respectively . when written in terms of the dimensionless
renormalized coupling constant , the normalization function @xmath124 up to the desired order as a power series in @xmath138 has the form @xmath139 .
we prefer to define the function @xmath140 , and write the bare dimensionless coupling constant as @xmath141 $ ] .
the diagrammatic expansion of the derivative of the bare vertex part @xmath142 is given by ( see appendix a ) [ 25 ] @xmath143,\label{25b}\\ & & b_{3}=-\frac{(n+2)(n+8)}{648 \epsilon^{2}}\bigl[1-\frac{\epsilon}{4 } + \frac{3\epsilon}{2 } \tilde{w}_{0}(0,\tilde{r})\bigr].\label{25c}\end{aligned}\ ] ] consider eq .
( [ 24b ] ) above up to @xmath144 .
it implies that @xmath145 and @xmath146 .
we come back later to evaluate @xmath147 . for the time being
the simplest task is to determine @xmath148 and @xmath149 from eq .
( [ 24c ] ) .
the diagrammatic expansion of the @xmath150 bare vertex part up to two - loops at zero external momenta reads [ 26 ] @xmath151 , \label{26a } \\ & & a_{1 } = \frac{(n+8)}{6\epsilon}\bigl[1-\frac{\epsilon}{2 } + \epsilon \tilde{\zeta}_{0}(\tilde{r})\bigr ] , \label{26b}\\ & & a_{2}^{(1 ) } = \frac{(n^{2}+6n+20)}{36 } \bigl\{\frac{1}{\epsilon^{2}}(1- \epsilon + 2\epsilon \tilde{\zeta}_{0}(\tilde{r}))\bigr\ } , \label{26c}\\ & & a_{2}^{(2 ) } = \frac{(5n+22)}{18\epsilon^{2}}\bigl(1-\frac{\epsilon}{2 } + 2\epsilon \tilde{\zeta}_{0}(\tilde{r})\bigr ) , \label{26d}\end{aligned}\ ] ] replacing this into its normalization condition , expanding @xmath152 and using the value of @xmath153 just determined , the identification of terms of the same order in @xmath138 implies that the divergences in this bare vertex part are eliminated as long as [ 27 ] @xmath154 , \label{27a } \\ & & a_{2 } = \bigl[\frac{(n+8)}{6 \epsilon}\bigr]^{2 } \bigl[1 + 2\epsilon \tilde{\zeta}_{0}(\tilde{r})\bigr ] -\frac{(2n^{2}+41n+170)}{72\epsilon}.\label{27b}\end{aligned}\ ] ] we now compute @xmath147 from the diagrammatic expansion .
we just need the values of @xmath155 and @xmath156 replaced in the normalization condition eq.([24b ] ) . after some algebra
we find @xmath157.\ ] ] at this point we introduce the beta function @xmath158,\end{aligned}\ ] ] whose ultraviolet fixed point of the coupling constant is defined by @xmath159 .
exactly at this point the theory is scale invariant , and universal quantities like critical exponents can be computed .
we then find @xmath160 \bigl\{1+\bigl[\frac{(9n+42)}{(n+8)^{2}}+\frac{1}{2 } -\tilde{\zeta}_{0}(\tilde{r})\bigr]\epsilon \bigr\}.\ ] ] the wilson functions [ 31 ] @xmath161 , \label{31a}\\ & & \bar{\gamma}_{\phi^{2}}(u)= - \beta(u ) \bigl(\frac{\partial\ln \bar{z}_{\phi^{2}}}{\partial u}\bigr ) = \epsilon u [ c_{1 } + ( 2 c_{2 } - c_{1}^{2 } - a_{1 } c_{1})u],\label{31b}\end{aligned}\ ] ] are related to the critical exponents @xmath162 and @xmath163 .
the first one is obtained up to three - loop order from the relation @xmath164\bigr\}.\ ] ] the diagramatic expansion of the bare vertex part @xmath51 at zero external momenta is given by [ 33 ] @xmath165,\label{33b}\\ & & c_{2}^{(1 ) } = \frac{(n+2)^{2}}{36\epsilon^{2}}\bigl[1- \epsilon + 2 \epsilon \tilde{\zeta}_{0}(\tilde{r})\bigr],\label{33c}\\ & & c_{2}^{(2 ) } = \frac{(n+2)}{12 \epsilon^{2 } } \biggl[1-\frac{\epsilon}{2 } + 2 \epsilon \tilde{\zeta}_{0}(\tilde{r})\biggr].\label{33d}\end{aligned}\ ] ] following the same reasoning using the normalization condition for the composite field along with the results from appendix a , @xmath166 and @xmath167 can be obtained and expressed in the form [ 34 ] @xmath168,\label{34a}\\ & & c_{2 } = \frac{(n+2)(n+5)}{36 \epsilon^{2 } } - \frac{(n+2)(2n+13)}{72\epsilon } + \frac{(n^{2}+7n+10)}{18 \epsilon } \tilde{\zeta}_{0}(\tilde{r } ) .
\label{34b}\end{aligned}\ ] ] the exponent @xmath163 can be obtained using the identity @xmath169 .
therefore , up to two - loop order the critical index is given by @xmath170 the results for the exponents are identical to those from the bulk using the massive framework .
plus , the agreement of the zero quasi - momenta renormalization approach to the primitively divergent vertex part for @xmath11 with the unified approach using nonzero external quasi - momenta in the renormalization algorithm for @xmath11 corroborates that the normal ordering procedure utilized in the finite size correction in the latter is consistent , since the physical results are the same in both methods even though the intermediary results have quite a different structure .
indeed , the correction disappears `` miraculously '' in the exponents .
moreover , in the limit @xmath15 the finite size correction diverges linearly in the limit @xmath120 in the normal ordered expression of the unified approach as well as in the present approach with null quasi - momenta ( which does not require any normal ordering prescription ) .
this momentum space treatment is a reliable guide to unravel simple properties of finite critical systems . in order to have at hand a simple birds - eye view of such systems ,
it is time to tackle the problem from the perspective of massless fields .
the one - particle irreducible @xmath171 ) formulation for finite size systems has no ambiguity , provided the critical system is kept away from the dimensional crossover region . in general , for finite values of @xmath0 the `` scaling variable '' @xmath172 in the massless regime ( @xmath173 ) is governed by the limit @xmath174 .
nonetheless , this limit of the scaling variable is not sufficient to invalidate @xmath16-expansion results . indeed , in ref .
@xcite it was recognized that @xmath9 and @xmath10 have the same critical exponents as the bulk system using arguments based purely on green s functions results when the massless limit @xmath174 is taken .
however , it was widely believed that in this limit the @xmath16-expansion results for the critical exponents from @xmath8 and @xmath11 were invalid .
this belief proved incorrect .
a direct update on the theory of finite size effects has begun with the consistent evaluation of critical exponents utilizing @xmath7 vertex functions with massless descriptions for @xmath8 and @xmath11 in refs .
we now examine the massless one - loop integrals of the four - point vertex part in an analogous treatment presented in the massive theory .
the criterion for the breakdown of perturbation theory results should be unveiled within this picture . from the results in appendix b for massless integrals we find [ 36 ] @xmath175\big|_{sp } = \frac{\kappa^{-\epsilon}}{\epsilon } \bigl[1 + \frac{\epsilon}{2 } + \epsilon \hat{\zeta}_{0}(\hat{r } ) \bigr ] , \label{36a}\\ & & \hat{\zeta}_{0}(\hat{r } ) = \frac{1}{2 } \bigl[\int_{0}^{1 } f_{\frac{1}{2 } } ( 0 , \hat{r}^{-1}\sqrt{x(1-x ) } ) + \pi \hat{r}\bigr ] , \label{36b}\\ & & \hat{r}=\frac{\tilde{\sigma}}{\kappa},\label{36c}\end{aligned}\ ] ] where @xmath176 is defined by @xmath177 .
the finite size correction @xmath178 corresponds to the massless theory where @xmath179 is the typical external momenta scale . the first term of the finite size correction in the limit @xmath1 was already shown to be zero for @xmath8 ( analogous to @xmath11 here ) and @xmath9 in ref .
@xcite ( see eqs .
( 53)-(59 ) by setting @xmath180 and @xmath181 therein ) . as @xmath182 in this limit ,
@xmath183 consistent with the same limit in the massive formulation .
consider the @xmath15 limit .
just as worked out in the massive theory , at zero external quasi - momenta we do not need to devise any `` normal ordering '' and we obtain @xmath184 which is divergent , since @xmath185 , therefore invalidating @xmath16-expansion perturbative results .
except for a bigger coefficient in the dominant divergence ( linear in @xmath186 ) in comparison with the massive theory , the behaviors in both limits are entirely consistent with those coming from the massive theory .
surface effects coming from the linear divergence either in the massive or massless theories have the same origin : half of the correction comes from the diagonal integral @xmath119 and half come from the translation invariant symmetry breaking contribution coming from the integral @xmath118 .
the @xmath8 contribution associated with the integral @xmath119 has half the coefficient of the neumann linear divergence in the limit @xmath187 .
the mechanism of dominance of @xmath11 in the bulk - surface transition ( instead of the dominance of @xmath8 condition in that structural phase transition ) is consistent with the analysis performed in the massive case .
the coefficient of the linear divergence in the massless case is approximately three times bigger than its analogue in the massive theory , since the fluctuations are wilder at the transition temperature , justifying the enhancement of surface effects at the critical point as expected .
needless to say , the unified approach for @xmath11 when normal ordering of the finite size correction is taken into account shows exact agreement within the same physical limits in comparison with the simpler approach described in the present work .
this proves that renormalization using either zero or nonzero external quasi - momenta for @xmath11 is immaterial , so long as the one - loop contribution of the four - point vertex part is concerned .
let us conclude the proof by performing the renormalization and evaluation of critical exponents at higher - loop orders . by borrowing the arguments utilized in the massive setting , multiplicative renormalization is transliterated in the form ( @xmath122 ) :
@xmath188 dimensional regularization will be employed herafter and we forget about the cutoff @xmath189 henceforth .
just as before we adopt the convention @xmath130 and @xmath131 .
the normalization conditions for the primitively divergent bare vertex parts are defined by : [ 39 ] @xmath190 where @xmath191 is defined by @xmath192 , @xmath193 and @xmath194 ( see @xcite for more details ) . here , we write the dimensionful bare and renormalized coupling constant in terms of their dimensionless counterparts as @xmath195 and @xmath196 , respectively . we write power series in @xmath138 of the renormalization functions , namely , @xmath139 , @xmath140 , and the bare dimensionless coupling constant as @xmath141
$ ] . using the results listed in appendix b ,
we conclude that the diagrammatic expansion of the derivative of the bare vertex part @xmath142 reads [ 40 ] @xmath197,\label{40b}\\ & & b_{3}=-\frac{(n+2)(n+8)}{648 \epsilon^{2}}\bigl[1 + 2\epsilon - 3 \epsilon \hat{w}(\kappa,\hat{r})\bigr].\label{40c}\end{aligned}\ ] ] substitution of the last equation into eq .
( [ 39b ] ) above up to @xmath144 leads to @xmath145 and @xmath198 $ ] .
we save the calculation of @xmath147 for later . from eq .
( [ 39c ] ) we shall compute @xmath148 and @xmath149 . at the symmetry point ,
the @xmath150 bare vertex part up to two - loops at zero external quasi - momenta is very similar to the massive vertex function , but with different coefficients .
we have [ 41 ] @xmath199 , \label{41a } \\ & & a_{1 } = \frac{(n+8)}{6\epsilon}\bigl[1+\frac{\epsilon}{2 } + \epsilon \hat{\zeta}_{0}(\hat{r})\bigr ] , \label{41b}\\ & & a_{2}^{(1 ) } = \frac{(n^{2}+6n+20)}{36 } \bigl\{\frac{1}{\epsilon^{2}}(1 + \epsilon + 2\epsilon \hat{\zeta}_{0}(\hat{r}))\bigr\ } , \label{41c}\\ & & a_{2}^{(2 ) } = \frac{(5n+22)}{18\epsilon^{2}}\bigl(1+\frac{3 \epsilon}{2 } + 2\epsilon \hat{\zeta}_{0}(\hat{r})\bigr ) , \label{41d}\end{aligned}\ ] ] the renormalization algorithm follows the same flow of reasoning in the massless theory . using eq .
( [ 41 ] ) into eq .
( [ 39c ] ) yields directly [ 42 ] @xmath200 , \label{42a } \\ & & a_{2 } = \bigl[\frac{(n+8)}{6 \epsilon}\bigr]^{2 } \bigl[1 + 2\epsilon \hat{\zeta}_{0}(\hat{r})\bigr ] + \frac{(2n^{2}+23n+86)}{72\epsilon}.\label{42b}\end{aligned}\ ] ] a direct calculation of @xmath147 is now possible from the diagrammatic expansion as explained in the massive case .
it is not difficult to show that @xmath201.\ ] ] the beta function @xmath202,\end{aligned}\ ] ] now has the infrared fixed point in parameter space that is independent of the initial values of the bare parameters .
the infrared fixed point of the coupling constant is defined by @xmath203 .
recall that universal quantities like critical exponents can be computed after the determination of @xmath204 , that is given by @xmath205 \bigl\{1+\bigl[\frac{(9n+42)}{(n+8)^{2}}-\frac{1}{2 } -\hat{\zeta}_{0}(\hat{r})\bigr]\epsilon \bigr\}.\ ] ] the wilson functions [ 46 ] @xmath161 , \label{46a}\\ & & \bar{\gamma}_{\phi^{2}}(u)= - \beta(u ) \bigl(\frac{\partial\ln \bar{z}_{\phi^{2}}}{\partial u}\bigr ) = \epsilon u [ c_{1 } + ( 2 c_{2 } - c_{1}^{2 } - a_{1 } c_{1})u],\label{46b}\end{aligned}\ ] ] at the infrared fixed point are related to the critical exponents @xmath162 and @xmath163 .
we obtain @xmath206\bigr\}.\ ] ] the diagramatic expansion of the bare vertex part @xmath51 at zero external momenta can be written in the form [ 48 ] @xmath207,\label{48b}\\ & & c_{2}^{(1 ) } = \frac{(n+2)^{2}}{36\epsilon^{2}}\bigl[1 + \epsilon + 2 \epsilon \hat{\zeta}_{0}(\hat{r})\bigr],\label{48c}\\ & & c_{2}^{(2 ) } = \frac{(n+2)}{12 \epsilon^{2 } } \biggl[1 + \frac{3 \epsilon}{2 } + 2 \epsilon \hat{\zeta}_{0}(\hat{r})\biggr].\label{48d}\end{aligned}\ ] ] by employing the normalization condition for the composite field in conjumination with the results from appendix b , @xmath166 and @xmath167 can be obtained .
we find [ 49 ] @xmath208,\label{49a}\\ & & c_{2 } = \frac{(n+2)(n+5)}{36 \epsilon^{2 } } + \frac{(n+2)(2n+7)}{72\epsilon } + \frac{(n^{2}+7n+10)}{18 \epsilon } \hat{\zeta}_{0}(\hat{r } ) .
\label{49b}\end{aligned}\ ] ] the identity @xmath209 leads to the result @xmath210 the whole process confirms the universal character of the exponents , since they do not depend on the renormalization scheme either using massless or massive fields .
although the massless integrals are different and have a distinct typical scale ( the external momenta scale @xmath179 , whereas in the massive case the proper scale is the bare mass @xmath61 at three - loop level ) the critical exponents independ of the typical scale where the integrals are computed .
the bottom line is : for finite @xmath0 and @xmath173 at the critical point where the field theory becomes massless ( @xmath174 ) , the physical results are well - defined and perturbation theory is really valid in the computation of critical indices for @xmath9 , @xmath8 , @xmath10 and @xmath11 , in stark contrast to earlier beliefs regarding finite size effects .
in this paper we discussed the problem of renormalization of fields in a layered geometry in momentum space , confined in a finite size slab of thickness @xmath0 composed of @xmath12-dimensional parallel plates of infinite extent .
we attacked the problem using null external quasi - momenta to begin with .
the critical exponents obtained with neuman boundary conditions applied in the limiting surfaces at @xmath13 coincide with those from the infinite system . considering purely finite size effects here , the intermediate results strengthen our understanding why the dominance of @xmath11 in the description of the bulk - surface transition takes place .
the present approach is simpler for @xmath11 in comparison with the recent unified approach proposed in ref .
@xcite : normal ordering of the finite size correction to the one - loop four point vertex part diagram is not necessary .
the critical exponents for @xmath11 just obtained in this work are equal to those in the unified framework , to the @xmath8 and @xmath9 exponents from ref .
@xcite : they are the same as those from the bulk system .
furthermore , the dimensional crossover criteria also coincide in both formalisms for @xmath11 , providing a firmer ground for the dominance of the onset of surface effects purely out of volume fields in that case .
our treatment of massless fields in the the region @xmath174 for finite values of @xmath0 and away from the dimensional crossover region has a well - defined perturbation theory : not only the exponents are the same , but the dimensional crossover criteria are coincident either in the massless or in the massive setting , with minor modifications .
the work just presented when collected with those from refs .
@xcite represents a direct modern update in the theory of finite - systems : either massive or massless renormalizations to this problem in momentum space might improve our knowledge of more difficult effects appearing simultanenously with finite size ones .
the study of critical amplitude ratios of several thermodynamic potentials require ingredients from the massive and massless framework _
simultaneously_. with the modern approach to finite size effects on systems close to their critical points , we could update old results @xcite with the techniques described in our recent series of papers on finite - systems .
moreover , the computation of these amplitude ratios is lacking for the @xmath9 case , and it would be a nice exercise to see how finite size effects show up for @xmath9 in these ratios .
thin films @xcite could then be studied entirely using the modern techniques in momentum space .
finite - systems with lifshitz bulk critical behavior could be studied within this new paradigm
. this would allow the computation of universal quantities like critical exponents @xcite and amplitude ratios @xcite .
ferroelectric materials exhibit lifshitz bulk behavior .
ferroeletric nanomaterials thin films might exhibit also ferromagnetism due to strain in the planes caused by the growth method and the substrate .
the theoretical understanding of finite size effects in lifshitz critical behavior could improve the understanding of more exotic boundary conditions , for instance , in ferroelectric semiconductor thin films @xcite .
another lifshitz system is realized in multiferroic materials , but are more dificult to explain due to the ferrolectric and ferromagnetic coupling .
this is a hot topic in the whole set of effects present in engineering nanomaterials @xcite .
we hope many other real physical finite - systems pertaining to other lifshitz criticalities could have the starting point to explain their exotic properties by commencing with a similar simple description in the future along the same lines of the finite effects described so far .
after that , the study of surface effects by either introducing external surface fields in the lagrangian density or by examining more carefully the dimensional crossover region in the present approach could shed new light in the mixing of finite size , interface and surface effects . perhaps including both ingredients could give a clue on which effect is dominant .
therefore , this strategy has the potential to provide a better insight on the bulk - surface transition . beyond static aspects of critical phenomena , dynamic phase transitions in thin ferromagnetic films
were also studied using monte carlo simulations for classical heisenberg spin systems with certain classes of anisotropy and different boundary conditions for the order parameter @xcite .
it might be interesting to study dynamical finite size states using field - theoretic methods applied to bulk systems @xcite .
another promising application for the field - theoretical techniques which have been developed for finite size systems concluded with the content of the present paper is the study of quantum finite - systems .
recent fundamentally simple ideas in determining a universal order parameter of quantum phase transitions utilizing finite size arguments @xcite have been put forth . in addition , studies on the modification of the topological phase transition mechanism induced by the finite size in interface - engineered @xmath211 thin films , using theoretical and experimental methods @xcite indicates that very complex behaviors in the quantum character of finite size effects are waiting to be described with rigorous quantum field theory methods .
it would be desirable to put together many theoretical tools to attack problems like these displaying simultaneously many non - trivial effects @xcite .
the conclusion of the set of papers dealing with finite size effects employing only bulk fields in momentum space can pave the way to a complete description of more intrincate phenomena associated to the miniaturization intrinsic to nanodevices .
we still have to envisage an efficient perturbation theory in the dimensional crossover region , perhaps using the smallness of @xmath0 dominating the poles in @xmath16 . in this approach
, we would have to consider all the integrals regular in @xmath16 since they depend on @xmath0 , to determine their asymptotic ( singular ) @xmath0-dependence and neglect all dimensional poles .
the logarithmic contributions in @xmath0 could be taken care of by postulating an inverse `` finite size cutoff '' @xmath212 dominating the dimensional poles in @xmath16 , or using another dimensional regularization where @xmath213 ( massive theory ) would be represented as dimensional poles in @xmath214 , whereas the linear divergence can be thought of as a single pole in @xmath215 in the same way quadratic ultraviolet divergences in standard field theory can be represented as a double pole in @xmath216 , at least in massive theories @xcite . the next problem would be to devise a renormalization group for small @xmath0 or alternatively using @xmath61 or @xmath179 as the running value since the @xmath0 dependence in the integral always come as the products @xmath217 or @xmath218 .
the resulting treatment is not simple to guess in a glance and we live this and the above topics to be investigated in the future .
mvss would like to thank financial support from capes , grant number 76640 and cnpq , grant number 141912/2012 - 0 .
jbsj thanks cnpq for financial support , grant number 142220/2007 - 8 .
mml acknowledges cnpq , grant number 232352/2014 - 3 for partial financial support .
since the integrals either massive or massless appearing in the present work were already discussed in the unified work from ref .
@xcite at nonnull external quasi - momenta , we shall simply list their solution in a suitable form to our purposes in the computation of the critical exponents .
we list firstly the results of the integrals contributing to the four - point vertex part . at one - loop level
the structure is very simple . in the two - loop graphs of the four - point vertex part we are interested at @xmath219 and more singular terms .
we note that the contributions of the four - point vertex part corresponding to the term @xmath220 is regular in @xmath16 .
it comes from the trivial two - loop graph eq .
( [ 13 ] ) and can be neglected .
similar remarks apply to the two - loop integral @xmath221 which is also regular in @xmath16 and are not going to be listed below .
we should take advantage of the facts observed here for the four - point vertex part and substitute directly in the discussion of the vertex function @xmath51 .
the expression of every integral was previously derived in the main text .
the results for the massive integrals corresponding to the four - point vertex part at zero external momenta and quasi - momenta are : [ a1 ] @xmath222 , \label{a1a}\\ & & \tilde{i}_{2}(0;0,0;\tilde{\sigma } , \mu ) = \mu^{-\epsilon } \bigl(\frac{\tilde{r}}{2}\bigr),\label{a1b}\\ & & i_{4}(0,0;0,0;\tilde{\sigma } , \mu)= \mu^{-2\epsilon } \bigl[\frac{1}{2 \epsilon^{2 } } \bigl(1-\frac{\epsilon}{2 } + \epsilon f_{\frac{1}{2 } } ( 0 , \tilde{r}^{-1})\bigr)\bigr ] , \label{a1c}\\ & & \tilde{i}_{4}(0;0,0,0;\tilde{\sigma } , \mu ) = \mu^{-2\epsilon } \bigl(\frac{\tilde{r}}{2 \epsilon}\bigr).\label{a1d}\end{aligned}\ ] ] we now write down the results for the two - point vertex function integrals associated with two- and three - loop contributions .
let us emphasize that the three - loop integrals @xmath223 and @xmath224 are regular in @xmath16 and their computation , therefore , will not concern us herein .
hence the relevant integrals of the two - point vertex parts which are going to be used in the normalization conditions as discussed in the text are actually their derivatives computed at zero external momenta .
if we perform a rescaling in all momenta in the form @xmath225 , the derivatives with respect with the new variable @xmath226 is related to the derivative with respect to @xmath93 as @xmath227 .
using this fact and expressing everything in terms of the scaled quantities , the two - loop integrals have the following solutions : [ a2 ] @xmath228 \bigl(1-\frac{\epsilon}{4 } + \epsilon w_{0}(\tilde{r})\bigr ) , \label{a2a}\\ & & w_{0}(\tilde{r})= g_{0}(\tilde{r } ) + h_{0}(\tilde{r } ) - 4 f_{0}^{'}(\tilde{r}),\label{a2b}\\ & & g_{0}(\tilde{r})= -\frac{1}{2 } - 2\int_{0}^{1 } \int_{0}^{1}dx dy ( 1-y)ln\bigl[(1-y)\tilde{r}^{-2 } + \frac{y\tilde{r}^{-2}}{x(1-x)}\bigr],\label{a2c}\\ & & h_{0}(\tilde{r } ) = 2 \int_{0}^{1 } \int_{0}^{1}dx dy ( 1-y)f_{\frac{1}{2}}\left(0,\sqrt{(1-y)\tilde{r}^{-2 } + \frac{y\tilde{r}^{-2}}{x(1-x)}}\right),\label{a2d}\\ & & f_{0}^{'}(\tilde{r } ) \equiv \frac { \partial f_{0,1}(k';i;\tilde{r } ) } { \partial k^ { ' 2 } } \bigl|_{(k^{'2}=0 , i=0)},\label{a2e}\\ & & f_{\alpha}(k',i;\tilde{r } ) \equiv \tilde{r}^{-2 \alpha}\int_{0}^{1 } dx f_{\frac{1}{2 } + \alpha}(xi,\tilde{r}^{-1 } \sqrt{x(1-x)(k^{'2 } + \tilde{r}^{2}i^{2})+1 } ) , \label{a2f}\\ & & f_{\alpha , \beta}(k',i,\tilde{r } ) \equiv \frac{1}{s_{d } } \tilde{r } \overset{\infty}{\underset{j=-\infty}{\sum } } \int d^{d-1}q ' \frac{f_{\alpha}(q'+k',j+i,\tilde{r})}{[q^{'2}+ \tilde{r}^{2}j^{2 } + 1]^{\beta}},\label{a2g}\\ & & \tilde{i}_{3}^ { ' } ( 0;0;\tilde{\sigma } , \mu ) \equiv \frac{\partial \tilde{i}_{3}(k';0,0;r)}{\mu^{2 } \partial k^{'2}}\bigl|_{k^{'2}=0 } = - \mu^{- 2 \epsilon}\bigl[\frac{\tilde{r}}{4 } \int_{0}^{1}dx \int_{0}^{1 } dy ( 1-y)\bigl ( 1-y + \frac{y}{x(1-x)}\bigr)^{-\frac{1}{2}}\nonumber\\ & & -\frac{1}{2 } \mathcal{f}_{0}^{'}(0;0,0;\tilde{r})\bigr ] .
\label{a2h}\end{aligned}\ ] ] in last equation we employed the definitions [ a3 ] @xmath229^{\beta}},\\ & & \mathcal{f}_{\alpha}^{'}(0;0,0;\tilde{r } ) \equiv \frac{\partial \mathcal{f}_{\alpha , 1}(k';0,0;\tilde{r})}{\partial k^ { ' 2}}\bigl|_{k^ { ' 2}=0}.\end{aligned}\ ] ] the same comments about the scaling of the momenta are in order here .
therefore , one finds [ a4 ] @xmath230 \bigl(1-\frac{\epsilon}{4 } + \frac{3 \epsilon}{2 } w_{0}(\tilde{r})\bigr ) , \label{a4a}\\ & & \tilde{i}_{5}^ { ' } ( 0;0,0,0;\tilde{\sigma } , \mu ) = \mu^{-3 \epsilon } \bigl(-\frac{1}{2 \epsilon}\bigr ) \bigl[\tilde{r } \int_{0}^{1}dx \int_{0}^{1 } dy ( 1-y)\bigl ( 1-y + \frac{y}{x(1-x)}\bigr)^{-\frac{1}{2}}\nonumber\\ & & - 2 \mathcal{f}_{0}^{'}(0;0,0;\tilde{r})\bigr].\label{a4b}\end{aligned}\ ] ] in order to make the connection with the content of the text we define the quantity @xmath231 this amount is going to be used in the renormalization of the two - point vertex .
however , it is demonstrated in the main text that it does not show up in the expression for the critical exponents .
in the previous appendix , most of the notation was already fixed . except for minor modifications , which will be briefly highlighted here ,
we follow the same flow of reasoning .
for instance , @xmath232 , where @xmath179 is the typical external momenta scale .
the external momenta @xmath233 is defined by @xmath234 .
recall that in the total contribution of the four - point vertex function other permutations of the external momenta must be included .
the four - point vertex part contributions relevant to our purposes are represented by the one- and two - loop massless integrals computed at the symmetry point defined in the main text . at nonvanishing external momenta at the symmetry point and zero external quasi - momenta ,
their results are written down below directly , namely [ b1 ] @xmath235^{\frac{1}{2}})\bigr)\bigr ] , \label{b1a}\\ & & \tilde{i}_{2}(p;0,0;\tilde{\sigma})\bigl|_{sp } = \kappa^{-\epsilon } \bigl(\frac { \pi \hat{r}}{2}\bigr),\label{b1b}\\ & & i_{4}(p , k_{3};0,0;\tilde{\sigma})\bigl|_{sp}=
\kappa^{-2\epsilon } \bigl[\frac{1}{2 \epsilon^{2 } } \bigl(1 + \frac{3 \epsilon}{2 } + \epsilon \int_{0}^{1 } dx f_{\frac{1}{2 } } ( 0 , \hat{r}^{-1}[x(1-x)]^{\frac{1}{2}})\bigr)\bigr ] , \label{b1c}\\ & & \tilde{i}_{4}(p , k_{3};0,0,0;\tilde{\sigma})\bigl|_{sp } = \kappa^{-2\epsilon } \bigl(\frac{\hat{r}}{2 \epsilon}\bigr).\label{b1d}\end{aligned}\ ] ] let us focus on the two - point vertex part diagrams .
just as in the massive case , provided the system is away from the `` dimensional crossover '' region where the smallness due to @xmath0 dominates over the dimensional poles in @xmath16 , we do not have to worry about regular contributions in @xmath16 in the the three - loop graph .
thus , we restrict ourselves to the presentation of the the results for @xmath236 and @xmath237 .
[ b2 ] @xmath240 \bigl(1 + \frac{5 \epsilon}{4 } - 2 \epsilon w(\hat{r})\bigr),\label{b2a}\\ & & w(\hat{r})= 2 \hat{f}_{0}^{'}(\hat{r } ) - \hat{\bar{f}}_{0}(\hat{r } ) , \label{b2b}\\ & & \hat{f}_{\alpha}(k , i,\tilde{\sigma})= \frac{\tilde{\sigma}^{-2 \alpha}}{s_{d } } \int_{0}^{1 } dx f_{\frac{1}{2 } + \alpha}\bigl(0 , \bigl[\bigl(\frac{k^{2}}{\tilde{\sigma}^{2 } } + i^{2}\bigr)x(1-x)\bigr]^{\frac{1}{2}}\bigr ) , \label{b2c}\\ & & \hat{f}_{\alpha,\beta}(k , i=0,\tilde{\sigma } ) = \frac{\tilde{\sigma}}{s_{d}}\overset{\infty}{\underset{j=-\infty}{\sum } } \int d^{d-1}q \frac{\hat{f}_{\alpha}(q+k , j,\tilde{\sigma})}{[q^{2}+ \tilde{\sigma}^{2}j^{2}]^{\beta}},\label{b2d}\\ & & \hat{f}_{0}^{'}(\hat{r } ) = \frac{\partial \hat{f}_{0,1}(k , i=0,\tilde{\sigma})}{\partial k^{2}}\bigl|_{k^{2 } = \kappa^{2}},\label{b2e}\\ & & \hat{\bar{f}}_{0}(\hat{r})= \int_{0}^{1 } dx ( 1-x ) f_{\frac{1}{2}}(0 , \sqrt{x(1-x)}\hat{r}^{-1 } ) , \label{b2f}\\ & & \frac{\partial i_{3}(k;0,\tilde{\sigma})}{\partial k^{2}}\bigl|_{k^{2}= \kappa^{2}}= - \kappa^{-2 \epsilon } \bigl[\frac{\hat{r}}{4 } \hat{h}_{0 } -\frac{\kappa^{2 \epsilon}}{2 } \hat{\mathcal{f}}_{0}^{'}(\hat{r})\bigr],\label{b2g}\\ & & \hat{h}_{0}= \int_{0}^{1 } dx ( 1-x)[x(1-x)]^{-\frac{1}{2 } } , \label{b2h}\\ & & \hat{\mathcal{f}}_{\alpha,\beta}(k;i , j;\tilde{\sigma})= \frac{\tilde{\sigma}}{s_{d } } \int \frac{d^{d-1}q \hat{f}_{\alpha}(q+k , j,\tilde{\sigma})}{[q^{2}+ \tilde{\sigma}^{2}i^{2}]^{\beta } } , \label{b2i}\\ & & \hat{\mathcal{f}}_{0}^{'}(\hat{r } ) \equiv \frac{\partial \hat{\mathcal{f}}_{0,1}^{(\tau)}(k , i=0,j=0;\tilde{\sigma } ) } { \partial k^{2}}\big|_{k^{2}=\kappa^{2}}. \label{b2j}\end{aligned}\ ] ] [ b4 ] @xmath245 , \label{b4a}\\ & & \tilde{i}_{5}^{'}(k,0,\tilde{\sigma})\bigl|_{k^{2}=\kappa^{2 } } = -\frac{\kappa^{-3 \epsilon}}{\epsilon } \bigl[\frac{\hat{r}}{2 } \hat{h}_{0 } - \hat{\mathcal{h}}_{0}\bigr].\label{b4b}\end{aligned}\ ] ] 99 fisher m. e. fisher , in _ critical phenomena , proceedings of the 51st enrico fermi summer school _ , varenna , italy , m. s. green , ed .
, academic press , new york , 1971 .
barber m. n. barber , in _ phase transitions and critical phenomena _ , edited by c. domb and j. l. lebowitz , ( academic , london , 1986 ) , vol .
10 , p. 76 .
privman v. privman , in _ finite size scaling and numerical simulations in statistical mechanics _ , edited by v. privman ( world scientific , singapore , 1990 ) , p. 1 .
gasparini f. m. gasparini , m. o. kimball , k. p mooney , and m. diaz - avila , _ rev .
phys . _ * 80 * , 1009 ( 2008 ) .
su l. sandoval and h. m. urbassek , _ nano lett . _ * 9 * , 2290 ( 2009 ) .
b c. borgs , _ nucl . phys . _
* b 384 * , 605 ( 1992 ) .
falicov l. m. falicov , d. t. pierce , s. d. bader , r. gronsky , k. b. hathaway , h. j. hopster , d. n. lambeth , s. s. p. parkin , g. prinz , m. salamon , i. k. schuller , and r. h. victora , _ j. mater .
_ * 5 * , 1299 ( 1990 ) .
jin x. w. jin , l. lu , s. b. mi , m. liu , and c. l. jia _ app .
lett . _ * 109 * , 031904 ( 2016 ) .
monsen a. monsen , j. e. boschker , f. maci , j. j. wells .
, p. nordblad , a. kent , r. mathieu , t. tybell , and e. wahlstr@xmath246 m , _ j. mag . mag .
mater . _ * 369 * , 197 ( 2014 ) .
schlom d. g. schlom , l. q. chen , x. pan , a. schmehl , and m. zuburchen , _ j. am . ceram . soc .
_ * 91 * , 2429 ( 2008 ) .
d h. w. diehl , in _ phase transitions and critical phenomena _ , edited by c. domb and j. l. lebowitz , ( academic , london , 1986 ) , vol .
10 , p. 76 .
hbbhbm m. hamedoun , k. bouslykhane , h. bakrim , a. hourmatallah , n. benzakour and r. masrour , _ j. mag . mag .
mater . _ * 301 * , 22 ( 2006 ) .
scott j. f. scott , _ nat .
_ * 6 * , 256 ( 2007 ) .
subra g. subramanyan , m. w. cole , n. x. sun , t. s. kalkur , n. m. sbrockey , g. s. tompa , x. guo , c. chen , s. p. alpay , g. a. rossetti , jr .
, k. dayal , l. q. chen , and d. g. schlom , _ j. appl .
_ * 114 * , 191301 ( 2013 ) .
kaplan r. j. kaplan , _ j. vac .
tech . _ * a 1 * , 551 ( 1983 ) .
hong s. hong , k. yang , and i. rhee , _ j. kor
. phys . soc . _
* 34 * , 254 ( 1999 ) .
ihzaz n. ihzaz , m. boudard , l. rapenne , h. roussel , s. pignard , and m. oumezzine , _ epj web of conferences _ * 29 * , 00024 ( 2012 ) .
nf1 a. m. nemirovsky and k. f. freed , _ j. phys . a : math gen . _
* 18 * , l319 ( 1985 ) .
nf2 a. m. nemirovsky and k. f. freed , _ nucl .
b _ * 270 * , [ fs16 ] , 423 ( 1986 ) .
bl j. b. silva jr . and m. m. leite , _ j. math
. phys . _ * 53 * , 043303 ( 2012 ) . sbl m. v. s. santos , j. b. silva jr . and m. m. leite , _ arxiv:1509.05793_. mockler1 b. a. scheibner , m. r. meadows , r. c. mockler , and w. j. osullivan , _ phys . rev
* 43 * , 590 ( 1979 ) .
mockler2 m. r. meadows , b. a. scheibner , r. c. mockler , and w. j. osullivan , _ phys .
lett . _ * 43 * , 592 ( 1979 ) .
jgh h. jang , m. j. grimson and c. k. hall , _ phys .
b _ * 67 * , 094411 , ( 2003 ) . cl p. r. s. carvalho and m. m. leite ,
_ j. math .
* 54 * , 093301 ( 2013 ) ; _ j. math .
phys . _ * 57 * , 119901 ( 2016 ) .
messias for a different alternative for the tree - level values of the primitively divergent vertex parts with equivalent results for the exponents , see criticalidade em tamanho finito : presena e ausncia de competio anisotrpica , ph.d .
thesis ( in portuguese ) by m. v. s. santos , df - ufpe , 2015 ( unpublished ) .
amit d. j. amit and v. martin - mayor , in _ field theory , the renormalization group and critical phenomena _ , 3rd ed .
( world scientific , singapore , 2005 ) .
lnc m. m. leite , a. m. nemirovsky and m. d. coutinho - filho , _ j. mag . mag .
mater . _ * 104 - 107 * , 181 ( 1992 ) .
lsc m. m. leite , m. sardelich and m. d. coutinho - filho , _ phys .
* 59 * , 2683 , ( 1999 ) .
f m. h. francombe , in _ physics of thin films : mechanic and dielectric properties _ , edited by m. h. francombe and j. l. vossen , ( academic , san diego , 1993 ) , p. 225 .
l1 m. m. leite , _ phys . rev .
b _ * 67 * , 104415 , ( 2003 ) ; _ phys . rev .
b _ * 80 * , 029904(e ) , ( 2009 ) .
l2 m. m. leite , _ phys .
a _ * 326 * , 281 , ( 2004 ) ; m. m. leite , _ phys .
b _ * 72 * , 224432 , ( 2005 ) .
cl1 p. r. s. carvalho and m. m. leite ,
* 324 * , 178 ( 2009 ) ; _ ann .
* 324 * , 2452 ( 2009 ) .
cl2 p. r. s. carvalho and m. m. leite , _ ann .
phys . _ * 325 * , 151 ( 2010 ) .
l3 m. m. leite , _ phys .
b _ * 61 * , 14691 ( 2000 )
. l4 m. m. leite , _ phys .
b _ * 68 * , 052408 ( 2003 ) . fl c. f. farias and m. m. leite , _ j. stat .
_ * 148 * , 972 ( 2012 ) .
sl m. i. sena jr . and m. m. leite , _ j. phys .
series _ * 574 * , 012170 ( 2015 ) .
ekm e. a. eliseev , s. v. kalinin , and a. n. morozovska , _ j. app
. phys . _ * 117 * , 034102 ( 2015 ) .
paz d. r. tilley , in _ ferroelectric thin films : synthesis and basic properties _ , edited by c. paz de araujo , j. f. scott and g. w. taylor , ( gordon and breach , amsterdam , 1996 ) , p. 11 .
lovesey s. w. lovesey , _ in condensed matter physics : dynamic correlations _ , 2nd ed .
( addison - wesley , new york,1986 ) .
szb q. q. shi , h. q. zhou and m. t. batchelor , _ sci .
rep _ * 5 * , 7673 ( 2015 ) .
sskbmo m. salehi , h shapourian , n. koirala , m. j. brahlek , j. moon and s. oh ,
_ nano lett . _
* 16 * , 5528 ( 2016 ) .
kleinert h. kleinert , and v. schulte - frohlinde , _ critical properties of @xmath45-theories _ , 1st ed .
( world scientific , singapore , 2001 ) . | _ the order parameter of a critical system defined in a layered parallel plate geometry subject to neumann boundary conditions at the limiting surfaces is studied .
we utilize a one - particle irreducible vertex parts framework in order to study the critical behavior of such a system .
the renormalized vertex parts are defined at zero external quasi - momenta , which makes the analysis particularly simple .
the distance between the boundary plates @xmath0 characterizing the finite size system direction perpendicular to the hyperplanes plays a similar role here in comparison with our recent unified treatment for neumann and dirichlet boundary conditions .
critical exponents are computed using diagrammatic expansion at least up to two - loop order and are shown to be identical to those from the bulk theory ( limit @xmath1 ) . _ |
the concept of universality is used to classify and characterise equilibrium states of matter . for example
, there are different types of order in a magnetic material separated by a second - order phase transition at which the relevant physical properties become independent of the microscopic details of the system .
this constitutes a type of universality and allows to characterise an extensive range of different phenomena in terms of just a few classes governed by the same critical properties . in view of the intensifying discussion on the dynamics of many - body systems it is a pressing question whether also far away from the thermal limit
the character of dynamical evolution can become independent of the microscopic details @xcite . for a closed system
this would imply that in approaching a critical configuration the evolution must become independent of the particular initial state the system has started from and critical slowing down in the actual time evolution is observed . as a consequence ,
different types of dynamical evolution could be distinguished by means of universality classes .
dynamical , self - organised criticality , related to pattern and order formation is a well - explored theme in classical statistical mechanics @xcite .
the growth of an order parameter in an initially non - ordered system for which the hamiltonian and the total energy let expect order to develop is commonly described in terms of defect nucleation and coarsening dynamics .
if this process occurs close to an equilibrium configuration , it is described in terms of so - called dynamical critical phenomena , in particular linear - response formulations in the vicinity of equilibrium phase transitions @xcite .
we emphasize that such descriptions potentially fail when the system is excited sufficiently far from an equilibrium configuration .
phase - ordering phenomena after domain nucleation are being studied increasingly also in quantum many - body systems , in particular in quantum gases and fluids .
these systems exhibit , at shorter wave lengths , quantum nature arising from their microscopic constituents , e.g. , cold atoms , but , on the spatial scales of interest for macroscopic ordering phenomena , can be described in terms of classical field theory .
related to this , the equilibrium temperature in these situations is sufficiently low such that dynamical critical phenomena at so - called quantum phase transitions can be studied .
systems are tuned across these transitions by means of , e.g. , coupling parameters present in the model .
described in terms of non - linear field models , the quantum many - body systems bear the possibility of defect formation @xcite . for two - component bose gases , domain structures @xcite have been observed in experiments with different hyperfine species of ultracold @xmath0rb atoms @xcite .
in addition to these , the system can develop angular - momentum carrying vortices and skyrmions @xcite .
closely related dynamics appears in spin-1 gases , see refs .
@xcite . for discussions of the dynamics near quantum phase transitions see , e.g. , refs .
@xcite .
universality and scaling near second - order phase transitions is commonly described by means of the renormalization group ( rg ) .
critical phenomena arise as fixed points at which the rg flow of effective couplings of a particular model , vanishes . varying the microscopic parameters of a model
, their subsequent rg flow can pass close by such a fixed point and become critically slowed down . as a consequence , a wide range of momenta governs the macroscopic correlations of the system , and scaling behaviour , i.e. , power laws in momentum appear . here
we study the equilibration dynamics of a two - component bose gas after a quench to a far - from - equilibrium state .
we evaluate correlation functions in view of a corresponding universal evolution in real time .
our results show that long before the system reaches its final equilibrium state it can approach a non - thermal fixed point @xcite at which correlations exhibit characteristic infrared scaling laws .
non - thermal fixed points can be derived as stationary scaling solutions of real - time renormalisation - group flows @xcite . in the system studied here they are found to exist irrespective of whether the two spin components are miscible or the system is in the immiscible regime where the components separate in the low - energy limit . the type of universal dynamics near a non - thermal fixed point in which we are interested here , has been studied in detail for a single - component bose gas , in one , two , and three spatial dimensions @xcite . according to this
, the approach of the fixed point is characterised by the creation and dynamical evolution of ( quasi-)topological defects @xcite .
it was shown that equilibration trajectories passing closely by the fixed point lead far away from equilibrium configurations @xcite and experience critical slowing down . in the vicinity of the fixed point ,
power - law and logarithmic decay behaviour in time of the density of vortices was found , corresponding to a growth of the characteristic scale of the pattern @xcite .
the above studies reveal that the universal dynamics near a non - thermal fixed point is in many respects different from the dynamical critical phenomena known in the vicinity of equilibrium fixed points @xcite .
the phenomena predicted at non - thermal fixed points are closely related to classical ( wave ) turbulence @xcite .
wave turbulence theory has been extended by field - theoretic methods to quantum systems @xcite . in the limit of long wavelengths where simple kinetic approximations used to describe weak wave turbulence fail ,
the non - perturbative methods are used to evaluate high - order correlations .
this allows to predict strong - wave - turbulence scaling laws at non - thermal stationary points of the equations of motion for the correlation functions which could be confirmed by numerical simulations .
non - thermal fixed points have been demonstrated for relativistic @xmath1-component scalar theories @xcite , relativistic fermi - bose systems @xcite , non - abelian gauge theories @xcite as well as higgs models @xcite . in this article , we present a numerical analysis of possible non - thermal - fixed - point scalings in a two - component bose gas in two spatial dimensions tuned to either side of the miscible - immiscible transition .
we prepare unstable initial conditions to drive the system away from equilibrium and follow the ensuing evolution towards thermalisation . extending upon the results presented in ref .
@xcite we make use of quantum - turbulence and spin decompositions to trace back the emerging scaling exponents to the presence of various transient domain structures , vortices and skyrmion excitations .
similar universal features are expected in the evolution of one - dimensional systems @xcite . in sect .
[ sec : model ] we summarise the main properties of the two - component bose gas and specify our numerical methodology .
we furthermore discuss the system in the spin - fluid representation which will be used extensively throughout the paper and use this to exhibit the symmetry properties across the transition between the miscible and immiscible phases .
we finally summarise the analytical results on non - thermal fixed points relevant for the dynamics studied in the following .
[ sec : simulation ] contains our numerical results .
it is divided in detailed discussions of the evolution in the immiscible and immiscible parameter regimes of the model as well as of the dynamics at the transition point .
we present our conclusions in sect .
[ sec : summary ] .
in the following section we will present results of quasi - classical simulations of the dynamics of a dilute two - component bose gas in two spatial dimensions .
this system is described by the hamiltonian density ( @xmath2 ) @xmath3 where @xmath4 is the mass of the atoms , the sum over the field index @xmath5 is implied , and @xmath6 the ratio between the inter - species coupling @xmath7 and the intra - species interaction constant @xmath8 .
the coupling @xmath8 as well as the mass are chosen to be the same for both components .
since we will consider mainly near - degenerate states where the infrared modes of the system are strongly populated we will use quasi - classical simulation techniques @xcite to numerically evaluate the time - evolution of correlation functions . for this , initial field configurations @xmath9 are sampled from gaussian wigner distributions and then propagated according to the coupled classical equations of motion derived from the hamiltonian ( [ eq : action ] ) , [ eq : gp ] @xmath10 at each time during the evolution expectation values can be obtained from ensemble averages over the set of paths sampled .
we have applied the following rescalings to obtain dimensionless quantities in the system of coupled gross - pitaevskii - type equations ( [ eq : gp ] ) , using the lattice constant of the computational grid @xmath11 : @xmath12 , @xmath13 , and @xmath14 . here
, @xmath15 is the mean total particle density on an @xmath16 simulation grid of linear size @xmath17 .
the healing length derived from the total density reads @xmath18 and can be used to map the lattice constant to a value in physical units . while we will always solve numerically the full set ( [ eq : gp ] ) of equations of motion , it is instructive to concentrate for a moment on the degrees of freedom which describe the relative evolution of the two components . writing the fields in the polar representation @xmath19 the relative degrees of freedom
are given by the local phase difference @xmath20 as well as the local density difference @xmath21 .
with the help of the pauli matrices @xmath22 the schwinger representation of angular momentum is defined as @xmath23 ( sum over repeated indices implied ) .
this results in a three - component vector of ( pseudo-)spin densities @xmath24 for @xmath25 which encodes the relative degrees of freedom , [ eq : spinrep ] @xmath26 where the modulus corresponds to the total density @xmath27 . for convenience ,
we apply the redefinition @xmath28 such that @xmath29 . using the above representation , the total energy @xmath30 derived from eq .
( [ eq : action ] ) can be written as @xmath31\bigr]\ , , \end{aligned}\ ] ] which exhibits the role of the relative degrees of freedom and their coupling to the global ones @xcite .
the quantity @xmath32 is the conserved total particle current associated with the global @xmath33 phase symmetry of eq .
( [ eq : action ] ) , i.e. , its invariance under a global shift of the total phase @xmath34 .
thus @xmath35 can not be expressed using just the spin densities but contains also the total phase @xmath36 .
we obtain @xmath37}(s^y\nabla s^x - s^x\nabla s^y)\,.\ ] ] the representation ( [ eq : energy ] ) of the energy shows that the two - component bose gas can be equally well described as a spin - carrying fluid with density @xmath38 and a ( conserved ) quasi - particle current @xmath35 . for a fluid at rest , i.e. , @xmath39 and @xmath40 , the spin system thereby assumes the form of a classical nonlinear sigma model ( nl@xmath41 m ) with a mass term @xmath42 $ ] , whereas in general a current @xmath43 leads to a highly non - trivial coupling between internal and hydrodynamic degrees of freedom .
the two - component bose gas ( [ eq : action ] ) is well - known to possess two different ground states depending on the value of the parameter @xmath44 @xcite . in the immiscible regime , @xmath45 ,
the inter - species interaction energy overcomes the intra - species interaction .
hence , in the ground state of the system the spatial overlap of the components is minimised .
this becomes immediately clear from the expression ( [ eq : energy ] ) since for @xmath45 the mass term in the nl@xmath41 m , @xmath42 $ ] , gives a positive contribution to the total energy for all possible spin configurations , similar to the case of a heisenberg ferromagnet .
thus configurations with @xmath46 are preferred which leads not only to a spin - polarised ground state but also to the formation of domains with oppositely aligned @xmath47 due to the discrete @xmath48 symmetry @xmath49 of the energy ( [ eq : energy ] ) . on the contrary , in the miscible regime , @xmath50 ,
spin configurations with @xmath51 have lowest potential energy .
thus , the zero - temperature ground state is characterised by spatially mixed components with @xmath52 .
note that in both regimes the dynamic evolution is constrained by separate exact particle conservation in each component as can be inferred from the structure of interactions in eq .
( [ eq : action ] ) . in this article
we will investigate the dynamics of spin and fluid degrees of freedom when driven out of equilibrium , in both , the miscible and immiscible regimes and , in particular , close to the transition point .
it is known for the non - equilibrium dynamics of a one - component bose gas that features of turbulence become manifest during the evolution towards thermal equilibrium . in particular , it has been numerically demonstrated in one , two and three dimensions that the single - particle momentum distribution develops universal scaling behaviour during a turbulent stage @xcite . here , we are interested in the corresponding observable for the two - component bose gas @xmath53 where @xmath54 denotes the angular average in the two dimensional momentum space . in the presence of strong wave turbulence , self similar solutions of the form
@xmath55 are expected .
in addition to numerical calculations analytic expressions for the scaling exponents @xmath56 can be found in the context of a generic @xmath57 symmetric scalar quantum field theory @xcite .
for the ultraviolet ( uv ) momentum region , where occupation numbers are low and thus the theory is weakly interacting , kinetic quantum boltzmann theory is sufficient and results in a weak - wave - turbulence exponent @xmath58 @xcite . in the infrared
( ir ) region , high occupation numbers render the theory strongly interacting and therefore non - perturbative methods have to be used .
an infinite resummation of a certain class of feynman diagrams was shown to yield a scaling exponent in the ir , @xmath59 which corresponds to an inverse cascade of particles from intermediate momenta towards the ir @xcite .
finally , note that also the thermalised system is characterised by a scaling solution for @xmath60 with an exponent @xmath61 , corresponding to a rayleigh - jeans momentum distribution .
in the following , we study the time evolution of a two - component bose gas in the miscible and immiscible regimes , starting from on - average unpolarised nonequilibrium initial states
. the parameters of our simulation are chosen such that the final state is close to the groundstate of the system . making use of dynamical instabilities , we drive the system far from equilibrium and study the transient properties of ensemble averaged correlation functions .
a special focus is set on the relation between scaling properties and their microscopic origin .
( upper half ) and of the incompressible kinetic energy density @xmath62 ( lower half ) in the immiscible regime .
numerical parameters are @xmath63 , @xmath64 , @xmath65 and @xmath66 .
panels ( a ) to ( d ) correspond to snapshots of a single run at different grid times @xmath67 .
note that only a @xmath68 subsection of the computational grid is displayed in order to highlight the emerging structures .
quickly after the start of the simulation domains with oppositely oriented spin emerge as an isotropic pattern , and incompressible energy is created along the domain borders , see panels ( a ) . during the ensuing evolution towards the equilibrium state these domains merge until the system reaches a state in which only two large domains exist , cf .
panels ( d ) . on top of the coarse domain structure point - like domains become visible during the intermediate stage of the evolution which are revealed to be skyrmionic defects by the accompanying pattern of incompressible energy .
see the main text for further details .
, title="fig:",scaledwidth=50.0% ] ( upper half ) and of the incompressible kinetic energy density @xmath62 ( lower half ) in the immiscible regime . numerical parameters are @xmath63 , @xmath64 , @xmath65 and @xmath66 .
panels ( a ) to ( d ) correspond to snapshots of a single run at different grid times @xmath67 .
note that only a @xmath68 subsection of the computational grid is displayed in order to highlight the emerging structures .
quickly after the start of the simulation domains with oppositely oriented spin emerge as an isotropic pattern , and incompressible energy is created along the domain borders , see panels ( a ) . during the ensuing evolution towards the equilibrium state these domains merge until the system reaches a state in which only two large domains exist , cf . panels ( d ) . on top of the coarse domain structure point - like domains
become visible during the intermediate stage of the evolution which are revealed to be skyrmionic defects by the accompanying pattern of incompressible energy .
see the main text for further details . , title="fig:",scaledwidth=50.0% ] ) at different time steps of the evolution , in dimensionless units defined in the main text .
numerical parameters correspond to fig .
[ fig : immdens ] and the spectra are averaged over 100 runs .
after the onset of domain formation a scaling regime with @xmath69 forms at intermediate momenta . at a time
@xmath70 this region begins to move towards the ir while a new scaling region with @xmath71 develops in the uv .
the latter signals the onset of thermalisation of the uv modes . at late times
a stable bimodal power law behaviour has formed with a scaling exponent @xmath72 in the uv and an ir scaling exponent @xmath73 .
scaledwidth=49.0% ] for the simulations in the immiscible regime we choose @xmath66 and initial unpolarised configurations , @xmath74 , and add quantum noise to each mode .
specifically , we take only the zero mode of both complex fields to be initially occupied with a macroscopic number of particles @xmath75 , where the @xmath76 are gaussian distributed complex numbers with random phase , @xmath77 and @xmath78 .
we choose the number of particles in each component to be equal on average , @xmath79 .
initial configurations of such type are far from the energetically preferred state and therefore lead to instabilities that drive the system dynamically towards spatial demixing @xcite . a typical time evolution of one realisation of the described initial configurations is depicted in fig .
[ fig : immdens ] .
we follow the evolution until the spin system resembles its energetic ground state configuration , implying the total energy to be close to the ground - state energy .
this means that , consistent with particle conservation in each component , only two domains exist which are separated by a nearly straight domain wall .
compared to the full simulation time the demixing process occurs on a much faster time scale which is given by the energy of the fastest growing unstable mode , @xmath80 .
the main part of the evolution is dominated by merging of domains which leads to a coarse - graining of the domain structure .
+ to show the relation of the above results to the approach of a non - thermal fixed point , we consider the evolution in terms of the single - particle spectrum eq .
( [ eq : sinpartspec ] ) , see fig .
[ fig : immspec ] .
we find that already at an early stage of the evolution a power - law distribution has developed in the ir regime , @xmath81 .
this scaling behaviour terminates at a decreasing infrared cutoff scale @xmath82 given by the mean domain size @xmath83 .
the cut - off at the uv end at @xmath84 is found to be approximately set by the width of the domain walls , i.e. , the spin healing length @xmath85 . in between these scales
, a scaling region with an exponent of @xmath73 emerges .
consistent with the observed spatial growth of the domains of uniform spin polarisation , the ir limit of the scaling regime shifts in time towards lower momenta . in the ` far ' uv , quasi - classical thermalisation of the distribution sets in which is signalled by the appearance of the rayleigh - jeans scaling exponent @xmath72 . during the whole time evolution ,
the ir scaling exponent stays at @xmath73 .
this exponent stands in contrast with the strong - wave - turbulence prediction for a non - relativistic field theory in @xmath86 dimensions which is @xmath87 , cf .
( [ eq : scalingexp ] ) and ref .
+ ) , in comparison to the full spectrum at grid time @xmath88 , with numerical parameters corresponding to fig .
[ fig : immdens ] . averages are taken over 100 runs .
the black dots show the full spectrum which is also displayed in fig .
[ fig : immspec ] while open coloured symbols refer to the different parts of the decomposition , the incompressible and compressible parts of the classical hydrodynamic kinetic energy @xmath89 ( red squares ) and @xmath90 ( blue circles ) , the quantum pressure part @xmath91 ( grey triangles ) and the spin pressure part @xmath92 ( green stars ) .
the ir scaling behaviour @xmath93 is dominated by incompressible flow generated by vortex excitations with a scaling exponent @xmath94 .
spin excitations form domain walls which also give an important contribution in the same momentum regime but with a scaling exponent @xmath95 , such that the sum appears to follow a scaling law with @xmath73 .
see main text for further details.,scaledwidth=49.0% ] in order to gain further insight into the interplay between the domain structure and other excitations we decompose the energy along the lines introduced in refs .
@xcite , see also @xcite . according to the representation ( [ eq : energy ] ) of the energy , the kinetic part of the original gross - pitaevskii model ( [ eq : action ] ) can be written in the following way @xmath96 \ , .
\end{aligned}\ ] ] the velocity field @xmath97 is defined via the total particle current @xmath98 , similar to the convenient choice for the one - component case . in this decomposition , the first and the last term are the quantum - pressure and the classical hydrodynamic components , respectively . in contrast to a single - component fluid , the second term of eq .
( [ eq : energydec ] ) adds a pressure - like contribution to the kinetic energy which is produced by internal excitations only .
in addition , the velocity field @xmath97 and thus the corresponding part of the kinetic energy can be decomposed in a compressible and an incompressible part , @xmath99 , with @xmath100 and @xmath101 .
in such a decomposition , the effects of wave - like and vortical excitations show up in different parts . for sound waves ,
the compressible part becomes the important one while vortices appear in the incompressible part of the decomposition @xcite .
based on eq .
( [ eq : energydec ] ) , radial energy spectra in momentum space which correspond to the respective parts of the energy can be defined , [ eq : radenergyspec ] @xmath102 here we have introduced additional velocities @xmath103 and @xmath104 for the sake of a closed representation . finally , the energy spectra can be converted to occupation - number spectra in the sense of eq .
( [ eq : sinpartspec ] ) by multiplication with a factor @xmath105 , @xmath106 @xcite . a decomposition of such type in comparison to the full occupation spectrum at a late stage of the evolution ( @xmath107 ) is shown in fig .
[ fig : immspec_decomp ] corresponding to the evolution depicted in fig .
[ fig : immspec ] .
we find that compressible and pressure excitations dominate the occupation spectrum in the uv region following a thermal distribution @xmath108 while they give a negligible contribution except for the spin pressure component @xmath92 towards lower momenta .
the main contribution to the spectrum in the ir momentum regime is provided by incompressible excitations @xmath89 which is characteristic for quantum turbulence whereas the spin excitations @xmath92 overtake in a regime of intermediate momenta .
thereby , the incompressible spectrum shows an @xmath109 scaling over approximately one decade for low momenta which is generated by coherent vortical flows @xmath110 around topological defects . to elaborate on this , the lower part of fig .
[ fig : immdens ] shows incompressible energy densities @xmath111 for a single run .
here we observe that for later stages of the evolution incompressible energy is mainly distributed around isolated points which correspond to either vortices in one component or vortices filled with the other component , the latter being a variant of a so - called skyrmion @xcite .
these defects get created during the merging process of the domains and , persisting due to their topological nature , give the main contribution to the incompressible excitation spectrum with the scaling exponent @xmath94 .
for very late stages of the time evolution it can be seen that those skyrmionic defects can also be created or annihilated by excitations of the domain walls and thus the point defects are present during the whole simulation .
the spin excitation spectrum , on the other hand , displays a scaling of @xmath112 in the low- and intermediate - momentum regimes due to the existence of domain walls . under the assumption @xmath39
, the spectrum @xmath92 can be related to the fourier transform of the correlation function of the spin order parameter @xmath113 , @xmath114 with @xmath115 denoting the fourier transform , and therefore @xmath116 .
hence , in the regime @xmath117 where the scaling behaviour is dominated by a single domain wall @xmath118 follows , e.g. , from an ensemble of configurations with @xmath119{\mathbf{e}}_{\perp})$ ] , @xmath120 , with random position @xmath121 and orientation of vector @xmath122 normal to the wall .
this feature is similar to the scaling induced by solitons in one - dimensional bose gases , where a phase jump occurs in the bosonic field @xmath123 and induces a scaling @xmath124 , see ref .
@xcite .
since in fig .
[ fig : immspec_decomp ] the two contributions @xmath92 and @xmath89 are of comparable magnitude in an intermediate momentum range the sum of all contributions , which gives the full spectrum , appears to follow the scaling law @xmath69 in the ir . however , the momentum range in our simulations is limited by an ir cut - off determined by the grid size .
we expect that , on a much larger grid , the two superimposed scaling powers to separate such that in the limit @xmath125 a clear exponent @xmath94 arises while in an intermediate momentum region @xmath95 dominates . for the simulations in the miscible regime we used initial field configurations
which are spatially homogeneous in the density of both components and unpolarised , @xmath126 except for quantum noise .
in contrast to the immiscible regime , configurations of such type are close to the energetically favoured state of the spin system .
therefore a special mechanism is needed to drive the spin system towards a non - equilibrium state .
for this we exploit the so - called counter - superflow instability ( csi ) which is a hydrodynamic instability known to exist in miscible ultracold two - component gases @xcite .
both gas components are chosen to be spatially homogenenous in density but possess constant counter - directed current fields .
hence , each realisation has the form @xmath127 , with @xmath128-directed constant velocity fields @xmath129 . a bogoliubov de gennes analysis of this initial state reveals a critical instability for certain choices of the counterflow velocity @xmath130 . in particular , if @xmath130 exceeds a critical value , @xmath131 , unstable momentum modes exist in the system and momentum exchange between the two superfluid components is possible @xcite . for the dynamical simulations we choose the initial states to be @xmath132 as before ,
there are random complex numbers added to each momentum mode such that @xmath77 and @xmath78 to generate an appropriate ensemble of initial field configurations .
the relative interaction strength is set to @xmath133 and the initial counterflow velocity is set to @xmath134 such that @xmath135 . in the early stages of the time evolution the dynamical onset of the discussed instability generates planar spin waves , i.e. , kinks in the @xmath47 field which propagate along the counterflow direction .
a typical spin configuration for this early stage of the evolution is depicted in the inset of fig .
[ fig : mspecasy_onset ] . since these ( pseudo-)kink solutions are energetically unfavourable in the miscible regime they decay very fast , thereby driving the system through a phase of turbulent evolution . fig . [
fig : mdens ] , upper set , panel ( a ) shows that after the initial evolution stage the spin system develops an isotropically @xmath47-polarised form that is similar to the domain structure which builds up in the immiscible regime . instead of a coarsening
, however , the polarised structure decays in the following evolution towards thermal equilibrium ( see panels ( b ) to ( d ) in the upper part of fig .
[ fig : mdens ] ) due to the lack of topologically protected kink solutions in the miscible regime .
during the initial evolution we observe also vortex anti - vortex pairs in both components being created by the instability .
these defects survive for a much longer timescale than the intermediate isotropic spin polarisation .
the late stage of the evolution is therefore dominated by vortex pair dynamics , with algebraic decay in time as found in simulations of one - component gases @xcite .
panel ( a ) in the lower part of fig .
[ fig : mdens ] shows that the intermediate spin configurations support also isotropic and smooth distributions of incompressible kinetic energy @xmath136 , similar to a classical turbulent flow . with the decay of the spin polarisation also
the incompressible energy vanishes in the major parts of the simulation space and instead concentrates around the persistent vortical excitations ( see panels ( b ) and ( c ) in the lower part of fig . [
fig : mdens ] ) .
consequently , in the late stage of the evolution incompressible energy is distributed only in the vicinity of point defects , i.e. , vortices , as is expected for quantum turbulent flows ( see panel ( d ) in the lower part of fig .
[ fig : mdens ] ) .
+ ( upper half ) and of the incompressible kinetic energy density @xmath136 ( lower half ) in the miscible regime .
numerical parameters are @xmath63 , @xmath64 , @xmath137 , and @xmath133 , with @xmath138 .
panels ( a ) to ( d ) correspond to snapshots of a single run at different grid times @xmath139 .
after the onset of the counter - superflow - instability , isotropic @xmath47-polarised spin configurations emerge at intermediate times which are accompanied by uniform distributions of incompressible energy @xmath140 ( see panel ( a ) ) . during the thermalisation process
, the polarisation decays in most of the spatial grid , thereby revealing the existence of vortex anti - vortex pairs in both components which have @xmath141 at the centre of their cores .
these point defects persist during the whole run time of the simulation and dominate the dynamical features during the late stage ( panel ( d ) ) .
consequently , the incompressible energy concentrates around vortical excitations during the later stage of the evolution ( panels ( b d ) ) , passing from classical to quantum turbulence.,title="fig:",scaledwidth=49.0% ] ( upper half ) and of the incompressible kinetic energy density @xmath136 ( lower half ) in the miscible regime .
numerical parameters are @xmath63 , @xmath64 , @xmath137 , and @xmath133 , with @xmath138 .
panels ( a ) to ( d ) correspond to snapshots of a single run at different grid times @xmath139 .
after the onset of the counter - superflow - instability , isotropic @xmath47-polarised spin configurations emerge at intermediate times which are accompanied by uniform distributions of incompressible energy @xmath140 ( see panel ( a ) ) . during the thermalisation process
, the polarisation decays in most of the spatial grid , thereby revealing the existence of vortex anti - vortex pairs in both components which have @xmath141 at the centre of their cores .
these point defects persist during the whole run time of the simulation and dominate the dynamical features during the late stage ( panel ( d ) ) .
consequently , the incompressible energy concentrates around vortical excitations during the later stage of the evolution ( panels ( b d ) ) , passing from classical to quantum turbulence.,title="fig:",scaledwidth=49.0% ] ) , at different grid times during the onset stage of the counter - superflow - instability ( csi ) .
the numerical parameters correspond to fig .
[ fig : mdens ] , and the spectra are averaged over 100 runs .
initially only modes @xmath142 corresponding to the counterflow velocity are macroscopically occupied ( marked by the vertical dashed line ) .
the csi triggers the growth of certain unstable modes which then act as a source positioned at an intermediate momentum scale .
this initiates a typical cascading process in momentum space , causing the transport of predominantly particles to the lower momentum modes and of energy to the higher momentum modes .
the inset shows a snapshot of a typical spin configuration @xmath143 at @xmath144 .
one finds that the csi results in planar spin waves propagating in the direction of the initial counterflow which we chose pointing into the @xmath145-direction.,scaledwidth=49.0% ] also in the miscible regime , signals of turbulent evolution should become manifest in scaling solutions for the single - particle momentum spectrum ( [ eq : sinpartspec ] ) as well as of its different hydrodynamic contributions according to the decomposition of the radial kinetic energy density ( [ eq : radenergyspec ] ) . as is shown in fig .
[ fig : mspecasy_onset ] , initially only modes @xmath142 corresponding to the counterflow velocity are macroscopically occupied ( marked by the vertical dashed line ) . in the early stage of the time evolution
these modes feed two processes simultaneously : due to the csi unstable modes grow exponentially in the low - momentum regime .
scattering between particles leads to power - law growth of momentum modes for @xmath146 .
ultimately , this results in the build - up of an inverse cascade which transports particles to the lower momentum modes . at grid time
@xmath70 ( last timestep in fig .
[ fig : mspecasy_onset ] and first in fig .
[ fig : mspecasym ] ) a scaling with @xmath95 appears , corresponding to randomly distributed ( pseudo-)kinks in the densities and thus in @xmath47 ( panel ( a ) in fig .
[ fig : mdens ] ) . fig .
[ fig : mspecasym ] depicts the further time evolution of the single - particle spectrum . after the intermediate , kink - dominated stage the spectrum develops a bimodal form with @xmath147 in the ir and @xmath108 in the uv .
this scaling is well known from vortex and vortex pair dynamics in a two dimensional one - component gas @xcite .
consistent with the field theoretical predictions and the existence of a non - thermal fixed point , the system stays for the majority of the simulated time in this vortex - dominated stage .
+ ) at different grid times , in the miscible regime .
numerical parameters correspond to those in fig .
[ fig : mdens ] , and the spectra are averaged over 100 runs . with vanishing @xmath47 polarisation also the scaling distribution @xmath148 , which is seen at intermediate times over a wide momentum interval changes towards a bimodal form .
while high momentum modes thermalise , forming a rayleigh - jeans distribution @xmath108 , the ir regime is dominated by vortical excitations , implying @xmath147.,scaledwidth=49.0% ] ) , in comparison with the full spectrum at grid time @xmath149 .
numerical parameters correspond to those in fig .
[ fig : mdens ] . averages are taken over 100 runs .
the black dots show the full spectrum also displayed in fig .
[ fig : mspecasym ] .
open coloured symbols refer to the different parts of the decomposition : the incompressible and compressible parts of the classical hydrodynamic kinetic energy @xmath89 ( red squares ) and @xmath90 ( blue circles ) , the quantum pressure part @xmath91 ( grey triangles ) and the spin pressure part @xmath92 ( green stars ) .
other than in the immiscible regime the spin pressure contribution @xmath150 develops a scaling @xmath147 in the ir region which adds up with the incompressible energy distribution @xmath151 to give an ir scaling exponent @xmath94 . see main text for further details.,scaledwidth=49.0% ] , divided into its @xmath152-projection @xmath153 and @xmath154-projection part @xmath155 as defined in eq .
( [ eq : spinspec_split ] ) , at four different time steps .
the time slices as well as the numerical parameters correspond to fig .
[ fig : mdens ] , and averages are taken over 100 runs . in the intermediate stage
both quantities show the scaling behaviour @xmath156 at intermediate momenta ( panel ( a ) ) which is expected for kink solutions . during the following time evolution ( panels ( b)(d ) ) the scaling of @xmath157 is overtaken by a thermal distribution up to a small intermediate momentum region . in contrast
, the distribution of @xmath155 develops the bimodal scaling form which is characteristic for vortical flows in two dimensions with an ir scaling exponent @xmath94 ( panel ( d ) ) .
see the main text for further details.,scaledwidth=49.0% ] it is again instructive to decompose the spectrum according to the different parts of the radial kinetic energy density , eq .
( [ eq : radenergyspec ] ) .
the result is shown in fig .
[ fig : mspecasy_decomp ] for the latest time step in figs .
[ fig : mdens ] and [ fig : mspecasym ] . similar to the immiscible regime and in agreement with simulations for one - component gases
the quantum pressure contribution @xmath158 and the compressible energy contribution @xmath159 are negligible for low momenta .
they contribute mainly to the uv modes , with @xmath72 , reflecting a thermal distribution .
the incompressible part of the spectrum @xmath151 on the contrary displays the strong - wave turbulent scaling distribution @xmath160 in the ir due to the incompressible flow around vortices and is negligible for the uv region .
in contrast to the immiscible regime , the spin - pressure contribution @xmath150 follows here also the @xmath161 scaling for low - momentum modes , identical to that of the full single - particle spectrum .
however , in a small intermediate momentum region the spin - pressure contribution deviates from the @xmath161 scaling , below the scale where the thermal distribution @xmath162 sets in .
+ since the spin order parameter of the model ( [ eq : energy ] ) is not entirely @xmath163 symmetric for @xmath164 a better insight in the behaviour of the related correlation function @xmath165 is gained by splitting it into a part parallel to and orthogonal to the @xmath152-projection of the spin , @xmath166 and @xmath167 , respectively .
note that for @xmath168 the spin - pressure distribution and the correlation function are proportional ( see eq .
( [ eq : strucfunc ] ) ) .
therefore we can define also a splitting of the spin pressure according to the symmetry of the spin order parameter , [ eq : spinspec_split ] @xmath169 @xmath170 such that the distribution @xmath157 is only sensitive to the @xmath47 polarisation while @xmath155 reflects spin correlation in the @xmath154 plane of the spin state space .
due to the strong polarisation in the immiscible regime the influence of the latter onto the particle spectrum or the spin pressure is negligible whereas in the miscible regime both parts are present and each undergo a different time evolution . fig .
[ fig : mspecasy_spin ] depicts the spin pressure contribution separated into @xmath157 and @xmath155 according to eq .
( [ eq : spinspec_split ] ) , at four different times corresponding to the snapshots shown in fig .
[ fig : mdens ] . in the intermediate stage
( see panel a in fig .
[ fig : mspecasy_spin ] ) both distributions follow a scaling behaviour @xmath112 in a momentum range @xmath171 which is typical for randomly distributed ( pseudo-)kinks .
since kinks in @xmath47 are not favoured by the ground state properties of the miscible regime spin domains decay here instead of growing and the scale @xmath172 does not move towards lower momenta . on the other hand , thermalisation sets in for the uv modes and consequently the scaling regime with @xmath156 vanishes for @xmath153 during the thermalisation process up to a small intermediate range of momenta
( see panel ( d ) in fig .
[ fig : mspecasy_spin ] ) .
note that vortices lead to @xmath47-polarised patches in their core area and thus preserve the @xmath173 scaling of @xmath157 .
however , the distribution @xmath155 inherits its dynamical evolution during the late stage from the dynamics of vortices . identifying the @xmath154-projection of the spin with a complex field
@xmath174 the relation @xmath175 holds and @xmath176 .
thus , for the later stages of the evolution , @xmath155 behaves like the correlation function of a two - dimensional complex order parameter which contains randomly distributed vortex anti - vortex pairs .
their dynamics imprints the ir scaling behaviour @xmath177 onto the distribution @xmath155 ( panel ( d ) in fig .
[ fig : mspecasy_spin ] ) , in complete analogy to the vortex binding - unbinding dynamics in a one - component gas in two dimensions . + ( upper half ) and of the incompressible kinetic energy density @xmath136 ( lower half ) .
the ratio @xmath44 of couplings is tuned to the transition point between the miscible and immiscible regimes ( @xmath178 ) .
besides this , numerical parameters and depicted times are the same as for fig .
[ fig : mdens ] .
note that the incompressible energy density has been amplified by a factor of @xmath179 for @xmath149 in panel ( d ) .
similar to the situation deep in the miscible regime ( compare with fig .
[ fig : mdens ] ) , after the onset stage of the counter - superflow instability @xmath47-polarised spin configurations emerge at intermediate times which are accompanied by uniform distributions of incompressible energy @xmath140 ( panel ( a ) ) .
however , here the pseudo - domain structure of @xmath47 undergoes a coarse - graining process similar to the situation in the immiscible regime ( compare with fig .
[ fig : immdens ] ) , instead of decaying in time . ,
title="fig:",scaledwidth=49.0% ] ( upper half ) and of the incompressible kinetic energy density @xmath136 ( lower half ) .
the ratio @xmath44 of couplings is tuned to the transition point between the miscible and immiscible regimes ( @xmath178 ) . besides this , numerical parameters and depicted times are the same as for fig . [
fig : mdens ] . note that the incompressible energy density has been amplified by a factor of @xmath179 for @xmath149 in panel ( d ) .
similar to the situation deep in the miscible regime ( compare with fig .
[ fig : mdens ] ) , after the onset stage of the counter - superflow instability @xmath47-polarised spin configurations emerge at intermediate times which are accompanied by uniform distributions of incompressible energy @xmath140 ( panel ( a ) ) .
however , here the pseudo - domain structure of @xmath47 undergoes a coarse - graining process similar to the situation in the immiscible regime ( compare with fig .
[ fig : immdens ] ) , instead of decaying in time . , title="fig:",scaledwidth=49.0% ] ) , at different grid times , at the transition point @xmath178 .
numerical parameters correspond to those in fig .
[ fig : mdens_sym ] , and the spectra are averaged over 100 runs .
similar to the miscible regime randomly distributed and orientated pseudo - kinks at intermediate times imply a domain - wall scaling @xmath148 over a wide range of momenta .
however , during the following evolution this scaling is retained and the region shifted towards the lower momentum modes while thermalisation of high momentum modes sets in . at the latest time step
we find a momentum distribution with an almost stationary scaling @xmath148 in the ir.,scaledwidth=49.0% ] ) , in comparison with the full spectrum at grid time @xmath149 .
numerical parameters correspond to those in fig .
[ fig : mdens_sym ] . averages are taken over 100 runs .
the black dots show the full spectrum which is also displayed in fig .
[ fig : mspecsym ] while open coloured symbols refer to the incompressible and compressible parts of the hydrodynamic kinetic energy , @xmath89 ( red squares ) and @xmath90 ( blue circles ) , respectively , to the quantum pressure part @xmath91 ( grey triangles ) and the spin pressure part @xmath92 ( green stars ) .
in contrast to the situation deep in the miscible and immiscible regimes the spin - pressure contribution @xmath92 dominates the spectrum of excitations for low momenta with a domain - wall scaling similar to that of @xmath180 , @xmath156 for @xmath181 . in this regime ,
incompressible energy excitations are found to be less relevant .
see the main text for further details.,scaledwidth=49.0% ] separated into its @xmath152-projection @xmath153 and @xmath154-projection @xmath155 as defined in eq .
( [ eq : spinspec_split ] ) , at four different time steps . those as well as the numerical parameters correspond to fig .
[ fig : mdens_sym ] , and averages are taken over 100 runs . for the intermediate and late times of evolution
depicted here the @xmath154-projection follows the form of the @xmath152-projection , thereby developing the same domain - wall scaling in the ir , @xmath182 for @xmath181 .
confer the main text for further details.,scaledwidth=49.0% ] in view of our findings in the immiscible and miscible regimes discussed in the previous sections it is particularly interesting to study the dynamic evolution of the two - component bose gas directly at the transition point , i.e. , for @xmath178 , where no state of mixture is preferred over the other from the perspective of their potential energies .
this particular choice of @xmath44 restores full @xmath183 symmetry in the model ( [ eq : action ] ) .
hence , taking the field - theoretic perspective , we simulate the dynamic evolution of a non - relativistic @xmath184 theory with @xmath185 field components , for which in turn predictions drawn within a @xmath186-expansion @xcite should apply even better . as in sect . [
sec : miscregime ] the counter - superflow - instability ( csi ) mechanism is used to drive the spin system away from equilibrium during the early evolution .
therefore , we use the same ensemble of initial field configurations as for the simulations presented in [ sec : miscregime ] . for comparability with the miscible regime ,
the same numerical parameters including the initial counterflow velocity @xmath187 are chosen , although at the transition point any initial counterflow is unstable , i.e. , @xmath188 . starting from this numerical preset a typical time evolution of a single realisation
is shown in fig .
[ fig : mdens_sym ] .
we concentrate again on the spin density @xmath47 and the distribution of hydrodynamic incompressible energy @xmath140 . in complete analogy to the evolution in the miscible regime csi - generated pseudo kinks in @xmath47 develop into an isotropic structure of domain - like patches at intermediate times .
this is accompanied by an isotropic distribution of incompressible energy ( see panel ( a ) in fig . [
fig : mdens_sym ] and compare to panel ( a ) in fig .
[ fig : mdens ] ) .
however , in the following stage of evolution these structures in @xmath47 undergo a coarse - graining process , i.e. , they grow , very similar to the dynamic evolution in the immiscible regime , while incompressible energy is decaying ( panels ( b c ) in fig . [
fig : mdens_sym ] ) .
we emphasise that even though configurations with large patches of @xmath141 are energetically favourable for @xmath178 , the curvature @xmath189 at interfaces between them tends to level out .
hence , also at the transition point the pseudo - domain structure in @xmath47 would be expected to decay back to the unpolarised state , especially since interfaces are not protected by means of topological constraints .
however , most remarkably , the spin system does not reach or even come close to the unpolarised state over the simulated time span , in contrast to the situation for a miscible system ( see fig .
[ fig : mdens_sym]d and compare with fig .
[ fig : mdens]d ) .
in addition , there is no nucleation of stable one - component vortices with coherent hydrodynamic vortical flow around them during the early or intermediate stages .
this qualitative observation of stable domain - like structures that slow down the evolution of the spin system towards the unpolarised state and therefore also the thermalisation process of the whole system is supported on a quantitative level by the time evolution of the single - particle momentum distribution , depicted in fig .
[ fig : mspecsym ] . at intermediate times after the onset stage of the csi the spectrum has developed a scaling @xmath148 in a momentum region @xmath181 consistent with randomly distributed and randomly orientated pseudo - kinks . during the late stages the scale connected to the characterisic size of the pseudo - domain structure @xmath190 moves towards the ir , thus signalling a coarse - graining of the structure .
in contrast to the immiscible case , here the intrinsic scale of interfaces @xmath191 is not fixed by a topological solution .
this circumstance allows for a reduction of interface energy by broadening , and therefore the scale @xmath192 moves also towards lower momenta .
however , the defect - induced scaling @xmath148 is retained between those two scales during the whole simulated time span . at the latest time step in fig .
[ fig : mspecsym ] , a slowly evolving spectrum has formed that shows thermal scaling @xmath108 for high momenta and @xmath193 in the low-@xmath194 regime , with @xmath195 .
thus , at the transition point @xmath196 we find a deviation in the ir scaling exponent from the field - theoretic prediction in @xmath86 dimensions , @xmath197 .
we attribute this to the absence of dominant vortical excitations . in fig .
[ fig : mspecsy_decomp ] , a decomposition of the single - particle spectrum shown in fig .
[ fig : mspecsym ] into hydrodynamic and spin - related degrees of freedom is shown for the latest time step .
we find that as before the quantum - pressure and compressible - energy contributions are negligible for low momenta and contribute equally to the thermal tail for high momenta , together with the spin excitations . in congruence with our findings for the ir dynamics at the transition point
so far , the spin - pressure component @xmath92 gives the dominant contribution for low momenta , thereby reflecting the defect scaling of @xmath180 , @xmath156 for @xmath181 .
in contrast to the situation deep in the miscible and immiscible regimes here the incompressible energy excitations are below the spin - pressure contribution within a wide range of ir momenta .
nevertheless , the incompressible occupation spectrum develops a scaling @xmath160 in the ir , which hints at vortical excitations or flow hidden under the spin polarisation .
we remark that due to the enlarged ground state manifold of the @xmath183 symmetric model vortices in a single component loose their topological stability since they can simply ` unwind ' .
nevertheless , animations @xcite of the incompressible energy density show that vortices indeed are created in our simulations but persist only on very short timescales before they decay .
this is enough to produce a visible ir scaling behaviour of @xmath147 in the spectrum of incompressible excitations .
focusing on the main contribution to the spectrum of excitations , fig .
[ fig : mspecsy_spin ] depicts the time evolution of the two projections @xmath157 and @xmath155 of the spin pressure . at the transition point
, the @xmath154-projection follows the functional form of the @xmath47-projection during the intermediate and late stages of evolution , in accordance with the @xmath163 symmetry in the spin system for @xmath178 .
thus , the @xmath154-projection develops also the defect scaling in the ir , and consequently the corresponding spin densities @xmath198 and @xmath199 develop domain - like structures as well .
we have discussed the non - equilibrium dynamical evolution of a near - degenerate two - component bose gas towards equilibration , following dynamical instabilities induced by the initial conditions .
the main results concern the investigation of non - thermal - fixed - point behaviour across the miscible - immiscible transition .
note , that the underlying concept of topological defects determining bulk features of correlation functions in far - from - equilibrium situations is very general .
it is easily imaginable that multi - component field theories with more than two components show behaviour similar to the one outlined here .
this requires the generation of ( quasi-)topological configurations far from thermal equilibrium and their slow decay , going together with an increase of coherence and defect separation @xcite . under these conditions ,
an inverse particle cascade is generated , and the associated power laws can be found from the scaling properties of the respective single defects .
new interesting features that are readily accessible in experiment are expected for ultra cold spinor gases , with spin 1/2 @xcite or higher @xcite .
the transition between different types of transient non - equilibrium order can be controlled by changing the symmetry properties of the hamiltonian and thus topology and local conservation laws of the system .
this offers interesting prospects for far - from - equilibrium dynamical transitions in very different areas of physics .
much work has been done recently concerning the question whether a system on the way to thermalization or general equilibration can approach , underway , non - equilibrium states with characteristics of thermal or generalised gibbs ensembles , so - called prethermalisation @xcite , and non - thermal fixed points @xcite .
prethermalisation is also discussed in the context of the question under which conditions a closed interacting quantum system can thermalise , cf .
, e.g. , @xcite .
prethermalisation in the sense of refs .
@xcite can by definition be seen as a mean - field ( gaussian ) non - thermal fixed point reached through dephasing of quasiparticle modes occupied in the initial state . in contrast , the non - thermal fixed points studied here represent stationary configurations where interactions play a vital role .
these fixed points are reached not long after the initial mean - field dephasing period through scattering processes which redistribute particles to form the non - thermal algebraic momentum spectra .
it is expected that within a general renormalisation - group theory of far - from - equilibrium critical phenomena @xcite , prethermalisation phenomena and non - thermal fixed points are treated on equal footings . _
we thank j. berges , s. diehl , s. erne , p. kevrekidis , l. mclerran , e. nicklas , m. k. oberthaler , j. m. pawlowski , j. schole , d. sexty , and c. wetterich for discussions .
this work was supported by deutsche forschungsgemeinschaft ( ga677/7,8 ) , the university of heidelberg ( cqd ) , and the helmholtz association ( ha216/emmi ) . | quasi - stationary far - from - equilibrium critical states of a two - component bose gas are studied in two spatial dimensions . after the system has undergone an initial dynamical instability it approaches a non - thermal fixed point . at this critical point
the structure of the gas is characterised by ensembles of ( quasi-)topological defects such as vortices , skyrmions and solitons which give rise to universal power - law behaviour of momentum correlation functions .
the resulting power - law spectra can be interpreted in terms of strong - wave - turbulence cascades driven by particle transport into long - wave - length excitations . scaling exponents are determined on both sides of the miscible - immiscible transition controlled by the ratio of the intra - species to inter - species couplings . making use of quantum turbulence methods
, we explain the specific values of the exponents from the presence of transient ( quasi-)topological defects . |
the production of high transverse momentum jets by quasi real photons on protons is one of the major processes to gain further insight into the interactions of photons with quarks and gluons .
experimental results by the h1 @xcite and zeus @xcite collaborations at hera have been presented , and more and better data are expected to come out soon .
a subsample of these events are so called _ large rapidity gap events _ , which have been discovered recently in photoproduction and deep inelastic electroproduction in both hera experiments @xcite .
their properties were found to be inconsistent with the dominant dynamical mechanism for the production of jets in photoproduction and deep
inelastic electroproduction , where colour is transferred between the produced quark and gluon jets and the proton remnant .
whereas the majority of the events clustered around @xmath1 , a second class of events , with @xmath2 , were observed , i.e. , they had a large gap in rapidity between the fast moving proton , which in the moment is still an assumption , and the rest of the hadronic final state .
( here , @xmath3 is the rapidity measured along the incoming proton direction in the laboratory frame . ) the same type of events had been observed already some time ago in @xmath4 scattering @xcite .
they were interpreted as being due to diffractive hard scattering , following a suggestion of ingelman and schlein @xcite . in this interpretation , which also applies to the large rapidity gap events at hera
, the proton emits a pomeron which acts as a virtual target for the incoming hadron ( in the case of @xmath4 scattering ) and the quasi real or highly virtual photon ( in the case of @xmath0 scattering ) , respectively .
the incoming proton stays intact or becomes a low mass state , so that there is no colour flow from the proton to the other final state particles .
similarly to ordinary hadrons , mesons and baryons , the incoming pomeron is supposed to have a quark and/or a gluon structure , which is probed through electroweak or strong hard scattering processes and expressed by a pomeron structure function .
models based on different assumptions concerning this quark
gluon structure have been used for calculations of @xmath4 and @xmath0 scattering processes ( see , e.g. , @xcite ) .
based on these models , bruni and ingelman have developed a monte carlo program @xcite to study the event characteristics , which is already widely used by the hera collaborations to interpret their large
rapidity gap data .
if this interpretation is valid , a major task is to disentangle the quark and gluon distributions of the pomeron . as is the case in usual hard
scattering processes , one process is not sufficient to isolate the various parton densities .
several processes must be considered simultaneously .
inelastic diffractive @xmath0 scattering primarily measures the quark distribution of the pomeron , while the gluon structure function appears as a higher order qcd correction .
photoproduction of jets , however , is sensitive to both the quark and gluon densities already in the leading order ( lo ) of qcd . in lo , photoproduction has two components : a direct one and a resolved one .
the latter mechanism has similar features as jet production in diffractive hadron
hadron scattering .
the hadron structure function is just replaced by the photon structure function .
all these processes can be studied in detail with the monte carlo program cited above .
recently , a number of authors advocated the notion of a direct pomeron coupling to quarks and gluons @xcite .
donnachie and landshoff developed a model in which two gluons emitted coherently from the proton couple to the quark antiquark system produced by the incoming real or virtual photon @xcite .
the quark and antiquark thus give rise to two jets associated with the hard scattering .
such configurations give a delta function in the pomeron structure function in the same way as direct photoproduction is equivalent to a delta function term in the photon structure function of resolved photoproduction . since in the donnachie landshoff model the second gluon of the pomeron is attached directly to the hard scattering , or to an outgoing parton , the result is higher twist , i.e. , considered as a function of @xmath5 , it falls off more strongly than in the case of a factorizable direct pomeron coupling .
subsequently , several groups @xcite suggested other mechanisms which break the usual parton
model factorization and are effectively proportional to a delta function term .
these mechanisms , called _ coherent hard diffraction _ @xcite or _
lossless diffractive jet production _
@xcite , have the property that the total incoming longitudinal momentum of the pomeron is delivered to the jet system .
both mechanisms produce leading twist contributions and are not present in deep
inelastic electron
proton scattering , i.e. , in quasi real photoproduction , which we are interested in here primarily , they would occur only in the resolved process .
in addition , it is said explicitly that the cfs mechanism @xcite leads to a delta function term only in the gluon density of the pomeron .
so , these mechanisms for creating a quasi direct pomeron coupling are very special and their validity is very difficult to assess .
they are just conjectures and not proven consequences of qcd . under these circumstances
, it is a challenge to look at experimental data in the diffractive region and to investigate whether such a direct pomeron coupling is present . to that end , it is necessary to establish features which are characteristic for a direct pomeron coupling to quarks and/or gluons in the data .
this is the purpose of this work . in section 2 , we formulate the model and give the relevant formulas for calculating the cross section for the production of jets in @xmath0 scattering with almost real photons .
there , we specify the structure functions of the photon and the pomeron and write down the expression for the pomeron flux .
if a direct pomeron coupling exists , this has also consequences for the diffractive contribution of the deep
inelastic structure function @xmath6 , which in retrospect modifies the input for the photoproduction cross section . in section 3 ,
we present numerical results for the three scenarios : ( i ) resolved pomeron , ( ii ) resolved and direct pomeron , ( iii ) resolved and direct pomeron plus point like component in the pomeron structure function .
the last section is reserved for the conclusions and some outlook to future work .
the cross section for the diffractive @xmath0 scattering process , depicted in fig . [ dia ] , is obtained from the following parton model formula : [ d3s ] & & ( epjet+x ) + & = & _ a , b , c , ddx_g_/e(x _ ) dx_a g_a/(x_a ) dx_ip g_ip / p(x_ip ) dx_b g_b / ip(x_b ) + & & ( a+bc+d ) ( + + ) , where @xmath7 and @xmath5 are the rapidity and transverse momentum of the jet , respectively .
@xmath8 , @xmath9 and @xmath10 are the familiar mandelstam variables for the partonic 22 scattering process @xmath11 .
@xmath12 stands for the usual weizscker williams formula @xcite , which gives the flux of the quasi real photons .
the exact form will be specified later .
@xmath13 is an integral over the pomeron flux factor @xmath14 , [ ipflux ] g_ip / p(x_ip ) = _
t_1^t_2 dtf_ip / p(x_ip , t ) , where @xmath15 is the momentum transfer to the proton line . @xmath16 , with @xmath17 being the proton mass , is the kinematical upper boundary , while @xmath18 is determined by the experimental conditions .
for @xmath14 we use the ingelman
schlein ansatz @xcite , [ flux ] f_ip / p(x_ip , t ) = = ( ae^t + b e^t ) , with the parameters @xmath19 , @xmath20 , @xmath21 , @xmath22 and @xmath23 , which were obtained by fitting to the data of the diffractive cross section and the pomeron
proton total cross section .
other functional forms of @xmath14 are given in @xcite ; they are numerically equivalent in the region of small @xmath24 , which dominates the diffractive cross section . for the unknown pomeron structure functions , @xmath25
, we make the ansatz [ ipsf ] g_u / ip(x ) & = & g_/ip(x ) = g_d / ip(x ) = g_/ip(x ) = ( 1-x ) , + g_s / ip(x ) & = & g_/ip(x ) = ( 1-x ) , g_c / ip(x ) = g_/ip(x ) = 0 , + g_g / ip(x ) & = & ( 1-x ) . the gluon component fulfills _ 0 ^ 1 dxx g_g / ip(x ) = , and the quark components have the normalization _
^ 1 dxx g_q / ip(x ) = .
the sum of the gluon and quark parts obeys the sum rule _ 0 ^ 1 dxx ( g_g / ip(x)+_q , g_q
/ ip(x ) ) = 1 , which is usually assumed also by other authors @xcite .
since the exchanged pomeron is not a physical particle , there is no rigid justification for this sum rule , although some motivation can be given .
the quark content of the pomeron can be measured in diffractive deep
inelastic @xmath0 scattering ( ddis ) .
preliminary data from the h1 and zeus collaborations exist , which will be used for comparisons later .
the gluon content of the pomeron contributes to ddis through the order @xmath26 process @xmath27 and is therefore suppressed .
this means that , in a first approximation , ddis is determined by the quark and antiquark components of the pomeron structure function .
the gluon component enters when the pomeron structure function is evolved to larger @xmath28 , which , however , we shall not perform in the crude study presented here . for the calculation of the photoproduction of jets , we need the hard scattering cross sections @xmath29 in ( [ d3s ] ) .
we consider four components , dd , dr , rd , rr .
the first letter indicates whether the incoming quasi
real photon interacts directly ( d ) or as a resolved photon ( r ) with the incoming quarks and gluons .
the second letter characterizes the pomeron interaction ; d stands for the direct coupling of the pomeron and r for a resolved pomeron described by the structure functions in ( [ ipsf ] ) . the resolved
pomeron hard scattering cross sections are the usual ones . in the case of a direct photon coupling , we have the photon gluon fusion ( @xmath30 ) cross section , where a gluon inside the pomeron is struck , and the gluon compton ( @xmath31 ) cross section , where a quark or antiquark is pulled out of the pomeron . for a resolved photon and a resolved pomeron ( rr ) , we have the well known 22 quark and gluon scattering cross sections . for the calculation of the cross sections with a direct pomeron coupling ,
we restrict ourselves to a direct coupling with quarks ( antiquarks ) of the form [ lint ] _ int = c(x ) _
q(x)^(x ) . here
@xmath32 is the coupling constant . in ( [ lint ] )
, we assumed that the pomeron , @xmath33 , couples to quarks like a vector particle .
this contradicts the fact that the pomeron behaves like a @xmath34 exchange , whereas ( [ lint ] ) transforms invariantly under @xmath35 only if @xmath33 has @xmath36 .
it is an old observation , however , that , in soft processes like diffractive scattering , the pomeron couples to quarks effectively rather like an isoscalar photon , i.e. , with a constant @xmath37 coupling but with a regge signature factor , which endows it with even @xmath35 parity @xcite . with this ansatz ,
donnachie and landshoff were able to give a realistic description of all high energy elastic and diffractive cross sections applying the additive quark model and including less dominant regge exchanges @xcite .
pomeron exchange with the coupling of ( [ lint ] ) also correctly predicts the approximate helicity conservation that is observed experimentally .
with such a coupling , we obtain the following hard scattering cross section of @xmath38 , written in terms of the invariants , @xmath39 , @xmath40 , @xmath41 , introduced earlier : [ dsdt ] ( ipq ) = 6e^2e_q^2 c^2 ( + + ) , where @xmath42 is the electric charge of @xmath43 in units of the positron charge , @xmath44 . here
, @xmath15 is the momentum transfer at the proton vertex : @xmath45 , if the pomeron is timelike with mass @xmath46 , while @xmath47 in our application .
of course , ( [ dsdt ] ) is identical to the cross section of @xmath48 , where @xmath49 is an off shell photon with momentum squared @xmath15 . introducing the transverse momentum @xmath5 and the rapidity @xmath7 of the outgoing jet
, we have y = , = ( -t)^2 , + = - + t , so that [ dsdp ] ( ipq ) = 6e^2e_q^2 c^2 . for @xmath50 ,
the cross section decreases only like @xmath51 , i.e. , it exhibits a rather mild @xmath5 dependence .
furthermore , the rapidity is completely fixed by @xmath5 and @xmath39 . under the assumption that the pomeron couples like a scalar particle ( this would not allow the description of elastic and diffractive scattering proposed in @xcite ) , ( [ dsdp ] ) would only change slightly to ( ipq ) = 6e^2e_q^2 c^2 , i.e. , for @xmath52 the result coincides with the one for the vector coupling . in the following , we shall assume that the direct pomeron coupling is characterized by ( [ lint ] ) , i.e. , we shall employ ( [ dsdp ] ) for the hard collision cross section . in our model for the direct pomeron coupling , the pomeron behaves like a real photon target for @xmath53 . similarly to @xmath54 scattering , the @xmath38 cross section also contributes to the pomeron structure function , i.e. , to deep
inelastic @xmath0 scattering .
then the incoming photon in ( [ dsdp ] ) is off shell with invariant mass squared @xmath55 . in inelastic @xmath56 scattering , this represents the contribution of the _ point like _ photon in the photon structure function .
similarly , we have a _ point like _ contribution to the quark part of the pomeron structure function , which is obtained from ( [ dsdp ] ) by integrating over @xmath57 . in the form of ( [ dsdp ] )
, the cross section has a collinear singularity at @xmath58 . to remove it ,
we render the quark masses , @xmath59 , nonvanishing as usual @xcite and obtain [ poi ] g_q / ip^pl(x , q^2 ) & = & c^2 \ { + & & + } , with & = & .
the quark masses are chosen as @xmath60 , @xmath61 ( i.e. , equal to the constituent masses ) and @xmath62 .
@xmath63 is the momentum squared of the off
shell incoming photon . in a more realistic treatment ,
the mass parameter is replaced by @xmath64 .
it is known that ( [ poi ] ) is a reasonable approximation for the case of the point
like part of the photon structure function . for large @xmath28 , we have [ poia ] g_q / ip^pl(x , q^2 ) = c^2 . as expected , the point
like part of the quark distribution of the pomeron dominates at large @xmath28 .
this result is unusual for the structure of such a complicated object as the pomeron . as a first guess
, one would expect that the quark distribution function of the pomeron would behave more like the one of a meson , a @xmath65 or a @xmath66 meson say , which is supposed to have no point like component . whether this is true
can be tested by inspecting the data and trying to eventually place an upper limit on the coupling @xmath32 .
furthermore , the @xmath67 dependence of @xmath68 is quite different from the behaviour of the corresponding structure function of a resolved pomeron in ( [ ipsf ] ) .
to lo in @xmath26 , only the quark distribution of the pomeron enters into the deep
inelastic @xmath69 structure function @xmath70 , which is [ f2ip ] f_2^ip(x , q^2 ) = _ q e_q^2x . in fig .
[ plikec ] , we have plotted the point like distribution function for the three quark masses @xmath71 , @xmath72 , @xmath73 and the four values @xmath74 , @xmath75 , @xmath76 , @xmath77 . notice the logarithmic singularity at @xmath78 .
apart from that , @xmath68 falls off with @xmath67 increasing and increases with @xmath28 even at large @xmath67 , in contrast to what one expects for a hadron like object . for calculating @xmath79 ,
we have chosen @xmath80 .
this value is somewhat smaller than the one deduced from fits to total cross sections and elastic scattering data by donnachie and landshoff @xcite .
figure [ plikec]d shows the total contribution of the point like part to @xmath81 for the same @xmath28 values , i.e. , the sum of the @xmath82 , @xmath83 , @xmath84 , @xmath32 contributions with the appropriate charge factors .
the dents in the curves are caused by the charm threshold . under the assumption that factorization is applicable
, @xmath85 can be used to calculate the diffractive contribution to the deep
inelastic structure function @xmath86 of the proton .
the relation is [ f2diff ] f_2^diff(x , q^2 ) = _
x^x_0 dx_ip_t_1^t_2 dt f_ip / p(x_ip , t ) f_2^ip(x
/ x_ip , q^2 ) . here
@xmath87 is the pomeron flux factor given by ( [ flux ] ) .
the diffractive contribution is concentrated at small @xmath67 , and we shall mostly take @xmath88 as in the experimental analyses by the h1 and zeus collaborations .
@xmath89 is defined below ( [ ipflux ] ) , and we choose @xmath90 .
the results for @xmath86 at @xmath91 , @xmath92 , @xmath93 , @xmath94 are compared to preliminary h1 data @xcite in fig .
we see that the contribution without direct pomeron ( @xmath95 ) is too small .
of course , this can be changed by increasing @xmath96 in relation to @xmath97 in ( [ ipsf ] ) ; in its present form , the relation is 1:3 . with our choice of ( [ ipsf ] ) ,
the agreement with the data is improved significantly when we include the pointlike contribution with @xmath80 . clearly , fig .
[ f2df ] only demonstrates that @xmath80 is a consistent value for the direct pomeron coupling when the resolved pomeron is described by ( [ ipsf ] ) . that our curves for @xmath80 lie somewhat above the data points can be tolerated ,
since the data are obtained with the cut @xmath2 , i.e. , only a fraction , presumably not more than 5070% , of the diffractive contribution is included in the data points .
as expected from the cut on the invariant mass @xmath98 , i.e. , @xmath88 , the data for the diffractive part of @xmath6 are nonzero only for @xmath99 . in the case of @xmath86 , we could , in principle , accommodate a larger value of @xmath32 by correspondingly scaling down @xmath96 relative to @xmath97 . by fitting simultaneously @xmath86 and the jet
photoproduction cross sections to the data , it should be possible to place an upper bound on @xmath32 .
we emphasize that the proton momentum transfer is neglected in ( [ poi ] ) and hence in ( [ f2ip ] ) and in the @xmath69 structure function in ( [ f2diff ] ) .
this is justified , since in ( [ f2diff ] ) the integral over @xmath15 is dominated by the minimum @xmath24 , which is @xmath100 . of course
, it would be worthwhile to consider @xmath86 in ( [ f2diff ] ) for fixed @xmath15 and to study the full @xmath15 dependence .
we now turn to the jet cross sections of diffractive photoproduction .
the calculation of the one and two jet inclusive cross section is straightforward and proceeds from ( [ d3s ] ) . in this formula ,
the integration limits must be specified in terms of the external variables .
for the one jet inclusive cross section , they are @xmath7 , @xmath5 and @xmath101 , the energy squared of the @xmath0 centre - of - mass ( c.m . ) system . for the case of rr ,
i.e. , resolved @xmath102 and resolved @xmath103 , the incoming parton momenta are @xmath104 and @xmath105 , so that @xmath106 , @xmath107 and @xmath108 ( see fig . [ dia ] ) .
it is clear that all momentum fractions are limited in general to the interval @xmath109 $ ] . in our formulas , we have chosen the @xmath0 c.m . system , and the momentum of the incoming electron is taken along the positive @xmath110 direction .
the transformation to the hera system will be done when we present the numerical results in the next section . from phase space , we have as the integration limits in ( [ d3s ] ) x_^min & = & , x_a^min = , + x_ip^min & = & , x_b^min = . as usual
, the functions @xmath111 stand for the probability to find a parton @xmath112 with momentum fraction @xmath113 in parton @xmath114 , where @xmath112 and @xmath114 may be also the electron , photon or pomeron .
@xmath115 is described by an improved weizscker
williams function @xcite g_/e(x ) = , where @xmath116 and @xmath117 .
we adopt the photon structure functions from @xcite .
the contribution due to a direct photon and a resolved pomeron ( dr ) is calculated from ( [ d3s ] ) using @xmath118 and the hard scattering cross sections of @xmath31 and @xmath30 .
experimentally , the rapidities are limited by @xmath119 . in the one jet inclusive cross section ,
the integration over the rapidity of the second jet goes beyond this limit . to incorporate the limit
, we need to consider the two jet cross section .
it is obtained from [ d3s2 ] & & ( ep j_cj_d+x ) + & = & _ a , b , c , d
_ x_^min^x_^max dx_g_/e(x _ ) _
x_ip^min^x_ip^max dx_ipg_ip / p(x_ip ) x_ag_a/(x_a)x_bg_b / ip(x_b ) ( abcd ) , where x_a & = & ( e^y_c+e^y_d ) , + x_b & = & ( e^-y_c+e^-y_d ) .
so , for fixed @xmath120 and @xmath121 , one can directly extract information on the structure functions of the photon and the pomeron by varying @xmath5 , @xmath122 and @xmath123 .
the lower limits of the @xmath120 and @xmath121 integrals are x_^min & = & ( e^y_c+e^y_d ) , + x_ip^min & = & ( e^-y_c+e^-y_d ) .
clearly , @xmath5 , @xmath122 and @xmath123 are limited by the conditions @xmath124 and @xmath125 .
the rapidities @xmath122 and @xmath123 are defined in such a way that the incoming electron travels in the positive @xmath110 direction , i.e. , @xmath126 and @xmath127 . in the numerical discussion
, we shall flip the sign of the rapidity so as to be in conformity with hera standards .
the one jet inclusive cross section is obtained by identifying @xmath128 and integrating ( [ d3s2 ] ) over @xmath129 $ ] , where @xmath130 and @xmath131 . in our case , the two jet cross section is the integral over @xmath123 with the lower limit @xmath132 , so that the rapidity gap is always determined by @xmath133 . in the following ,
this cross section is denoted by @xmath134 , while the inclusive one jet cross section is called @xmath135 . as the last point
, we write down the cross section for the dd process , with direct photon and direct pomeron in the initial state .
this amounts to putting @xmath136 in ( [ d3s ] ) , so that the inclusive one jet cross section is [ ddone ] ( 1 jet ) = _ c , d _ x_^min^x_^max dx_g_/e(x _ ) x_ip^2 e^y g_ip / p(x_ip ) ( ipcd ) , where x_ip= , i.e. , @xmath121 is fixed by @xmath7 , @xmath5 and @xmath120 . on the other hand , the condition @xmath137 determines the lower bound of integration in ( [ ddone ] ) , x_^min & = & . in the numerical analysis
, we shall exclude @xmath120 values outside the interval @xmath138 $ ] .
then , we shall have @xmath139 . for fixed @xmath5 ,
the allowed rapidity interval is fully determined by the condition @xmath140 .
in this section , we present results for the one and two jet cross sections as defined in the last section with kinematical constraints as used in the data analyses by h1 and zeus .
the results are given for the hera frame with @xmath141 , @xmath142 and the positive @xmath110 axis pointing in the incoming proton direction .
we choose @xmath88 for the upper cut on @xmath121 and use the structure functions for electron , photon and pomeron that were specified in the previous section .
figure [ yds ] shows @xmath143 for the one jet inclusive ( a c ) and for the two jet cross section ( d f ) .
the cross sections are plotted as a function of rapidity at @xmath144 .
we shall first concentrate on one jet inclusive production . in fig .
[ yds]a , the direct pomeron coupling , @xmath32 , is put to zero .
therefore , we have only results for dr and rr . in fig . [ yds]a , we give also the one jet inclusive cross section for the usual photoproduction of jets for comparison .
as is well known , at @xmath144 , the resolved photon cross section ( r ) dominates the direct photon cross section ( d ) .
the diffractive jet cross sections are limited to rapidity values @xmath145 .
( the precise interval will be specified below . )
this is due to the @xmath146 cut or , experimentally , the cut on @xmath98 , the invariant mass of the diffractively produced final state .
the dr and rr components have cross sections of the same order .
the dr cross section exceeds the rr one at negative rapidities . in fig .
[ yds]b , the direct pomeron coupling is added with @xmath80 .
this produces a significant cross section in dd , i.e. , with direct photon and direct pomeron coupling .
this cross section is peaked for negative rapidities and is of similar magnitude as the dr component .
the rd component is small compared to all others .
we emphasize that , at @xmath144 , the direct pomeron predominantly contributes together with the
_ direct photon_. figure [ yds]c displays the inclusive one jet cross sections including also the point like part of the quark distribution of the pomeron given by ( [ poi ] ) , again with the normalization @xmath80 .
this affects only the dr and rr components . by comparing the respective curves in figs .
[ yds]b , c , we see that the point
like piece of the pomeron structure function leads to a modest increase of these cross sections , by some @xmath147 .
this moderate increase is explained by the fact that the gluon structure function of the pomeron is the dominant part in the dr and rr components . in figs .
[ yds]d f , we repeat the analyses of figs .
[ yds]a c for the two jet cross sections .
the restriction on the rapidity of the second jet leads to a decrease of the cross sections and limits the rapidity range in the electron direction . in the c.m .
frame , the lower edge of the rapidity spectrum now appears at @xmath148 , while in the one jet case it occurs at --.in both cases , the upper edge lies at .the rapidity in the laboratory frame is shifted by @xmath149 relative to the c.m .
> from these results it is obvious that the rapidity distribution for a fixed @xmath5 value near @xmath150 is not useful to distinguish the direct and resolved pomeron contributions . the two combinations dd @xmath151 rd and dr @xmath151 rr are very similar in their rapidity dependence . figures [ sumyds]a , b show the sum over the various components in fig . [ yds ] for the inclusive one jet and two jet cross sections , respectively . in fig .
[ sumyds]a , we compare our results with recent h1 data @xcite .
since these data are given as event rates , we can only compare the shapes .
the data point with lowest rapidity is fitted to the theoretical curve , which is in agreement with the other data points within the experimental errors . for simplicity , we choose the normalization factor to be 1/2 .
the @xmath5 dependence of the cross sections for fixed rapidity is more discriminative .
this can be seen from the results in fig .
[ pts ] . in figs .
[ pts]a c , we have plotted the inclusive one jet cross section as a function of @xmath5 for fixed @xmath152 , again for the three cases @xmath95 , @xmath80 and @xmath80 with point like pomeron structure function .
the dd component has a much slower fall off in @xmath5 as compared to the dr component and dominates the other components for @xmath153 .
this harder @xmath5 dependence is expected , since the hard scattering cross section of @xmath38 is proportional to @xmath51 . in the dr component , this dependence is softened through the pomeron structure functions .
the two jet cross sections show an almost identical pattern , since the additional restriction for the second jet is less effective , if @xmath7 is held fixed near the maximum of the rapidity distribution .
the shallow @xmath5 dependence of the dd ( and , to a minor extent , of the rd ) component is still visible in the sum over all components .
this is seen in figs .
[ sumpts]a , b for the inclusive one jet and two jet cross sections , respectively , when we compare the curves for @xmath80 and @xmath95 . in figs .
[ sumpts]a , b , we also compare our results with recent h1 @xcite and zeus data @xcite , respectively . similarly to fig .
[ sumyds]a , we can only compare the shapes .
the points with lowest @xmath5 are adjusted so as to match approximately with the theoretical curves , again using a normalization factor of 1/2 .
the data seem to show the shallow @xmath5 dependence characteristic for the combined direct pomeron contribution , dd @xmath151 rd .
however , this observation should be taken with a grain of salt .
firstly , these data are uncorrected . secondly , for a serious comparison , our model should include hadronization , which is certainly important for the low effective c.m .
energies in the @xmath154 channel .
therefore , we believe that it is premature to draw any firm conclusions from these comparisons .
there are two other features in the zeus data that are indicative of a direct pomeron coupling .
the @xmath155 distribution is peaked near @xmath156 , and the @xmath157 distribution , i.e. , the dependence on the pomeron structure function variable @xmath157 , has also a maximum near @xmath158 . but also here
, conclusions can be drawn only when corrected data are available and hadronization effects are included in our model .
in this work , we studied the effect of a direct pomeron coupling to quarks , which leads to lossless diffractive jet production .
we considered @xmath0 scattering under hera conditions with almost real photons .
we tried to conform to the experimental conditions of the h1 and zeus experiments as much as possible .
only lo processes were investigated , and the hadronization of the parton jets was neglected . the most important signature of a direct pomeron coupling is the @xmath5 distribution in inclusive one jet and two jet production .
it leads to a much flatter @xmath5 distribution than in the case of a resolved pomeron .
the direct photon coupling gives the dominant contribution at @xmath153 .
most of our results are model
dependent , since we have no a
priori knowledge about the strength of the pomeron coupling and about the relation of the quark to the gluon components of the pomeron structure function .
we considered the gluon part to be dominant over the quark part in the ratio @xmath159 .
so far , the existing preliminary experimental information on jet production with almost real photons in rapidity gap events observed by h1 and zeus is consistent with this assumption .
definite conclusions are not possible at this time , since these data are not yet corrected and our model does not contain hadronization corrections , which are likely to be important at these low energies in the @xmath154 c.m . system .
we are aware of the fact that the assumption of a direct pomeron coupling to quarks is quite unconventional , since the pomeron is a very complex object , which , at first hand , is not expected to behave similarly to a photon . on the other hand
, theory can not guide us .
so , only data can tell whether our assumptions are realistic .
there exist several modifications of the ideas presented in this work .
firstly , the direct pomeron coupling to quarks could contain an additional form factor which depends on @xmath5 . in this case
, we would expect that the @xmath5 dependence of the dd component is not as flat as in fig .
therefore , it would be much harder to detect such a direct coupling .
secondly , the direct coupling of the pomeron might be to gluons and not to quarks or to both . for a direct gluon coupling of the pomeron ,
the direct photon contribution does not exist in lo .
therefore , such a coupling could be seen only in connection with a resolved photon .
of course , for this case , there is also no point like component in the pomeron structure function .
if direct pomeron couplings both to quarks and to gluons exist , it will be difficult to disentangle the two .
it is conceivable that , when the experimental range of @xmath121 can be increased , also other regge exchanges between the primary proton vertex and the hard scattering process will become relevant .
such exchanges , like @xmath160 , @xmath161 , etc . , have been considered in the past as a method for extracting the deep
inelastic structure of reggeons @xcite when the incoming photon is highly virtual .
such exchanges are also possible in @xmath0 scattering with almost real photons . in this case , a direct @xmath160 ( @xmath161 , etc . )
coupling to quarks is even more natural than a direct pomeron coupling to quarks . actually , in a recent publication @xcite by the e683 collaboration at fermilab on jet production by real photons in the fixed
target energy range , it is reported that the energy flow in the photon ( forward ) direction is very similar in @xmath162 and @xmath163 reactions at comparable energies . since @xmath162 processes at such low c.m .
energies are dominated by _
photoproduction , this result can be interpreted by assuming that also the pion has an appreciable direct coupling to quarks , in contrast to nave expectations .
similar results were reported earlier by the e609 collaboration @xcite for hard @xmath163 collisions at @xmath164 on a fixed target .
these authors found evidence for two jet events with little or no energy flow in the forward direction .
they interpreted these results in terms of a higher twist process suggested by berger and brodsky @xcite . in this model ,
the pion structure function has a delta function component with a form factor that is related to the electromagnetic form factor of the pion . except for these higher twist factors , which reduce the cross section and
produce a steeper @xmath5 dependence , the cross section has the same characteristics as the one of jet production in @xmath162 collisions with a direct photon .
it is possible that the pomeron , too , has such a higher twist component .
this means that , independently of whether the pomeron model with a leading twist direct coupling proposed in this work is true or not , it might be reasonable to look for lossless diffractive jet production in @xmath0 collisions , which could be a signature for a higher twist direct coupling .
we remark that the c.m .
energies in the @xmath162 collisions considered in this work are of the same order as those in the @xmath163 fixed target experiments .
we would like to thank g. wolf for useful comments on the zeus data on diffractive photoproduction of jets and g. ingelman for carefully reading the manuscript .
one of us ( bak ) is indebted to the kek theory group for the warm hospitality extended to him during his visit , when this work was finalized .
j. dainton , preprint ral94012 ( 1994 ) , also in _ proceedings of the xvi international symposium on lepton and photon interactions _ , ithaca , new york , 1993 , edited by p. drell and d. rubin ( aip press , new york , 1994 ) p. 87 ; a. de roeck , h1 collaboration , preprint desy 94005 ( 1994 ) , also in _ proceedings of the international europhysics conference on high energy physics _ , marseille , france , 1993 , edited by j. carr and m. perrottet ( editions frontires , gif - sur - yvette , 1994 ) p. 791 .
p. bruni and g. ingelman , pompyt 1.0 ( to be published ) ; preprint desy 93187 ( 1993 ) , also in _ proceedings of the international europhysics conference on high energy physics _ ,
marseille , france , 1993 , edited by j. carr and m. perrottet ( editions frontires , gif - sur - yvette , 1994 ) p. 595 ; see also h. jung , rapgap ; preprint desy 93182 ( 1993 ) .
r. devenish , lecture delivered at the _ mini school on diffraction at hera _ , desy , may 5 , 1994 ; g. wolf , talk presented at the _ international workshop on deep inelastic scattering ( dis ) and related subjects _ , february 611 , 1994 , eilat , israel .
j. botts , j.g .
morfin , j.f .
owens , j. qiu , w.k .
tung and h. weerts , cteq collaboration , phys .
* b304 * ( 1993 ) 159 , now superseded by j. botts , h.l .
lai , j.g .
morfin , j.f .
owens , j. qiu and w.k .
tung , cteq2 collaboration , preprint msuhep93/28 ( 1993 ) ; k. charchua , comp .
commun . 69 ( 1992 ) 360 .
sullivan , phys .
* d5 * ( 1972 ) 1732 ; g. kramer , in _ proceedings of the seminar on @xmath44@xmath165 and @xmath44@xmath44 storage rings _ , edited by j.k .
bienlein , i. dammann and h. wiedemann , preprint desy 73/66 ( 1973 ) , p. 3 ; g. altarelli , n. cabibbo , l. maiani and r. petronzio , nucl .
* b92 * ( 1975 ) 413 ; n.s .
craigie and g. schierholz , nucl .
phys . * b100 * ( 1975 ) 125 . the @xmath67 dependence of the point like distribution function , @xmath68 , for various values of @xmath28 and for the three quark masses a ) @xmath167 , b ) @xmath168 , c ) @xmath169 ; the @xmath170weighted sum , @xmath171 , where @xmath172 , is shown in d ) .
@xmath86 compared to preliminary h1 data .
the solid curves represent only the contributions from quarks in the resolved pomeron , while for the dashed lines the point like contribution with @xmath80 is included .
the rapidity distributions of the one jet [ a)c ) ] and two jet [ d)f ) ] cross sections for fixed @xmath144 in the @xmath0 laboratory system . here
@xmath7 is defined to be positive for jets travelling in the proton direction .
figures a ) , d ) show the non diffractive cross sections with a direct ( d ) and resolved ( r ) photon in comparison with the dr and rr contributions in our model .
due to @xmath95 , there are no dd or rd contributions . in figs .
b ) , e ) , these contributions are included for @xmath80 but without the point like ( pl ) component of the pomeron .
this contribution is switched on finally in figs .
c ) , f ) .
the @xmath7 distributions of the a ) one jet and b ) two jet cross sections for fixed transverse momentum @xmath144 after summing up the various contributions shown in fig . 4 .
for a direct comparison with data , the one jet events from h1 with suitable normalization are also shown .
the @xmath5 spectra of the a ) one jet and b ) two jet cross sections for fixed rapidity @xmath152 after summing up the various contributions shown in fig .
[ pts ] . for a direct comparison with data ,
the one jet events from h1 and the two jet events from zeus with suitable normalizations are also shown .
like in figs .
4a ) , d ) , we have also plotted the non diffractive result d+r . | we investigate in detail the effect of a direct pomeron coupling to quarks on the production of jets in @xmath0 scattering with almost real photons .
jet production via a direct pomeron coupling is compared with the resolved pomeron mechanism .
we consider both direct and resolved photoproduction .
rapidity and transverse momentum distributions are calculated and compared with preliminary h1 and zeus data . -3 mm 16.6 cm 23.cm |
the large mass - to - light ratios ( @xmath24 ) of galaxy clusters ( @xmath29 , e.g. , * ? ? ?
* ) indicate that they contain large quantities of dark matter ( dm hereafter ) .
the nature and distribution of dm is the current subject of much theoretical work in cosmology , with detailed simulations yielding different expectations for the amount and distribution of dm in cluster cores ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
x - ray observations of the density and temperature of the hot intracluster medium ( icm ) gas probe the mass of a galaxy cluster , under the assumption of hydrostatic equilibrium .
such data potentially provide constraints on cluster dm simulations , and thus test dm theory ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
prior generations of x - ray satellites have provided a wealth of cluster observations , from which we have begun to build a picture of large - scale radial temperature variations ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , as well as two dimensional temperature maps of disturbed systems ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , detailed temperature and density profiles at the smallest scales ( @xmath30 ) exist for only a few systems , such as virgo @xcite , which exhibit irregularities in their cores .
while gravitational lensing studies provide a unique and important probe of dm in cluster cores @xcite , they generally can not be applied to nearby systems , and they may also be contaminated by other sources of mass along the line of sight . the advent of the _ chandra _ and _ xmm - newton _ satellites now allows us to measure the icm properties with simultaneous spatial resolution comparable to optical studies , thus providing a completely independent mass estimator over the same spatial scales .
the main criticism levelled at x - ray studies is the veracity of the hydrostatic equilibrium assumption in the actual clusters under study .
several groups have now obtained x - ray constraints on the dm profiles of clusters of galaxies , which are apparently consistent with the cdm paradigm ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
such systems are either too distant for a detailed analysis at @xmath31 , or they contain morphological disturbances indicating possible departures from hydrostatic equilibrium , especially at @xmath32@xmath3 kpc .
although simulations suggest that the x - ray analysis of such clusters is generally unaffected @xcite , some authors argue that mass estimates will be biased at small radii in such systems , which may reflect the majority of clusters ( see especially * ? ? ?
abell 2029 ( a2029 ) is a nearby ( @xmath33 ) , well - studied cluster of galaxies which provides an excellent opportunity to probe the dark matter content of a massive object .
it has a very high x - ray flux and luminosity , as well as a hot icm , indicating a massive system .
we have previously analyzed the _ chandra _ observations of a2029 in ( * ? ? ?
* paper 1 hereafter ) , presenting the temperature and fe abundance data .
it exhibits a very regular optical and x - ray morphology , and is an excellent example of a relaxed system with no evidence of disturbances ( e.g. , shock fronts , filaments , or `` cold fronts '' ) present in other systems . in a morphological analysis of 59 clusters
, @xcite found the global x - ray morphology of a2029 to be among the most regular in the sample . though it contains a wide angle tail ( wat ) radio source ,
there are no coincident x - ray `` holes '' , such as those found in the clusters hydra a @xcite , and perseus @xcite .
there is no optical or x - ray spectroscopic evidence for a cooling flow , though the x - ray temperature drops to @xmath34kev in the central @xmath35 ( 7@xmath3 kpc ) .
thus , this system is almost uniquely well - suited for analysis of its mass distribution since the hydrostatic equilibrium assumption should apply with high accuracy . in the current paper
we present estimates of the gas density and temperature profiles ( [ sec_obs ] ) , the total mass and the dm density profile ( [ sec_mtot ] ) , as well as the gas mass ( [ sec_gas ] ) . in [ sec_dis ]
we make a comparison with the stellar mass profile .
we discuss the implications of our analysis and present our conclusions in [ sec_conc ] . throughout this paper , we assume a cosmology of h@xmath36 @xmath37 km s@xmath38 mpc@xmath38 , @xmath39 , and @xmath40 , implying a luminosity distance to a2029 of 347 @xmath3 mpc and an angular scale of 1.45 kpc arcsec@xmath38 .
a2029 was observed by the _ chandra _
observatory for 20ks on the acis s3 chip at a focal plane temperature of -120 c ( see paper 1 for details ) .
we have reanalyzed the data in a similar manner to paper 1 , but using the latest version of the _ chandra _ calibration ( caldb 2.15 ) .
briefly , the data were processed to mitigate the effects of charge transfer inefficiency ( cti ) using the aciscticorrector.1.37 software @xcite .
we then subtracted the available source - free extragalactic sky background maps using the make_acisbg software created by maxim markevitch ( see , e.g. * ? ? ?
we have also applied the latest correction to the arf files to account for an apparent time - dependent degradation in qe ( using the `` corrarf '' routine for a description of the effect and links to the software . ] ) .
this correction has a small but significant effect on our fitted temperatures at larger radii , resulting in systematically lower values than our previous analysis ( see [ subsec_sys ] regarding the effect on the mass profile ) .
we extracted spectra in concentric , circular annuli centered on the peak of the x - ray emission .
we have used a different set of annuli than that used in paper 1 , to optimize the constraints on the shape of the mass profile in the center as well as the slope of the temperature profile at all radii .
this results in a total of 7 annuli , though this choice does not affect our results ( see [ subsec_sys ] ) .
our primary analysis relies on spectra in the energy range 0.7 - 8.0 kev , but we have explored the effects of using a lower limit of 0.5 or 1.0 kev ( see [ subsec_sys ] ) . using xspec , we fit the extracted spectra with the apec plasma emission model absorbed by galactic neutral hydrogen . we adopt the weighted - average galactic value of n@xmath41 @xmath42 obtained from the w3n@xmath43 heasarc tool .
the model normalization ( from which we calculate the electron density , @xmath44 , and gas mass density , @xmath45 ) , the gas temperature , @xmath46 , and the fe abundance were allowed to be free parameters in the fits , with all other elements tied to fe in their solar ratios . to properly recover the three - dimensional properties of the x - ray emitting icm
, we have performed a spectral deprojection analysis using the xdeproj code of @xcite .
we start at the outside working our way to the core in an `` onion - peeling '' method which accounts for the cumulative projected emission from the outer annuli . for details of our deprojection technique see @xcite and @xcite . to estimate the uncertainties on the fitted parameters we simulated spectra for each annulus using the best - fitting models and fit the simulated spectra in exactly the same manner as the actual data . from 100 monte carlo simulations
we compute the standard deviation for each free parameter which we quote as the `` 1@xmath47 '' error . in figure [ fig_rho_t ]
we present the radial gas density and temperature data for a2029 ( left and right panels , respectively ) .
the horizontal bars indicate the widths of the annuli used , and are not error bars .
note the very small statistical errors in both @xmath45 and @xmath48 , which indicate the precise spectral constraints obtained from these data .
since we wish to fit smooth , parameterized functions to the data ( see below ) , we choose to assign an emission - weighted effective radius , @xmath49 , to each annulus @xmath50 ( see ,
e.g. , * ? ? ?
* ) : @xmath51^{2/3}. \label{eq_re}\ ] ] while @xmath49 should strictly be iteratively calculated from the fitted density profile , it is easy to show that eq .
[ eq_re ] is an excellent approximation to the correct @xmath49 over the entire profile ( e.g. , * ? ? ?
* see also [ subsec_sys ] ) .
we make the assumptions of hydrostatic equilibrium and spherical symmetry , such that the enclosed mass is @xmath52 , \label{eq_mtot}\ ] ] where @xmath53 is boltzmann s constant , @xmath54 is the constant of gravitation , @xmath55 is the mean atomic weight of the gas ( taken to be 0.62 ) , and @xmath56 is the atomic mass unit . to obtain the instantaneous logarithmic derivatives necessary to evaluate eq .
[ eq_mtot ] , we fit parameterized models to both the density and temperature . by parameterizing the @xmath45 and @xmath48 data
we derive a mass profile that may be smoother than the true mass distribution .
this approach , therefore , is best suited for interpreting average properties of @xmath57 , such as its radial slope and comparison with dm simulations ( also smooth ) , which are the focus of the present paper .
key advantages of this method are that it is simple to implement , and the mass profile is straightforward to interpret in terms of the input @xmath45 and @xmath48 profiles .
we initially fit the gas density data with the ubiquitous @xmath5-model : @xmath58^{-3\beta/2 } , \label{eq_beta}\ ] ] where @xmath59 is the central gas density , @xmath60 is the core radius , and @xmath61 is the slope of the profile at @xmath62 . the result is overlaid on the data as a dotted curve ( fig .
[ fig_rho_t ] , left panel ) . due to a peak in the profile at @xmath63@xmath3 kpc ( the first 3 data points )
, the @xmath5-model does not provide an acceptable fit ( see table [ tab_rho_t ] below ) .
table [ tab_rho_t ] : gas density and temperature fits [ cols="<,>,^,<,^ " , ] for the cusp model , we find @xmath64 g @xmath65 . for the double-@xmath5 model , @xmath66 g @xmath65 and @xmath67 g @xmath65 .
[ tab_rho_t ] we explored two additional models : ( 1 ) the ` cusp ' model , which is a modified @xmath5 model given by @xmath68^{-3\beta/2 + \alpha_{\rho}/2 } , \label{eq_cusp}\ ] ] where @xmath69 , and the @xmath70 parameter allows a steepening of the profile at @xmath71 , and ( 2 ) a double-@xmath5 model ( e.g. , * ? ? ?
* ; * ? ? ?
* ) given by @xmath72 where @xmath73 and @xmath74 are each given by eq .
[ eq_beta ] .
the double-@xmath5 and cusp models both provide satisfactory fits to the data ( dashed and solid curves , respectively , figure [ fig_rho_t ] , left panel ) , though the reduced @xmath75 is slightly improved for the double-@xmath5 model .
we present the results of the gas density fits in table [ tab_rho_t ] .
it is apparent that both the cusp and the double-@xmath5 models are sensitive to a break at @xmath76 , and that the parameters for the first component of the double-@xmath5 model are not well constrained .
we have chosen the cusp model as our `` reference '' fit for the rest of our analysis for two reasons : ( 1 ) it provides a similar quality fit with two fewer free parameters than the double-@xmath5 model ; ( 2 ) it yields a positive mass in the inner shell , which the @xmath5 models do not ( see [ subsec_sys ] ) .
the temperature data are generally increasing , and are fairly well - fitted by a simple power - law @xmath77 , with @xmath78 ( overlaid as a dashed line ) .
the power - law fit misses the central data point entirely , and predicts a steeply rising slope at the outer point .
a somewhat better fit to the data is obtained with a ( * ? ? ?
* b&m hereafter ) profile ( which adds an additional free parameter ) : @xmath79^{\alpha_t } , \label{eq_tber}\ ] ] where @xmath80 is the asymptotic temperature at large radii .
we have overlaid the two fitted models to the temperature data in figure [ fig_rho_t ] ( _ right panel _ ) . the b&m model ( _ solid curve _ ) provides an improvement in the fit to the data vs. the power - law model ( _ dashed line _ ) at the smallest and largest radii .
the parameters of the fits are given in table [ tab_rho_t ] . from the table
we see that the core radius is not well constrained , and that the temperature asymptotes to @xmath81kev at large radii .
previous measurements of the temperature of a2029 at large radii @xcite , indicate a gas temperature of @xmath82 kev , suggesting a turnover in @xmath48 between @xmath83 @xmath3 kpc .
future work incorporating data at larger radius will thus require a more sophisticated temperature model .
we note that as we found in paper 1 , the icm of a2029 is apparently a single - phase gas at all radii . even within a radius of @xmath84 ,
a single apec model provides an excellent fit to the data , and there is simply no spectral evidence to support an additional component such as a second temperature , a cooling flow model , or a power - law .
the b&m model provides a very good approximation to the @xmath48 data except for underestimation of the fourth data point .
we can not determine if this is a statistical fluctuation , a systematic error in our analysis , or real structure in the @xmath48 profile .
the last case would introduce a correction factor to the mass in that region , otherwise the temperature , and thus the mass profile appears quite smooth ( see [ subsec_sys ] ) . at each radius @xmath50
, we have calculated the total enclosed gravitating mass @xmath85 according to eq .
[ eq_mtot ] , using various density and temperature model fits .
the logarithmic derivatives of @xmath45 and @xmath48 are evaluated at @xmath86 , and @xmath87 is interpolated from the fitted @xmath48 model .
we estimate statistical errors @xmath88 on the mass data as follows : for each monte carlo simulation @xmath89 of the @xmath45 and @xmath48 data ( [ subsec_bin ] ) , we obtain a pair of fitted profiles ( @xmath90 and @xmath91 ) from which we calculate a set of mass values @xmath92 .
we thus obtain 100 mass values at each radius , from which we calculate the standard deviation and report it as the `` 1@xmath47 '' error .
for the reference pair of gas models ( the cusp model for @xmath45 and the b&m model for @xmath48 ) , we present the total enclosed mass of a2029 in the left panel of figure [ fig_mass_compare ] ( _ open circles _ ) .
the profile has a nearly constant slope , with a slight flattening in the inner 3 data points . at our final data point
, we obtain a total enclosed mass of @xmath93 within @xmath11@xmath3 kpc .
the mass calculated using the power - law @xmath48 model is also shown ( _ open squares _ ) .
we note that there is significant uncertainty in the mass at the innermost data point : while the temperature error is relatively high ( @xmath94 ) , the slopes of both @xmath45 and @xmath48 are even less well - constrained .
in fact , using the double-@xmath5 model @xmath45 fit obtains a flat slope at this point , resulting in a negative mass value ( see [ subsec_sys ] ) . to analyze the shape of the mass profile , we fit parameterized models to the best fit mass values . to estimate errors on the parameters of these mass models , we obtain a mass fit @xmath95 for each simulated mass data set @xmath96 . as above
, we therefore obtain 100 values of each parameter in a given mass model , from which we calculate a standard deviation .
we have overlaid in figure [ fig_mass_compare ] ( _ left panel _ ) power - law fits to both mass profiles . the reference mass data are well - fitted by a power - law @xmath97 , with slope @xmath17 ( _ solid line _ ) , over the entire mass range .
the mass data obtained from using the power - law @xmath48 model are slightly higher below @xmath98@xmath3 kpc , and are also well - fitted by a power - law with @xmath99 ( _ dashed line _ ) .
although the power - law is a good visual fit to the mass data , we have examined whether @xmath75 is reduced by instead using a broken power - law ( bpl ) model : @xmath100 where @xmath101 is the radius of the break between inner and outer slopes .
if all the data points are included in the fit , then we obtain @xmath102 @xmath3 kpc , with @xmath103 , and @xmath104 consistent with @xmath105 given above ( these results assumed a cusp @xmath45 model , and either the b&m or power - law @xmath48 model ) . however , if the inner data point is excluded , then the bpl fits are consistent with the single power - law fits given above .
since the inner data point is subject to large systematic error due to the chosen @xmath45 parameterization ( as above , see also
[ subsec_sys ] ) , we can conclude there is no evidence for a significant break in the logarithmic radial mass profile . to account for more gradual deviations from a power law ,
we have also fitted the mass data with the ( * ? ? ?
* nfw hereafter ) profile , @xmath106^{-1}$ ] , the @xcite model , @xmath107^{-1}$ ] , and the ( * ? ? ?
* m99 hereafter ) model , @xmath108^{-1}$ ] . integrating these density profiles obtains mass models for fitting ; analytic forms can be found in the literature ( see , e.g. , * ? ? ?
* ; * ? ? ?
these fits are presented in the next section . assuming the cluster to be dominated by dm ( a point we argue below ) , the fitted total mass profile corresponds to an implicit dm density distribution .
for any power - law fit , @xmath109 , where @xmath110 . for the power - law fit to the reference mass profile
we therefore observe a dark matter density slope of @xmath111 . in figure
[ fig_mass_compare ] , ( _ right panel _ )
we again show the total enclosed mass profile ( _ open circles _ ) . to compare with other clusters , as well as theoretical expectations , it is convenient to scale the radius in terms of the virial radius , @xmath112 .
we calculate one popular predicted value of @xmath112 as a function of emission - weighted global temperature , using the form given by @xcite , eq . 9 , normalized to the 2.0 - 9.5 kev band mass - temperature relation given by @xcite , which we convert to our cosmology .
if we extract one spectrum in this band from the entire region within @xmath113 which we observe with _
chandra _ , we measure a single - temperature fit of @xmath114 kev .
we thus obtain @xmath115@xmath3 mpc for a2029 .
we have plotted the upper axis of fig .
[ fig_mass_compare ] ( _ right panel _ ) in units of @xmath112 , which shows that we are examining the dark matter profile on a scale from @xmath116 .
we have overlaid fits to the mass profile of a2029 from three different mass models : a power - law ( _ dashed line _ ) , an nfw mass model ( _ solid curve _ ) , and an m99 model ( _ dotted curve _ ) .
while the power - law model provides a good overall fit ( @xmath75/dof @xmath117 ) , the nfw model is preferred ( @xmath75/dof @xmath118 ) .
the data are more closely approximated by the nfw profile except for a @xmath119 difference at the innermost data point , which has additional systematic uncertainties ( as noted above ) rendering the discrepancy insignificant ( see [ subsec_sys ] ) .
the hernquist profile fit ( not shown ) is nearly identical to the nfw fit ( @xmath75/dof @xmath120 ) .
the m99 model does not provide an acceptable fit to the overall profile ( @xmath75/dof @xmath121 ) , though its small radius slope is compatible with the inner 3 data points .
however , if fit solely to these points , it falls well below the remainder of the mass profile , which is better constrained .
we note that although the nfw and hernquist models improve @xmath75 significantly , the relative differences between these models vs. the power - law mass model are small ( @xmath122 between 17 and 260 @xmath3 kpc ) . unlike the bpl model , which presents a sharp break , the gradual change in the logarithmic slope of the nfw or hernquist profiles better quantifies the small deviations of the mass data from a pure power - law . for the nfw profile ,
we find a scale radius @xmath123@xmath3 kpc , and a concentration parameter @xmath15 .
this allows us to calculate the value of @xmath112 expected from the nfw model ( @xmath124 ) , for which we obtain @xmath112(nfw)@xmath125@xmath3 mpc , in good agreement with @xmath112 as predicted by the mass - temperature relation given above .
several other authors find that cdm halos ( e.g. , nfw or m99 profiles ) are also consistent with their x - ray cluster observations at @xmath126 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , while gravitational lensing studies report conflicting results in cluster cores ( i.e. , * ? ? ?
* ; * ? ? ?
the hydra a cluster exhibits @xmath127 over a similar range of radii to our analysis @xcite , however , there are prominent interactions between the x - ray gas and the radio source in that system , implying significant deviations from hydrostatic equilibrium .
thus the dm profile inferred for a2029 ( which is comparatively undisturbed in this regime ) , provides evidence that the hydra a results are robust , and is an important independent confirmation of cdm predictions .
we have explored several alternative data reduction and analysis choices to investigate systematic errors in the mass profile .
we have calculated density , temperature and mass profiles in four different sets of annuli : ( 1 ) the original 11 annuli used in paper 1 , ( 2 ) 13 annuli , obtained by dividing the inner 2 annuli of paper 1 into 4 smaller annuli , ( 3 ) 5 annuli , using much larger bins , ( 4 ) 7 annuli , using a combination of the very small annuli from ( 2 ) in the center with the larger , higher s / n annuli from ( 3 ) .
binning choice ( 4 ) is the reference choice for this paper , which shows the most detail in the core , and provides the highest s / n for the temperature estimates at larger radii .
for all 4 choices of binning , the resulting mass profiles are not perceptibly different from the reference results .
we have also analyzed the data in energy ranges of @xmath128 , @xmath129 , and @xmath130 kev , and repeated the entire experiment using a mekal plasma model rather than the apec reference model ; in all cases the slope of the fitted mass profile is within the @xmath131 errors of the reference fit .
the effect on the mass profile of our choice of eq .
[ eq_re ] to calculate @xmath49 can be tested by instead using 2 extreme cases for the estimate of @xmath49 : assumed density slopes of @xmath132 and @xmath133 , which easily bracket all the instantaneous slopes of the observed density profile ( see also * ? ? ?
we find that the mass profile does not vary perceptibly with the choice of @xmath49 .
our reference analysis is to use the same period c background files as in paper 1 .
variations in the x - ray flux of the extragalactic sky between our target and the background templates may render them inaccurate .
we experimented with modifying the background flux by changing the effective exposure time of the templates to @xmath134 of its nominal value .
while this can affect the estimated temperature in the outermost annulus , the overall temperature profile does not vary beyond the @xmath131 errors , nor does the slope of the mass profile .
we have corrected the data using the `` corrarf '' routine ( [ sec_obs ] ) , which results in lower @xmath48 values than we reported in paper 1 , by up to 16% in the outermost bin .
this has the effect of slightly flattening the fitted @xmath48 profile , as well as the mass profile ( the uncorrected profile has overall slope @xmath135 ) .
the calibration of the acis instrument below 1 kev is ongoing , but it is reportedly now accurate at the @xmath136 level ( see footnote [ foot_cor ] ) . in figure [ fig_mass_compare ] ( _ left panel _ ) we show the total mass values obtained from each of the two temperature profiles shown in figure [ fig_rho_t ] ( in both cases , we use the cusp fit to @xmath45 ) .
the mass values all overlap in their @xmath131 error bars , as do the slopes of the mass profiles .
we have also explored the choice of the model which is fit to @xmath45 .
the overall power - law fits to the mass profiles derived from the cusp and double-@xmath5 profiles are indistinguishable .
however , we note that the double-@xmath5 model results in a negative ( and unphysical ) mass value in the central bin ( due to its large statistical uncertainty , it does not affect the overall profile fit ) . the single @xmath5 model ( which is an unacceptable fit to the @xmath45 data ) also obtains a very flat slope for the inner 3 data points which results in negative mass values over that region . to obtain a positive mass , eq .
[ eq_mtot ] requires a negative sum for the logarithmic derivatives : inspection of figure [ fig_rho_t ] shows that the slopes of neither @xmath45 nor @xmath48 are well - constrained at the radius of the innermost data point .
the cusp and double-@xmath5 models diverge in this regime , but the data do not distinguish between them .
if the gas was in fact isothermal at this radius our mass estimate would be positive for all 3 @xmath45 fits , but without much higher s / n data ( at higher spatial resolution ) we can not infer such a state for the gas .
_ we note that exclusion of the innermost data point has no perceptible affect on the mass profile fit .
_ the use of parameterized functions for @xmath45 and @xmath48 has the effect of smoothing those data , in turn resulting in a smoother mass profile ( [ subsec_profiles ] ) .
we have also calculated the mass profile by using a simple `` point - to - point '' estimation of the @xmath45 and @xmath48 slopes where the logarithmic derivatives needed for eq .
[ eq_mtot ] are taken to be the slope between each pair of adjacent points .
the mass values are significantly different only in the fourth and fifth bins , which lie above and below the reference data points , respectively .
nonetheless , the average slope of the mass profile is not altered beyond the @xmath131 limits of the reference fit ( as one would expect , since the reference @xmath45 and @xmath48 models provide good fits to the data ) . aside from the innermost data point , the slopes of @xmath45 and @xmath48 are well constrained regardless of the parameterization , and the overall mass profile is robust .
we have assumed spherical symmetry for a2029 , though its mass distribution exhibits an average ellipticity of 0.5 @xcite within @xmath137@xmath3 mpc inferred from the x - rays and optical light .
however , we have measured the spherically averaged mass distribution , which is quite insensitive to the ellipticity of the system ( including the specific case of a2029 , @xcite ; see also @xcite ) .
this approach is also useful for comparison with simulations which also present a spherically averaged profile ( i.e. , nfw , m99 ) .
we have calculated the total gas mass @xmath139 in each spherical shell directly and summed them to obtain an enclosed gas mass profile ( i.e. , we do not interpolate the @xmath45 values from the fitted profile as we did for @xmath48 in the total mass calculation ; the results with either method are unchanged ) . in figure [ fig_mass_compare ]
( _ right panel _ ) we have overlaid the total enclosed gas mass profile ( _ open diamonds _ ) .
we see that the gas mass at the center of the cluster is @xmath140 of the total mass , rising to @xmath141 at our last measured data point .
this may safely be regarded as a lower limit , as simulations and other measurements suggest a higher asymptotic value for the gas mass fraction in clusters , f@xmath142 ( e.g. , * ? ? ?
* ) , while we only have observations at @xmath143 .
assuming that a2029 contains proportions of dark matter and baryonic matter equal to their universal values , one may place an upper limit on @xmath138 with f@xmath144 . given the universal baryon mass density from big bang nucleosynthesis calculations and recent observations of the deuterium abundance ( @xmath145(bbn ) ) , one finds @xmath146 , where f@xmath147 is the universal baryon mass fraction .
given @xmath148 ( see , e.g. , * ? ? ?
* ) , @xmath149 , and f@xmath150 f@xmath151 from this work , we estimate @xmath152 , in agreement with current estimates ( for a recent review , see * ? ? ?
massive elliptical galaxies such as ngc 720 @xcite , and ngc 4636 @xcite , are apparently dm - dominated all the way into their cores .
this situation is likely to exist in a cd cluster such as a2029 , except at possibly the smallest scales ( @xmath153kpc ) , where the cd stellar mass component may be more important @xcite .
we have found that the gas in a2029 is only a few percent of the total mass in the core of the system . but are we observing the stellar mass ( e.g. , the cd ) ?
@xcite suggests the turnover between stellar and dm - domination should occur at a few tens of kpc ; but although we are specifically observing this regime , we see no significant break in the mass profile ( see [ subsec_mass ] ) .
the outlying fourth @xmath48 point and the central @xmath45 peak over a single @xmath5-model are intriguing , but the balance of the data suggest that any transition from stellar - dominated to dm - dominated regimes is quite smooth in this system .
we have estimated the contribution of stellar matter to the system , by using optical observations of the cd galaxy in a2029 and its diffuse halo .
@xcite conducted a detailed optical analysis of a2029 , and observed that the surface brightness of the cd+halo was well - fitted by a @xcite profile out to @xmath154 .
they report a total @xmath155-band luminosity of @xmath156 for the cd+halo , which is 41% of the total light of the cluster within @xmath157 radius .
we assume the cd+halo to be the vast majority of the light within its effective radius @xmath158 ( 76@xmath3 kpc ) . using the @xcite model as an approximate deprojection of the @xmath159 law
, we normalize to the total light of @xcite , converting to our cosmology , and the @xmath160 band , and obtain a luminosity profile . in figure
[ fig_ml ] , we present the total cluster mass - to - light ratio ( @xmath161 ) profile of the core of a2029 , based on the cd+halo light profile .
it is consistent with a constant @xmath162 within 20@xmath3 kpc , and rises rapidly at larger radius .
dynamical stellar analyses estimate the total masses of galactic systems , but can not directly measure the stellar mass .
attempts to resolve this with population synthesis techniques yield highly uncertain results for the stellar mass - to - light ratio , @xmath163 @xcite .
using these bounding values for @xmath164 we plot in figure [ fig_mass_compare ] ( right panel ) associated stellar mass profiles ( _ dot - dashed curves _ ) for the cd+halo system .
the data indicate that at @xmath165 ( @xmath166@xmath3 kpc ) the stellar mass can potentially dominate the system , depending on the highly uncertain assumed @xmath164 . the stellar material outweighs the gas mass up to a radius between @xmath167 ( @xmath168@xmath3 kpc ) .
if we assert a plausible value , @xmath169 for the system , we would conclude that the cluster is dm dominated down to the smallest scales measured here , which is consistent with a single nfw mass component ( and also with the lack of a significant break in the total mass profile , [ subsec_mass ] ) .
alternatively , the gas density peak over the @xmath5-model at @xmath6@xmath3 kpc , or the outlying @xmath48 data point may indicate the presence of a stellar mass component in excess of ( or differing from ) an nfw dm halo .
an interesting comparison can be made with the optical velocity dispersion profile of a2029 . as noted by @xcite , ic1101 ( the cd galaxy in a2029 )
is one of the only brightest cluster galaxies observed to have an increasing velocity dispersion , @xmath170 , @xcite .
we may regard the stars in the cd galaxy ( from which @xmath170 is measured ) to be simply another family of tracer particles ( as the hot icm gas particles ) bound to the same gravitational potential .
thus , the effective velocity of the icm gas particles may be expected to mirror that of the stars .
we estimate the velocity dispersion for the icm gas as @xmath171 .
the temperature data yield @xmath172 , while the cd velocity dispersion profile of @xcite , measured over the range @xmath173 kpc reveals @xmath174 .
this correspondence is not required for the condition of hydrostatic equilibrium , but the similarity of the two species is striking .
a2029 is one of the few systems where the shape of both the x - ray and optical velocity dispersions can be measured in detail : the consistency between them is further evidence that we are observing a dynamically relaxed system where all the mass components are in equilibrium with the gravitational potential .
we have analyzed high spatial resolution _
chandra _ data of the a2029 cluster of galaxies , obtaining well - constrained icm gas density and temperature profiles on scales of @xmath4 ( @xmath175@xmath3 kpc ) .
fitting smooth functions to these profiles , we obtain mass profiles for a2029 , measuring the shape of the total mass profile at unprecedented accuracy to a very small fraction of the virial radius .
our results are insensitive to most details of the data reduction , but are closely tied to the well - constrained temperature and density distributions .
we find that the shape of the inferred dark matter density at @xmath31 is consistent with the nfw parameterization of cdm halos , but apparently incompatible with that of m99 , though we note that individual objects may be expected to show significant scatter from a mean dm halo profile @xcite .
the consistency of the nfw model with the mass profile all the way down to @xmath19 indicates that there is no need to modify the standard cdm paradigm to fit the dm distribution in this cluster , consistent with x - ray observations of other clusters at larger radii ( see [ subsec_dm ] ) .
this result contrasts with the strong deviations from the cdm predictions observed in the rotation curves of low surface brightness galaxies ( e.g. , * ? ? ?
* ) , and dwarf galaxies ( e.g. , * ? ? ?
* ) which inspired the self - interacting dm model @xcite to explain the relatively flat core density distributions in these galaxies .
hence in light of the good agreement with the nfw profile in clusters , the deviations observed on small galaxy scales do not seem to imply a fundamental problem with the general cdm paradigm .
instead , it is likely that the numerical simulations do not currently account properly for the effects of feedback processes on the formation and evolution of small halos .
our analysis suggests that a2029 is dominated by a single mass ( i.e. , dm ) component at all measured scales below @xmath176 , or that any transition from a stellar mass dominated component in the cd and a dm component is quite gradual .
we also observe a rising gas fraction from @xmath140 to @xmath177 in a2029 , obtaining an upper limit to @xmath178 , consistent with other current studies . , m. , ponman , t. j. , nulsen , p. e. j. , bautz , m. w. , burke , d. j. , david , l. p. , davis , d. , donnelly , r. h. , forman , w. r. , jones , c. , kaastra , j. , kellogg , e. , kim , d .- w . ,
kolodziejczak , j. , mazzotta , p. , pagliaro , a. , patel , s. , van speybroeck , l. , vikhlinin , a. , vrtilek , j. , wise , m. , & zhao , p. 2000 , , 541 , 542 | we have used a high spatial resolution _
chandra _ observation to examine the core mass distribution of the unusually regular cd cluster abell 2029 .
this bright , nearby system is especially well - suited for analysis of its mass distribution under the assumption of hydrostatic equilibrium : it exhibits an undisturbed , symmetric x - ray morphology , and a single - phase intracluster medium ( icm ) . from the deprojected temperature and density profiles we estimate the total mass , and the contributions of the gas and dark matter ( dm ) components from @xmath0 to @xmath1 ( @xmath2@xmath3 kpc , @xmath4 ) .
the gas density profile is not adequately described by a single @xmath5-model fit , due to an increase in the density at the center ( @xmath6@xmath3 kpc , @xmath7 ) , but it is well fitted by either a double @xmath5-model , or a `` cusped '' @xmath5-model .
the temperature data increase as a function of radius and are well - fitted by a @xcite profile and approximately by a power - law @xmath8 , with @xmath9 . using the fitted profiles to obtain smooth functions of density and temperature , we employed the equation of hydrostatic equilibrium to compute the total enclosed mass as a function of radius .
we report a total mass of @xmath10 within @xmath11@xmath3 kpc , using the chosen parameterization of gas density and temperature .
the mass profile is remarkably well described down to @xmath12 by the @xcite ( nfw ) profile , or a @xcite profile , over 2 decades of radius and 3 decades of mass . for the nfw model
, we measure a scale radius @xmath13@xmath3 kpc ( @xmath14 ) and concentration parameter @xmath15 .
the mass profile is also well - approximated by a simple power - law fit ( @xmath16 ) , with @xmath17 ( corresponding to a logarithmic density profile slope of @xmath18 ) .
the density profile is too shallow to be fitted with the profile described by @xcite .
the consistency with nfw down to @xmath19 is incompatible with the flattened core dm profiles predicted for self - interacting dm ( e.g. , @xcite ) , and thus contrasts with the strong deviations from cdm predictions observed in the rotation curves of low surface brightness galaxies and dwarf galaxies .
this suggests that while cdm simulations may adequately describe objects of cluster mass , they do not currently account properly for the formation and evolution of smaller halos . assuming that the cd dominates the optical cluster light within its effective radius ( @xmath20 , 76@xmath3 kpc ) , we observe a total mass - to - light ratio @xmath21 at @xmath22@xmath3 kpc , rising rapidly to @xmath23 beyond 200@xmath3 kpc .
the consistency with a single nfw mass component , and the large @xmath24 suggest the cluster is dm - dominated down to very small radii ( @xmath25 ) .
we observe the icm gas mass to rise from @xmath26 of the total mass in the center to @xmath27 at the limit of our observations .
this provides an upper limit to the current matter density of the universe , @xmath28 . |
in order to understand the nature of the gottfried sum rule violation @xcite two different drell - yan experiments were performed @xcite .
the integrated result for the asymmetry from a more complete fermilab experiment @xcite is @xmath4 \ ; dx$ ] = 0.09 @xmath5 0.02 , to be compared with the nmc result : @xmath4 \ ; dx$ ] = 0.148 @xmath5 0.039 .
the nmc integral asymmetry apears slightly bigger .
it was suggested recently @xcite that the difference can be partly due to large higher - twist effects for @xmath6 .
recently the hermes collaboration @xcite used semi - inclusive unpolarized production of pions to extract the asymmetry .
we discuss briefly a disturbing role of some nonpartonic processes which cloud the extraction of the asymmetry from semi - inclusive dis .
in the quark - parton model ( see fig.1a ) the generalized semi - inclusive structure function can be written as @xmath7 where the sum runs over the quark / antiquark flavours @xmath8 , @xmath9 are quark distribution functions and @xmath10 are so - called fragmentation functions .
the isospin and charge conjugation symmetries allow to reduce the number of fragmentation functions to two : favoured @xmath11 and unfavoured @xmath12 . for strange quarks does not enter the quantity analyzed here ( [ exp_extract ] ) . ] in the qpm one can combine semi - inclusive cross sections for the production of @xmath13 and @xmath14 on proton and neutron targets to isolate a quantity sensitive to the flavour asymmetry @xcite @xmath15 - [ 1 + r(x , z ) ] } { j(z ) [ 1 - r(x , z ) ] + [ 1 + r(x , z ) ] } \ ; , \label{exp_extract}\ ] ] where @xmath16 and @xmath17 . in the absence of other mechanisms
the equation can be used to extract the @xmath18-dependence of the difference @xmath2 . in order to demonstrate the effect of nonpartonic components on the extraction of @xmath2 asymmetry we need to fix the effective fragmentation functions which the main partonic term will be calculated with .
most of the model fragmentation functions were constructed in the context of @xmath19 pion production data , where amount of @xmath13 and @xmath14 produced is equal , and do not manage to describe separately multiplicity distributions of positive and negative pions in dis @xcite .
besides separate yields of @xmath13 and @xmath14 the qpm formula ( [ exp_extract ] ) directly depends also on the ratio of unfavoured and favoured fragmentation functions .
surprisingly only a rather old field - feynman parametrization @xcite provides a good representation of the available @xmath20 pion production data in the hermes kinematical region @xcite .
this parametrization will be used in the following analysis .
for small @xmath21 , as in the case of the hermes experiment , some mechanisms of nonpartonic origin may become important too .
for instance the virtual photon can interact with the nucleon via its intermediate hadronic state .
such a mechanism can be described within the vector dominance model ( vdm ) .
the photon could also fluctuate into a pair of pions , where both or one of them interact with the nucleon .
some exclusive processes can produce pions directly or as decay products of heavier mesons . to our best knowledge none of such processes
has been investigated in the literature .
their influence on the extracted @xmath2 asymmetry also remains unknown . here for illustration we discuss only two of them .
let us start from the vdm component ( see fig.1b ) .
it was shown recently that the inclusion of the vdm contribution and a related modification of the partonic component help to understand the behaviour of structure functions @xmath22 and @xmath23 at small @xmath21 and broad range of bjorken-@xmath18 @xcite .
this model was confirmed by a recent analysis of the @xmath21-dependence of the world data for @xmath6 @xcite .
the model for inclusive structure functions @xcite can be generalized to semi - inclusive production of pions : @xmath24 the second sum above runs over vector mesons @xmath25 and @xmath26 decribes a correction factor due to finite fluctuation time of virtual photon into vector mesons for large @xmath18 @xcite .
the inclusive cross section for pion production in vector meson ( @xmath27 ) scattering off proton and neutron is not known experimentally . in analogy to the total cross section the pion production inclusive cross section @xmath28 can be estimated as : @xmath29 \ ; , \nonumber \\ \sigma(\rho^0
n \rightarrow \pi^{\pm } x ) & \approx & 1/2 \ ; [ \sigma(\pi^+ n \rightarrow \pi^{\pm } x ) + \sigma(\pi^- n \rightarrow \pi^{\pm } x ) ] \ ; .\end{aligned}\ ] ] experimental data from the abbcchw collaboration @xcite at @xmath30 = 8 , 16 gev correspond approximately to the range of the hermes experiment @xcite . however , the experimental spectra for @xmath31 contain components due to peripheral processes , which are specific , different for different beams .
we wish to note that peripheral processes in the @xmath32 and @xmath33 reactions do not contribute to the @xmath34 reaction and should be eliminated ; only nondiffractive components for @xmath35 reactions should be taken into account .
this requires a physically motivated parametrization of the @xmath36 data .
therefore we have parametrized the experimental differential cross sections for four different reactions @xmath31 from @xcite as a sum of central and peripheral components @xmath37 the details of the analysis will be given elsewhere @xcite . because the cm - energy of the abbcchw collaboration is very similar to that of the hermes experiment , we believe that the parametrization is suitable in the limiting range of energy relevant for the hermes experiment @xcite .
the cross sections on the neutron can be obtained from those on the proton by assuming isospin symmetry for the hadronic reactions .
the analysis of experimental data @xcite combined with the assumption of isospin symmetry strongly indicate that for the _ nondiffractive _ components @xcite @xmath38 this automatically means that the vdm contribution modifies the r.h.s .
of eq.([exp_extract ] ) . in fig.2
we show a modification of the measured hermes quantity @xmath39 due to the vdm component . in the present calculation
the photon - proton cm energy was fixed at the average hermes value @xmath40 = 5.0 gev@xmath41 and the quark fragmentation component was rescalled by a factor @xmath42 ( see eq.([su_semi ] ) ) .
the solid line represents @xmath39 obtained directly from the parton distributions @xcite . as can be seen from the figure the r.h.s .
of eq.([exp_extract ] ) clearly deviates from the partonic result .
thus , if not canceled by other effects , the quark flavour asymmetry extracted from semi - inclusive experiments in the simple qpm approach seems to be highly overestimated . the exclusive meson production @xmath44 is not included in the fragmentation formalism and may also modify the extraction of @xmath2 asymmetry .
the pion exclusive channels ( @xmath45 ) contribute at @xmath46 1 , i.e. outside of the range of the hermes kinematics and will be ignored here .
the pions from decays of light vector mesons can be important in the context of the @xmath2 asymmetry .
the production of @xmath43 mesons ( @xmath47 ) seems to be of particular importance .
first of all the @xmath48 channel is know to be the dominant exclusive channel in the @xmath49 scattering . secondly , because @xmath3 decays predominantly into two pions it will produce pions with @xmath50 .
a detailed calculation @xcite shows that the dispersion of the decay - pion @xmath51-distribution is large and therefore this effect could be observed at large @xmath51 where hadronization rate is rather small . below as an example we shall consider the @xmath3 elastic production only . the elastic @xmath3-production contribution ( diagram ( c ) in fig.1 ) to semi - inclusive structure function can be written formally as @xmath52 the cross section for proton and neutron target and their difference can be estimated within the regge approach as well as in a qcd inspired quark - exchange model .
it is not clear a priori what is the applicability range of these models . in this short note
we shall try to understand the elastic @xmath3 meson production only within the regge phenomenology .
this requires an analysis of relevant experimental data for the proton and deuteron targets simultaneously .
while for the proton target there are data in quite a broad kinematical range of @xmath18 and @xmath21 @xcite ( although slightly different from the hermes kinematics ) , there is almost no data for the deuteron target .
we have parametrized the existing experimental data for exclusive @xmath3 production by means of the following simple regge - inspired reaction amplitude @xmath53 \cdot c_{is}(t ) \left ( \frac{s}{s_0 } \right)^{-1/2 } \nonumber \\ & \pm & \left [ \frac{-1+i}{\sqrt{2 } } \right ] \cdot c_{iv}(t ) \left ( \frac{s}{s_0 } \right)^{-1/2 } % \right ] ) \nonumber \\ & & \cdot \left [ \frac{m_{\rho}^2}{m_{\rho}^2+q^2 } \right ] \delta_{\lambda_{n ' } \lambda_{n } } \delta_{\lambda_{v } \lambda_{\gamma } } \label{regge_amplitude}\end{aligned}\ ] ] with `` @xmath54 '' for proton and `` @xmath55 '' for neutron , respectively . in practical application
we shall assume the same t - dependence of @xmath56 , @xmath57 and @xmath58 and take @xmath59 .
the free parameters in eq.([regge_amplitude ] ) i.e. @xmath60 , @xmath56 and @xmath61 have been fitted to the existing experimental data for @xmath62-integrated differential cross section for @xmath3 production on hydrogen @xcite . in this fit
we have limited to the experimental data with @xmath63 3 gev ( to avoid resonances ) and @xmath64 10 gev@xmath41 ( above genuine hard qcd processes should reveal ) and fixed the slope parameter @xmath65 = 6 gev@xmath66 in exponential @xmath62-distribution which is known experimentally .
all other details of the analysis will be discussed in @xcite . from the fit to the proton data we have obtained only a sum of the coefficients @xmath61 .
the separation into the isoscalar and isovector contributions can not be done in a model independent way .
the size of the isovector @xmath67-exchange contribution was estimated long ago for total photoproduction cross section ( see for instance @xcite ) .
our regge model can be applied to both real and virtual photoproduction and exclusive as well as inclusive case @xcite .
we have used the data for real photoproduction total cross section and the data for exclusive @xmath68 photoproduction to estimate the size of the isovector amplitude @xcite i.e. the strength of @xmath67-reggeon exchange .
although the difference of the cross sections for exclusive @xmath3 photoproduction on the neutron / proton targets calculated with resulting amplitudes is small , its effect on the @xmath69 ratio extracted by the hermes collaboration , shown in fig.3 , is not negligible at all .
we show in the figure the `` measured '' hermes quantity as a function of bjorken-@xmath18 for a few values of @xmath51 for a fixed @xmath40 = 5 gev . as in the case of the vdm contribution
the qpm term was modified by the factor @xmath42 according to our prescription for inclusive structure functions @xcite .
the semi - inclusive production of pions was recently used to determine the @xmath70 asymmetry in the nucleon sea . in the present analysis
we have investigated a few effects beyond the quark - parton model which may influence the so - extracted asymmetry . according to our estimation the interaction of the resolved hadron - like component of the photon with the nucleon leads to an artificial enhancement of the measured @xmath2 asymmetry in the region of small @xmath18 .
this enhancement depends on @xmath51 very weakly .
the elastic production of @xmath3 meson is equally important .
this effect , however , modifies the measured @xmath2 asymmetry in the opposite direction from the resolved photon component , but the two effects cancel only within a narrow interval of @xmath51 .
the elastic @xmath3 production makes the r.h.s . of eq.([exp_extract ] ) @xmath51-dependent invalidating somewhat averaging done in @xcite .
here we have only shortly discussed two effects .
a more detailed analysis of these two and other effects will be presented elsewhere @xcite .
finally we wish to conclude that nonperturbative effects beyond qpm may substantially disturb the extraction of the @xmath2 asymmetry from semi - inclusive production of pions in dis .
such an extraction requires a carefull combined analysis including many nonpartonic effects together with main partonic term .
in addition we would like to point out that some of the effects discussed in the present paper may also influence the extraction of the polarized quark distributions from semi - inclusive production of pions in dis .
* acknowledgments * we are indebted to the members of the hermes collaboration for discussions of their recent results .
this work was partially supported by the german - polish dlr exchange program , grant number pol-028 - 98 .
dakin et al .
, phys . rev .
* d8 * ( 1973 ) 687 ; + p. joos et al .
, nucl . phys . *
b113 * ( 1976 ) 53 ; + c. del papa et al .
, phys . rev .
* d19 * ( 1979 ) 1303 ; + d.g .
cassel et al . , phys .
* d24 * ( 1981 ) 2787 ; + i. cohen et al .
, phys . rev .
* d25 * ( 1982 ) 634 ; + w.d .
shambroom et al . , phys . rev .
* d26 * ( 1982 ) 1 ; + j.j .
aubert et al .
( em collaboration ) , phys .
b161 * ( 1985 ) 203 ; + m.r .
adams et al .
( e665 collaboration ) , z. phys . * c74 * ( 1997 ) 237 ; + j. breitweg et al .
( zeus collaboration ) , eur .
* c6 * ( 1999 ) 603 ; + c. adloff et al .
( h1 collaboration ) , hep - ex/9902019 , submitted to eur . | we discuss the role of some nonpartonic effects which lead to @xmath0 and @xmath1 and may therefore modify the conclusion on the @xmath2 asymmetry extracted from semi - inclusive production of pions in dis .
quantitative estimations for resolved photon and exclusive @xmath3 are given as examples .
the results are discussed in the context of recent hermes data . |
quantum information is built around two - state quantum systems , qubits @xcite , which are manipulated by applying single- or few - qubit gates @xcite , and can be entangled as required by a quantum algorithm @xcite . in order to perform a sustainable quantum computation and benefit from quantum codes @xcite
, an error correction @xcite must take place to battle decoherence @xcite .
error correction , a quantum algorithm itself , rely on multiqubit entanglement and must be performed on a sufficiently coherent system to be successful @xcite .
therefore quantum operations involving multiple qubits must be sufficiently fast to enable continuous error correction .
although alternative concepts exist @xcite , in many physical implementations each qubit is a separate physical system @xcite isolated form other qubits . as the result ,
single qubit operations are easy to perform , while entangling operations are hard the qubits must be set to interact temporarily . in this investigation
we focus on systems where interactions are mediated via cavity modes ( as , e.g. , in superconducting transmon systems @xcite ) . in these systems
multiqubit gates are performed via simpler one- and two - qubit gates @xcite .
thus , temporal complexity ( speed of operation ) of a typical entangling multiqubit gate @xcite grows rapidly with the number of qubits , @xmath0 .
substantial effort has been made to reduce the gate operation time by optimizing two - qubit representations of multiqubit gates @xcite . although some speed - up ( a constant factor ) can be achieved this way @xcite , the complexity class type of functional dependence on @xmath0remains the same in most cases .
we propose an approach to mutiqubit entangling operations that relies on accessible extended hilbert space in which there is a set of always interacting ( non - local ) states involving multiple qubit systems .
unlike in the concept of qudits multistate quantum units we discuss excited states that can naturally remain interacting ( non - local ) when the qubit states are isolated from each other .
auxiliary excited states have been widely used to perform single- or two - qubit entangling gates in different qubit architectures @xcite via optical transitions using , e.g. , @xmath1-system physics [ see fig . [
fig : mqb - scheme1](a ) ] . with this study
we do not aim to suggest utility of non - local auxiliary states , which has been widely accepted @xcite , rather we aim to demonstrate that the available structure of multiple auxiliary states [ see fig .
[ fig : mqb - scheme1](b ) ] can be used to perform multi - qubit operations more effectively .
specifically , we show that multiqubit operations can be performed via quantum walks @xcite through such excited states : quantum data is still encoded in qubits , facilitating existing quantum codes , but the gate is performed by a quantum algorithm in an extended hilbert space .
we show that the combination of two factors : ( i ) _ non - locality of higher - energy states _ and ( ii ) representation of multiqubit gates via a _ quantum walk _
, can speed - up multiqubit operations potentially reducing their complexity class .
non - local states provide extra physical connections between qubits and quantum walks explore multiple quantum trajectories enabled by these connections at once .
this approach can contribute to on - going research on gate compression schemes and can advance quantum algorithms , including error correction , reducing the fault - tolerance threshold .
it can also be of interest in relation to other physical @xcite and biophysical @xcite problems . extended hilbert space : ( a ) a schematic illustration of an entangling operation done via a non - local auxiliary state @xmath2 .
( b ) a schematic illustration of entangling operation on multiple qubits using multiple non - local auxiliary states to optimize ( compress ) the operation . , scaledwidth=80.0% ] many physical implementations of quantum computing systems have multiple well - defined quantum states that are routinely addressed in experiments @xcite .
when such systems interact , these states form superpositions that are no longer localized in each individual system but can spread over a number of them [ see fig . [
fig : mqb - scheme1](b ) ] .
the degree of non - locality of such superpositions is generally different at different energy levels .
this non - local network of states can be engaged by transferring quantum amplitudes from qubit states to higher energy states temporarily .
as we will demonstrate , one of the most effective ways to accomplish this is to design continuous - time quantum walks @xcite through these networks of states . in what follows ,
we first introduce terminology .
we , then , show how multiqubit gates can be performed via pulse sequences addressing always interacting auxiliary states .
this approach , when optimized , can provide some gate compression but is limited because the structure of nonlocal states involving multiple qubits is not used .
the pulse - sequence approach is used further as a benchmark to suggest increased efficiency of the proposed quantum - walk - based scheme .
in order to illustrate the concept we will focus on one of the simplest systems necessary : a collection of four - state quantum systems coupled to a single cavity mode of frequency @xmath3 [ see fig .
[ fig : mqb - scheme1](b ) ] .
such configuration can be realized in a number of currently most advanced qubit architectures @xcite .
generalization to larger number of states and other types of interaction is possible .
we set each individual four - state qubit system ( qs ) to encode a qubit via the two lowest energy states @xmath4 , and use two additional states , @xmath5 , as auxiliary states .
the hamiltonian is @xmath6 where @xmath7 here constant energies @xmath8 and @xmath9 correspond to states @xmath10 and @xmath11 respectively , @xmath12 ( @xmath13 ) is the cavity mode creation ( annihilation ) operators , and the superscript @xmath0 enumerates qss .
external pulse control enters via @xmath14 .
other choices for cavity - qubit coupling , e.g. , to @xmath15 , can also be made @xcite .
this hamiltonian ( without @xmath16 ) can be diagonalized to @xmath17 where states @xmath18 approach non - interacting states @xmath19 , @xmath20 , etc .
, when @xmath21 , and index @xmath0 runs over the complete set of states as defined by hamiltonian ( [ eq : h ] ) .
when all transitions in qss are sufficiently detuned away from the resonance with @xmath3 ( _ off - resonance _ ) or have zero ( or negligible ) matrix elements , the cavity field can not mediate the interaction effectively . in this regime all transitions
@xmath22{a}\!{|{e_i , jk ...
}\rangle},\quad { |{i , e_jk ... }\rangle}\!\xleftrightarrow[]{a'}\!{|{e_i , e_jk ... }\rangle},\quad ... \end{aligned}\ ] ] with the same @xmath23 are identical irrespective of the values of other indexes ( @xmath24 ) .
this means that each qs is isolated and is not affected by the state of the rest of the system only single - qubit manipulations are possible .
when , on the other hand , the cavity mode is _ in resonance _ with at least one of the transitions in each qs and the corresponding matrix elements are non - negligible , the degeneracy can be lifted a change in the state of one of the qubits can affect the entire system .
as the result , entanglement can be modified , but single- ( or few- ) qubit operations are difficult to perform .
we will label the superposition states @xmath18 at @xmath25 the same way as the states from which they originate when @xmath26 is adiabatically increased from zero . for the purpose of this investigation we will focus on the simplest case , when the cavity has non - negligible and identical matrix elements for transitions between states in each individual qs .
this assumption simplifies the discussion because in this case it suffices to only discuss frequency resonance conditions and detunings that are independent of the details of qubit - cavity interactions in each individual physical system . on the other hand , this imposes the limit on how many qubits can be coupled this way until spectral crowding introduces unacceptable levels of cross - talk noise .
this limit can be partially or completely lifted by using multiple cavities , each coupled to only few qss , as will be briefly discussed at the end of this paper . in the systems described by hamiltonian of type ( [ eq : h ] ) and with the assumptions stated above , a combination of both resonance and off - resonance limits an _ intermediate resonance regime _
( irr ) @xcite is possible . in this regime ,
the cavity mode is somewhat away from resonance with every qubit system , and the decrease of qubit - cavity interactions ( degree of non - locality ) is different for different types of qubit - cavity states .
specifically , the cavity may be sufficiently far off resonance with qss , such that transitions @xmath27 and @xmath28 are indistinguishable for different @xmath29 and @xmath30 respectively , while transitions in group @xmath27 remain distinguishable from those in group @xmath28 .
we can quantify differences in the transition frequencies for transitions in each group by @xmath31 , and differences in transition frequencies between transitions in group @xmath27 relative to that in group @xmath28 by @xmath32 , and so on .
we can also quantify the detuning of qubit transitions that belong to different qubits from each other by @xmath33 , i.e. , @xmath34 for @xmath35 and specific choice of @xmath36 .
when @xmath37 , both @xmath31 and @xmath32 vanish .
the irr is achieved at some finite values of @xmath33 ( and an appropriate choice of @xmath3 @xcite ) when @xmath38 while the irr , as obtained in ref . ,
is specific to qs coupled via cavity modes , a similar structure with condition ( [ eq : irr ] ) can be present when interactions between qss occur via other mechanisms .
in this section we give a brief overview of a standard @xmath1-system approach to two - qubit quantum gates to introduce concepts used in the other sections . this approach ( and its variations )
is used in different qubit architectures , e.g. , in ion traps @xcite or in quantum dots @xcite , to manipulate entanglement .
the key ingredient is a non - local auxiliary state that has non - zero matrix element for transitions involving at least one of the basis states in the two - qubit system , e.g. , @xmath39 , as in fig .
[ fig : mqb - scheme1](a ) .
the state is non - local if it is formed as the result of interaction between qubits , either direct or mediated via a cavity field .
the non - locality is manifested via the fact that transition @xmath40 is different and distinguishable from @xmath41 or @xmath42 . in order to perform the simplest entangling operation ,
a cz ( control - z ) gate , it suffices to send a @xmath43 pulse in resonance with @xmath40 transition .
resonance pulses perform a rotation according to @xmath44 where @xmath45 is the integral of the pulse amplitude . when @xmath46 the amplitude is returned back to the original state in the computational basis with a phase factor of @xmath47 . as the result
, the initial two - quit state is altered as @xmath48 which is equivalent to application of the cz gate .
several pulse fields using both legs of the @xmath1-system can be used to create more complex entangling two - qubit rotations @xcite . despite the simplicity of this scheme ,
the existence of suitable `` bus '' state @xmath2 in systems with multiple qubits is hardware - dependent and may be difficult to have permanently in many architectures .
all @xmath49 transitions in a three qubit system ( schematically ) . transitions
are grouped by the qubit at which transition occurs . when transitions are local , each transition is indistinguishable from the other in the same group .
a similar diagram can be obtained if @xmath15 transitions are of interest ( as in sec .
[ sec : ccz - tr ] ) by inverting indexes 0 and 1 . ] three - qubit entangling gates performed via a sequence of pulses .
( a ) a three qubit toffoli gate .
( b ) cascaded cnot gate performed by local resonant pulses . ] in this section we demonstrate how entanglement can be manipulated in systems under irr conditions using a pulse - sequence approach .
we begin with the simplest multiqubit entangling gate a three - qubit toffoli gate @xcite .
the toffoli gate can be decomposed into @xmath50 or represented by six cnot gates @xcite . here
@xmath51 is a hadamard operation @xcite applied to @xmath23-th qubit , and @xmath52 is a control - not operation . the control - z gate with one ( two ) control qubits and one target qubit is denoted by cz ( ccz ) .
note that control - z operations with arbitrary number of control qubits , [ c]@xmath53z , are diagonal identity matrices with , e.g. , one diagonal entry set to @xmath54 .
therefore , to perform the entangling part of the toffoli gate one must add a phase of @xmath47 to the amplitude residing on , e.g. , one of the basis states .
the three - qubit basis states are shown in fig .
[ fig:3qb - mqb - a ] schematically .
all @xmath55 transitions in fig .
[ fig:3qb - mqb - a ] are grouped by the qubit in which transition takes place . in the irr
all transitions within each group are indistinguishable ( @xmath56 ) . if we apply a _
resonant _ pulse to transition @xmath57 it will also affect states @xmath58 , @xmath59 , and @xmath60 in exactly the same fashion ( solid lines in fig .
[ fig:3qb - mqb - a ] ) , and , hence , will be completely local to qs-1 . however , if a local pulse is applied to , e.g. , the first qubit such that some part of the population remains in the excited auxiliary state after the pulse , the next pulse applied to , e.g. , the second qubit is no longer local because two of its transitions ( from @xmath61 and @xmath62 ) affect only a portion of the population .
the population left on the excited auxiliary states by the first pulse can not be transfered to higher energy states by the second pulse , because @xmath32 is not negligible due to condition ( [ eq : irr ] ) .
a three - qubit ccz gate can be constructed based on three pulses described above .
each pulse induces a transition in a pair of states according to eq .
( [ eq : lam:2pi ] ) .
we choose to apply @xmath63 and @xmath46 pulses as shown in figs .
[ fig:3qb - mqb - b](a ) and ( b ) . because the combination of two @xmath47 pulses completely returns the population back to the initial state adding a phase of @xmath47 , a set of pulses shown in fig .
[ fig:3qb - mqb - b](a ) will change the sign of the amplitudes that reside on all states but @xmath64 , realizing a ccz gate ( up to a negligible common phase factor ) .
similarly , pulse sequence shown in fig .
[ fig:3qb - mqb - b](b ) will realize a czz gate ( two cascaded cz gates ) .
pulse scheme for multiqubit entangling gates performed via a sequence of pulses . ] the above pulse - sequence approach can now be easily generalized to larger number of qubits , fig .
[ fig:3qb - mqb - c ] , and arbitrary phase accumulation for the states .
the latter can be achieved if in any sequence of two consecutive @xmath47 pulses , the pulse frequency is detuned from resonance by some value @xmath65 . in the case of @xmath66-shaped pulses
the resulting phase accumulation , @xmath67 , is @xmath68 where @xmath69 is the pulse bandwidth @xcite .
if we denote a sequence of @xmath47 pulses applied to @xmath70-th qs by bits of a binary number @xmath71 such that the value @xmath72 corresponds to a @xmath47 pulse and @xmath73 to free evolution ( no pulses ) , the pulses will not interfere if @xmath74 where @xmath75 is a bit - wise classical and operation .
this set of pulses can perform any diagonal multiqubit gate .
for example , @xmath76 will perform multiqubit toffoli , [ c]@xmath77z , gate ; @xmath78 , @xmath79 will accomplish a cascade of cz gates , or c[z]@xmath77 . a single quantum fourier transform cascade @xcite
can be performed with the same sequence of pulses as c[z]@xmath77 , detuning each pair of @xmath80 pulses from the exact resonance condition to accumulate a phase @xmath81 .
numerical simulations of a system of three transmons @xcite in a cavity . schematic energy structure of each transmon is shown in panel ( a ) .
representative parts of the combined numerically - obtained spectrum are shown in panels ( b ) through ( f ) .
the dependence on cavity frequncy @xmath3 is emphasized in panel ( c ) for reference .
the transmon frequency is @xmath82 ghz ( anharmonicity is @xmath83 ghz ) .
the frequencies of the second and third transmons are scaled by @xmath84 and @xmath85 ; @xmath86 mhz . ] note that pulse schemes in figs .
[ fig:3qb - mqb - b ] and [ fig:3qb - mqb - c ] are optimized versions of multiqubit gates represented via two - qubit entangling cz gates , introduced in @xcite for similar architectures .
thus , temporal complexity of the gates still scales as @xmath87 , that is , approximately @xmath0 two - qubit cz gates ( or pulses realizing those gates ) are needed to perform the operation .
thus , the ( temporal ) complexity class is the same as in traditional cnot representation : the best representation of [ c]@xmath77z gate via two - qubit entangling gates requires @xmath88 individual cz gates @xcite .
therefore these optimized pulse sequences can be used to benchmark quantum - walk - based gates introduced further against known cnot - base decompositions . as an example of application of the above scheme
, we numerically model a system of three transmons @xcite interacting via a single microwave cavity mode .
we model the @xmath89 sequence and compute fidelity @xcite of the resulting ccz gate operation .
for this example we can describe transmons approximately as anharmonic oscillators [ see fig .
[ fig:3qb - mqb - d](a ) ] with energies @xcite @xmath90 and anharmonicity @xmath91 defined as @xmath92 where @xmath93 the states with @xmath94 correspond to states @xmath95 , @xmath96 , @xmath97 , @xmath98 , respectively . in transmon
qubits the cavity mode ( due to either 3d or planar transmission - line cavity ) can couple strongly to consecutive transition in each transmon .
we chose the frequency of the cavity such , that it is near resonance , but somewhat detuned from @xmath15 transition in each transmon @xcite to realize irr condition ( [ eq : irr ] ) .
as it is the case for all transmon qubit systems , multi - qubit computational basis states are scattered over the large part of the spectrum with multiple other states laying in between due to very small anharmonicity in these systems . in order to correctly account for all possible relevant transitions available in this system we used 5 states to represent each of the three qubits and at least 5 states to represent the cavity mode .
some parts of the complete numerical spectrum ( over 600 states ) are shown in figs .
[ fig:3qb - mqb - d](b)-(f ) , where we plot energies as a function of cavity mode frequency @xmath3 .
states beyond @xmath97 were not activated by external pulses , but were included to numerically confirm that the population remains at lower energy states when the gate is effective .
quantum evolution in systems with many accessible transitions across different frequency range becomes less stable numerically .
implicit integration techniques can partially mitigate this problem , however development of appropriate probability - conserving numerical approaches are needed to correctly account for large number of accessible transitions without artificial cutoffs . in order to guarantee acceptable levels of numerical error , we chose anharmonicity that is about three times larger than what is expected for a transmon @xcite .
transmons with large coherence times @xcite are designed with even smaller anharmonicities , @xmath99 ghz , to suppress charge fluctuation noise @xcite . due to @xmath43-periodic conditions on the phase of the superconducting order parameter , the system can
, in principle , tunnel to the other identical minima .
this tunneling is negligible in the transmon limit@xcite due to large potential barrier separating the minima and the approximate description ( [ eq : transmon ] ) is well justified .
we adopt this approximation in our numerical modeling to investigate the regime consistent with the experimentally used set of parameters . in figure [ fig:3qb - mqb - e ]
we show numerically computed fidelity of the ccz gate operation as a function of the cavity mode frequency and the bandwidth of the pulses . a train of four @xmath47 pulses of the same bandwidth @xmath69 , as prescribed by the scheme outlined in fig .
[ fig:3qb - mqb - c ] , were taken to realize the gate suggested in fig .
[ fig:3qb - mqb - b](a ) .
gaussian pulse shapes with appropriate temporal cut - offs were taken for all pulses .
a plateau of high fidelity of the gate operation shrinks and disappears when the pulse bandwidth approaches @xmath26 , and cavity - induced splittings are no longer distinguished by the pulses .
this behavior is consistent with the qualitative analysis drawn from the level diagram shown in fig .
[ fig:3qb - mqb - a ] and discussed above . at smaller
@xmath69 ( not shown ) the plateau must descent and eventually vanish when @xmath69 approaches the decoherence rates .
numerically computed fidelity of the three - qubit entangling gate ccz performed using the pulse - train approach outlined in fig .
[ fig:3qb - mqb - c ] in a system of three transmons in a cavity ( see fig .
[ fig:3qb - mqb - d ] ) .
fidelity is shown as a function of the cavity mode frequency @xmath3 and the pulse bandwidth @xmath69 .
parameters are the same as in fig .
[ fig:3qb - mqb - d ] . ]
note that in the above examples a sequence of non - overlapping pulses was used .
this is an artificial constraint . when a larger hilbert space is involved and pulses are allowed to overlap , the entire gate can be executed with a single multicolor pulse of comparable strength per fourier harmonic .
this suggests a potential change in the ( temporal ) complexity class for the multiqubit gates . in what follows ,
we demonstrate how evolution governed by fully overlapped pulses can be designed to perform entangling quantum gates .
a network of transitions in a multi - state multiqubit system can be viewed as a graph , with each node describing a single state , @xmath18 , of the diagonalized hamiltonian ( [ eq : h_diag ] ) , and edges describing accessible transitions . this graph can have disconnected parts as , e.g. , in the case of a three - qubit system with only @xmath100 transitions allowed [ fig .
[ fig : graphs](a ) ] . quantum evolution through such graph becomes non - trivial when external ( control ) pulse @xmath14 has harmonics that enable certain transitions @xmath101 for simplicity we consider resonant @xmath102 frequencies only . in this case
the common pulse profile @xmath103 enters only via @xmath104 and the entire evolution operator in the rotating frame is @xmath105.\end{aligned}\ ] ] here @xmath106 is the complex hermitian adjacency matrix composed of fourier amplitudes @xmath107 entering eq .
( [ eq : v ] ) . evolution operator ( [ eq : u ] ) describes propagation of amplitude due to a continuous time quantum walk @xcite a continuous time quantum evolution in a single - particle quantum system initially set up to occupy only one of the quantum states available .
a graph corresponding to such quantum walk is a set of nodes ( states ) with edges ( connections ) representing elements in adjacency matrix @xmath106 , which have the physical meaning of dimensionless rabi frequencies set up by external control pulses . to obtain the desired @xmath108 , corresponding to a quantum gate , e.g. , ccz , in the qubits computational basis
, we analytically solve quantum walk through the graph and chose appropriate @xmath106 based on that solution . in order to produce a gate , we need quantum evolution that satisfies specific initial and final conditions : the walk must start and end ( at time @xmath109 ) on one of the qubit basis states .
the _ free parameters _ to be adjusted are the amplitudes of the fourier harmonics of the control pulse ( elements of @xmath106 matrix ) .
the gate is non - trivial if non - trivial phases are accumulated during the walk .
note that fourier harmonics @xmath106 play the role of pulse amplitudes if we compare quantum walk approach to the pulse train approach outlined in fig .
[ fig:3qb - mqb - c ] . therefore magnitudes of entries in @xmath106 must be compared to the magnitudes of pulses in figs .
[ fig:3qb - mqb - b ] and [ fig:3qb - mqb - c ] .
the substaintial difference however stems from the fact that when the gate is performed via quantum walks , the `` sequence '' of pulses is `` applied '' in _ frequency _ space rather than in _ time _ and the entire gate is executed in a _ single _ pulse ( concurrently ) . in this case the system is left to explore multiple quantum mechanical trajectories involving entangled states at once , as opposed to being guided ( classically ) through a single trajectory as it happens for pulse - sequence schemes in figs .
[ fig:3qb - mqb - b ] and [ fig:3qb - mqb - c ] . in most cases , @xmath106 and the corresponding matrix , @xmath110 , of frequencies @xmath102 entering eq .
( [ eq : v ] ) are not unique . in what follows
we give several specific configurations to illustrate a three - qubit ccz gate and discuss systems with larger number of qubits in the last section .
for the purpose of illustration we discuss a system with only one transition , @xmath49 , per qubit system accessible to external control pulses and show @xmath111 transitions [ dashed lines in fig .
[ fig : graphs](a ) ] to demonstrate the more general structure of the graph .
in order to understand how many different rabi frequencies are available for adjustments , we recall that in systems under the irr condition ( [ eq : irr ] ) , transitions between different levels [ @xmath23,@xmath112 , and @xmath113 in fig . [
fig : graphs](a ) ] are spectrally distinct and can be addressed independently @xcite .
connections within each level correspond to transition in the same qs and are indistinguishable and can not be adjusted independently . the numbering convention for rabi frequencies in fig .
[ fig : graphs ] is set to indicate distinguishable frequencies .
networks of states .
( a ) a complete set of graphs for a three - qubit system .
( b - d ) sub - sections of ( a ) activated to perform ccz gate .
labels refer to the corresponding rabi frequencies in @xmath106 . in panels ( b ) , ( c ) , and
( d ) , we drop the subscript in @xmath97 for clarity . ] as an example , we construct a control pulse of five harmonics activating connection 1 on level @xmath23 , and connections 2 and 3 on level @xmath112 and @xmath113 [ solid lines in fig . [
fig : graphs](a)]all are transitions of type @xmath49 in different qubit systems if labeled in the limit @xmath21 .
this produces four non - trivial graphs , each connected to one of the qubit basis states [ fig .
[ fig : graphs](b)-(d ) ] .
a ccz gate is realized if , e.g. , one of the qubit basis states accumulates a phase of @xmath47 for the amplitude residing on it .
if we chose @xmath114 , the phases accumulated for amplitudes residing on @xmath60 , @xmath58 , and @xmath59 at time @xmath109 can be easily found from diagonalization of respective adjacency matrices .
they are @xmath115 for ( b ) , and @xmath116 for ( c ) graphs .
all trivial ( one state ) graphs accumulate no phase ( in rotating frame ) .
two rabi frequencies , @xmath117 and @xmath118 , must be adjusted to ensure that the phase accumulated for state @xmath61 is an even multiple of @xmath47 .
we can set @xmath119 transforming the balloon graph ( d ) into a linear chain of 5 states @xmath120 with rabbi frequencies @xmath121 the corresponding adjacency matrix has four non - zero eigenfrequencies @xmath122 where @xmath123 with @xmath124 . by setting @xmath125 and @xmath126 with integers
@xmath127 , we set the evolution operator @xmath128 for graph ( d ) to identity . for @xmath129 and @xmath130
we find @xmath131 note that an arbitrary number of @xmath43 phase can be added for each @xmath132 , i.e. , @xmath0 and @xmath70 can be arbitrary non - equal integers .
this means that the choice of rabi frequencies coming due to external pulse control and realizing ccz gate via quantum walks is not unique .
both , the pulse train and the quantum walks approach , as described above , rely on spectrally distinct states and can suffer from spectral degeneracies and reduction of the effective interaction strength that can occur in larger systems of qubits interacting via a single quantum resonator .
considering a system of transmons as an example , it is natural to assume that the condition that allows to avoid interference due to spectral crowding is @xmath133 where superscripts refer to different qubit systems and the frequencies are defined in eq .
( [ eq : omegas ] ) .
note that all pulses are designed to distinguish energies @xmath134 in order to manipulate entanglement .
for the purpose of estimating the above condition , we can assume that qs parameters are scaled as @xmath135 .
this leads to @xmath136 , where we recall that @xmath137 . as the result ,
condition ( [ eq : multi : cond ] ) becomes @xmath138 , where @xmath139 is the number of qubits .
in realistic systems of transmons one can have @xmath140 , @xmath141 , and can set the scaling factor to about @xmath142 .
this produces the limit of @xmath143 .
in addition , the effective interaction must be kept larger than all decoherence rates @xmath144.\end{aligned}\ ] ] this limits usable coupling strength between qss at the opposite ends of the spectrum as @xmath145 $ ] .
the above limitation on @xmath139 stems from the fact that we have considered a simplified model to introduce a new approach for entanglement - manipulating gates based on quantum walks .
specifically , the operation of the gate was demonstrated analytically based on spectral selectivity that relies on frequency detuning . in actual physical systems selectivity of pulses
is determined not only by detuning , but also by the magnitude of the transition matrix elements present for the given pulse field distribution and the wave functions involved in the transition of interest .
in other words , if the pulse field does not overlap with the states involved in the given transition it does not affect that transition even if it is in or near resonance with the pulse frequency .
this makes transitions corresponding to different edges of the graph , e.g. , in fig .
[ fig : graphs](a ) , individually addressable even when the corresponding transition frequencies are near resonance with each other .
furthermore , in many architectures scalable qubit registers would involve multiple cavities carrying interaction between different parts of the system , rather then a single quantum resonator coupled to every qs .
this introduces larger number of addressable auxiliary states , hence generating larger graphs with addressable edges . as the result
, restrictions ( [ eq : multi : cond ] ) and ( [ eq : multi : geff ] ) no longer apply to the entire qubit register , although , they can still be important locally .
this situation is ideal for the proposed approach of using continuous time quantum walks to manipulate entanglement in multiqubit systems .
investigation of quantum walk properties in this context , particularly `` return '' quantum walks discussed here , is an interesting direction of further research , which , however , goes beyond the scope of the present paper . finally , we note that the evolution through networks of auxiliary states is also affected by decoherence .
this can potentially limit sizes of the graphs involved in entangling gates . at the same time , small decoherence can , in some cases , enhance quantum propagation through networks @xcite .
i am grateful to q. khan , w. d. thacker , and d. s. wisbey for helpful discussions .
this work was supported by slu and slu hpc .
h. paik , d. i. schuster , l. s. bishop , g. kirchmair , g. catelani , a. p. sears , b. r. johnson , m. j. reagor , l. frunzio , l. i. glazman , s. m. girvin , m. h. devoret , and r. j. schoelkopf , phys .
lett . * 107 * , 240501 ( 2011 ) . | a quantum computing system is typically represented by a set of non - interacting ( local ) two - state systems qubits .
many physical systems can naturally have more accessible states , both local and non - local .
we show that the resulting non - local network of states connecting qubits can be efficiently addressed via continuous time quantum random walks , leading to substantial speed - up of multiqubit entanglement manipulations .
we discuss a three - qubit toffoli gate and a system of superconducting qubits as an illustration .
_ keywords _ : entanglement , quantum information , quantum gates |
understanding the evolution of cooperation among selfish individuals in human and animal societies remains a grand challenge across disciplines .
evolutionary games are employed frequently as the theoretical framework of choice in order to interpret the emergence and survival of cooperative behavior @xcite . the prisoner s dilemma game , in particular ,
has attracted considerable interest @xcite as the essential yet minimalist example of a social dilemma . in the original two - person one - shot game
the two players have two strategies to choose from ( cooperation and defection ) , and their payoffs depend on the simultaneous decision of both .
if they choose to cooperate they will receive the highest collective payoff , which will be shared equally among them .
mutual defection , on the other hand , yields the lowest collective payoff . yet to defect is tempting because it yields a higher individual payoff regardless of the opponent s decision .
it is thus frequently so that both players choose not to cooperate , thus procreating the inevitable social dilemma . in reality , however , interactions may be repeated and the reputation of players compromised @xcite .
additionally , individuals may alter with whom they interact @xcite , and different behaviors may be expressed when participants in a social interaction occupy different roles @xcite .
such and similar considerations have been very successful in elucidating why the unadorned scenario of total defection is often at odds with reality @xcite , where it is clear that both humans and animals cooperate to achieve what would be impossible by means of isolated efforts .
mechanisms supporting cooperation identified thus far include kin selection @xcite as well as many others @xcite , and there is progress in place aimed at unifying some of these approaches @xcite .
probably the most vibrant of all in recent years have been advances building on the seminal paper by nowak and may @xcite , who showed that spatial structure may sustain cooperation without the aid of additional mechanisms or strategic complexity .
although in part anticipated by hamilton s comments on viscous populations @xcite , it was fascinating to discover that structured populations , including complex and social networks @xcite , provide an optimal playground for the pursuit of cooperation .
notably , a simple rule for the evolution of cooperation on graphs and social networks is that natural selection favors cooperation if @xmath7 , where @xmath8 is the benefit of the altruistic act , @xmath9 is its cost , while @xmath10 is the average number of neighbors @xcite .
this is similar to hamilton s rule stating that @xmath11 should be larger than the coefficient of genetic relatedness between individuals @xcite .
in fact , on graphs and social networks the evolution of altruism can thus be fully explained by the inclusive fitness theory since the population is structured such that interactions are between genetic relatives on average @xcite . according to the `` best takes all '' rule
@xcite players are allowed to adopt the strategy of one of their neighbors , provided its payoff is higher than that from the other neighbors as well as from the player aspiring to improve by changing its strategy . based on this relatively simple setup , it was shown that on a square lattice cooperators form compact clusters and so protect themselves against being exploited by defectors .
the `` best takes all '' strategy adoption rule is , however , just one of the many possible alternatives that were considered in the past .
other examples include the birth - death and imitation rule @xcite , the proportional imitation rule @xcite , the reinforcement learning adoption rule @xcite , or the fermi - function based strategy adoption rule @xcite .
the latter received substantial attention , particularly in the physics community , for its compatibility with the monte carlo simulation procedure and the straightforward adjustment of the level of uncertainty governing the strategy adoptions @xmath6 .
however , with this rule the potential donor of the new strategy is selected uniformly at random from all the neighbors . this is somewhat untrue to what can be observed in reality , where in fact individuals typically aspire to their most successful neighbors rather than just somebody random . in this sense
the `` best takes all '' rule seems more appropriate , although it fails to account for errors in judgment , uncertainty , external factors , and other disturbances that may vitally affect how we evaluate and see our co - players . here
we therefore propose a simple tunable function that interpolates between the `` best takes all '' and the random selection of a neighbor in a smooth fashion by means of a single parameter @xmath0 . in this sense
the parameter @xmath0 acts as an aspiration parameter , determining to what degree neighbors with a higher payoff will be considered more likely as potential strategy sources than other ( randomly selected ) neighbors .
aiming to further disentangle the role of aspirations , we also consider two types of players , denoted by type @xmath1 and @xmath2 , respectively . while players of type @xmath1 conform to the aspirations imposed by the value of the aspiration parameter @xmath0 , type @xmath2 players choose whom to potentially copy uniformly at random irrespective of @xmath0 .
we denote the fraction of type @xmath1 and @xmath2 players by @xmath3 and @xmath4 , respectively .
this additional division of players into two groups is motivated by the overwhelming evidence indicating that heterogeneity , almost irrespective of its origin , promotes cooperative actions .
most notably associated with this statement are complex networks , including small - world networks @xcite , random regular graphs @xcite , scale - free networks @xcite , as well as adaptive and growing networks @xcite .
furthermore , we follow the work by mcnamara _ et al . _ on the coevolution of choosiness and cooperation @xcite , in particular by omitting the separation of the population on two types of players and introducing the heterogeneity by means of normally distributed individual aspiration levels that are then also subject to evolution .
at present , we thus investigate how aspirations on an individual level affect the evolution of cooperation .
having something to aspire to is crucial for progress and betterment .
but how high should we set our goals ?
should our role models be only overachievers and sports heroes , or is it perhaps better to aspire to achieving somewhat more modest goals ?
here we address these questions in the context of the evolutionary prisoner s dilemma game and determine just how strong and how widespread aspirations should be for cooperation to thrive best . as we will show , a strong drive to excellence in the majority of the population may in fact act detrimental on the evolution of cooperation , while on the other hand , properly spread and heterogeneous aspirations may be just the key to fully eliminating the defectors
. we will show that this holds irrespective of the structure of the underlying interaction network , as well as irrespective of the level of uncertainty by strategy adoptions @xmath6 .
in addition , the presented results will be contrasted with the output of a simple coevolutionary model , where individual aspirations will also be subject to evolution by means of natural selection .
we will conclude that appropriately tuned aspirations may be seen as a universally applicable promoter of cooperation , which will hopefully inspire new studies along this line of research .
depending on the interaction network , the strategy adoption rule and other simulation details ( see _ e.g. _ @xcite ) , there always exists a critical cost - to - benefit ratio @xmath12 at which cooperators in the prisoner s dilemma die out .
this is directly related to hamilton s rule stating that natural selection favors cooperation if @xmath11 is larger than the coefficient of genetic relatedness between individuals @xcite . if the aspiration parameter @xmath13 ( note that then the division of players to those of type @xmath1 and those of type @xmath2 is irrelevant ) , @xmath14 , and the interaction network is a square lattice , then , in our case , @xmath15 . in what follows
, we will typically set @xmath16 slightly below this threshold to @xmath17 and examine how different values of @xmath0 , @xmath3 , @xmath6 , as well as different interaction networks influence the outcome of the prisoner s dilemma game .
it is instructive to first examine characteristic snapshots of the square lattice for different values of @xmath0 and @xmath3 .
results presented in fig .
[ fig1 ] hint to the conclusion that heterogenous aspiration to the fittest promotes cooperation , although the details of this claim depend somewhat on the value of the aspiration parameter @xmath0 . for small values of @xmath0
it is best if all the players , _ i.e. _ @xmath18 , aspire to their slightly ( note that @xmath0 is small ) fitter neighbors and thus none actually choose the potential strategy sources uniformly at random .
this can be deduced from the top three panels of fig .
[ fig1 ] if compared from left to right . for large @xmath0 , however , it is best if only half of the players , _
i.e. _ @xmath5 , aspire to their most fittest neighbors , while the other half chooses their role models randomly .
this can be observed if one compares the bottom three panels of fig .
[ fig1 ] with one another , although the difference in the overall density of cooperators ( depicted green and blue ) between the middle and the right panel is fairly small .
finally , the role of the aspiration parameter is more clear cut since larger @xmath0 clearly favor the cooperative strategy if compared to small @xmath0 .
this can be observed if comparing the snapshots presented in fig .
[ fig1 ] vertically . since the snapshots presented in fig .
[ fig1 ] can be used primarily for an initial qualitative assessment of the impact of heterogeneous aspirations , we present in fig .
[ fig2 ] the fraction of cooperators @xmath19 ( left ) and the critical cost - to - benefit ratio @xmath20 ( right ) in dependence on @xmath3 for different values of @xmath0 .
it can be observed that the promotion of cooperation for the optimal combination of the two parameters , being @xmath21 and @xmath5 , is really remarkable .
the fraction of cooperators rises from @xmath22 to @xmath23 , while the critical cost - to - benefit ratio rises a full order of magnitude from @xmath15 to @xmath24 . as tentatively deduced from the lower three snapshots in fig .
[ fig1 ] , it can also be observed that for high values of @xmath0 an intermediate fraction of type @xmath1 players is optimal for the evolution of cooperation .
conversely , for low @xmath0 the fraction of cooperators @xmath19 and the critical cost - to - benefit ratio @xmath20 both increase monotonously with increasing @xmath3 .
if , however , selecting a particular value of @xmath3 , then the impact of the aspiration parameter @xmath0 is always such that cooperation is the more promoted the larger the value of @xmath0 .
this can be observed clearly from both panels , and indeed seems like the main driving force behind the elevated levels of cooperation .
fine - tuning the fraction of players making use of the aspiration to the fittest ( from @xmath18 downwards since the @xmath25 limit trivially returns the random selection of potential strategy sources ) at high @xmath0 can rise the cooperation level further , but more in the sense of minor adjustments , similarly as was observed in the past for the impact of uncertainty by strategy adoptions @xcite or the impact of noise @xcite .
aiming to generalize the validity of our results , we present in fig .
[ fig3 ] the fraction of cooperators @xmath19 in dependence on @xmath3 for different values of @xmath0 as obtained on the random regular graph ( left ) and the small - world network ( right ) .
the goal is to test to what extend above conclusions hold also on interaction networks other than the square lattice , in particular such that are more complex and spatially heterogenous . if comparing the obtained results with those presented in the left panel of fig .
[ fig2 ] , it seems save to conclude that they are to a very large degree qualitatively identical . some differences nevertheless can be observed .
the first is that what constitutes a high @xmath0 limit is a bit higher on complex networks than on the regular lattice .
note that for @xmath26 the optimal fraction of type @xmath1 players is practically still @xmath27 . even for @xmath28
the bell - shaped dependence on @xmath3 is far less pronounced than on the square lattice , and the optimal @xmath3 ( the peak of @xmath19 ) is closer to @xmath29 than @xmath30 .
the second difference is , looking relatively to the starting point at @xmath31 , that the promotion of cooperation due to positive @xmath0 and @xmath3 is somewhat less prolific .
this is , however , not that surprising since complex networks in general promote cooperation already on their own @xcite , and thus secondary promotive mechanisms may therefore become less expressed .
aside from these fairly mild differences though , we can conclude that heterogenous aspirations do promote cooperation irrespective of the underlying interaction network , and that the details of the promotive effect are largely universal and predictable .
next , we proceed with examining how positive values of @xmath0 and @xmath3 fare under different levels of uncertainty by strategy adoptions .
the latter can be tuned via @xmath6 [ see eq .
( 3 ) ] , which acts as a temperature parameter in the employed fermi strategy adoption function @xcite . accordingly ,
when @xmath32 all information is lost and the strategies are adopted by means of a coin toss .
note that this aspect has thus far not received any attention here as @xmath14 was fixed .
the matter is not trivial to address because uncertainty and noise can have a rather profound impact on the evolution of cooperation @xcite , and thus care needs to be exercised .
the safest and most accurate way to approach the problem is by means of phase diagrams .
since we have two additional parameters ( @xmath0 and @xmath3 ) against which we want to test the impact of @xmath6 , we determined full @xmath33 phase diagrams for six characteristic combinations of @xmath0 and @xmath3 on the square lattice .
obtained results are presented in fig .
notably , the phase diagram presented in the top left panel of fig . [ fig4 ] is well - known , implying the existence of an optimal level of uncertainty for the evolution of cooperation , as was previously reported in @xcite . in particular , note that the @xmath34 transition line is bell shaped , indicating that @xmath35 is the optimal temperature at which cooperators are able to survive at the highest value of @xmath16 .
importantly though , this phenomenon can only be observed on interaction topologies lacking overlapping triangles @xcite .
interestingly , increasing @xmath0 from @xmath36 ( top row ) to @xmath37 ( bottom row ) completely eradicates ( as do interaction networks incorporating overlapping triangles ) the existence of an optimal @xmath6 , and in fact qualitatively reverses the dependence .
the @xmath34 transition line has an inverted bell - shaped outlay , indicating the existence of the worst rather than an optimal temperature @xmath6 for the evolution of cooperation .
the qualitative changes are less profound if @xmath0 is kept constant at @xmath36 ( top row ) and @xmath3 increases ( from left to right ) .
still , however , the bell - shaped outlay of the @xmath34 transition gives way to a monotonically increasing curve , saturating only for high @xmath6 .
these qualitative changes in the phase diagrams imply that increasing the aspiration parameter @xmath0 or the fraction of players abiding to it ( type @xmath1 ) effectively alters the interaction network .
while the square lattice obviously lacks overlapping triangles and thus enables the observation of an optimal @xmath6 for small enough values of @xmath0 and @xmath3 ( or a combination thereof , as is the case in the top left panels ) , trimming the likelihood of who will act as a strategy source and how many players will actually aspire to their fittest neighbors seems to effectively enhance linkage among essentially disconnected triplets and thus precludes the same observation .
it is instructive to note that a similar phenomenon was observed recently in public goods games , where the joint membership in large groups was also found to alter the effective interaction network and thus the impact of uncertainly on the evolution of cooperation @xcite . in terms of the facilitation of cooperation
, however , it can be concluded that the promotive impact of positive values of @xmath0 and @xmath3 prevails irrespective of @xmath6 . by comparing the extend of pure @xmath38 and mixed @xmath39 regions for different pairs of the two parameters
, we can observe that for small values of @xmath0 ( top panels in fig .
[ fig4 ] ) it is best if all the players , _ i.e. _ @xmath18 , aspire to their slightly ( note that @xmath0 is small ) fitter neighbors , while for large @xmath0 ( bottom panels in fig . [ fig4 ] ) it is best if only approximately half of the players , _ i.e. _ @xmath5 , aspire to their most fittest neighbors .
the same conclusions were stated already upon the inspection of results presented in figs .
[ fig2 ] and [ fig3 ] , and with this we now affirm that not only is the promotion of cooperation via heterogeneous aspirations robust against differences in the interaction networks , but also against variations in the uncertainty by strategy adoptions .
it remains of interest to elucidate why then cooperative behavior is in fact promoted by positive values of @xmath0 and @xmath3 . to provide answers
, we show in fig .
[ fig5 ] time courses of @xmath19 for different characteristic combination of the two main parameters that we have used throughout this work .
what should attract the attention is the fact that in the most early stages of the evolutionary process ( note that values of @xmath19 were recorded also in - between full iteration steps ) it appears as if defectors would actually fare better than cooperators .
this is actually what one would expect , given that defectors are , as individuals , more successful than cooperators and will thus be chosen more likely as potential strategy donors if @xmath0 is positive .
this should in turn amplify their chances of spreading and ultimately result in the decimation of cooperators ( indeed , only between 20 - 30 % survive ) .
quite surprisingly though , the tide changes fairly fast , and as one can observe from the presented time courses , frequently the more so the deeper the initial downfall of cooperators .
we argue that for positive values of @xmath0 and @xmath3 a negative feedback effect occurs , which halts and eventually reverts what appears to be a march of defectors toward dominance .
namely , in the very early stages of the game defectors are able to plunder very efficiently , which quickly results in a state where there are hardly any cooperators left to exploit .
consequently , the few remaining clusters of cooperators start recovering lost ground against weakened defectors .
crucial thereby is the fact that the clusters formed by cooperators are impervious to defector attacks even at high values of @xmath16 because of the positive selection towards the fittest neighbors acting as strategy sources ( occurring for @xmath40 ) . in a sea of cooperators
this is practically always another cooperator rather than a defector trying to penetrate into the cluster .
this newly identified mechanism ultimately results in fairly widespread cooperation that goes beyond what can be warranted by the spatial reciprocity alone ( see _ e.g. _ @xcite ) , and this irrespective of the underlying interaction network and uncertainty by strategy adoptions .
finally , it is instructive to examine whether an optimal intermediate value of @xmath41 , determining the aspiration level of player @xmath42 , can emerge spontaneously from an initial array of normally distributed values .
this would imply that natural selection indeed favors individuals with a specific aspiration level , which would in turn extend the credibility of thus far presented results that were obtained primarily in a top - down manner [ by optimizing a population - level property ( cooperation ) by means of an appropriate selection of parameters determining the aspiration level of individuals ] .
for this purpose we omit the division of the population on players of type @xmath1 and @xmath2 , and initially assign to every player a value @xmath41 that is drawn randomly from a gaussian distribution with a given mean @xmath43 and standard deviation @xmath44 .
then if player @xmath42 adopts the strategy from player @xmath45 also @xmath41 becomes equal to @xmath46 ( see methods for details ) .
results obtained with this alternative coevolutionary model are presented in fig .
[ fig6 ] . it can be observed that the initial gaussian distribution sharpens fast around an intermediate value of @xmath47 , which then gradually becomes more and more frequent in the population as the natural selection spontaneously eliminates the less favorable values that warrant a lower individual fitness .
the final state is a population where virtually all players have an identical aspiration level @xmath48 , and accordingly , the outcome in terms of the stationary density of cooperators is equal to that obtained with the original model having @xmath18 and @xmath49 . in this sense
the preceding results are validated and their generality extended by means of a bottom - up approach entailing a spontaneous coevolution towards an intermediate individually optimal aspiration level .
we note , however , that with this simple coevolutionary model the result that heterogeneous aspirations promote cooperation is not exactly reproduced .
further studies on more sophisticated models incorporating coevolving aspirations are required to arrive spontaneously at a heterogeneous distribution of individual aspiration levels .
inspirations for this can be found in the recent review on coevolutionary games @xcite , and we are looking forward to further developments in this direction .
we have shown that heterogenous aspiration to the fittest , _
i.e. _ the propensity of designating the most successful neighbor as being the role model , may be seen as a universally applicable promoter of cooperation that works on different interaction networks and under different levels of stochasticity . for low and moderate values of the aspiration parameter
@xmath0 cooperation thrives best if the total population abides to aspiring to the fittest . for large values of @xmath0 , however , it is best if only approximately half of the players persuasively attempt to copy their most successful neighbors while the rest chooses their opponents uniformly at random .
the optimal evolution of cooperation thus requires fine - tuning of both , the density of players that are prone to aspiring to the fittest , as well as the aspiration parameter that determines how fit a neighbor actually must be in order to be considered as the potential source of the new strategy .
in addition , by studying an alternative model where individual aspiration levels were also subject to evolution , we have shown that an intermediate value of the aspiration level emerges spontaneously through natural selection , thus supplementing the main results by means of a coevolutionary approach .
notably , the extensions of the prisoner s dilemma game we have considered here seem very reasonable and are in fact easily justifiable with realistic examples .
for example , it is a fact that people will , in general , much more likely follow a successful individual than somebody who is just struggling to get by . under certain adverse circumstances , like in a state of rebelion or in revolutionary times , however , it is also possible that individuals will be inspired to copy their less successful partners or those that seem to do more harm than good . in many ways
it seems that the ones who are satisfied with just picking somebody randomly to aspire to are the ones that are most difficult to come by . in this sense
the rather frequently adopted random selection of a neighbor , retrieved in our case if @xmath13 ( or equivalently @xmath25 ) , seems in many ways like the least probable alternative . in this sense
it is interesting to note that our aspiring to the fittest becomes identical to the frequently adopted , especially in the early seminal works on games on grids @xcite , `` best takes all '' adoption rule if @xmath18 , @xmath50 in eq .
( 2 ) , and @xmath51 in eq . ( 3 ) .
although in our simulations we never quite reach the `` best takes all '' limit , and thus a direct comparison with the seminal works is somewhat circumstantial , we find here that the intermediate regions of heterogenous aspirations offer fascinating new insights into the evolution of cooperation , and we hope that this work will inspire future studies , especially in terms of understanding the emergence of successful leaders in societies via a coevolutionary process @xcite .
an evolutionary prisoner s dilemma game with the temptation to defect @xmath52 ( the highest payoff received by a defector if playing against a cooperator ) , reward for mutual cooperation @xmath53 , the punishment for mutual defection @xmath54 , and the sucker s
payoff @xmath55 ( the lowest payoff received by a cooperator if playing against a defector ) is used as the basis for our simulations . without loss of generality
the payoffs can be rescaled as @xmath56 , @xmath57 , @xmath58 and @xmath54 , where @xmath59 is the cost - to - benefit ratio @xcite .
for positive @xmath16 we have @xmath60 , thus strictly satisfying the prisoner s dilemma payoff ranking . as the interaction network , we use either a regular @xmath61 square lattice , the random regular graph ( rrg ) constructed as described in @xcite , or the small - world ( sw ) topology with an average degree of four generated via the watts - strogatz algorithm @xcite .
each vertex @xmath42 is initially designated as hosting either players of type @xmath62 or @xmath2 with the probability @xmath3 and @xmath4 , respectively .
this division of players is performed uniformly at random irrespective of their initial strategies and remains unchanged during the simulations .
according to established procedures , each player is initially also designated either as a cooperator ( @xmath63 ) or defector ( @xmath64 ) with equal probability .
the game is iterated forward in accordance with the sequential simulation procedure comprising the following elementary steps .
first , player @xmath42 acquires its payoff @xmath65 by playing the game with all its neighbors .
next , we evaluate in the same way the payoffs of all the neighbors of player @xmath42 and subsequently select one neighbor @xmath45 via the probability @xmath66 where the sum runs over all the neighbors of player @xmath42 .
importantly , @xmath41 is the so - called selection or aspiration parameter that depends on the type of player @xmath42 according to @xmath67 evidently , if the aspiration parameter @xmath13 then irrespective of @xmath3 ( density of type @xmath1 players ) the most frequently adopted situation is recovered where player @xmath45 is chosen uniformly at random from all the neighbors of player @xmath42 . for @xmath40 and @xmath68 , however , eqs .
( 1 ) and ( 2 ) introduce a preference in all players of type @xmath1 ( but not in players of type @xmath2 ) to copy the strategy of those neighbors who have a high fitness , or equivalently , a high payoff @xmath69 .
lastly then , after the neighbor @xmath45 that is aspired to by player @xmath42 is chosen , player @xmath42 adopts the strategy @xmath70 from the selected player @xmath45 with the probability @xmath71},\ ] ] where @xmath6 denotes the amplitude of noise or its inverse ( @xmath72 ) the so - called intensity of selection @xcite .
irrespective of the values of @xmath0 and @xmath3 one full iteration step involves all players @xmath73 having a chance to adopt a strategy from one of their neighbors once .
an alternative model , allowing for individual @xmath41 values to be subject to evolution as well , entails omitting the division of the population on two types of players and assigning to every individual an initial @xmath41 value that is drown randomly from a gaussian distribution having mean @xmath43 and standard deviation @xmath44 , as was done recently in @xcite , for example .
subsequently , if player @xmath42 adopts the strategy from player @xmath45 following the identical procedure as described above for the original model , then the value of @xmath41 changes to that of @xmath46 as well .
the key question that we aim to answer with this model is whether a specific aspiration level is indeed optimal for an individual to prosper , and if yes , does the selection pressure favor it spontaneously .
essentially , we are interested in the distribution of @xmath41 values after the stationary fraction of strategies in the population is reached . a link with the original model can be established by considering in this case @xmath3 to equal one and @xmath74 .
results of computer simulations were obtained on populations comprising @xmath75 to @xmath76 individuals , whereby the fraction of cooperators @xmath19 was determined within @xmath77 full iteration steps after sufficiently long transients were discarded . moreover , since the heterogeneous preferential selection of neighbors may introduce additional disturbances , final results were averaged over up to @xmath78 independent runs for each set of parameter values in order to assure suitable accuracy .
we have benefited substantially from the insightful comments of the editor james arthur robert marshall and an anonymous referee , which were instrumental for elevating the quality of this work .
helpful discussions with professor lianzhong zhang are gratefully acknowledged as well . while the bottom row features results for @xmath28 . in both rows the fraction of type @xmath1 players @xmath3 is @xmath80 , @xmath30 and @xmath37 from left to right .
cooperators of type @xmath1 and @xmath2 are colored green and blue , respectively .
defectors of type @xmath1 and @xmath2 , on the other hand , are colored red and yellow . if comparing the snapshots vertically , it can be observed that larger values of @xmath0 ( top @xmath36 , bottom @xmath37 ) clearly promote the evolution of cooperation .
the scenario from left to right via increasing the fraction of type @xmath1 players is not so clear cut . for @xmath81 ( top row )
we can conclude that larger @xmath3 favor cooperative behavior , as clearly the cooperators flourish more and more from the left toward the right panel . for @xmath28 ( bottom row )
, however , it seems that for @xmath82 ( bottom middle ) cooperators actually fare better then for both @xmath83 ( bottom left ) and @xmath84 ( bottom right ) .
hence , the conclusion imposes that for higher @xmath0 values an intermediate ( rather than the maximal , as is the case for lower @xmath0 ) fraction of type @xmath1 players ( those that aspire to their most fittest neighbors only ) is optimal for the evolution of cooperation .
results in all panels were obtained for @xmath85 and @xmath14.,width=595 ] in dependence on the fraction of type @xmath1 players @xmath3 for different values of the aspiration parameter @xmath0 .
right panel depicts the critical cost - to - benefit ratio @xmath86 at which cooperators die out , _
i.e. _ @xmath87 , in dependence on @xmath3 for different values of @xmath0 .
results in both panels convey the message that low values of @xmath0 require a high fraction of type @xmath1 players for cooperation to flourish .
conversely , higher values of @xmath0 sustain cooperation optimally if only half ( @xmath5 ) of the players aspires to their most fittest neighbors while the rest chooses whom to potentially imitate uniformly at random .
optimal conditions for the evolution of cooperation thus require @xmath0 and @xmath3 to be fine - tuned jointly .
depicted results in both panels were obtained for @xmath85 and @xmath14.,width=416 ] in dependence on the fraction of type @xmath1 players @xmath3 for different values of the aspiration parameter @xmath0 for the rrg .
right panel depicts @xmath19 in dependence on @xmath3 for different values of @xmath0 for the watts - strogatz sw network with the fraction of rewired links equalling @xmath88 .
these results are in agreement with those presented in fig .
[ fig2 ] , supporting the conclusion that the impact of heterogenous aspirations on the evolution of cooperation is robust against alterations of the interaction network . as on the square lattice ,
low , but also intermediate , values of @xmath0 require @xmath84 for cooperation to thrive , while higher values of @xmath0 sustain cooperation optimally only if @xmath89 .
depicted results in both panels were obtained for @xmath85 and @xmath14.,width=416 ] phase diagrams for the square lattice . *
top row depicts results for the aspiration parameter @xmath81 while the bottom row features results for @xmath28 . in
both rows the fraction of type @xmath1 players @xmath3 is @xmath80 , @xmath30 and @xmath37 from left to right .
the outline of panels thus corresponds to the snapshots presented in fig .
thin blue and thick red lines mark the border between stationary pure @xmath38 and @xmath64 phases and the mixed @xmath39 phase , respectively . in agreement with previous works @xcite ,
it can be observed that for @xmath81 and @xmath83 ( top left ) there exists an intermediate uncertainty in the strategy adoption process ( an intermediate value of @xmath6 ) for which the survivability of cooperators is optimal , _ i.e. _ @xmath90 is maximal .
conversely , while the borderline separating the pure @xmath38 and the mixed @xmath39 phase for all the other combinations of @xmath0 and @xmath3 exhibits a qualitatively identical outlay as for the @xmath81 and @xmath83 case , the @xmath34 transition is qualitatively different and very much dependent on the particularities players aspirations .
note that in all the bottom panels there exist an intermediate value of @xmath6 for which @xmath90 is minimal rather than maximal , while towards the large @xmath6 limit @xmath90 increases , saturating only for @xmath91 ( not shown ) . in the top middle and right panel , on the other hand , the bell - shaped outlay of the @xmath34 transition gives way to a monotonically increasing curve , again saturating only for @xmath91 .
it can thus be concluded that , while the aspiration based promotion of cooperation is largely independent of @xmath6 , the details of phase transition are very much affected , which can be attributed to an effective alterations of the interaction network due to preferred strategy sources ( see also main text for details).,width=585 ] ( solid lines ) and @xmath28 ( dashed lines ) , each for three different fractions of type @xmath1 players @xmath3 , as depicted on the figure .
the crucial feature of all time courses is the initial temporary downfall of cooperators , which sets in for all depicted combinations of @xmath0 and @xmath3 .
quite remarkably , what appears to become an ever faster extinction eventually becomes a rise to , at least in some cases , near - dominance .
note that the horizontal axis is logarithmic and that values of @xmath19 were recorded also in - between full iteration steps to ensure a proper resolution .
depicted results were obtained for @xmath85 and @xmath14.,width=280 ] drawn from a gaussian distribution with the mean @xmath92 and standard deviation @xmath93 .
the main panel depicts the distributions @xmath94 of individual aspiration levels as recorded at @xmath95 ( solid black line ) , @xmath96 ( dashed red line ) and @xmath97 ( dotted green line ) full iteration steps .
the fixation towards a dominant average value @xmath74 due to natural selection is evident since the interval of @xmath47 values still present in the population becomes more and more narrow as time progresses .
the inset shows the convergence of @xmath0 ( solid gray line ) and @xmath19 ( dotted blue line ) .
the initial temporary downfall of cooperators , followed by the rise to near - dominance , is well - expressed also in the coevolutionary setup , and the stationary density agrees well with the results obtained by means of the original model with @xmath84 and @xmath28 ( compare with the dashed cyan line in fig .
[ fig5 ] ) .
note that in the inset the horizontal axis is logarithmic and that values of @xmath0 and @xmath19 were recorded also in - between full iteration steps to ensure a proper resolution .
depicted results were obtained for @xmath85 and @xmath14.,width=280 ] | to be the fittest is central to proliferation in evolutionary games .
individuals thus adopt the strategies of better performing players in the hope of successful reproduction . in structured populations the array of those that are eligible to act as strategy sources is bounded to the immediate neighbors of each individual . but which one of these strategy sources should potentially be copied ?
previous research dealt with this question either by selecting the fittest or by selecting one player uniformly at random .
here we introduce a parameter @xmath0 that interpolates between these two extreme options . setting @xmath0 equal to zero returns the random selection of the opponent , while positive @xmath0 favor the fitter players .
in addition , we divide the population into two groups .
players from group @xmath1 select their opponents as dictated by the parameter @xmath0 , while players from group @xmath2 do so randomly irrespective of @xmath0 .
we denote the fraction of players contained in groups @xmath1 and @xmath2 by @xmath3 and @xmath4 , respectively . the two parameters @xmath0 and
@xmath3 allow us to analyze in detail how aspirations in the context of the prisoner s dilemma game influence the evolution of cooperation .
we find that for sufficiently positive values of @xmath0 there exist a robust intermediate @xmath5 for which cooperation thrives best .
the robustness of this observation is tested against different levels of uncertainty in the strategy adoption process @xmath6 and for different interaction networks .
we also provide complete phase diagrams depicting the dependence of the impact of @xmath0 and @xmath3 for different values of @xmath6 , and contrast the validity of our conclusions by means of an alternative model where individual aspiration levels are subject to evolution as well .
our study indicates that heterogeneity in aspirations may be key for the sustainability of cooperation in structured populations . |
turing proved in 1936 that undecidability exists by showing that the halting problem is undecidable @xcite .
rice extended the set of known undecidable problems to cover all questions of the form `` does the partial function computed by the given program have property @xmath0 '' , where @xmath0 is any property that at least one computable partial function has and at least one does not have @xcite .
for instance , @xmath0 could be `` returns @xmath1 for all syntactically correct c++ programs and @xmath2 for all remaining inputs . '' in other words , it may be impossible to find out whether a given weird - looking program is a correct c++ syntax checker .
these results are basic material in such textbooks as @xcite .
on the other hand , imperfect halting testers are possible . for any instance of the halting problem , a _ three - way tester _ eventually answers `` yes '' , `` no '' , or `` i do nt know '' .
if it answers `` yes '' or `` no '' , then it must be correct .
we say that the `` i do nt know '' instances are _ hard instances _ for the tester .
also other kinds of imperfect testers have been introduced , as will be discussed in section [ s : variants ] .
assume that @xmath3 is a tester . by turing s proof
, it has a hard instance @xmath4 .
if @xmath4 is a halting instance , then let @xmath5 be `` if the input is @xmath4 , then reply ` yes ' , otherwise run @xmath3 and return its reply '' .
if @xmath4 is non - halting , then let @xmath5 be `` if the input is @xmath4 , then reply ` no ' , otherwise run @xmath3 and return its reply '' . by construction
, @xmath5 is a tester with one fewer hard instances than @xmath3 has . by turing s proof
, also @xmath5 has a hard instance .
let us call it @xmath6 .
it is hard also for @xmath3 .
this reasoning can be repeated without limit , yielding an infinite sequence @xmath3 , @xmath5 , of testers and @xmath4 , @xmath6 , of instances such that @xmath7 is hard for @xmath3 , , @xmath8 but not for @xmath9 , .
therefore , every tester has an infinite number of hard instances , but no instance is hard for all testers . a program that answers `` i do nt know '' for every program and input is a three - way tester , although it is useless .
a much more careful tester simulates the given program on the given input at most @xmath10 steps , where @xmath11 is the joint size of the program and its input .
if the program stops by then , then the tester answers `` yes '' . if the program repeats a configuration ( that is , a complete description of the values of variables , the program counter , etc . ) by then , then the tester answers `` no '' .
otherwise it answers `` i do nt know '' . with this theoretically possible but in practice unrealistic tester ,
any hard halting instance has a finite but very long running time . the proofs by turing and rice may leave the hope that only rare artificial contrived programs yield hard instances .
one could dream of a three - way tester that answers very seldom `` i do nt know '' .
this publication analyses this issue , by surveying and proving results that tell how the proportion of hard instances behaves when the size of the instances grows without limit .
section [ s : def ] presents the variants of the halting problem and imperfect testers surveyed , together with some basic results and notation .
earlier research is discussed in section [ s : related ] .
the section contains some proofs to bring results into the framework of this publication .
section [ s : proglang ] presents some new results in the case that a program has many copies of all big sizes , or information can be packed densely inside the program .
it is not always assumed that the program has access to the information .
a natural example of such information is dead code , such as ` if(1==0)then { } ` . in section [ s : c++ ] , results are derived for c++ programs with inputs from files .
section [ s : conclusions ] briefly concludes this publication .
this publication is a significantly extended version of @xcite .
the papers @xcite are otherwise essentially the same , but three proofs were left out from @xcite because of lack of space . in the present publication , theorems [ t : ill - eof ] and [ t : hard - sd ] and corollaries [ c : nolimit1 ] and [ c : nolimit2 ] are new results lacking from @xcite . furthermore , @xcite incorrectly claimed the opposite of theorem [ t : hard - sd ] .
the present publication fixes this error and also a small error in proposition [ p : syntax ] .
the literature on hard instances of the halting problem considers at least three variants of the halting problem : e : : does the given program halt on the _ empty _ input @xcite , s : : does the given program halt when given _ itself _ as its input ( * ? ? ?
* ; * ? ? ?
* ) , and g : : does the given program halt on the _ given _ input ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
each variant is undecidable .
variant g has a different notion of instances from others : program input pairs instead of just programs .
a tester for g can be trivially converted to a tester for e or s , but the proportion of hard program
input pairs among all program input pairs of some size is not necessarily the same as the similar proportion with the input fixed to the empty one or to the program itself .
the literature also varies on what the tester does when it fails .
three - way testers , that is , the `` i do nt know '' answer is used implicitly by @xcite , as it discusses the union of two decidable sets , one being a subset of the halting and the other of the non - halting instances . in _
generic - case decidability _
@xcite , instead of the `` i do nt know '' answer , the tester itself fails to halt .
yet another idea is to always give a `` yes '' or `` no '' answer , but let the answer be incorrect for some instances @xcite .
such a tester is called _
approximating_. one - sided results , where the answer is either `` yes '' or `` i do nt know '' , were presented in @xcite . for a tester of any of the three variants , we say that an instance is _ easy _ if the tester correctly answers `` yes '' or `` no '' on it , otherwise the instance is _
these yield altogether nine different sets of testers , which we will denote with three - way(x),generic(x ) , and approx(x ) , where x is e , s , or g. some simple facts facilitate carrying some results from one variant of testers to another .
[ p:3way- > ] for any three - way tester there is a generic - case tester that has precisely the same easy `` yes''-instances , easy `` no''-instances , hard halting instances , and hard non - halting instances .
there also is an approximating tester that has precisely the same easy `` yes''-instances , at least the same easy `` no''-instances , precisely the same hard halting instances , and no hard non - halting instances ; and an approximating tester that has at least the same easy `` yes''-instances , precisely the same easy `` no''-instances , no hard halting instances , and precisely the same hard non - halting instances .
a three - way tester can be trivially converted to the promised tester by replacing the `` i do nt know '' answer with an eternal loop , the reply `` no '' , or the reply `` yes '' . [ p : gen->gen ] for any generic - case tester there is a generic - case tester that has at least the same `` yes''-instances , precisely the same `` no''-instances , no hard halting instances , and precisely the same hard non - halting instances . in parallel with the original tester , the instance is simulated .
( in turing machine terminology , parallel simulation is called `` dovetailing '' . )
if the original tester replies something , the simulation is aborted .
if the simulation halts , the original tester is aborted and the reply `` yes '' is returned .
[ p : finite ] for any @xmath12 and tester @xmath13 , there is a tester @xmath8 that answers correctly `` yes '' or `` no '' for all instances of size at most @xmath14 , and similarly to @xmath13 for bigger instances .
because there are only finitely many instances of size at most @xmath14 , there is a finite bit string that lists the correct answers for them . if @xmath15 , @xmath8 picks the answer from it and otherwise calls @xmath13 .
( we do not necessarily know what bit string is the right one , but that does not rule out its existence . )
we use @xmath16 to denote the set of characters that are used for writing programs and their inputs .
it is finite and has at least two elements .
there are @xmath17 character strings of size @xmath11 .
if @xmath18 and @xmath19 are in @xmath20 , then @xmath21 denotes that @xmath18 is a prefix of @xmath19 , and @xmath22 denotes proper prefix .
the size of @xmath18 is denoted with @xmath23 .
a set @xmath24 of finite character strings is _ self - delimiting _ if and only if membership in @xmath24 is decidable and no member of @xmath24 is a proper prefix of a member of @xmath24 .
the _ shortlex ordering _ of any set of finite character strings is obtained by sorting the strings in the set primarily according to their sizes and strings of the same size in the lexicographic order . not necessarily all elements of @xmath20 are programs .
the set of programs is denoted with @xmath25 , and the set of all ( not necessarily proper ) prefixes of programs with @xmath26 .
so @xmath27 .
for tester variants e and s , we use @xmath28 to denote the number of programs of size @xmath11 .
then @xmath29 . for tester variant g
, @xmath28 denotes the number of program
input pairs of joint size @xmath11 .
we will later discuss how the program and its input are paired into a single string .
the numbers of halting and non - halting ( a.k.a .
diverging ) instances of size @xmath11 are denoted with @xmath30 and @xmath31 , respectively .
we have @xmath32 .
if @xmath13 is a tester , then @xmath33 , @xmath34 , @xmath35 , and @xmath36 denote the number of its easy halting , hard halting , easy non - halting , and hard non - halting instances of size @xmath11 , respectively . obviously @xmath37 and @xmath38 .
the smaller @xmath34 and @xmath36 are , the better the tester is .
the _ failure rate _ of @xmath13 is @xmath39 .
when referring to all instances of size at most @xmath11 , we use capital letters .
so , for example , @xmath40 and @xmath41 .
nancy lynch @xcite used _ gdel numberings _ for discussing programs .
in essence , it means that each program has at least one index number ( which is a natural number ) from which the program can be constructed , and each natural number is the index of some program .
although the index of an individual program may be smaller than the index of some shorter program , the overall trend is that indices grow as the size of the programs grows , because otherwise we would run out of small numbers . on the other hand ,
if the mapping between the programs and indices is 11 , then the growth can not be faster than exponential .
this is because @xmath42 . with real - life programming languages ,
the growth is exponential , but ( as we will see in section [ s : c++model ] ) the base of the exponent may be smaller than @xmath43 . to avoid confusion ,
we refrain from using the notation @xmath44 , etc . , when discussing results in @xcite , because the results use indices instead of sizes of programs , and their relationship is not entirely straightforward .
fortunately , some results of @xcite can be immediately applied to programming languages by using the _ shortlex gdel numbering_. the shortlex gdel number of a program is its index in the shortlex ordering of all programs .
the first group of results of @xcite reveals that a wide variety of situations may be obtained by spreading the indices of all programs sparsely enough and then filling the gaps in a suitable way .
for instance , with one gdel numbering , for each three - way tester , the proportion of hard instances among the first @xmath14 indices approaches @xmath1 as @xmath14 grows . with another gdel numbering
, there is a three - way tester such that the proportion approaches @xmath2 as @xmath14 grows .
there even is a gdel numbering such that as @xmath14 grows , the proportion oscillates in the following sense : for some three - way tester , it comes arbitrarily close to @xmath2 infinitely often and for each three - way tester , it comes arbitrarily close to @xmath1 infinitely often . in its simplest form ,
spreading the indices is analogous to defining a new language spaciousc++ whose syntax is identical to that of c++ but the semantics is different .
if the first @xmath45 characters of a spaciousc++ program of size @xmath11 are space characters , then the program is executed like a c++ program , otherwise it halts immediately .
this does not restrict the expressiveness of the language , because any c++ program can be converted to a similarly behaving spaciousc++ program by adding sufficiently many space characters to its front .
however , it makes the proportion of easily recognizable trivially halting instances overwhelm . a program that replies `` yes '' if there are fewer than @xmath45 space characters at the front and `` i do nt know '' otherwise , is a three - way tester .
its proportion of hard instances vanishes as the size of the program grows . as a consequence of this and proposition [ p : finite ]
, one may choose any failure rate above zero and there is a three - way tester for spaciousc++ programs with at most that failure rate .
of course , this result does not tell anything about how hard it is to test the halting of interesting programs .
this is the first example in this publication of what we call _ an anomaly stealing the result_. that is , a proof of a theorem goes through for a reason that has little to do with the phenomenon we are interested in . indeed ,
the first results of @xcite depend on using unnatural gdel numberings .
they do not tell what happens with untampered programming languages .
even so , they rule out the possibility of a simple and powerful general theorem that applies to all models of computation
. they also make it necessary to be careful with the assumptions that are made about the programming language . to get sharper results , _ optimal gdel numberings _
were discussed in @xcite .
they do not allow distributing programs arbitrarily .
a gdel numbering is optimal if and only if for any gdel numbering , there is a computable function that maps it to the former such that the index never grows more than by a constant factor .
the most interesting sharper results are opposite to what was obtained without the optimality assumption . to apply them to programming languages , we first define a programming language version of optimal gdel numberings .
[ d : eof - data ] a programming language is _ end - of - file data segment _ , if and only if each program consists of two parts in the following way . the first part , called the _ actual program _ ,
is written in a self - delimiting language ( so its end can be detected ) .
the second part , called the _ data segment _ , is an arbitrary character string that extends to the end of the file .
the language has a construct via which the actual program can read the contents of the data segment .
the data segment is thus a data literal in the program , packed with maximum density .
it is not the same thing as the input to the program .
[ c : lyn6 ] for each end - of - file data segment language ,
@xmath46 @xmath47 let @xmath48 be the end - of - file data segment language , and let @xmath49 be any gdel numbering .
consider the following program @xmath50 in @xmath48 .
let @xmath51 and @xmath52 be the sizes of its actual program and data segment .
the actual program reads the data segment , interpreting its content as a number @xmath14 in the range from @xmath53 to @xmath54 .
then it simulates the @xmath14th program in @xmath49 .
the shortlex index of @xmath50 is at most @xmath55 , yielding @xmath56 , so @xmath57 , thus @xmath58 .
the shortlex numbering of @xmath48 is thus an optimal gdel numbering . from this
, proposition 6 in @xcite gives the claims .
a remarkable feature of the latter result compared to many others in this publication is that @xmath59 is chosen before @xmath13 .
that is , there is a positive constant that only depends on the programming language ( and not on the choice of the tester ) such that all testers have at least that proportion of hard instances , for any big enough @xmath11 . on the other hand ,
the proof depends on the programming language allowing to pack raw data very densely .
real - life programming languages do not satisfy this assumption .
for instance , c++ string literals ` ` can not pack data densely enough , because the representation of ` " ` inside the literal ( e.g. , ` " ` or ` 042 ` ) requires more than one character .
because of proposition [ p : finite ] , `` @xmath60 '' can not be moved to the front of `` @xmath61 '' .
the result can not be generalized to @xmath62 , @xmath63 , and @xmath64 , because the following anomaly steals it .
we can change the language by first adding ` 1 ` or ` 01 ` to the beginning of each program @xmath65 and then declaring that if the size of ` 1`@xmath65 or ` 01`@xmath65 is odd , then it halts immediately , otherwise it behaves like @xmath65 .
this trick does not invalidate optimality but introduces infinitely many sizes for which the proportion of hard instances is @xmath2 . in @xcite ,
the halting problem was analyzed in the context of programming languages that are _ frequent _ in the following sense : [ d : frequent ] a programming language is ( a ) _ frequent _ ( b ) _ domain - frequent _ , if and only if for every program @xmath65 , there are @xmath66 and @xmath67 such that for every @xmath68 , at least @xmath69 programs of size @xmath11 ( a ) compute the same partial function as @xmath65 ( b ) halt on precisely the same inputs as @xmath65 . instead of `` frequent ''
, the word `` dense '' was used in @xcite , but we renamed the concept because we felt `` dense '' a bit misleading .
the definition says that programs that compute the same partial function are common .
however , the more common they are , the less room there is for programs that compute other partial functions , implying that the smallest programs for each distinct partial function must be distributed more sparsely .
`` dense '' was used for domain - frequent in @xcite .
any frequent programming language is obviously domain - frequent but not necessarily vice versa . on the other hand ,
even if a theorem in this field mentions frequency as an assumption , the odds are that its proof goes through with domain - frequency . whether a real - life programming language such as c++ is domain - frequent , is surprisingly difficult to find out .
we will discuss this question briefly in section [ s : frequent ] . as an example of a frequent programming language ,
was mentioned in @xcite .
its full name starts with `` brain '' and then contains a word that is widely considered inappropriate language , so we follow the convention of @xcite and call it .
information on it can be found on wikipedia under its real name .
it is an exceptionally simple programming language suitable for recreational and illustrational but not for real - life programming purposes .
in essence , programs describe turing machines with a read - only input tape , write - only output tape , and one work tape .
the alphabet of each tape is the set of 8-bit bytes .
however , programs only use eight characters . as a side issue , a non - trivial proof was given in @xcite that only a vanishing proportion of character strings over the eight characters are programs .
that is , @xmath70 exists and is @xmath2 .
it trivially follows that if all character strings over the 8 characters are considered as instances and failure to compile is considered as non - halting , then the proportion of hard instances vanishes as @xmath11 grows .
the only possible compile - time error in is that the square brackets ` [ ` and ` ] ` do not match .
most , if not all , real - life programming languages have parentheses or brackets that must match .
so it seems likely that compile - time errors dominate also in the case of most , if not all , real - life programming languages .
unfortunately , this is difficult to check rigorously , because the syntax and other compile - time rules of real - life programming languages are complicated . using another , simpler line of argument
, we will prove the result for both c++ and in section [ s : syntax ] . in any event
, if the proportion of hard instances among all character strings vanishes because the proportion of programs vanishes , that is yet another example of an anomaly stealing the result .
it is uninteresting in itself , but it rules out the possibility of interesting results about the proportion of hard instances of size @xmath11 among all character strings of size @xmath11
. therefore , from now on , excluding section [ s : syntax ] , we focus on the proportion of hard instances among all programs or program input pairs . in the case of program
input pairs , the results may be sensitive to how the program and its input are combined into a single string that is used as the input of the tester . to avoid anomalous results ,
it was assumed in @xcite that this `` pairing function '' has a certain property called `` pair - fair '' .
the commonly used function @xmath71 is pair - fair . to use this pairing function ,
strings are mapped to numbers and back via their indices in the shortlex ordering of all finite character strings .
a proof was sketched in @xcite that , assuming domain - frequency and pair - fairness , @xmath72 that is , the proportion of wrong answers does not vanish .
however , this leaves open the possibility that for any failure rate @xmath73 , there is a tester that fares better than that for all big enough @xmath11 .
this possibility was ruled out in @xcite , assuming frequency and pair - fairness .
( it is probably not important that frequency instead of domain - frequency was assumed . )
that is , there is a positive constant such that for any tester , the proportion of wrong answers exceeds the constant for infinitely many sizes of instances : @xmath74 the third main result in @xcite , adapted and generalized to the present setting , is the following .
we present its proof to obtain the generalization and to add a detail that the proof in @xcite lacks , that is , how @xmath75 is made to halt for `` wrong sizes '' .
generic - case testers are not mentioned , because proposition [ p : gen->gen ] gave a related result for them .
[ t : ksz05 - 3 ] for each programming model and variant e , s , g of the halting problem ,
@xmath76 @xmath77 let @xmath78 .
consider the family @xmath75 of the programs of the following kind , where @xmath79 , @xmath80 , and @xmath81 . if @xmath82 , @xmath75 answers `` no '' in the case of approximating and `` i do nt know '' in the case of three - way testers .
if @xmath83 , @xmath75 simulates all instances of size @xmath11 until @xmath84 of them have halted .
if the simulation stage terminates , then if the given instance is among those that halted , @xmath75 answers `` yes '' , otherwise @xmath75 answers `` no '' or `` i do nt know '' .
thus an approximating @xmath75 has @xmath85 .
we prove next that some @xmath75 is the required tester .
let @xmath86 . then @xmath87 .
when @xmath83 , the simulation stage of @xmath88 terminates and the proportion of hard halting instances of @xmath88 is less than @xmath89 .
some @xmath81 is the @xmath90 for infinitely many values of @xmath11 .
furthermore , there is a smallest such @xmath14 .
we denote it with @xmath91 .
there also is a @xmath92 such that when @xmath83 , then @xmath93 . with these choices
, @xmath94 always halts . for a small enough @xmath59 and the approximating tester @xmath95 in theorem [ t : ksz05 - 3 ] , ( [ e : ksz05 ] ) implies that the failure rate of @xmath95 oscillates , that is , does not approach any limit as @xmath96 .
this observation is directly obtainable from lemma 23 in @xcite . for turing machines with one - way infinite tape and randomly chosen transition function , the probability of falling off the left end of the tape before halting or repeating a state approaches @xmath1 as the number of states grows @xcite .
the tester simulates the machine until it falls off the left end , halts , or repeats a state .
if falling off the left end is considered as halting , then the proportion of hard instances vanishes as the size of the machine grows .
this can be thought of as yet another example of an anomaly stealing the result .
formally , @xmath97 , that is , @xmath98 here x may be e , s , or g. although e was considered in @xcite , the proof also applies to s and g. comparing the result to theorem [ t : itself ] in section [ s : frequent ] reveals that the representation of programs as transition functions of turing machines is not domain - frequent . on the other hand , independently of the tape model , the proportion does not vanish exponentially fast @xcite .
like in @xcite , the proportion is computed on the transition functions , and not on some textual representations of the programs .
the proof relies on the fact that any turing machine has many obviously similarly behaving copies of bigger and bigger sizes .
they are obtained by adding new states and transitions while keeping the original states and transitions intact .
so the new states and transitions are unreachable .
they are analogous to dead code .
these copies are not common enough to satisfy definition [ d : frequent ] , but they are common enough to rule out exponentially fast vanishing .
generic - case decidability was used in @xcite , but the result applies also to three - way testers by proposition [ p:3way- > ] .
the results in @xcite are based on using weighted running times . for every positive integer @xmath99 , the proportion of halting programs that do not halt within time
@xmath100 is less than @xmath101 , simply because the proportion of times greater than @xmath100 is less than @xmath101 .
the publication presents such a weighting that @xmath59 is a computable constant .
assume that programs are represented as self - delimiting bit strings on the input tape of a universal turing machine .
the smallest three - way tester of variant e that answers `` yes '' or `` no '' up to size @xmath11 and `` i do nt know '' for bigger programs , is of size @xmath102 @xcite .
the assumption that the programming language is domain - frequent ( definition [ d : frequent ] ) makes it possible to use a small variation of the standard proof of the non - existence of halting testers , to prove that each halting tester of variant s has a non - vanishing set of hard instances . for three - way and generic - case testers
, one can also say something about whether the hard instances are halting or not . despite its simplicity ,
as far as we know , the following result has not been presented in the literature .
however , see the comment on @xcite in section [ s : ksz05 ] .
[ t : itself ] if the programming language is domain - frequent , then @xmath103 @xmath104 @xmath105 let the execution of @xmath0 with an input @xmath106 be denoted with @xmath107 . for any @xmath13 , consider the program @xmath108 that first tries its input @xmath109 with @xmath13 .
if @xmath110 replies `` yes '' , then @xmath111 enters an eternal loop .
if @xmath110 replies `` no '' , then @xmath111 halts immediately .
the case that @xmath110 replies `` i do nt know '' is discussed below .
if @xmath110 fails to halt , then @xmath111 can not continue and thus also fails to halt . by the definition of domain - frequent ,
there are @xmath112 and @xmath113 such that when @xmath114 , at least @xmath115 programs halt on precisely the same inputs as @xmath108 .
let @xmath116 be any such program .
consider @xmath117 .
if @xmath118 answers `` yes '' , then @xmath117 fails to halt
. then also @xmath119 fails to halt .
thus `` yes '' can not be the correct answer for @xmath118 .
a similar reasoning reveals that also `` no '' can not be the correct answer for @xmath118 .
so @xmath116 is a hard instance for @xmath13 .
nothing more is needed to prove the claim for approximating testers . in the case of generic - case testers ,
the hard instances make @xmath13 and thus @xmath108 fail to halt , so they are non - halting instances . in the case of three - way testers , all hard instances can be made halting instances by making @xmath108 halt when @xmath13 replies `` i do nt know '' .
this proves the claim @xmath120 .
the claim @xmath121 is proven by making @xmath108 enter an eternal loop when @xmath13 replies `` i do nt know '' .
these two proofs may yield different @xmath122 values , but the smaller one of them is suitable for both .
similarly , the bigger of their @xmath123 values is suitable for both .
the second claim of theorem [ t : itself ] lacks a @xmath34 part .
indeed , proposition [ p : gen->gen ] says that with generic - case testers , @xmath34 can be made @xmath2 . with approximating testers ,
@xmath34 can be made @xmath2 at the cost of @xmath36 becoming @xmath31 , by always replying `` yes '' .
similarly , @xmath36 can be made @xmath2 by always replying `` no '' .
the next theorem applies to testers of variant e and presents some results similar to theorem [ t : itself ] . to our knowledge , it is the first theorem of its kind that applies to the halting problem on the empty input .
it assumes not only that many enough equivalent copies exist but also that they can be constructed . on the other hand
, its equivalence only pays attention to the empty input .
[ d : cdf ] a programming language is _ computably empty - frequent _ if and only if there is a decidable equivalence relation `` @xmath124 '' between programs such that * for each program @xmath65 , there are @xmath67 and @xmath66 such that for every @xmath68 , at least @xmath69 programs of size @xmath11 are equivalent to @xmath65 , and * for each programs @xmath65 and @xmath125 , if @xmath126 , then either both or none of @xmath65 and @xmath125 halt on the empty input .
if @xmath126 , we say that @xmath125 is a _ cousin _ of @xmath65 .
it can be easily seen from @xcite that is computably empty - frequent .
[ t : ill - d - r1 ] if the programming language is computably empty - frequent , then @xmath127 the result also holds for generic - case testers but not for approximating testers . given any three - way tester @xmath13 ,
consider a program @xmath108 that behaves as follows .
first it constructs its own code and stores it in a string variable .
hard - wiring the code of a program inside the program is somewhat tricky , but it is well known that it can be done . with gdel numberings ,
the same can be obtained with kleene s second recursion theorem
. then @xmath108 starts constructing its cousins of all sizes and tests each of them with @xmath13 . by the assumption ,
there are @xmath112 and @xmath128 such that for every @xmath114 , @xmath108 has at least @xmath129 cousins of size @xmath11 .
if @xmath13 ever replies `` yes '' , then @xmath108 enters an eternal loop and thus does not continue testing its cousins .
if @xmath13 ever replies `` no '' , then @xmath108 halts immediately .
if @xmath13 replies `` i do nt know '' , then @xmath108 tries the next cousin .
if @xmath13 ever replies `` yes '' , then @xmath108 fails to halt on the empty input . by definition ,
also the tested cousin fails to halt on the empty input .
so the answer `` yes '' would be incorrect .
similarly , if @xmath13 ever replies `` no '' , that would be incorrect .
so @xmath13 must reply `` i do nt know '' for all cousins of @xmath108 .
they are thus hard instances for @xmath13 . because there are infinitely many of them ,
@xmath108 does not halt , so they are non - halting .
to prove the result for generic - case testers , it suffices to run the tests of the cousins in parallel , that is , go around a loop where each test that has been started is executed one step and the next test is started .
if any test ever replies `` yes '' or `` no '' , @xmath108 aborts all tests that it has started and then does the opposite of the reply . a program that always replies `` no '' is an approximating tester with @xmath130 for every @xmath131 .
the results in this section and section [ s : ksz05 ] motivate the question : are real - life programming languages domain - frequent ?
for instance , is c++ domain - frequent ?
unfortunately , we have not been able to answer it .
we try now to illustrate why it is difficult . given any c++ program , it is easy to construct many longer programs that behave in precisely the same way , by adding space characters , line feeds ( denoted with @xmath132{->}(2,1.5)(2,.5)(0,.5 ) \end{pspicture}}$ ] ) , comments , or dead code such as ` if(0!=0)`\{ }. it is , however , hard to verify that many enough programs are obtained in this way .
for instance , it might seem that many enough programs can be constructed with string literals .
we now provide evidence that suggests ( but does not prove ) that it fails .
any program of size @xmath11 can be converted to @xmath133 identically behaving programs of size @xmath134 by adding \{`char*s=\sigma ; ` } to the beginning of some function , where @xmath135{->}(2,1.5)(2,.5)(0,.5 ) \end{pspicture}}\})^k$ ] .
( the purpose of \ { and } is to hide the variable ` s ` , so that it does not collide with any other variable with the same name . )
more programs are obtained by including escape codes such as ` " ` to @xmath136 .
however , it seems that this is a vanishing instead of at least a positive constant proportion when @xmath137 . in the absence of escape codes
, it certainly is a vanishing proportion .
this is because one can add \{`char*s=\sigma,*t=\rho ; ` } instead , where @xmath138 .
without escape codes , this yields @xmath139 programs . when @xmath137 , @xmath140 . that is , although string literals can represent information rather densely
, they do not constitute the densest possible way of packing information into a c++ program ( assuming the absence of escape codes ) .
a pair of string literals yields an asymptotically strictly denser packing .
similarly , a triple of string literals is denser still , and so on . counting the programs in the presence of escape codes
is too difficult , but it seems likely that the phenomenon remains the same .
so string literals do not yield many enough programs .
it seems difficult to first find a construct that does yield many enough programs , and then prove that it works . in this section
we prove a theorem that resembles theorem [ t : ill - d - r1 ] , but relies on different assumptions and has a different proof .
we say that a three - way tester is _
@xmath11-perfect _ if and only if it does not answer `` i do nt know '' when the size of the instance is at most @xmath11 .
the following lemma is adapted from @xcite .
[ l : n - o(1 ) ] each programming language has a constant @xmath141 such that the size of each @xmath11-perfect three - way tester of variant e or s is at least @xmath142 .
let @xmath143 be any @xmath11-perfect three - way tester of variant e or s. consider a program @xmath50 that constructs character strings @xmath109 in shortlex order and tests them with @xmath143 until @xmath144 replies `` i do nt know '' .
if @xmath144 replies `` yes '' , @xmath50 simulates @xmath109 before trying the next character string .
when simulating @xmath109 , @xmath50 gives it the empty input in the case of variant e and @xmath109 as the input in the case of s. the reply `` i do nt know '' eventually comes , because otherwise @xmath143 would be a true halting tester .
as a consequence , @xmath50 eventually halts . before halting , @xmath50 simulates at least all halting programs of size at most @xmath11 .
the time consumption of any simulated execution is at least the same as the time consumption of the corresponding genuine execution .
so the execution of @xmath50 can not contain properly a simulated execution of @xmath50 .
@xmath50 does not read any input , so it does not matter whether it is given itself or the empty string as its input .
therefore , the size of @xmath50 is bigger than @xmath11 . because the only part of @xmath50 that depends on @xmath11 is @xmath143 ,
there is a constant @xmath141 such that the size of @xmath143 is at least @xmath142 . in any everyday programming language
, space characters can be added freely between tokens .
motivated by this , we define that a _ blank character _ is a character that , for any program , can be added to at least one place in the program without affecting the meaning of the program .
[ t : ill - eof ] let x be e or s. if the programming language is end - of - file data segment and has a blank character , then @xmath145 assume first that tester @xmath13 is a counter - example to the @xmath62-claim .
that is , for every @xmath73 , @xmath13 has infinitely many values of @xmath11 such that @xmath146 .
if @xmath13 uses its data segment , let the use be replaced by the use of ordinary constants , liberating the data segment for the use described in the sequel .
let @xmath147 be the following program . here
@xmath99 is a constant inside @xmath147 represented by @xmath148 characters , and @xmath149 is the content of the data segment of @xmath147 interpreted as a natural number @xmath149 in base @xmath43 .
let @xmath51 and @xmath52 be the sizes of the actual program and data segment of @xmath147 .
we have @xmath150 .
let @xmath109 be the input of @xmath147 .
the program @xmath147 first computes @xmath11 : = @xmath151 . if @xmath152 , then @xmath147 adds blank characters to @xmath109 , to make its size @xmath11 .
next , if @xmath153 , then @xmath147 replies `` i do nt know '' and halts
. otherwise @xmath147 gives @xmath109 ( which is now of size precisely @xmath11 ) to @xmath13 .
if @xmath110 replies `` yes ''
or `` no '' , then @xmath147 gives the reply as its own reply and halts .
otherwise @xmath147 constructs each character string @xmath106 of size @xmath11 and tests it with @xmath13 .
@xmath147 simulates in parallel those @xmath106 for which @xmath154 returns `` i do nt know '' until @xmath149 of them have halted ( with @xmath106 or the empty string as the input , as appropriate ) .
then it aborts those that have not halted .
if @xmath109 is among those that halted , then @xmath147 replies `` yes '' , otherwise @xmath147 replies `` no '' . for each @xmath155
, there are infinitely many values of @xmath11 such that @xmath156 . for any such @xmath11 we have latexmath:[${\overline h_t}(n ) < p(n)|\sigma|^{-k } \leq representing @xmath34 .
therefore , there is @xmath147 such that @xmath158 and @xmath159 .
it is an @xmath11-perfect three - way tester of size @xmath160 .
a big enough @xmath99 yields a contradiction with lemma [ l : n - o(1 ) ] .
the proof of the @xmath63-claim is otherwise similar , but @xmath147 counts the number @xmath161 of those @xmath106 for which @xmath154 returns `` i do nt know '' , and simulates the @xmath106 until @xmath162 of them have halted .
the @xmath62-claim and @xmath63-claim are combined into a single claim by choosing the smaller @xmath122 and bigger @xmath123 provided by their proofs . in this section
we show that if dead information can be added extensively enough , a tester of variant e with an arbitrarily small positive failure rate exists , but the opposite holds for variant s. the reason for the result on variant e is that as the size of the programs grows , a bigger and bigger proportion of programs consists of copies of smaller programs .
this phenomenon is so strong that to obtain the desired failure rate , it suffices to know the empty - input behaviour of all programs up to a sufficient size .
an _ end - of - file dead segment language _ is defined otherwise like end - of - file data segment language ( definition [ d : eof - data ] ) , but the actual program can not read the data segment .
this is the situation with any self - delimiting real - life programming language , whose compiler stops reading its input when it has read a complete program .
any end - of - file dead segment language is frequent and computationally domain - frequent .
[ t : easy - sd ] for each end - of - file dead segment language , @xmath163 the result also holds with approximating and generic - case testers .
let @xmath164 denote the number of programs whose dead segment is not empty .
we have @xmath165 , so @xmath166 . for each @xmath131 ,
so @xmath168 grows as @xmath11 grows .
these imply that there is @xmath169 such that @xmath170 from below when @xmath96 . because there are programs , @xmath171 . for every @xmath73 we have @xmath172 , so there is @xmath173 such that @xmath174 . on the other hand , @xmath175 .
these imply @xmath176 . here
@xmath177 is the number of those programs of size @xmath11 whose actual program is of size less than @xmath178 .
the behaviour of a program on the empty input only depends on its actual program .
let @xmath179 be the size of the actual program .
consider a three - way tester that looks the answer from a look - up table if @xmath180 and replies `` i do nt know '' if @xmath181 ( cf.proposition [ p : finite ] ) .
it has @xmath182 , implying the claim .
proposition [ p:3way- > ] generalizes the result to approximating and generic - case testers .
the above proof exploited the fact that the correct answer for a long program is the same as the correct answer for a similarly behaving short program .
this does not work for testers of variant s , because the short and long program no longer get the same input , since each one gets itself as its input .
although the program does not have direct access to its dead segment , it gets it via the input .
this changes the situation to the opposite of the previous theorem .
[ t : hard - sd ] for each end - of - file dead segment language , @xmath183 @xmath184[l]{{\textnormal{generic(s ) } } } : \forall n_0 \in \mathbb{n } : \exists n \geq n_0
: \frac{{\overline d_t}(n)}{p(n ) } \geq c\textrm { , and}\hspace{13mm}\ ] ] @xmath184[l]{{\textnormal{approx(s ) } } } : \forall n_0 \in \mathbb{n } : \exists n \geq
n_0 : \frac{{\overline h_t}(n)+{\overline d_t}(n)}{p(n ) } \geq c\textrm { .}\hspace{7mm}\ ] ] we prove first the claims on three - way and generic - case testers .
let us recall the overall idea of the proof of theorem [ t : itself ] .
in that proof , for any tester @xmath13 , a program @xmath108 was constructed that gives its input @xmath109 to @xmath13 .
if @xmath110 replies `` yes '' , then @xmath111 enters an eternal loop .
if @xmath110 replies `` no '' , then @xmath111 halts immediately . to prove that a three - way tester has many hard ( a ) halting ( b ) non - halting instances , in the case of the
`` i do nt know '' reply , @xmath111 was made to ( a ) halt immediately ( b ) enter an eternal loop .
all programs that halt on the same inputs as @xmath108 were shown to be hard instances for @xmath13 . for each @xmath11 that is greater than a threshold that may depend on @xmath13 , the existence of at least @xmath129
such programs was proven , where @xmath122 may depend on @xmath13 but not on @xmath11 .
we now apply the same idea , but , to get a result where the same constant @xmath59 applies to all testers @xmath13 , we no longer construct a separate program @xmath108 for each @xmath13 . instead , we construct a single program @xmath50 , which obtains @xmath13 from the size of the input of @xmath50 .
( a similar idea appears in @xcite . ) to discuss this , for any @xmath185 , let @xmath186 be the program whose shortlex index is @xmath14 .
let @xmath187 , where @xmath188 is the biggest square number that is at most @xmath14 .
the essence of @xmath189 is that as @xmath14 gets the values @xmath1 , @xmath190 , @xmath191 , , @xmath189 gets each value @xmath1 , @xmath190 , @xmath191 ,
infinitely many times .
one more idea needs to be explained before discussing the details of @xmath50 .
let @xmath16 be partitioned to @xmath192 and @xmath193 of sizes @xmath194 and @xmath195 .
let @xmath179 be the size of the actual program of @xmath50 .
for each @xmath196 , by modifying the dead segment , @xmath197 programs are obtained that have the same actual program as @xmath50 . for @xmath198 ,
let @xmath199 be the set of those of them whose dead segment ends with a character in @xmath200 .
we have @xmath201 . because @xmath202 , by choosing @xmath203 we get @xmath204 .
the program @xmath50 first checks that its input @xmath109 is a program with a non - empty dead segment .
if it is not , then @xmath50 halts immediately .
otherwise , @xmath50 constructs @xmath205 by going through all character strings in the shortlex order until @xmath206 programs have been found .
then @xmath50 constructs every program @xmath106 that has the same size , has the same actual program , and belongs to the same @xmath199 as @xmath109 .
then @xmath50 executes the @xmath207 in parallel until any of the following happens .
if any @xmath207 replies `` yes '' , then @xmath50 enters an eternal loop . if any @xmath207 replies `` no '' , then @xmath50 aborts the remaining @xmath207 and halts .
if every @xmath207 replies `` i do nt know '' , then @xmath50 halts if @xmath208 , and enters an eternal loop if @xmath209 .
if none of the above ever happens , then @xmath50 fails to halt .
recall that @xmath179 is the size of the actual program of @xmath50 . for any tester @xmath13
, there are infinitely many @xmath11 such that @xmath196 and @xmath210 is @xmath13 . for any such @xmath11 , there are @xmath197 programs @xmath116 of size @xmath11 that have the same actual program as @xmath50 .
let @xmath211 be any of them .
the execution of @xmath212 starts @xmath213 for at least @xmath214 distinct @xmath116 .
if @xmath213 replies `` yes '' , then @xmath13 claims that @xmath119 halts . then also @xmath215 halts , because @xmath50 halts on the same inputs as @xmath116 , since they have the same actual program .
furthermore , @xmath212 halts , because @xmath50 only looks at the size , actual program , and @xmath199-class of its input , and @xmath211 and @xmath116 agree on them .
but the halting of @xmath212 is in contradiction with the behaviour of @xmath50 described above .
therefore , no @xmath213 can reply `` yes '' . for a similar reason
, none of them replies `` no '' either . in conclusion ,
all at least @xmath214 distinct @xmath116 are hard instances for @xmath13 .
if @xmath13 is a three - way tester , it replies `` i do nt know '' for all of them . depending on whether @xmath216 or @xmath217 , they are hard halting or hard non - halting instances
. if @xmath13 is a generic - case tester , it halts on none of these hard instances .
therefore , also @xmath212 and @xmath218 fail to halt .
so they all are hard non - halting instances . in the case of approximating testers ,
@xmath50 is modified such that it lets all @xmath207 run into completion and counts the `` yes''- and `` no''-replies that they give .
if the majority of the replies are `` no '' , then @xmath50 halts , otherwise @xmath50 enters an eternal loop . for the same reasons as above , @xmath212 halts if and only if @xmath215 halts if and only if @xmath119 halts
. so at least half of the replies are wrong .
finally , we prove a corollary of the above theorem that deals with _ the halting problem itself _ , not with imperfect testers .
imperfect testers are used in the proof of the corollary , but not in the statement of the corollary .
[ l : limit ] let x be any of e , s , and g , and let @xmath219 be any total computable function from natural numbers to integers . if @xmath220 then @xmath221 does not exist .
assume that @xmath222 and @xmath73 .
let @xmath223 .
there is an @xmath224 of the form @xmath225 such that @xmath149 is an integer , @xmath226 when @xmath227 , and @xmath228 .
there also is @xmath229 such that when @xmath230 , then @xmath231 .
a tester @xmath13 that disobeys the formula is obtained as follows .
if @xmath232 , @xmath13 replies `` i do nt know '' .
if @xmath230 , @xmath13 simulates all instances of size @xmath11 until @xmath233 have halted .
if the given instance is among those that halted , then @xmath13 replies `` yes '' and otherwise `` i do nt know '' .
we have @xmath234 .
[ c : nolimit1 ] consider variant s of the halting problem and any end - of - file dead segment language .
then @xmath235 does not exist .
the proof of lemma [ l : limit ] can be modified to approximating testers with @xmath236 . by ( [ e : ksz05 ] )
, the limit fails to exist also in the framework of @xcite .
we first show that among all character strings of size @xmath11 , those that are not c++ programs that is , those that yield a compile - time error dominate overwhelmingly , as @xmath11 grows .
in other words , a random character string is not a c++ program except with vanishing probability .
the result may seem obvious until one realizes that a c++ program may contain comments and string literals which may contain almost anything .
we prove the result in a form that also applies to .
c++ is not self - delimiting . after a complete c++ program , there may be , for instance , definitions of new functions that are not used by the program .
this is because a c++ program can be compiled in several units , and the compiler does not check whether the extra functions are needed by another compilation unit .
even so , if @xmath65 is a c++ program , then @xmath237 is definitely not a c++ program and not even a prefix of a c++ program .
similarly , if @xmath65 is a program , then @xmath238}$ ] is not a prefix of a program .
[ p : syntax ] if for every @xmath239 there is @xmath240 such that @xmath241 , then @xmath242 let @xmath243 .
obviously @xmath244 .
assume first that for every @xmath245 , there is @xmath246 such that @xmath247 for every @xmath248 . because @xmath249 , we get @xmath250 as @xmath96 .
in the opposite case there is @xmath245 such that @xmath251 for infinitely many values of @xmath11 .
let they be @xmath252 . because @xmath253 is not a prefix of any program , @xmath254 .
for the remaining values of @xmath11 , obviously @xmath255 @xmath256 .
these imply that when @xmath257 , we have latexmath:[$0 \leq p(n)/|\sigma|^n \leq q(n)/|\sigma|^n \leq q(n_i ) / |\sigma|^{n_i } \leq ( 1 - \varepsilon / @xmath96 .
consider a tester @xmath13 that replies `` no '' if the compilation fails and `` i do nt know '' otherwise .
if compile - time error is considered as non - halting , then proposition [ p : syntax ] implies that @xmath259 , @xmath260 , @xmath261 , and @xmath262 when @xmath96 . as we pointed out in section [ s : ksz05 ] , this is yet another instance of an anomaly stealing the result .
the model of computation we study in this section is program
input pairs , where the programs are written in c++ , and the inputs obey the rules stated by the linux operating system .
furthermore , @xmath16 is the set of all 8-bit bytes . to make firm claims about details ,
it is necessary to fix some language and operating system .
the validity of the details below has been checked with c++ and linux .
most likely many other programming languages and operating systems could have been used instead .
there are two deviations from the real everyday programming situation .
first , of course , it must be assumed that unbounded memory is available .
otherwise everything would be decidable .
( however , at any instant of time , only a finite number of bits are in use . )
second , it is assumed that the programs do not contain comments .
this assumption needs a discussion .
comments are information that is inside the program but ignored by the compiler .
they have no effect to the behaviour of the compiled program .
we show next that most long c++ programs consist of a shorter c++ program and one or more comments .
[ l : clc++ ] there are at most @xmath263 comment - less c++ programs of size @xmath11 . everywhere inside a c++ program excluding comments , it is either the case that ` @ ` or the case that the new line character @xmath132{->}(2,1.5)(2,.5)(0,.5 ) \end{pspicture}}$ ] can not occur next .
that is , for every character string @xmath18 , either @xmath264 or @xmath265{->}(2,1.5)(2,.5)(0,.5 ) \end{pspicture}}$ ] is not a prefix of any comment - less c++ program .
( perhaps surprisingly , there indeed are places that are outside comments and where any byte except @xmath132{->}(2,1.5)(2,.5)(0,.5 ) \end{pspicture}}$ ] can occur . )
[ l : allc++ ] if @xmath266 , then there are at least @xmath267 c++ programs of size @xmath11 .
let @xmath268 , and let @xmath269 .
consider the character strings of the form ` int main()`\{`/*\alpha\beta*/ ` } where @xmath18 consists of @xmath270 space characters and @xmath19 is any string of the form @xmath271 , where @xmath272 for @xmath273 .
each such string is a syntactically correct c++ program of size @xmath11 .
their number is @xmath274 .
the proportion of comment - less c++ programs among all c++ programs of size @xmath11 approaches @xmath2 , when @xmath96 .
let @xmath275 . by lemmas [ l : clc++ ] and [ l : allc++ ] ,
the proportion is at most + @xmath276 , when @xmath96 . as a consequence , although comments are irrelevant for the behaviour of programs , they have a significant effect on the distribution of long c++ programs .
to avoid the risk that they cause yet another anomaly stealing the result , we restrict ourselves to c++ programs without comments . this assumption does not restrict the expressive power of the programming language , but reduces the number of superficially different instances of the same program .
the input may be any finite string of bytes .
this is how it is in linux .
although not all such inputs can be given directly via the keyboard , they can be given by directing the so - called standard input to come from a file .
there is a separate test construct in c++ for detecting the end of the input , so the end of the input need not be distinguished by the contents of the input .
there are @xmath277 different inputs of size @xmath11 .
the sizes of a program and input are the number of bytes in the program and the number of bytes in the input file .
this is what linux reports .
the size of an instance is their sum .
analogously to section [ s : frequent ] , the size of a program is additional information to the concatenation of the program and the input .
this is ignored by our notion of size .
however , the notion is precisely what programmers mean with the word .
furthermore , the convention is similar to the convention in ordinary ( as opposed to self - delimiting ) kolmogorov complexity theory @xcite .
[ l : pc++ ] with the c++ programming model in section [ s : c++model ] , @xmath278 , the number of different program
input pairs of size @xmath11 is at most @xmath279 + the next theorem says that with halting testers of variant g and comment - less c++ , the proportions of hard halting and hard non - halting instances do not vanish .
[ t : ill - h - c ] with the c++ programming model in section [ s : c++model ] , @xmath280 we prove first the @xmath281 part and then the @xmath282 part .
the results are combined by picking the bigger @xmath123 and the smaller @xmath122 .
there is a program @xmath108 that behaves as follows .
first , it gets its own size @xmath283 from a constant in its program code . the constant uses some characters and thus affects the size of @xmath108 .
however , the size of a natural number constant @xmath149 is @xmath284 and grows in steps of zero or one as @xmath149 grows .
therefore , by starting with @xmath285 and incrementing it by steps of one , it eventually catches the size of the program , although also the latter may grow .
then @xmath108 reads the input , counting the number of the characters that it gets with @xmath286 and interpreting the string of characters as a natural number @xmath109 in base @xmath43 .
we have @xmath287 , and any natural number in this range is possible .
let @xmath288 .
next @xmath108 constructs every program input pair of size @xmath11 and tests it with @xmath13 . in this way
@xmath108 gets the number @xmath33 of easy halting pairs of size @xmath11 .
then @xmath108 constructs again every pair of size @xmath11 .
this time it simulates each of them in parallel until @xmath289 of them have halted .
then it aborts the rest and halts .
it halts if and only if @xmath290 .
( it may be helpful to think of @xmath109 as a guess of the number of hard halting pairs . ) among the pairs of size @xmath11 is @xmath108 itself with the string that represents @xmath109 as the input .
we denote it with @xmath291 .
the time consumption of any simulated execution is at least the same as the time consumption of the corresponding genuine execution .
so the execution of @xmath291 can not contain properly a simulated execution of @xmath291 .
therefore , either @xmath291 does not halt , or the simulated execution of @xmath291 is still continuing when @xmath291 halts . in the former case , @xmath292 . in the latter case
@xmath291 is a halting pair but not counted in @xmath293 , so @xmath294 . in both cases , @xmath295 .
as a consequence , no natural number less than @xmath296 is @xmath34 .
so @xmath297 . by lemma
[ l : pc++ ] , @xmath298 .
so for any @xmath299 , we have @xmath300 .
the proof of the @xmath301 part is otherwise similar , except that @xmath108 continues simulation until @xmath302 pairs have halted .
( now @xmath109 is a guess of @xmath36 , yielding a guess of @xmath30 by subtraction . )
the program @xmath108 gets @xmath28 by counting the pairs of size @xmath11 whose program part is compilable .
it turns out that @xmath303 , so @xmath109 can not be @xmath36 , yielding @xmath304 .
next we adapt the second main result in @xcite to our present setting , with a somewhat simplified proof and obtaining the result also for three - way and generic - case testers .
[ t : indep ] with the c++ programming model in section [ s : c++model ] , @xmath305 @xmath306[l]{{\textnormal{generic(g ) } } } : \forall n_0 \in \mathbb{n } : \exists
: \frac{{\overline d_t}(n)}{p(n ) } \geq c\textrm { , and}\hspace{13mm}\ ] ] @xmath306[l]{{\textnormal{approx(g ) } } } : \forall n_0 \in \mathbb{n } : \exists n \geq
\frac{{\overline h_t}(n)+{\overline d_t}(n)}{p(n ) } \geq c\textrm { .}\hspace{7mm}\ ] ] the proof follows the same strategy as the proof of theorem [ t : hard - sd ] , but differs in some technical details . to prove the claim for three - way testers , for any character string @xmath18 , let
@xmath307 if @xmath18 is the empty string , and otherwise @xmath308 is the value of the least significant bit of the last character of @xmath18 . for any character strings @xmath18 and @xmath19 , let @xmath309 if and only if @xmath310 and @xmath311 .
for any size @xmath11 greater than @xmath2 , `` @xmath312 '' has two equivalence classes , each containing @xmath313 character strings . for any @xmath185 ,
let @xmath186 be the program whose shortlex index is @xmath14 .
there is a program @xmath50 that behaves as follows .
we denote its execution on input @xmath18 with @xmath314 .
please observe that if @xmath309 , then @xmath315 behaves in the same way as @xmath314 .
first @xmath314 finds the program @xmath316 , where @xmath317 , where @xmath188 is the biggest square number that is at most @xmath14 .
then @xmath314 goes through , in the shortlex order , all @xmath318 , until any of the termination conditions mentioned below occurs or @xmath314 has gone through all of them . for each @xmath19
, it runs @xmath316 on @xmath19 .
we denote this with @xmath319 .
if @xmath319 fails to halt , then @xmath314 never returns from it and thus fails to halt .
if @xmath319 halts replying `` yes '' , then @xmath314 enters an eternal loop , thus failing to halt .
if @xmath319 halts replying `` no '' , then @xmath314 halts immediately .
if @xmath319 halts replying `` i do nt know '' , then @xmath314 tries the next @xmath19 .
it is not important what @xmath314 does if @xmath319 halts replying something else . if @xmath319 halted replying `` i do nt know '' for every @xmath19 such that @xmath309 , then @xmath314 checks whether @xmath307 .
if yes , then @xmath314 enters an eternal loop , otherwise @xmath314 halts .
now let @xmath320 be any three - way tester that tests whether program @xmath321 halts on the input @xmath322 .
how the two components @xmath321 and @xmath322 of the input of @xmath13 are encoded into one input string is not important .
there is a program that has @xmath50 hard - coded into a string constant , inputs @xmath19 , calls @xmath323 , and gives its reply as its own reply . let @xmath14 be the shortlex index of this program , so the program is @xmath186 .
there are infinitely many positive integers @xmath92 such that @xmath324 .
let @xmath92 be such , and let @xmath18 be any character string of size @xmath92 .
so @xmath316 is @xmath186 .
if , during the execution of @xmath314 , @xmath325 ever replies `` yes '' or `` no '' , then the same happens during the execution of @xmath315 , because @xmath315 behaves in the same way as @xmath314 ( the fact that @xmath325 was called implies @xmath309 ) .
but that would be incorrect by the construction of @xmath50
. therefore , @xmath323 replies `` i do nt know '' for every @xmath19 of size @xmath92 . as a consequence
, @xmath13 has at least @xmath326 hard instances of size @xmath327 . if @xmath328 , then half of them are halting and the other half non - halting , thanks to the @xmath307 test near the end of @xmath50 . by lemma [ l : pc++ ] ,
so if @xmath329 , then @xmath330 the program @xmath50 does not depend on @xmath11 , so letting latexmath:[$c = 1/(2 the proof for generic - case testers is otherwise similar , but the @xmath19 are tried in parallel and @xmath323 fails to halt for every @xmath19 of size @xmath92 .
all hard instances are non - halting .
the @xmath50 for approximating testers lets each @xmath319 continue until completion , counts the numbers of the `` yes''- and `` no''-replies they yield , and then does the opposite of the majority of the replies .
application of lemma [ l : limit ] to this result yields the following .
[ c : nolimit2 ] with the c++ programming model in section [ s : c++model ] , @xmath332 does not exist .
this study did not cover all combinations of a programming model , variant of the halting problem , and variant of the tester .
so there is a lot of room for future work .
the results highlight what was already known since @xcite : the programming model has a significant role . with some programming models , a phenomenon of secondary interest dominates the distribution of programs , making hard instances rare .
such phenomena include compile - time errors and falling off the left end of the tape of a turing machine .
many results were derived using the assumption that information can be packed very densely in the program or the input file .
sometimes it was not even necessary to assume that the program could use the information .
it sufficed that the assumption allowed to make many enough similarly behaving longer copies of an original program .
intuition suggests that if the program can access the information , testing halting is harder than in the opposite case .
a comparison of theorem [ t : easy - sd ] to theorem [ t : hard - sd ] supports this intuition .
corollaries [ c : nolimit1 ] and [ c : nolimit2 ] and the comment after corollary [ c : nolimit1 ] tell that the proportion of _ all _ ( not just hard ) halting instances has no limit with end - of - file dead segment languages and variant s of the halting problem , with the c++ model and variant g , and in the framework of @xcite .
it must thus oscillate irregularly as the size of the program grows irregularly because of lemma [
l : limit ] .
this is not a property of various notions of imperfect halting testers , but a property of the halting problem itself .
i thank professor keijo ruohonen for helpful discussions , and the anonymous reviewers of splst 13 and acta cybernetica for their helpful comments .
the latter pointed out that proposition [ p : syntax ] had been formulated incorrectly .
khler , s. , schindelhauer , c. , and ziegler , m. on approximating real - world halting problems . in likiewicz ,
m. and reischuk , r. , editor , _ proc .
15th fundamentals of computation theory _ , lecture notes in computer science 3623 , pages 454466 , 2005 .
springer .
schindelhauer , c. and jakoby , a. the non - recursive power of erroneous computation . in pandu
rangan , c. , raman , v. , and ramanujam , r. , editors , _ proc .
19th foundations of software technology and theoretical computer science _ , lecture notes in computer science 1738 , pages 394406 , 1999 .
springer .
valmari , a. sizes of up - to-@xmath11 halting testers . in halava , v. , karhumki ,
j. , and matiyasevich , y. , editors , _ proceedings of the second russian finnish symposium on discrete mathematics _ , tucs lecture notes 17 , pages 176183 , turku , finland , 2012 . valmari , a. the asymptotic behaviour of the proportion of hard instances of the halting problem . in kiss , . ,
editor , _ proceedings of splst 13 , 13th symposium on programming languages and software tools _ , pages 170184 , szeged , hungary , 2013 . | although the halting problem is undecidable , imperfect testers that fail on some instances are possible .
such instances are called _ hard _ for the tester .
one variant of imperfect testers replies `` i do nt know '' on hard instances , another variant fails to halt , and yet another replies incorrectly `` yes '' or `` no '' .
also the halting problem has three variants : does a given program halt on the empty input , does a given program halt when given itself as its input , or does a given program halt on a given input .
the failure rate of a tester for some size is the proportion of hard instances among all instances of that size .
this publication investigates the behaviour of the failure rate as the size grows without limit .
earlier results are surveyed and new results are proven .
some of them use c++ on linux as the computational model .
it turns out that the behaviour is sensitive to the details of the programming language or computational model , but in many cases it is possible to prove that the proportion of hard instances does not vanish .
* keywords : * halting problem , three - way tester , generic - case tester , approximating tester * acm computing classification system 1998 : * f.1.1 models of computation computability theory * mathematics subject classification 2010 : * 68q17 computational difficulty of problems |
galaxies grow from small mass systems , the first structures to be unstable after recombination are of the order of a globular cluster mass ( @xmath0 10@xmath1 m@xmath2 ) . then , according to the hierarchical scenario of galaxy formation , small systems merge to form larger and larger systems , and in the same time , small structures accrete gas mass and dark matter from larger gaseous structures in the shape of filaments in the cosmic web . the relative importance of these two essential ways to build galaxies , through mergers or external gas accretion , is still unprecised in the numerical simulations , since it depends on many unknown parameters of the baryonic physics , gas dissipation , , star formation efficiency , feedback , gas re - heating , etc .. however the dynamical history of galaxies strongly depends on these processes , and it might be possible to find constraints on them from the observations , locally and at various redshifts , of the dynamical states of galaxies .
interaction with massive companions and mergers tend to heat the stellar component of galaxies , and to form the spheroids , either increase the mass of the central bulge , or even transform the system in a giant ellipticals in case of major mergers . on the contrary
, gas accretion can replenish the young disk in spiral galaxies , and rejuvenate spiral or bar waves , and reduce the bulge - to - disk ratio .
the thickness of galaxy disks , and their ability to maintain spiral structure and asymmetries is a tracer of their history , in terms of interactions or gas accretion .
i will review here the recent progress in obtaining these constraints , both on the observational side , which provides statistics of the dynamical state of spiral galaxies , and on the theoretical side , which provides the interpretation through predictions of this dynamical state under several hypothesis about the past history of galaxies .
section 2 reviews the star formation history of galaxies in the field , and section 3 the constraints from the dynamics : bars , warps , polar rings and asymmetries .
section 4 considers the same phenomena in clusters , and conclusions about secular evolution as a function of environment are drawn in section 5 .
models of the chemical evolution of the milky way suggest that the observed abundances of metals require a continuous infall of gas with metallicity about 0.1 times the solar value .
this can solve the well - known g - dwarf problem , i.e. the observational fact that the metallicities of most long - lived stars near the sun lie in a relatively narrow range ( -0.6 @xmath3 [ fe / h ] @xmath3 + 0.2 , rocha - pinto & maciel 1996 ) .
the infall of gas is also supported by the constant or increasing star formation rate ( sfr ) scenario inferred from the local distribution of stars ( e.g. haywood et al .
other abundance problems require also an infall rate integrated over the entire disk of the milky way of a few solar mass per year at least ( casuso & beckman 2001 ) .
this infall dilutes the enrichment arising from the production of heavy elements in stars , and thereby prevents the metallicity of the interstellar medium from increasing steadily with time .
some of this gas could come from the high velocity clouds ( hvc ) infalling onto our galaxy disk ( wakker et al 1999 ) .
the high - velocity clouds ( hvcs ) observed in the galactic neighbourhood , at least those not included into the magellanic stream more akin to tidal debris , have been proposed to be remnants of the formation of the galaxies in the local group ( blitz et al 1999 ) . with distances
much larger than previously assumed ( i.e. 1 mpc instead of 100 kpc ) , their gas masses could represent a large fraction of the total baryonic mass of the local group .
this hypothesis is supported by observational evidence that their kinematical centre is the local group barycentre ( blitz et al 1999 ) . within this hypothesis
, hvcs can well explain the evolution of the light elements in the galaxy , and the g - dwarf problem ( lopez - corredoira et al 1999 ) .
the present infall rate of gas is estimated to be 7.5 mo / yr ( blitz et al 1999 ) .
this is the right order of magnitude to double the baryonic mass of the galaxy in 10 gyr time - scale .
the hvcs , whose properties are very similar to the higher redshift ly@xmath4 forest clouds , may thus form a significant constituent of baryonic , and of non - baryonic , dark matter .
it is now possible to investigate the star formation evolution in nearby galaxies as well : worthey & espana ( 2004 ) with hst colour - magnitude diagrams derived a stellar abundance distribution for m31 .
the results are quite similar to what is observed in the solar neighbourhood and they concluded that closed - box models in m31 suffer a g - dwarf problem even more severe than in the milky way . the star formation rate in the milky way disk has remained of the same order of magnitude over the galaxy life - time ( rana & wilkinson 1986 ) , although some temporal fluctuations can be identified ( rocha - pinto et al 2000 ) .
this appears to be a general results for spiral galaxies in the middle of the hubble sequence ( e.g. kennicutt 1983 , kennicutt et al 1994 ) .
, by nagamine et al ( 2004 ) .
the mass of the galaxy in @xmath5 m@xmath2 and its rest - frame v - band magnitude are indicated at right.,width=377 ] to maintain the same order of magnitude for star formation rate requires external gas accretion , since an isolated galaxy must have an exponentially decreasing star formation rate , even taking into account stellar mass loss .
numerical simulations of galaxy evolution , in a general cosmological frame , have shown that indeed a typical spiral galaxy in average environment , maintains a constant level of star formation rate , with superposed bursts ( tissera 2000 , nagamine et al 2004 ) .
while simulations of isolated galaxy mergers reveal how star formation is triggered by interaction and mergers ( e.g. mihos & hernquist 1996 ) , they can not maintain a constant average level of star formation , as in cosmological simulations .
figures [ tissf2 ] and [ naga - f8 ] shows several histories of star formation for galaxy - like objects at @xmath6 , from tissera ( 2000 ) and at @xmath7 , nagamine et al .
the history has been obtained by reconstructing the merging tree , in identifying the progenitors of the present galaxy with the particles it is made of in the simulation , a companion being defined by having at least 10% of the main galaxy mass ( the accretion of baryonic clumps are assimilated to external gas accretion ) .
each time such a companion enters the main halo , an up arrow is drawn , and each merger is traced by a down arrow .
several phenomena can be emphasized : a large fraction of gas is accreted between mergers , and this contribute an essential part of the star formation .
the tidal interaction triggers bursts of star formation , which in general are delayed with respect to the first passage , but the bursts are not only fuelled by the accreted companion gas , but mainly from the main galaxy gas driven in by the tidal forces . in that sense ,
the external gas accretion in between mergers also fuels the merger - triggered starbursts .
observations of stellar populations in a large sample of local galaxies ( sdss , heavens et al 2004 , jimenez et al 2004 ) have shown that massive galaxies have formed most of their stars at early times , while dwarf galaxies are forming their stars efficiently only now .
only intermediate mass galaxies could have in average maintained their star formation rate over a hubble time .
the high efficiency of star formation in massive galaxies underlines the role of environment , and the availability of high gas accretion .
the rate of gas consumption is then high , and the galaxies run out of gas quickly . on the contrary ,
far from deep potential wells , dwarf galaxies need billion years to reach the threshold of star formation , and convert their gas only now . in semi - analytic models , an essential free parameter to reconstruct merger trees and mass assembly in galaxies , in the fraction of accretion , and
this can change considerably the final angular momentum of the system ( wechsler et al 2002 ) .
figure [ sp99-f5 ] illustrates the time evolution of the baryonic content of halos that correspond to a `` local group '' ( @xmath8 km / s ) sized halo at @xmath6 .
this evolution corresponds to the best fit of the observations , i.e. the luminosity function of galaxies and the tully - fisher relation .
the star formation rate has been calibrated , so that the star formation efficiency is constant with redshift ( somerville & primack 1999 ) .
note that the fraction of available gas in the form of cold phase remains quite high until @xmath6 .
220 km / s at @xmath6 , in a standard cdm model , computed with semi - analytic models by somerville & primack ( 1999 ) .
, width=377 ] as far as the global gas content is concerned , constraints can be found in the frequency and column densities of the damped lyman-@xmath4 systems ( dla ) . already at high redshift ,
the models are limited by the observations , which then constrain the star formation rate ( fig [ sp98-f6 ] ) .
a strong starburst mode is required in addition to quiescent star formation mode , as triggered by the frequent mergers around @xmath9 .
note that the observational points only correspond to hi gas of column density larger than 2 10@xmath10 @xmath11 , ignoring molecular or ionised gas , that could bring the model closer to observations at low redshift .
is from local hi observations ( zwaan et al 1997 ) .
, width=377 ]
several dynamical processes can be naturally explained with external gas accretion , and could be used as constraints on the amount of gas available , they are summarised in fig [ dyn - gas ] . from observations and numerical simulations in the last decades , secular evolution has been shown to be driven by bars and spirals ( e.g. kormendy & kennicutt 2004 ) .
in particular , the bar gravity torques produce radial gas inflow , generating star formation . dynamical instabilities then regulate themselves , since the gas inflow itself can destroy the bar .
the bar is destroyed by two main mechanisms : first the central mass concentration built after the gas inflow , destroys the orbital structure sustaining the bar , scatter particles and push them on chaotic orbits ( hasan et al 1990 , 1993 , hozumi & hernquist 1999 ) .
second , the gas inflow itself weakens the bar , since the gas loses its angular momentum to the stars forming the bar ( bournaud & combes 2004 ) .
this increases the angular momentum of the bar wave , in decreasing the eccentricity of the orbits .
this bar destruction is reversible , since the central mass concentration is then not strong enough to prevent bar formation
. secular evolution then includes several bar episodes in a galaxy life - time .
a spiral galaxy rich in gas ( at least 5% of the disk mass ) is unstable with respect to bar formation .
gravity torques are then efficient to drive the matter inwards .
the galaxy morphological type evolves towards early - types , the mass is concentrated , the bulge is developped , through horizontal and vertical resonances .
this weakens the bar , and when the galaxy becomes again axi - symmetric , gas can be accreted from the outer parts by viscosity ( bournaud & combes 2002 ) .
the gas accretion , if significant with respect to the disk mass , can reduce the bulge - to - disk ratio ( by replenishing the disk ) , and make the galaxy disk unstable again to a new bar .
several bars can successively produce secular evolution of spiral galaxies , if there is enough external gas accretion .
this can be used to quantify the amount of accretion in a galaxy life - time . to estimate the number of bars that has occured in a typical galaxy , and the time spent in a bar phase , block et al ( 2002 )
have estimated quantitatively the frequency of bars and @xmath12 spiral components in a sample of 163 spiral galaxies , observed in the near - infrared ( eskridge et al 2002 ) . in this band , it is more easy to identify the bars in the old stellar component , with limited dust extinction .
the bar strength @xmath13 has been estimated from the gravitational potential , derived from the nir images , assuming constant mass - to - light ratios .
the main surprising result in the histogram of bar strength is the hole at low values ( lack of axi - symmetric galaxies ) , and a long tail at high values , corresponding to a large number of strongly barred galaxies ( see also whyte et al 2002 , buta et al 2004 ) .
it can be concluded from the comparison of bar life - times given by numerical simulations , and the observation of bar frequency , that bars have to be reformed , and this can only be through significant accretion of external gas ( see bournaud & combes , this volume ) .
most spiral galaxies reveal a warped disk , particularly spectacular in the hi component ( briggs 1990 ) , although it is still frequent in the optical disk ( reshetnikov & combes 1998 ) .
many physical mechanisms have been studied to explain the origin of these warps ( see the review by binney 1992 ) , but the high frequency of warps is only accounted for by external gas and dark matter accretion .
matter accreted in the outer parts of a dark matter halo , with an angular momentum misaligned with that of the inner parts , car re - oriented the whole system in 7 - 10 gyr , and the corresponding disk will show an integral - sign warp during the process ( jiang & binney 1999 ) .
intergalactic gas accretion is sufficient to produce a torque corresponding to the observed warps , with the infall amplitude already required by evolutionary chemical models ( lopez - corredoira et al 2002 ) . during some transient phase , a u - shape warp
is predicted , combining m=0 and m=1 perturbations , and its frequency corresponds to the observations .
the galaxy has just to move in an accretion flow of average density of 10@xmath14 @xmath15 , and velocity 100km / s .
bullock et al ( 2001 ) have explored the angular momentum distribution in a large number of dark matter haloes , from a cosmological simulation .
they found an almost universal form , with a moderate misalignment in the outer parts , that could also favor warps in the baryonic matter .
however , both the halo growth through merger of smaller units , and quiescent and continuous accretion are able to produce the same universal profile , so this will not be discriminant .
polar ring galaxies are particular objects , composed of a primary galaxy in general of early - type ( lenticular or elliptical ) , and a polar ring or disk , where the matter is orbiting in a nearly perpendicular plane .
the polar material is quite gas - rich , but also composed of stars , of age larger than a few gyrs , implying that the polar disk is stable .
indeed , in any of the various scenarios advanced to account for this structure , the bulk of the stars form later from the gas settled in the polar plane .
the probability for a galaxy to have a polar ring has been estimated to @xmath0 5% , given the detection biases , that these systems are better detected edge - on ( whitmore et al 1990 ) .
the two main scenarios to form polar rings are mergers of two galaxies , or external gas accretion , this gas coming either from a passing - by companion , or from a cosmic gaseous filament . in the merging scenario ( bekki 1997 , 1998 ) , a special geometry is required for the two interacting system , it must be a head - on collision at low velocity , with the two galaxy planes nearly perpendicular to each other .
in the accretion scenario , gas may be accreted from an unbound companion passing by ( schweizer et al .
1983 , reshetnikov & sotnikova 1997 ) , in a nearly polar orbit . the companion must be gas - rich , and may also be more massive than the primary .
but gas could also come from a gaseous cosmic filament , during the long formation period of a giant galaxy , since the various filaments have not the same orientation .
this is the extreme case of the warp formation mechanism , invoked in the last subsection , where the various accreted gas have not aligned angular momentum .
intermediate cases between warps and polar rings , where rings of gas orbit at around 45@xmath16 , are observed ( for instance ngc 660 ) .
polar rings around galactic - like objects are frequently seen in cosmological simulations ( semelin & combes 2004 ) . from a large number of numerical simulations ,
exploring the various geometrical or mass parameters of the two scenarios , it has been shown that the accretion scenario is more likely to produce polar rings , corresponding to observations , than the merger scenario ( bournaud & combes 2003 ) .
in addition to this statistical argument , some prototypical cases , like ngc 4650a , support the accretion hypothesis , since they do not possess a halo of very old stars , expected in the merger scenario .
the observation of extended disks of neutral hydrogen ( through the hi line at 21 cm ) around most spiral disks , has revealed a large frequency of asymmetries and lopsidedness . in their compiled sample of 1700 galaxies ,
richter & sancisi ( 1994 ) found that at least 50% of spiral galaxies are lopsided , and have a characteristic signature in their global hi spectrum .
this percentage is even higher in late - type galaxies , where matthews et al ( 1998 ) found a frequency of 77% of hi distorted profiles .
the asymmetries affect also the stellar disk , as shown by rix & zaritsky ( 1995 ) and zaritsky & rix ( 1997 ) on near - infrared images of galaxies .
about 20 - 30% of galaxies have a significantly perturbed stellar disk , quantified by the @xmath17 fourier term in their potential or density distribution ( kornreich et al 1998 , combes et al 2004 ) .
since most of these galaxies are isolated without obvious sources of perturbations ( e.g. wilcots & prescott 2004 ) , the interpretation was searched in a possible long life - time of these features .
baldwin et al . (
1980 ) proposed that the lopsidedness comes from @xmath17 kinematic waves build from off - center elliptical orbits that may persist a long time against differential precession .
although this can prolonge the life - time significantly , since the winding out by differential precession is quite long in the outer parts of galaxies , they conclude that it is still not sufficient to explain the high frequency of lopsidedness in neutral gas disks . alternatively
, the perturbations could come from recent minor mergers , explaining the non - correlation with the presence of companions ( zaritsky & rix 1997 ) . in this hypothesis
, lopsidedness can be used to constrain the merger frequency , which appears very high .
indeed , walker et al ( 1996 ) have estimated through n - body simulations that the life - time of perturbations are of the order of the gyr .
a high merging frequency may enter in conflict with other observations , for instance the presence of thin stellar disks in spiral galaxies ( toth & ostriker 1992 ) .
external gas accretion could solve the problem of the high frequency of lopsidedness , since it is likely that in many cases , the accretion is asymmetric .
the relative role of galaxy merging and external gas accretion must depend strongly on environment .
in particular , in rich clusters , mergers have considerably increased the fraction of spheroids and ellipticals , and tidal interactions and ram pressure have stripped and heated the cold gas around galaxies , so that external accretion will be reduced or suppressed . to quantify theses effects ,
we consider successively the influence of environment on star formation rate , morphological types , and dynamical features , such as bars or warps .
high resolution images with the hst , followed by spectroscopic surveys in a dozen galaxy clusters at redshift around 0.3 - 1 .
have allowed to follow galaxy evolution and in particular their star formation history as a function of time , and also position in the cluster ( oemler et al 1997 , dressler et al .
1999 , poggianti et al 1999 ) .
rich clusters of galaxies , and especially their dense cores , are well known to be dominated by early - type galaxies without star formation , or even passively evolving and anemic spirals , conspicuous for their low star - formation rate . to find star formation it is necessary to look back in time , at redshift at least larger than 0.2 , where there exists a larger fraction of blue galaxies ( b - o effect , butcher & oemler 1978 , 1984 ) .
the recent hst images have shown that these blue galaxies are disk galaxies in majority perturbed by tidal interactions , strongly suggesting that galaxy interactions are still triggering star formation activity around @xmath18 ( lavery & henry 1988 ) .
there is thus a strong evolution effect in clusters in a recent past .
moreover , detailed spectroscopic studies have shown that @xmath19 clusters possess a large fraction of peculiar galaxies , likely post - starburst , called e+a ( or k+a ) , devoid of emission lines ( and therefore with no current star formation ) , but very strong balmer absorption lines .
( dressler et al 1999 , poggianti et al 1999 ) .
this means that they have a large fraction of a stars , implying that the galaxy was experiencing a strong starburst that has just been abruptly interrupted .
these galaxies are in majority disk - dominated .
their fraction is about 20% , much larger than in the field , and in the clusters at @xmath6 ( dressler et al 1999 ) .
that star formation is quenched in clusters is revealed by the low fraction ( @xmath0 10% ) of h@xmath4 emitters ; star formation and morphological evolution in cluster galaxies appear to be largely decoupled ( couch et al 2001 ) . on the contrary ,
the star formation is continuing in groups .
about 55% of galaxies have been cataloged in groups at @xmath20 in the 2df and sdss surveys , and the h@xmath4 detection rate and equivalent width varies strongly and continuously with galaxy density ( balogh et al 2004 ) .
the star formation rate , as traced by the h@xmath4 line , is strongly dependent on the density of galaxies ( lewis et al 2002 , 2df survey ) .
there is a tight correlation between sfr and local projected density , as soon as the density is above @xmath21 = 1 galaxy / mpc@xmath22 , independent of the size of the structure . in clusters ,
the field star formation rate is only recovered at about 3 times the virial radius .
gmez et al ( 2003 ) find also a strong sfr-@xmath21 relation with the early data release of the sdss , the sf - quenching effect being even more noticeable for strongly star - forming galaxies .
the same break of the sfr-@xmath21 relation is observed at 1 galaxy / mpc@xmath22 .
this relation is somewhat linked to the morphological type - density ( t-@xmath21 ) relation , but can not be reduced to it , the spatial distribution of star forming galaxies is also quite clear in clusters .
there is a clear radial gradient , the star formation being more active in the outer parts ( balogh et al 1999 ) .
there is a smooth transition from a blue , disk - dominated galaxy population in the outskirts , to a red , bulge - dominated , galaxy population in the cluster cores .
although this is influenced by the various transformation processes occuring in clusters ( tidal interaction , mergers , harrassment , ram - pressure .. ) , this has been mainly interpreted in terms of the building up of the clusters themselves , i.e. gas - rich galaxies continue to fall into the cluster , and contribute to the star - formation activity there , but then their gas replenishment is stopped when these galaxies arrive into the core ( diaferio et al .
2001 ) . in a model
where the morphology of galaxies are assumed to depend only on their merger history , where mergers lead to spheroid and bulges , and subsequent gas cooling replenishes disks , it is possible to reproduce the morphological gradient , and star - formation gradient observed .
galaxies in clusters are much sooner devoid of their gas , and their gas fueling is stopped , as shown in figure [ diaf - f1 ] .
m@xmath2 in clusters ( top ) and in the field ( bottom ) , from semi - analytical models by diaferio et al ( 2001 ) .
the dashed lines correspond to sfr proportional to the cold gas mass over the dynamical time .
the solid lines include in addition a dependency of the sfr with redshift as ( 1+z)@xmath23.,width=377 ] there is a strong morphological segregation as a function of projected density @xmath21 at @xmath6 ( dressler 1980 ) , essentially the fraction of spirals fall from two thirds to less than one third to the benefit of lenticulars and ellipticals .
this segregation has also been observed at @xmath24 ( or 5gyrs ago ) , where the fraction of ellipticals is the same as at @xmath6 ( dressler at al 1997 ) .
the difference is in the fraction of lenticulars , which is less than at @xmath6 .
this suggests that the formation of elliptical galaxies occurs before the formation of rich clusters , probably in the loose - group phase or earlier .
lenticulars are generated in large numbers only after cluster virialization , through the various galaxy transformation processes , tidal interactions , mergers , ram - pressure stripping and sudden halt in gas fueling ( poggianti et al 2001 ) .
these results have been confirmed and precised with the local sloan survey by goto et al ( 2003 ) ; at least two mechanisms are required to explain the morphological segregation : first at low density in the outskirts of the cluster , the gas supply is stopped , and star formation halted .
second at high density , in the cluster core , galaxy interactions and merging must explain the high frequency of early - types .
the segregation is also found quite similar at @xmath18 and locally , confirming that ellipticals in clusters have formed earlier .
this corresponds to numerical simulations , which show that the halos located inside clusters form earlier than isolated halos of the same mass ( gottloeber et al 2001 ) .
many aspects of galaxy morphology in clusters can be explained by evolutionary processes ( see the reviews by moore 2003 , combes 2003 ) , but the fact that the gas supply is halted plays in all these processes a major role ( shioya et al 2004 and fig [ bek - f2 ] ) : this gas shortening could happen in two ways , either gradual from gas stripping , and star formation rate dropping subsequently , or abruptly , through a starburst triggered by an interaction , and lack of gas supply afterwards .
the second way is necessary to explain the high frequency of post - starburst systems ( k+a ) in distant clusters , and the low frequency of strong h@xmath4 line emitters .
equal to 3.2 the scale radius of the cluster @xmath25 = 230kpc ( solid line ) . by comparison
, the dashed line corresponds to a comparable orbit in a group with @xmath25 = 62kpc , and the dotted line , with @xmath26 = 0.3 @xmath25 in this group ( from bekki et al 2002).,width=377 ] spiral galaxies are globally less abundant in clusters , and are preferentially located in the outskirts .
it has long been known that clusters are the site of a particular class of spirals , called `` anemic '' , with low star formation , smooth arms , and low gas content ( van den bergh 1976 ) . in average ,
anemic galaxies have an hi deficiency of a factor 4 , but no co emission deficiency , their spiral structure is fuzzy and short - lived , since new stars with low velocity dispersion are no longer formed out of the gas ( elmegreen et al 2002 ) . the frequency of bars in cluster spirals has been studied in order to test the influence of tidal interactions by andersen ( 1996 ) for the virgo cluster . barred galaxies appear to be more centrally concentrated in the cluster core than unbarred galaxies .
this is inferred from both velocity dispersions which are quite different . on the contrary ,
barred or unbarred lenticulars have the same radial distribution in the cluster .
andersen ( 1996 ) suggests that tidal triggering by the cluster itself is the most likely source of the enhanced fraction of barred spirals in the cluster center .
the same phenomenon is observed in coma ( thompson 1981 ) .
bars occur twice as often in the core region of coma as they do in the outer parts of the cluster .
this is certainly the consequence of tidal interactions triggering bar instability ( gerin et al .
1990 , miwa & noguchi 1998 , berentzen et al 2004 ) . at a global scale
however , there does not appear to be any significant difference between the frequency of bars in clusters and in the field ( van den bergh , 2002 ) . because bars are triggered by galaxy interactions in the cluster cores
, this implies that the suppression of gas supply in clusters has a tendency to reduce bar frequency to compensate .
it is difficult to observe warps around spiral galaxies in clusters , because of their characteristic general hi deficiency .
about two thirds of galaxy clusters are hi deficient ( solanes et al .
gas is particularly stripped in the outer parts of galaxies , where warps develop .
the environment of polar ring galaxies seems to be similar to that of normal galaxies ( brocca et al 1997 ) .
there is no evidence of more companions for example .
some polar ring galaxies are observed in clusters ( taniguchi et al .
1986 ) . as for asymmetries
, they are very frequent in clusters , but this has to be attributed to galaxy interactions ; in a hubble time , each galaxy suffers about 10 encounters with an impact parameter less than 10kpc , due to the importance of substructures ( gnedin 2003 ) . in clusters of galaxies as a function of @xmath27 , the average density inside the radius r ( normalised to the critical density ) .
the filled circles , empty triangle and crosses are the data , while the squares connected by a dashed line are the theoretical predictions ( cosmological simulations from several groups ) , with an assumed baryon fraction of f@xmath28 = 0.16 ; other models ( with or without stellar feedback ) are shown with f@xmath28 = 0.20 ( small squares ) , from sadat & blanchard ( 2001),width=377 ]
the mass assembly of galaxies occurs through two main processes : hierarchical merging of smaller entities , and more diffuse gas accretion .
the relative importance of the two processes can not be easily found by cosmological simulations , since many physical parameters such as gas dissipation , star formation and feedback , are still unknown .
however , star forming histories in galaxies ( age , kinematics and metallicity ) , and also the dynamical states of galaxies ( bars , spirals , warps , polar rings , asymmetries .. ) can constrain the role of the two processes .
chemical evolution of normal spiral galaxies requires gas infall of low - metallicity gas , in the proportion of a few solar masses per year .
the developpment of bars and spirals in galaxies constrain the amount of external gas accretion at about 10 m@xmath2/yr .
bars are transient features in galaxies , they exert gravity torques on the gas component , and drive mass to the center
. a new bar can reform in disks that have been replenished in gas .
the large frequency of gaseous warps in the outskirts of galaxies require also continuous gas accretion , with misaligned angular momentum . when the accreted gas has perpendicular angular momentum , a polar ring can be formed .
asymmetric gas accretion could be responsible for frequent lopsided galaxies
. in rich environments , the evolution is occuring at a much quicker pace .
galaxy interactions and mergers are much more frequent , and happen earlier in the hubble time .
the star formation activity triggered by the mergers must have occurred early in dense groups that coalesce then in galaxy clusters .
most of the star formation is then quenched after redshift @xmath29 1 .
the only activity remaining comes from the infall of field spiral galaxies in the outer parts of the clusters .
secular evolution through external gas accretion is slowed down or halted , since galaxies are stripped from their gas , at large galactocentric radii .
most of the gas has been shock - heated at the virial temperature of the clusters , and is observed in x - rays .
the mass fraction in hot gas reaches the baryonic fraction in the outer parts ( e.g. sadat & blanchard 2001 , fig [ sadat01 ] , valageas et al .
contrary to the field , it appears that hierarchical merging has dominated secular evolution in rich galaxy clusters .
andersen v. : 1996 aj 111 , 1805 baldwin j. , lynden - bell d. , sancisi r. : 1980 , mnras 193 , 313 balogh , m. , eke , v. , miller , c. et al . : 2004 , mnras 348 , 1355 balogh , m.l . ,
morris , s.l . , yee , h. k. c. et al : 1999 , apj 527 , 54 bekki , k. 1997 , apj , 490l , 37 bekki , k. 1998 , apj , 499 , 635 bekki , k. , couch w. , shioya y. : 2002 , apj 577 , 651 berentzen , i. , athanassoula , e. , heller , c. h. , fricke , k. j. : 2004 mnras 347 , 220 binney j. : 1992 ara&a 30 , 51 blitz l. , spergel d. , teuben p.
et al . : 1999 , apj 514 , 818 block d. , bournaud f. , combes f. , puerari i. , buta r. : 2002 , a&a 394 , l35 bournaud f. , combes f. : 2004 , a&a letter , in press bournaud f. , combes f. : 2003 , a&a 401 , 817 bournaud f. , combes f. : 2002 , a&a 392 , 83 briggs f.h . :
1990 apj 352 , 15 brocca , c. , bettoni , d. , galletta , g. : 1997 a&a 326 , 907 bullock , j. s. , dekel , a. , kolatt , t. s. et al . : 2001 , apj 555 , 240 buta r. , laurikainen e. , salo h. : 2004 , aj 127 , 279 butcher , h. , oemler , a. : 1978 , apj 219 , 18 butcher , h. , oemler , a. : 1984 , apj 285 , 426 casuso e. , beckman j.e .
: 2001 , apj 557 , 681 combes f. , jog c. , bournaud f. , puerari i. : 2004 , a&a in prep combes f. : 2003 , iau symp .
217 , recycling intergalactic and interstellar matter , asp conf series , ed .
p - a . duc et al . ( astro - ph/0308293 )
couch w.j . , balogh , m.l .
, bower , r.g .
et al . : 2001 , apj 549 , 820 diaferio a. , kauffmann g. , balogh m.l .
: 2001 , mnras 323 , 999 dressler , a. : 1980 apj 236 , 351 dressler , a. , oemler a. , couch w.j .
et al . : 1997 , apj 490 , 577 dressler , a. , smail , i. , poggianti , b. et al . : 1999 apjs 122 , 51 elmegreen d.m .
, elmegreen b.g .
, frogel j.a .
et al : 2002 aj 124 , 777 eskridge , p. b. , frogel , j. a. , pogge , r. w. , et al . : 2000 , aj 119 , 536 gerin , m. , combes , f. , athanassoula , e. : 1990 a&a 230 , 37 gnedin , o. : 2003 , apj 582 , 141 and apj 589 , 752 gmez , p. , nichol , r. , miller , c. et al . : 2003 , apj 584 , 210 goto , t. , yamauchi , c. , fujita , y. et al . : 2003 , mnras , 346 , 601 gottloeber , s. , klypin , a. , kravtsov , a. v. : 2001 , apj 546 , 223 hasan , h. , & norman , c.a . : 1990 , apj 361 , 69 hasan , h. , pfenniger , d. , norman , c. : 1993 , apj 409 , 91 haywood m. , robin a.c . , creze m. : 1997 , a&a 320 , 428 & 440 heavens , a. , panter , b. , jimenez , r. , dunlop , j. : 2004 , nature , 428 , 625 hozumi s. , hernquist l. : 1999 , in `` galaxy dynamics '' , asp conf series , ed .
d. r. merritt , m. valluri , and j. a. sellwood , vol 182 , p.259 jiang i - g . , binney j. : 1999 mnras 303 , l7 jimenez r. , panter b. , heavens a. , verde l. : 2004 , mnras , preprint kennicutt , r. c. , tamblyn , p. , congdon , c. e. : 1994 , apj 435 , 22 kennicutt , r. c. : 1983 , apj 272 , 54 kormendy j. , kennicutt , r. c. : 2004 , araa , in press kornreich , d. a. , haynes , m. p. , lovelace , r. v. e. : 1998 , aj 116 , 2154 lavery r.j . , henry j.p . : 1988 , apj 330 , 596 lewis , i. , balogh , m. , de propris , r. et al . 2002 : mnras 334 , 673 lopez - corredoira , m. , beckman , j. e. , casuso , e. : 1999 a&a 351 , 920 lopez - corredoira , m. , betancort - rijo , j. , beckman , j. e. : 2002 , a&a 386 , 169 matthews , l. d. , van driel , w. , gallagher , j. s. : 1998 , aj 116 , 2196 mihos j.c . , hernquist l. : 1996 , apj 464 , 641 miwa , t. , noguchi , m. : 1998 apj 499 , 149 moore , b. : 2003 , in clusters of galaxies : probes of cosmological structure and galaxy evolution , carnegie observatories symposium iii .
( astro - ph/0306596 ) nagamine k. , springel v. , hernquist l. , machacek m. : 2004 , mnras 350 , 385 oemler , a. , dressler , a. , butcher , h. : 1997 apj 474 , 561 poggianti , b. , bridges t.j . ,
carter , d. et al . : 2001 apj 563 , 118 poggianti , b. , smail , i. , dressler , a. et al . : 1999 apj 518 , 576 rana , n. c. , wilkinson , d. a. : 1986 mnras , 218 , 497 reshetnikov v. , combes f. : 1998 a&a 337 , 9 reshetnikov , v. , sotnikova , n. 1997 , a&a , 325 , 933 richter o. , sancisi r. : 1994 , a&a 290 , 9 rix , h - w .
, zaritsky d. : 1995 , apj 447 , 82 rocha - pinto , h. j. , scalo , j. , maciel , w. j. , flynn , c. : 2000 , apj 531 , l115 rocha - pinto h.j . , maciel w.j . : 1996 , mnras 279 , 447 sadat r. , blanchard a. : 2001 a&a 371 , 19 schweizer , f. , whitmore , b. c. , rubin , v. c. 1983 , aj , 88 , 909 semelin b. , combes f. : 2004 , a&a in prep .
shioya y. , bekki k. , couch w. : 2004 , apj 601 , 654 solanes , j. , manrique , a. , garca - gmez , c. et al . : 2001 , apj 548 , 97 somerville r.s .
, primack j.r . : 1999 , mnras 310 , 1087 somerville r.s .
, primack j.r . : 1998 , proceedings of the xth rencontre de blois storrie - lombardi , l.j . ,
mcmahon , r.g .
, irwin m.j . : 1996 , mnras , 283 , l79 taniguchi , y. , shibata , k. , wakamatsu , k .-
1986 ap&ss 118 , 529 tissera p. : 2000
, apj 534 , 636 thompson l.a . : 1981 apj 244 , l43 toth g. , ostriker j.p . : 1992 , apj 389 , 5 valageas , p. , schaeffer , r. , silk , j. : 2002 a&a 388 , 741 van den bergh , s. : 2002 , aj 124 , 782 van den bergh , s. : 1976 apj 206 , 883 wakker b.p . , howk j.c .
, savage b.d .
et al . : 1999 nature , 402 , 388 walker , i. r. , mihos , j. c. ; hernquist , l. : 1996 , apj 460 , 121 wechsler , r. h. , bullock , j. s. , primack , et al . : 2002 , apj 568 , 52 wilcots , e. m. , prescott , m. k. m. : 2004 , aj 127 , 1900 whitmore , b. c. , lucas , r. a. 1990 , aj , 100 , 1489 whyte l.f .
, abraham r.g . , merrifield m.r.et al . : 2002 , mnras 336 , 1281 worthey g. , espana a.l . : 2004 , in _ origin and evolution of the elements _ , carnegie observatories astrophysics series , a. mcwilliam and m. rauch ed .
zaritsky d. , rix , h - w .
: 1997 , apj 477 , 118 zwaan m.a .
, briggs f.h . , sprayberry d. , sorar e. : 1997 , apj 490 , 173 | in the current galaxy formation scenarios , two physical phenomena are invoked to build disk galaxies : hierarchical mergers and more quiescent external gas accretion , coming from intergalactic filaments .
although both are thought to play a role , their relative importance is not known precisely .
here we consider the constraints on these scenarios brought by the observation - deduced star formation history on the one hand , and observed dynamics of galaxies on the other hand : the high frequency of bars and spirals , the high frequency of perturbations such as lopsidedness , warps , or polar rings .
all these observations are not easily reproduced in simulations without important gas accretion .
n - body simulations taking into account the mass exchange between stars and gas through star formation and feedback , can reproduce the data , only if galaxies double their mass in about 10 gyr through gas accretion .
warped and polar ring systems are good tracers of this accretion , which occurs from cold gas which has not been virialised in the system s potential .
the relative importance of these phenomena are compared between the field and rich clusters .
the respective role of mergers and gas accretion vary considerably with environment .
galaxy evolution
hubble sequence
accretion
star formation
interactions |
the broad - band infrared ( ir ) observations are well suited for the study of the evolution of snrs . besides the general advantage of suffering less extinction effects , there are diverse important phenomena related to supernova remnants ( snrs ) which are bright in the ir wavebands . in the near- and mid - ir regimes , for example , one can study the distributions of dense ejecta from core - collapse supernova explosions ( e.g. , * ? ? ? * ; * ? ? ?
* ) , circumstellar material produced by mass loss of the progenitors ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , shock interactions between snrs and nearby molecular clouds ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and hot dust synthesized in ejecta ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , along with the emission from the central compact objects ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . the far - ir observations of snrs , on the other hand , are relatively rare but very efficient to investigate cold dust which often contributes to a significant portion of the total mass of the dust in snrs .
first , it is studied that the swept - up dust in snrs is responsible for the most of cooling of snrs and plays a critical role in the evolutions of snrs in dense medium @xcite .
secondly , the far - ir observations of snrs appear to provide better means of studying the dust formed during the process of supernova explosion ( e.g. , * ? ? ?
* ; * ? ? ?
finally , if an snr is interacting with a nearby molecular cloud , its far - ir emission is expected to be significantly enhanced by the swept - up , inter - clump medium in the cloud or by the evaporation of dense clumps ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , although the reality of these processes is yet to be confirmed observationally .
kes 17 , known as [email protected] , is such an object in which broad - band ir observations can play an important role in our understanding of snrs . until recently this snr has only been studied in radio continuum and oh line observations @xcite , revealing broken radio shell structures in the west and south .
the recent @xmath2 observations , however , detected very bright ir emission in the 38 @xmath0 m range @xcite , which makes it in fact one of the brightest snrs in the wavelength range .
the follow - up spectroscopic observations in the 530 @xmath0 m range also detected many strong emission lines @xcite .
very recently , the xmm - newton x - ray observations detected centrally filled thermal x - ray emission inside the non - thermal radio shell , listing it as a member of the class of mixed - morphology snrs @xcite .
all these indicate that kes 17 is interacting with nearby molecular gas and that more ir observations are needed to better understand its nature . in this paper , we present extensive broad - band ir observations of kes 17 using the @xmath1 satellite , together with new analyses of archival @xmath2 and radio continuum data .
for the distance to kes 17 , only a lower limit of 8 kpc , corrected for the distance between the sun and the galactic center of 8.5 kpc , is known from the observations of line @xcite .
we adopt a distance of 8 @xmath9 kpc where @xmath9 is a scaling factor in this paper .
the @xmath1 imaging observations of kes 17 were carried out covering the mid - ir ( 1327 @xmath0 m ) and far - ir ( 50180 @xmath0 m ) bands on 2007 february 5 and 6 , respectively .
the journal of the @xmath1 observations is given in table [ tab_obssum ] .
the mid - ir observations were made by the infrared camera ( irc ) equipped with an si : as detector array of 256 @xmath5 256 pixels which produced two images of 10 @xmath5 10 centered on 15 @xmath0 m ( irc l15 ) and 24 @xmath0 m ( irc l24 ) @xcite .
the total on - source integration time was 196 s for both images .
the basic calibration and data handling such as dark subtraction , linearity fitting , distortion correction , flat fielding , image combination and astrometric measurement were performed by irc imaging data reduction pipeline version 20110225 .
the far - ir observations , on the other hand , were made by the far - infrared surveyor ( fis ) in two round - trip scans in the cross - scan shift mode @xcite .
the scan speed and shift length were 15 s@xmath10 and 240 , respectively , and the resulting imaging size was 40 @xmath5 12 elongated in the scan direction .
all the four fis band ( n60 , wide - s , wide - l , and n160 ; see table [ tab_obsimg ] for the details of the bands ) images were obtained simultaneously in a single observing run using ge : ga ( 20 @xmath5 2 pixels for n60 ; 20 @xmath5 3 pixels for wide - s ) and stressed ge : ga ( 15 @xmath5 3 pixels for wide - l ; 15 @xmath5 2 pixels for n160 ) detector arrays . the initial calibration and data handling such as glitch detection , dark subtraction , flat fielding and flux calibration were performed by fis slow - scan toolkit version 20070914 , followed by advanced image construction based on the refined sampling mechanism .
a summary of imaging bands used in this paper , including six @xmath1 bands ( irc l15 , irc l24 , fis n60 , fis wide - s , fis wide - l , and fis n160 ) and three @xmath2 bands ( irac 4.5 , irac 8 , and mips 24 ) , is given in table [ tab_obsimg ] .
the @xmath1 spectroscopic observations were carried out in the near - ir slit mode on 2008 august 8 and 2009 february 46 ( table [ tab_obssum ] ) , which produced grism spectra of r @xmath11 120 spectral resolving power in the 2.55.0 @xmath0 m wavelength range @xcite .
spectra of the two bright peaks and the narrow filament in the western shell of kes 17 previously identified in @xmath2 irac images @xcite were obtained .
the slit positions and their coordinates are presented in figure [ fig_ir ] and in table [ tab_slit ] , respectively .
the 5 @xmath5 0.8 slit was used for the target observations , whereas background spectra , which were subtracted from the target spectra later , were simultaneously obtained using the 3 @xmath5 1 slit positioned 1 apart from the targets .
the data calibration and handing were performed by irc spectroscopy toolkit version 20081015 .
previous @xmath2 studies detected bright ir emission of kes 17 in the irac bands between 3.6 and 8.0 @xmath0 m bands @xcite .
we used the @xmath2 irac images of 4.5 and 8.0 @xmath0 m bands in this paper .
we analyzed the @xmath2 archival data of mips 24 @xmath0 m band @xcite to make an image of @xmath12 size of kes 17 .
for the 24 @xmath0 m band image , we used the @xmath2 data in this paper , unless explicitly mentioned otherwise .
in addition , we obtained the @xmath2 irs 838 @xmath0 m and mips sed 5297 @xmath0 m spectra available from the @xmath2 data archive to estimate contributions of line emission in the wavelength ranges .
we also obtained a high - resolution ( 9.17 @xmath5 7.36 ) radio - continuum image of kes 17 using the australia telescope compact array ( @xmath13 ) archival data provided by the australia telescope online archive .. we synthesized the radio - continuum image of kes 17 from the 12-h exposure 1.4 ghz data obtained with the 1.5a array configuration on 2004 march 24 after calibration of the data with 1329@xmath14665 .
figure [ fig_ir ] presents the @xmath1 and @xmath2 near- to far - ir ( 4.5160 @xmath0 m ) band images of kes 17 , together with the @xmath13 20 cm radio - continuum image .
the most conspicuous feature in the ir images is the bright emission of @xmath3 5 in the western shell apparent in all of the ir bands .
the emission appears to be clumpy and filamentary in the near- and mid - ir bands up to 24 @xmath0 m , whereas it becomes unresolved in the longer ( 65160 @xmath0 m ) wavelength images due to increasingly large beam sizes .
this ir emission in the western shell partly overlaps that of the radio continuum emission .
in addition to the bright western shell , there is relatively weak mid - ir ( 15 and 24 @xmath0 m ) emission in the southern shell overlapping the radio continuum emission in this region .
there exists noticeable background emission in the direction of kes 17 in all the images longer than 4.5 @xmath0 m ( figure [ fig_ir ] ) , especially in the northeast . in order to identify the southern shell emission in the 15 and 24 @xmath0 m images more clearly
, we subtract the background emission from both the images as follows : first , given that both the 4.5 and 8 @xmath0 m emission of kes 17 is due to the same origin ( i.e. , emission from shocked ; see 4 for the details ) in the western shell , we obtain a linear correlation with a correlation coefficient @xmath15 @xmath11 0.86 between the 4.5 and 8 @xmath0 m emission of the shell . based on the correlation , we scale the 4.5 @xmath0 m emission to 8 @xmath0 m and subtract the scaled emission from the observed 8 @xmath0 m emission to create a background image at 8 @xmath0 m .
next , we obtain a linear correlation of @xmath15 @xmath11 0.89 between the background 8 @xmath0 m emission and the 24 @xmath0 m emission outside the boundaries of kes 17 , and then scale the 8 @xmath0 m emission to 24 @xmath0 m using the correlation .
we subtract the scaled emission from the observed emission at 24 @xmath0 m , which finally gives the background - removed 24 @xmath0 m image of kes 17 ( figure [ fig_mid ] ) .
the peak surface brightnesses of the western shell after the background subtraction are 6.1 @xmath16 1.5 and 6.4 @xmath16 1.5 mjy sr@xmath10 for 15 and 24 @xmath0 m bands , respectively .
we apply the same method to obtain the background - removed @xmath1 15 and 24 @xmath0 m images . in all these processes , we convolved the images of different beam sizes with that of the 24 @xmath0 m images . in figure [ fig_mid ]
we can now easily identify the existence of the southern shell at 15 and 24 @xmath0 m which overlaps that of the radio continuum .
for far - ir images of 65 , 90 , 140 , and 160 @xmath0 m , we estimate the background emission by taking the median value in the region between 4@xmath17 and 6@xmath17 radius from the center of kes 17 .
after background subtraction , the peak surface brightnesses of the western shell are 108 @xmath16 24 , 135 @xmath16 45 , 173 @xmath16 84 , and 117 @xmath16 55 mjy sr@xmath10 for 65 , 90 , 140 and 160 @xmath0 m bands , respectively .
table [ tab_flux ] presents the measured ( or upper limits ) ir fluxes of the western and southern shells between 15 and 160 @xmath0 m , together with the estimation of contributions of line emission in the bands .
the errors in the flux measurements are mainly caused by the uncertainties in the background measurements and also by those associated with the calibrations of the @xmath1 and @xmath2 observations .
we estimate the line contributions to the observed fluxes of the 15 and 24 @xmath0 m bands using the @xmath2 irs 838 @xmath0 m spectra .
we construct the irs spectra by subtracting the background obtained outside the shell structure and integrate them over the bandpass .
the estimated line contributions are 75 @xmath16 21 ( 15 @xmath0 m ) and 38 @xmath16 10 % ( 24 @xmath0 m for both @xmath1 and @xmath2 ) , respectively .
the line emission in these bands is dominated by and ionic lines .
we also estimate the contribution of line emission of the 60 and 90 @xmath0 m bands to be 16 @xmath16 8 and 3 @xmath16 2 % , respectively , using @xmath2 mips sed 5297 @xmath0 m spectra .
the line contributions in these bands are mostly due to [ ] at 63 @xmath0 m . for the 140 and 160 @xmath0 m bands , we expect the line contributions to be less than 5 % by scaling the flux of the [ ] 63 @xmath0 m line that we obtain above to those of [ ] 145 @xmath0 m and [ ] 153 @xmath0 m lines in a shocked region @xcite . figure [ fig_sed ] shows the flux distribution of the western shell excluding the line contributions .
the ir synchrotron flux expected from an extrapolation of the radio fluxes applying synchrotron power - law index of @xmath140.54 @xcite is less than a few percents of the measured ir values , supporting that the observed continuum is dominated by dust emission . in order to calculate temperatures of the dusts responsible for the mid- and far - ir emission of kes 17
, we perform the spectral energy distribution ( sed ) fits of the flux distribution ( figure [ fig_sed ] ) with two modified blackbody components . for the emissivity of the dusts in the sed fits ,
we adopt a mixture of carbonaceous and silicate interstellar grains @xcite . as a result
, we obtain acceptable fit ( reduced chi - square @xmath182 ) with dust temperature components of 79 @xmath16 6 k and 27 @xmath16 3 k ( figure [ fig_sed ] ) .
the mass of the hot ( 79 k ) dust is @xmath19 @xmath20 @xmath7 and that of the cold ( 27 k ) dust is 6.7 @xmath16 4.0 @xmath20 @xmath7 .
the luminosities of the dust emission are ( 9.5 @xmath16 6.1 ) @xmath5 10@xmath21 @xmath20 @xmath22 and ( 8.1 @xmath16 5.0 ) @xmath5 10@xmath23 @xmath20 @xmath22 for the hot and cold components , respectively .
it is worthwhile to note that the hot dust temperature could have been overestimated if there is any contribution from stochastically heated small grains to the 15 and 24 @xmath0 m continuum fluxes @xcite .
figure [ fig_spec ] shows near - ir ( 2.55.0 @xmath0 m ) @xmath1 spectra of five points ( a , b1 , b2 , c1 , c2 in figure 1 ) in the western shell of kes 17 obtained after subtraction of the background spectrum .
( see table 4 for their coordinates . )
we detect several lines in the spectra including the pure rotational transitions of 0 - 0 s(914 ) and ro - vibrational transitions of 1 - 0 o(3 , 5 , 6 ) .
table [ tab_line ] lists the observed intensities of the line emission .
figure [ fig_lpop ] presents the excitation diagram of the pure rotational and ro - vibrational lines of b2 between the upper energy level of 6,000 k and 20,000 k. for this we correct the extinction effect using the h column density of 3.6@xmath5 10@xmath25 @xmath26 estimated in previous x - ray observations of kes 17 and the galactic extinction curve @xcite .
the linear correlation between the upper energy level and the upper column density represented by a straight line in figure [ fig_lpop ] indicates the level population of the local thermodynamic equilibrium at the excitation temperature of 2200 @xmath16 400 k. the spectra of other positions , which have smaller number of detected lines , show similar excitation temperatures to that of b2 .
the ir emission of kes 17 shows the existence of both the western and southern shell structures .
the former is visible in all the ir bands , while the latter is visible only in the mid - ir bands of 15 and 24 @xmath0 m .
the near - ir emission of the western shell is dominated by line emission ( figure [ fig_spec ] ) , which suggests that kes 17 is interacting with molecular gas in this region , consistent with the results of previous observations ( e.g. , * ? ? ?
the distribution of the mid - ir emission overlaps that of the radio continuum emission in the western and southern shells ( figure [ fig_mid ] ) , supporting the interpretation that it represents swept - up dust at the boundary of the snr . on the other hand ,
the far - ir emission is similar to the near - ir emission with only the western shell being readily identifiable .
the coexistence of the far - ir emission from the cold dust and the near - ir line emission form shocked molecular gas in the western shell suggests that the bright far - ir emission of kes 17 in this region is also related to the interaction of the source with a nearby molecular cloud .
the far - ir luminosity of kes 17 ( @xmath3 8100 @xmath20 @xmath22 ; 3 ) makes it one of the most luminous snrs in the far - ir wavebands and the second ( after cassiopeia a ) galactic snr detected up to 160 @xmath0 m in the ir wavebands @xcite . for comparison ,
only eight snrs were detected by iras at 100 @xmath0 m with far - ir luminosity larger than kes 17 @xcite , five of them interacting with molecular clouds ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
our observations also strongly suggest that the far - ir emission of kes 17 is originated from the dust in a molecular cloud interacting with the snr .
however , molecular shocks producing strong lines in a dense molecular cloud can not directly generate bright far - ir dust continuum emission .
this is because , in the non - dissociative molecular shock with shock speeds of @xmath18 50 km s@xmath10 , the gas temperature does not rise above a few thousand degrees @xcite .
this is inadequate to heat the dust grains sufficiently high enough to emit bright far - ir continuum emission . instead ,
if a molecular cloud is clumpy , composed of clumps and inter - clump medium , then the far - ir emission may be produced by faster shocks propagating in the inter - clump medium of lower densities .
even though the density of inter - clump medium is lower than that of the h@xmath24 emitting clumps , it can still be high enough to contain large amount of dusts and reduce shock speeds significantly . in the
following we consider these possibilities .
when snr shocks propagate into a clumpy molecular cloud , hot gas in swept - up inter - clump medium behind shock front collisionally heats dusts to produce the ir emission .
the estimated dust mass of the western shell of kes 17 is @xmath27 m@xmath28 .
if the cloud initially occupies the entire western region with normal dust - to - gas ratio of 1 % , the pre - shock inter - clump density is @xmath29 @xmath30 . assuming that kes 17 is in a sedov phase with radius of @xmath3 7 @xmath9 pc , the shock velocity in the inter - clump medium is @xmath31 , where @xmath32 is the sn explosion energy in the unit of 10@xmath33 erg and @xmath34 is the density of the inter - clump medium . in this swept - up inter - clump gas ,
the ir surface brightness of the dust by strong shocks is @xcite @xmath35 around 90 @xmath0 m , which , for the obtained inter - clump density and velocity of @xmath3 40 @xmath30 and @xmath3 200 km s@xmath10 , gives comparable surface brightness to the observed value .
note that the calculation by @xcite expects the peak of dust emission power ( @xmath36 ) at the mid - ir wavelength ( @xmath37 m ) , while our observations indicate a somewhat flat @xmath36 peak around the 6590 @xmath0 m bands .
this is suggestive that there may be a significantly increased amount of large dusts contributing effectively in the far - ir regime if the shocked inter - clump medium is responsible for the observed emission of kes 17 .
alternatively , the bright far - ir emission of kes 17 may be the emission of dusts injected from an evaporating cloud @xcite .
if so , the expected ir luminosity of an evaporating cloud in the hot gas of an snr is @xcite @xmath38 where @xmath39 is the radius of the cloud and @xmath40 and @xmath41 are the density and temperature of the hot gas , respectively .
the x - ray emitting gas was recently detected inside the radio shell of kes 17 @xcite , suggesting that the evaporation should occur at the surface of an ir emitting cloud where it is in contact with the hot interior gas .
the temperature and density of the hot x - ray gas are @xmath42 k and @xmath43 @xmath30 , respectively @xcite .
we expect the ir luminosity of an evaporating cloud to be @xmath3 2,000 @xmath44 @xmath22 using the cloud radius of 3 @xmath9 pc ( @xmath45 ) at 90 @xmath0 m ( figure [ fig_ir ] ) .
this is somewhat smaller than the observed far - ir luminosity ( @xmath3 8000 @xmath22 ,
3 ) of kes 17 , indicating that the dust emission from an evaporating cloud may contribute only a small amount of the far - ir emission of kes 17 , although we can not rule out the possibility completely given the various uncertainties involved in the measurements .
another explanation of the bright far - ir emission of kes 17 may be that the dusts in a pre - shock molecular cloud heated by strong radiation of the snr shocks in a radiative phase emit the observed emission .
the shocks propagating into a molecular cloud at the western shell of kes 17 can easily become radiative for the pre - shock density of @xmath46 100 @xmath30 if gas cooling is efficient . taking a power - law temperature dependence of the cooling efficiencies ( @xmath47 ;
* ) , @xcite calculated a radiative shell forming at the radius @xmath48 with a shell velocity @xmath49 for a pre - shock density @xmath50 and a radiative shell forming time @xmath51 = 10,000 yrs .
this indicates that kes 17 likely has a radiative shell , if the pre - shock density is @xmath52 20 @xmath30 . using the shell forming values and a cloud radius of @xmath39 = 3 @xmath9 pc ,
the luminosity of a radiative shock at the western shell of kes 17 is expected to be @xcite @xmath53 which appears to be adequate to explain the observed luminosity of the western shell ( 3 ) . in this case
, we expect the peak of the dust emission power to be at the far - ir ( @xmath54 60 @xmath0 m ) region for the dusts of mixture of carbonaceous and silicate grains @xcite , consistent with the results of this study .
this can also explain the absence of the x - ray emission in the western shell of kes 17 as the lack of x - ray emitting high temperature gas in the radiative phase @xcite .
our near - ir ( 2.55.0 @xmath0 m ) spectra of the western shell of kes 17 indicate the excitation temperature of 2200 @xmath16 400 k for gas in this region ( figure [ fig_spec ] ) .
this is different from the previous results based on the @xmath2 observations where a mixture of two components of gas of @xmath3 300 k and @xmath3 1200 k was proposed to explain the observed intensity distribution of lines of the western shell in the 528 @xmath0 m range @xcite .
the discrepancy in the excitation temperature suggests that the gas in the western shell is in a thermal admixture as found in other gas interacting with snrs @xcite .
figure [ fig_lcal ] shows the excitation diagram of lines detected in both the @xmath2 and @xmath1 observations , covering the upper energy level of 100017,000 k. ( note that while the @xmath2 results represent the levels of @xmath46 7,000 k , the @xmath1 results do the levels of higher temperatures . )
we fit the observed excitation diagram with a thermal admixture model where the column density of gas is related to a power of its temperature of 1004,000 k : @xmath55 @xcite .
the ortho - to - para ratio is fixed to be 3 in the fit .
the best fit gives 3.0 @xmath16 0.1 , ( @xmath56 @xmath26 , and ( @xmath57 @xmath30 for the power index , column density , and number density , respectively . the solid curve in figure [ fig_lcal ] represents the best - fit model to the observed data , where we can identify that the model prediction matches the observed values well .
this column density is twice that of the two temperature model of the @xmath2 observations , while the number density is between those obtained for the warm and hot components of the @xmath2 results @xcite .
the obtained power index of 3.0 @xmath16 0.1 is comparable to what obtained from other sources @xcite , and is not far from 3.8 obtained for bow shocks @xcite .
if kes 17 is indeed interacting with clumpy molecular gas ( see 5.1 ) , it can be easy to develop a group of shocks around dense clumps with various shock velocities , which in turn naturally leads to an admixture of gas of different temperatures .
we show in this paper that kes 17 is one of the most luminous snrs in the ir wavebands , reaching the far - ir luminosity of @xmath3 8100 @xmath22 .
its ir emission is concentrated on the western and southern shell structures .
the western shell is bright in the mid- and far - ir continuum and the near - ir line emission , whereas the southern shell is visible only in the mid - ir continuum emission .
the far - ir continuum and near - ir line emission of the western shell is related to its interaction with nearby molecular gas .
it is apparent that there exists dense shocked molecular gas , excited to @xmath3 2,000 k , in this region that produces the observed line emission .
if the molecular gas is clumpy , the observed far - ir emission can be produced by the dust associated with inter - clump molecular gas of lower densities .
the far - ir emission may originate from the dust emission of the swept - up inter - clump molecular gas , the dust emission from an evaporating cloud , or the dust emission exposed to strong radiative shocks .
the mid - ir emission of kes 17 is bright in both the western and southern shells and overlaps radio continuum emission which forms a partly - broken circular shell structure .
this suggests that the mid - ir and radio continuum emission delineates the boundary of kes 17 made by the swept - up material .
the sed distribution of the ir emission of kes 17 indicates the existence of two dust components : hot ( 79 @xmath16 6 k ) component of @xmath19 @xmath20 @xmath7 and cold ( 27 @xmath16 3 k ) component of 6.7 @xmath16 4.0 @xmath20 @xmath7 .
the former is responsible for most of the mid - ir continuum emission , while the latter is for the far - ir emission .
the observed flux distribution of the lines , on the other hand , can be explained with a thermal admixture of the shocked molecular gas of different temperatures in the 1004,000 k range .
this thermal admixture of the shocked molecular gas may be a natural consequence of the interactions between kes 17 and clumpy molecular gas in which various shock velocities can easily produce thermal admixture of the molecular gas .
overall , it appears that combined far - ir continuum and near - ir spectroscopic observations are very useful to studying snrs interacting with molecular gas , as was the case for kes 17 in this paper , providing information for both the gas and dust components associated with the shock interactions .
this work is based on observations with @xmath1 , a jaxa project with the participation of esa .
this work is based in part on observations made with the @xmath2 space telescope , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
the australia telescope compact array is part of the australia telescope which is funded by the commonwealth of australia for operation as a national facility managed by csiro .
we thank all the members of the @xmath1 project .
we also thank an anonymous referee for constructive comments .
was supported by the early research award program ( era07 - 03 - 270 ) to d .- s.m . from ministry of research and innovation of the ontario provincial government .
s.m . acknowledges the support by nserc through discovery program 327277 .
c.k . was supported by the national research foundation of korea ( nrf ) grant ( nrf-2010 - 616-c00020 ) .
allen , m. g. , groves , b. a. , dopita , m. a. , sutherland , r. s. , & kewley , l. j. 2008 , , 178 , 20 arendt , r. g. , dwek , e. , & moseley , s. h. 1999 , , 521 , 234 barlow , m. j. et al .
2010 , , 518 , 138 burton , m. g. , brand , p. w. j. l. , geballe , t. r. , & webster , a. s. 1988 , , 235 , 161 carey , s. j. , et al .
2009 , , 121 , 76 caswell , j. l. , murray , j. d. , roger , r. s. , cole , d. j. , & cooke , d. j. 1975 , , 45 , 239 combi , j. a. et al .
2010 , , 523 , 76 cox , d. p. , shelton , r. l. , maciejewski , w. , smith , r. k. , plewa , t. , pawl , a. , ryczka , m. 1999 , , 524 , 179 draine , b. t. , 1981 , , 245 , 880 draine , b. t. , 2003 , , 41 , 241 draine , b. t. , 2011 , physics of the interstellar and intergalactic medium , ( princeton , nj ; princeton university press ) dwek , e. , 1981 , , 430 , 433 dwek , e. , 1986 , , 302 , 363 dwek , e. , petre , r. , szymkowiak , a. , & rice , w. l. 1987 , , 320 , l27
dwek , e. , et al .
2008 , , 676 , 1029 fazio , g. g. , et al .
2004 , , 154 , 10 frail , d. a. , goss , w. m. , reynoso , e. m. , giacani , e. b. , green , a. j. , & otrupcek , r. 1996 , , 111 , 1651 hewitt , john w. , rho , j. , andersen , m. , & reach , w. t. 2009 , , 694 , 1266 hollenbach , d. , & mckee , c. f. 1989 , , 342 , 306 kahn , f. d. 1976 , , 50 , 145 kaplan , d. l. , & moon , d .- s .
2006 , , 644 , 1056 kawada , m. , baba , h. , barthel , p. d. et al .
2007 , pasj , 59 , 389 keohane , j. w. , reach , w. t. , rho , j. , & jarrett , t. h. 2007 , , 654 , 938 koo , b .-
c . , & moon , d .- s .
1997 , , 485 , 263 koo , b .- c . , moon , d .- s . , lee , h .-
j . , & matthews , k. 2007 , , 657 , 308 lee , h .- g .
2005 , jkas , 38 , 385 lee , h .-
moon , d .- s . , koo , b .- c . , lee , j .-
& matthews , k. 2009 , , 691 , 1042 lee , h .-
c . , moon , d .- s . ,
sakon , i. , onaka , t. , jeong , w .- s . ,
kaneda , h. , nozawa , t. , & kozasa , t. 2009 , , 706 , 441 milne , d. k. , & dickel , j. r. 1975 , aujph , 28 , 209 moon , d .- s . ,
lee , j .- j . , eikenberry , s. s. , koo , b .- c . , chatterjee , s. , kaplan , d. l. , hester , j. j. , cordes , j. m. , gallant , y. a. , koch - miramond , l. 2004 , , 610 , 33 moon , d .- s . , koo , b .- c . , lee , h .- g . , matthews , k. , lee , j .-
j . , pyo , t .- s . ,
seok , j. y. , & hayashi , m. 2009 , , 703 , 81 neufeld , d. a. , hollenbach , d. j. , kaufman , m. j. , snell , r. l. , melnick , g. j. , bergin , e. a. , & sonnentrucker , p. 2007
, , 664 , 890 neufeld , d. a. & yuan , y. 2008 , , 678 , 974 neufeld , d. a. nisini , b , giannini , t , melnick , g. j. , bergin , e. a. , yuan , y , maret , s , tolls , v , gsten , r , & kaufman , m. j. 2009 , , 706 , 170 oliva , e. , moorwood , a. f. m. , & danziger , i. j. 1990 , , 240 , 453 onaka , t. , matsuhara , h. , wada , t. et al .
2007 , pasj , 59 , 401 ohyama , y. , onaka , t. , matsuhara , h. et al .
2007 , pasj , 59 , 411 reach , w. t. , & rho , j. 1999 , , 511 , 836 reach , w. t. , rho , j. , & jarrett , t. h. 2005 , , 618 , 297 reach , w. t. , rho , j. , tappe , a. , pannuti , t. g. , brogan , c. l. , churchwell , e. b. , meade , m. r. , babler , b. , indebetouw , r. , & whitney , b. a. 2006 , , 131 , 1479 rho , j. , kozasa , t. , reach , w. t. , smith , j. d. , rudnick , l. , delaney , t. , ennis , j. a. , gomez , h. , & tappe , a. 2008 , , 673 , 271 rieke et al .
2004 , , 154 , 25 saken , j. m. , fesen , r. a. , & shull , j. m. 1992 , , 81 , 715 shaver , p. a. , & goss , w. m. 1970 , aujpa , 14 , 133s shinn , j .- h . , koo , b .-
c . , burton , m. g. , lee , h .-
, & moon , d .- s .
2009 , , 693 , 1883 shinn , j .- h . , koo , b .- c . , burton , m. g. , lee , h .-
g . , & moon , d .- s .
2010 , adspr , 45 , 445 sibthorpe , b. , et al . 2010 , , 719 , 1553 wang , z. , chakrabarty , d. , & kaplan , d. l. 2006 , nature , 440 , 772 whiteoak , j. b. z. , & green , a. j. 1996 , , 118 , 329 l ccc 2007 .
5 . & 1400755 & irc imaging & l15 , l24 + 2007 . 2 .
& 1400754 & fis imaging & n60 , wide - s , wide - l , n160 + 2008 .
1420785(1,2 ) , 1420783(1,2 ) , 1421783 & irc spectroscopy & ng + 2009 . 2 .
4 . & 1420784 & irc spectroscopy & ng + 2009 . 2 .
6 . & 1421784 , 1421785 & irc spectroscopy & ng + l ccccc cccc & irac 4.5 & irac 8 & irc l15 & mips 24 & irc l24 & fis n60 & fis wide - s & fis wide - l & fis n160 + reference wavelength ( @xmath0 m ) & 4.49 & 7.87 & 15.0 & 23.68 & 24.0 & 65 & 90 & 140 & 160 + effective bandwidth ( @xmath0 m ) & 1.01 & 2.93 & 5.98 & 5.3 & 5.34 & 21.7 & 37.9 & 52.4 & 34.1 + fwhm ( @xmath58 ) & 1.72 & 1.98 & 5.7 & 6 & 6.8 & 37 & 39 & 58 & 61 l cc a & ( @xmath59 , @xmath60 ) & 1 @xmath5 308 s + b1 & ( @xmath61 , @xmath62 ) & 2 @xmath5 308 s + b2 & ( @xmath61 , @xmath63 ) & 1 @xmath5 308 s + c1 & ( @xmath64 , @xmath65 ) & 1 @xmath5 308 s + c2 & ( @xmath61 , @xmath66 ) & 1 @xmath5 308 s + background & ( @xmath67 , @xmath68 ) & 1 @xmath5 308 s + lccc @xmath1 15 @xmath0 m & 2.6 @xmath16 0.5 jy & 75 @xmath16 21 % & 1.2 @xmath16 0.2 jy + @xmath2 24 @xmath0 m & 2.2 @xmath16 0.4 jy & 38 @xmath16 10 % & 1.5 @xmath16 0.2 jy + @xmath1 24 @xmath0 m & 3.1 @xmath16 0.7 jy & 39 @xmath16 11 % & 1.0 @xmath16 0.3 jy + @xmath1 65 @xmath0 m & 75 @xmath16 15 jy & 16 @xmath16 8 % & @xmath184 jy + @xmath1 90 @xmath0 m & 90 @xmath16 27 jy & 3 @xmath16 2 % & @xmath188 jy + @xmath1 140 @xmath0 m & 104 @xmath16 44 jy & @xmath185 % & @xmath1853 jy + @xmath1 160 @xmath0 m & 75 @xmath16 31 jy & @xmath185 % & @xmath1834 jy + lc ccccc h@xmath24 1 - 0 o(3 ) & 2.80 & & 11.6 ( 4.8 ) & 9.7 ( 4.3 ) & 12.4 ( 4.7 ) & 2.8 ( 2.0 ) + h@xmath24 1 - 0 o(5 ) & 3.24 & 4.3 ( 3.1 ) & 10.6 ( 6.6 ) & 16.4 ( 4.5 ) & 10.3 ( 3.1 ) & 5.7 ( 2.9 ) + h@xmath24 1 - 0 o(6 ) & 3.50 & & & 4.2 ( 2.0 ) & 7.8 ( 4.0 ) & + h@xmath24 0 - 0 s(14 ) & 3.72 & & & 2.5 ( 1.9 ) & 2.9 ( 1.7 ) & + h@xmath24 1 - 0 o(7 ) & 3.81 & & 2.6 ( 1.4 ) & 5.8 ( 2.2 ) & 6.1 ( 3.3 ) & 2.2 ( 2.3 ) + h@xmath24 0 - 0 s(13 ) & 3.85 & & 6.9 ( 2.3 ) & 8.4 ( 2.2 ) & 7.7 ( 2.4 ) & 3.5 ( 1.4 ) + h@xmath24 0 - 0 s(12 ) & 4.00 & & & 3.6 ( 1.7 ) & 1.7 ( 1.4 ) & + h@xmath24 0 - 0 s(11 ) & 4.19 & 7.1 ( 2.5 ) & 16.0 ( 2.7 ) & 18.6 ( 2.3 ) & 17.1 ( 2.1 ) & 10.7 ( 2.2 ) + h@xmath24 0 - 0 s(10 ) & 4.41 & 8.5 ( 2.6 ) & 14.0 ( 2.4 ) & 14.5 ( 2.5 ) & 15.3 ( 2.5 ) & 7.8 ( 2.1 ) + h@xmath24 0 - 0 s(9 ) & 4.69 & 13.1 ( 3.7 ) & 35.1 ( 1.9 ) & 39.5 ( 2.9 ) & 37.5 ( 1.9 ) & 18.5 ( 1.7 ) + | we present the results of infrared ( ir ; 2.5160 @xmath0 m ) observations of the supernova remnant ( snr ) kes 17 based on the data obtained with the @xmath1 and @xmath2 satellites .
we first detect bright continuum emission of its western shell in the mid- and far - ir wavebands together with its near - ir molecular line emission . we also detect hidden mid - ir emission of its southern shell after subtraction of the background emission in this region .
the far - ir luminosity of the western shell is @xmath3 8100 @xmath4 , which makes kes 17 one of the few snrs of significant far - ir emission .
the fittings of the spectral energy distribution indicate the existence of two dust components : @xmath3 79 k ( hot ) and @xmath3 27 k ( cold ) corresponding to the dust mass of @xmath3 6.2 @xmath5 10@xmath6 @xmath7 and @xmath3 6.7 @xmath7 , respectively .
we suggest that the hot component represents the dust emission of the material swept up by the snr to its western and southern boundaries , compatible with the distribution of radio continuum emission overlapping the mid - ir emission in the western and southern shells .
the existence of hot ( @xmath3 2,000 k ) , shocked dense molecular gas revealed by the near - ir molecular line emission in the western shell , on the other hand , suggests that the cold dust component represents the dust emission related to the interaction between the snr and nearby molecular gas .
the excitation conditions of the molecular gas appear to be consistent with those from shocked , clumpy admixture gas of different temperatures .
we discuss three possibilities for the origin of the bright far - ir emission of the cold dust in the western shell : the emission of dust in the inter - clump medium of shocked molecular clouds , the emission of dust in evaporating flows of molecular clouds engulfed by hot gas , and the emission of dust of nearby molecular clouds illuminated by radiative shocks . |
the intriguing possibility that our universe is only a part of a higher dimensional spacetime has raised a lot of interest in physics community recently . in string theory
extra dimensions are necessary since superstring theory requires a ten - dimensional spacetime to be consistent from the quantum point of view .
the effect of string theory on the left hand side of field equations of gravity is usually investigated by means of a low energy effective action which describes gravity at the classical level @xcite .
this effective action consists of the einstein - hilbert action plus curvature - squared terms and higher powers as well , and in general give rise to fourth order field equations and bring in ghosts .
however , if the effective action contains the higher powers of curvature in particular combinations , then only second order field equations are produced and consequently no ghosts arise @xcite .
the effective action obtained by this argument is precisely of the form proposed by lovelock @xcite . on the other hand , the dynamics of d - branes and some soliton solutions of supergravity
is governed by the born - infeld action @xcite , and therefore if one want to couple electromagnetic field with gravity it is more suitable to put the energy momentum of born - infeld electromagnetic field on the right hand side of the field equations .
while the lovelock gravity was proposed to have field equations with at most second order derivatives of the metric @xcite , the nonlinear electrodynamics proposed , by born and infeld , with the aim of obtaining a finite value for the self - energy of a point - like charge @xcite .
the lovelock gravity reduces to einstein gravity in four dimensions and also in the weak field limit , while the lagrangian of the born - infeld ( bi ) electrodynamics reduces to the maxwell lagrangian in the weak field limit .
considering the analogy between the lovelock and the born - infeld terms , which are on similar footing with regard to string corrections on the gravity side and electrodynamic side , respectively , it is plausible to include both these corrections simultaneously .
here we restrict ourself to the first three terms of lovelock gravity .
the first two terms are the einstein - hilbert term with cosmological constant , while the third term is known as the gauss - bonnet term . in recent decades
, the exact solutions of einstein - gauss - bonnet gravity and their properties were studied .
for example , static spherically symmetric black hole solutions of the gauss - bonnet gravity were found in ref .
black hole solutions with nontrivial topology in this theory were also studied in refs .
nut charged black hole solutions of gauss - bonnet gravity and gauss - bonnet - maxwell gravity were obtained in @xcite . also rotating black hole solutions of gauss - bonnet - maxwell and born - infeld - gauss - bonnet gravity
have been studied in @xcite . currently , there exist some activities in the field of wormhole physics following , particularly , the seminal works of morris , thorne and yurtsever @xcite .
morris and thorne assumed that their traversable wormholes were time independent , non - rotating , and spherically symmetric bridges between two universes .
the manifold of interest is thus a static spherically symmetric spacetime possessing two asymptotically flat regions .
these kinds of * * wormholes could be threaded both by quantum and classical matter fields that violate certain energy conditions at least at the throat known as exotic matter .
on general grounds , it has recently been shown that the amount of exotic matter needed at the wormhole throat can be made arbitrarily small thereby facilitating an easier construction of wormholes @xcite .
lorentzian wormholes in spacetimes with more than four dimensions were analyzed by several authors @xcite .
in particular , wormholes in gauss - bonnet gravity were considered in ref .
@xcite . in this paper
we are looking for the @xmath0-dimensional horizonless solutions in born - infeld - gauss - bonnet gravity .
the motivation for constructing these kinds of solutions is that they may be interpreted as wormhole . here
, we will find these kinds of horizonless solutions in the born - infeld - gauss - bonnet gravity , and use the counterterm method to compute the conserved quantities of the system .
the outline of our paper is as follows . in sec .
[ long ] we give a brief review of the field equations of born - infeld - gauss - bonnet gravity , present a new class of static wormhole solutions which produces longitudinal magnetic field , and then we consider the properties of the solutions as well as the weak energy condition . in sec .
[ angul ] we endow these spacetime with global rotations and apply the counterterm method to compute the conserved quantities of these solutions .
finally , we finish our paper with some closing remarks .
we begin this section with a brief review of einstein - gauss - bonnet gravity in the presence of nonlinear born - infeld electromagnetic field .
the field equations of this theory in the presence of a negative cosmological constant may be written as @xmath1 @xmath2 where @xmath3 is the einstein tensor and @xmath4 is the divergence - free symmetric tensor @xmath5 in eq .
( [ geq ] ) @xmath6 is the born - infeld parameter with dimension of mass , @xmath7 where @xmath8 is electromagnetic field * * tensor , and @xmath9 is the born - infeld lagrangian given as @xmath10 in the limit @xmath11 , @xmath9 reduces to the standard maxwell form @xmath12 , while @xmath13 as @xmath14 .
equation ( [ geq ] ) does not contain the derivative of the curvatures , and therefore the derivatives of the metric higher than two do not appear . here , we want to obtain the @xmath0-dimensional solutions of eqs .
( [ geq ] ) and ( [ maxeq ] ) which produce longitudinal magnetic fields in the euclidean submanifold spans by the @xmath15 coordinates ( @xmath16 ) .
we assume that the metric has the following form : @xmath17 where @xmath18 and @xmath19 is a constant will be fixed later .
note that the coordinates @xmath20 have the dimension of length , while the angular coordinates @xmath21 and @xmath22 are dimensionless as usual and range in @xmath23 $ ] .
the motivation for this metric gauge @xmath24 and @xmath25 $ ] instead of the usual schwarzschild gauge @xmath26 $ ] comes from the fact that we are looking for a horizonless magnetic solution .
the electromagnetic field equation ( [ maxeq ] ) reduces to @xmath27 + ( n-1)f_{\psi r}^{3}(r)=0 , \label{fpr eq}\]]where the prime denotes a derivative with respect to the @xmath28 coordinate .
the solution of eq .
( [ fpr eq ] ) can be written as @xmath29 where @xmath30 is an arbitrary constant and @xmath31 equation ( [ fpr ] ) shows that @xmath28 should be greater than @xmath32 in order to have a real nonlinear electromagnetic field and consequently a real spacetime . to find the function @xmath33
, one may use any components of eq .
( [ geq ] ) .
the simplest equation is the @xmath34 component of these equations which can be written as @xmath35f^{^{\prime } } + ( n-2)l^{2}r^{n-5}[(n-3)(n-4)\alpha f - r^{2}]f \nonumber \\ & & + nr^{n-1}\}+4\beta ^{2}l^{2}r^{n-1}(1-\sqrt{1-\eta } ) = 0 .
\label{rrcom}\end{aligned}\ ] ] the solution of eq .
( [ rrcom ] ) , which also satisfies all the other components of the gravitational field equations ( [ geq ] ) , can be written as @xmath36 where @xmath37 , \left [ \frac{{3}n{-4}}{{2}n{-2}}\right ] , { \eta } \right ) \nonumber\\ & & + 4(n-2)(n-3)\alpha \left [ \frac{4\beta ^{2}\left ( \sqrt{1-\eta } -1\right)}{n(n-1 ) } -\frac{1}{l^{2}}- \frac{2ml}{r^{n } } \right ] , \label{gr}\end{aligned}\ ] ] in eq .
( [ gr ] ) @xmath38 is an integration constant which is related to geometrical mass of the spacetime and @xmath39,[c],z)$ ] is the hypergeometric function which may be defined as @xcite @xmath40 , \left [ \frac{{3}n{-4}}{{2}n% { -2}}\right ] , { b u^{2n-2 } } \right)=\frac{n-2}{u^{n-2}}\int \frac{u^{n-3}}{\sqrt{1-b u^{2n-2}}}du.\ ] ] one may note that the above asymptotically ads solution reduces to einstein - born - infeld solution when @xmath41 vanishes and reduces to the solution introduced in @xcite as @xmath42 goes to infinity .
when @xmath38 and @xmath30 are zero , the vacuum solution is @xmath43 equation ( [ fg0 ] ) shows that for a positive value of @xmath41 , this parameter should satisfies @xmath44 . also note that @xmath45 for the ads solution of the theory is @xmath46^{1/2 } \left ( 1-\sqrt{1-\frac{4(n-2)(n-3)\alpha } { l^{2}}}\right ) ^{-1/2 } , \label{leff}\ ] ] which reduces to @xmath47 as @xmath41 goes to zero . to have a real spacetime
, the function @xmath48 should be positive .
this occurs provided the mass parameter @xmath49 , where @xmath50 is the value of @xmath38 calculated from the equation @xmath51 given as : @xmath52 , \left [ \frac{{3}n{-4}}{{2}n{-2}}\right ] , { 1}\right ) -\frac{4\beta ^{2}}{n(n-1)}\right\}\ ] ] in this case the metric function @xmath33 is real for @xmath53 . for the case of @xmath54 , @xmath28 should be greater than @xmath55 in order to have a real spacetime , where @xmath55 is the largest real root of @xmath56 .
figure [ fig1 ] shows the zeros of @xmath48 and @xmath57 ( @xmath55 and @xmath58 respectively ) for @xmath59 . in order to study the general structure of this solution
, we first look for curvature singularities .
it is easy to show that the kretschmann invariant @xmath60 diverges at @xmath61 , it is finite for @xmath62 and goes to zero as @xmath63
. therefore one might think that there is a curvature singularity located at @xmath64 .
two cases happen . in the first case the function @xmath33has no real root greater than @xmath55 , and
therefore we encounter with a naked singularity which we are not interested in it .
so we consider only the second case which the function has one or more real root(s ) larger than @xmath65 . in this case
the function @xmath33 is negative for @xmath66 , and positive for @xmath67 where @xmath68 is the largest real root of @xmath69 .
indeed , @xmath70 and @xmath71 are related by @xmath72 , and therefore when @xmath70 becomes negative ( which occurs for @xmath73 ) so does @xmath71 .
this leads to an apparent change of signature of the metric from @xmath74 to @xmath75 , and therefore indicates that @xmath28 should be greater than @xmath68 .
thus the coordinate @xmath28 assumes the value @xmath76
the function @xmath33 given in eq .
( [ f(r ) ] ) is positive in the whole spacetime and is zero at @xmath78 .
the kretschmann scalar is a linear combination of the square of @xmath79 , @xmath80 and @xmath81 .
since these terms do not diverge in the range @xmath82 , one finds that the kretschmann scalar is finite .
also one can show that other curvature invariants ( such as ricci scalar , ricci square , weyl square and so on ) are functions of @xmath79 , @xmath80 and @xmath81 and therefore the spacetime has no curvature singularity in the range @xmath83 . in order to avoid conic singularity at @xmath78 in the @xmath84-section
, one may fix the factor @xmath85 $ ] in the metric .
now , we investigate the wormhole interpretation of the above solution .
it satisfies the so - called flare - out condition when @xmath78 .
this can be seen by embedding the 2-surface of constant @xmath86 , @xmath21 and @xmath15 s with the metric @xmath87 in an ( unphysical ) three - dimensional euclidean flat space , which has the metric @xmath88 in cylindrical coordinates .
the surface described by the function @xmath89 satisfies @xmath90 ^{2}}>0,\end{aligned}\ ] ] which show that it has the characteristic shape of a wormhole , as illustrated in figs . 1 and 2 of ref .
next we consider the weak energy condition ( wec ) for the above solution .
using the orthonormal contravariant ( hatted ) basis vectors @xmath91 the mathematics and physical interpretations become simplified .
it is a matter of straight forward calculations to show that the components of stress - energy tensor are @xmath92 ^{1/2}-1\right ) , \nonumber \\ \text { \ \ } t_{_{\widehat{r}\widehat{r } } } & = & t_{_{\widehat{\psi } \widehat{% \psi } } } = \frac{\beta ^{2}}{4\pi } \left ( 1-\left [ 1+\left ( \frac{f_{\psi r}}{% \gamma l\beta } \right ) ^{2}\right ] ^{-1/2}\right ) , \label{emtensor}\end{aligned}\ ] ] which satisfy the weak energy conditions : @xmath93
first , we want to endow our spacetime solution ( [ met1a ] ) with a global rotation . in order to add angular momentum to the spacetime
, we perform the following rotation boost in the @xmath94 plane @xmath95 where @xmath96 is a rotation parameter and @xmath97 . substituting eq .
( [ tr ] ) into eq .
( [ met1a ] ) we obtain @xmath98 where @xmath33 is the same as @xmath33 given in eq .
( [ f(r ) ] ) .
the non - vanishing components of electromagnetic field tensor are now given by @xmath99 because of the periodic nature of @xmath100 , the transformation ( [ tr ] ) is not a proper coordinate transformation on the entire manifold .
therefore , the metrics ( [ met1a ] ) and ( [ metr2 ] ) can be locally mapped into each other but not globally , and so they are distinct @xcite . note that this spacetime has no horizon and curvature singularity .
one should note that these solutions are different from those discussed in @xcite , which were electrically charged rotating black brane solutions in gauss - bonnet gravity .
the electric solutions have black holes , while the magnetic solution interpret as wormhole .
it is worthwhile to mention that this solution reduces to the solution of einstein - maxwell equation introduced in @xcite as @xmath41 goes to zero .
second , we generalize the above solution to the case of rotating solution with more rotation parameters .
the rotation group in @xmath101 dimensions is @xmath102 and therefore the number of independent rotation parameters is @xmath103 $ ] , where @xmath104 $ ] is the integer part of @xmath105 .
the generalized solution with @xmath106 $ ] rotation parameters can be written as @xmath107 where @xmath108 , @xmath109 is the euclidean metric on the @xmath110-dimensional submanifold and @xmath33 is the same as @xmath111 given in eq .
( [ f(r ) ] ) .
the non - vanishing components of electromagnetic field tensor are @xmath112 in general the conserved quantities are divergent when evaluated on the solutions . a systematic method of dealing with this divergence for asymptotically ads solutions of einstein gravity is through the use of the counterterms method inspired by the anti - de sitter conformal field theory ( ads / cft ) correspondence @xcite . for asymptotically ads solutions of lovelock gravity with flat boundary , @xmath113 ,
the finite energy momentum tensor is @xcite @xmath114 where @xmath115 is @xmath116 } l , \label{l}\ ] ] and @xmath117 . in eq .
( [ stress ] ) , @xmath118 is the extrinsic curvature of the boundary , @xmath119 is its trace , @xmath120 is the induced metric of the boundary , and @xmath121 is trace of @xmath122 @xmath123 one may note that when @xmath41 goes to zero , the finite stress - energy tensor ( [ stress ] ) reduces to that of asymptotically ads solutions of einstein gravity with flat boundary . to compute the conserved charges of the spacetime
, we choose a spacelike surface @xmath124 in @xmath125 with metric @xmath126 , and write the boundary metric in adm ( arnowitt - deser - misner ) form : @xmath127 where the coordinates @xmath128 are the angular variables parameterizing the hypersurface of constant @xmath28 around the origin , and @xmath129 and @xmath130 are the lapse and shift functions respectively .
when there is a killing vector field @xmath131 on the boundary , then the quasilocal conserved quantities associated with the stress tensors of eq .
( [ stress ] ) can be written as @xmath132 where @xmath133 is the determinant of the metric @xmath126 , and @xmath134 is the timelike unit normal vector to the boundary @xmath124 . in the context of counterterm method , the limit in which the boundary @xmath124 becomes infinite ( @xmath135 )
is taken , and the counterterm prescription ensures that the action and conserved charges are finite .
no embedding of the surface @xmath124 in to a reference of spacetime is required and the quantities which are computed are intrinsic to the spacetimes . for our case , the magnetic solutions of gauss - bonnet gravity , the first killing vector is @xmath136 , therefore its associated conserved charge is the total mass of the wormhole per unit volume @xmath137 , given by @xmath138 \gamma m. \label{mas}\ ] ] for the rotating solutions , the conserved quantities associated with the rotational killing symmetries generated by @xmath139 are the components of angular momentum per unit volume @xmath140 calculated as @xmath141 next , we calculate the electric charge of the solutions . to determine the electric field we should consider the projections of the electromagnetic field tensor on special hypersurfaces .
the normal to such hypersurfaces for the spacetimes with a longitudinal magnetic field is @xmath142 and the electric field is @xmath143 .
then the electric charge per unit volume @xmath140 can be found by calculating the flux of the electromagnetic field at infinity , yielding @xmath144 note that the electric charge is proportional to the magnitude of rotation parameters and is zero for the static solutions .
considering both the nonlinear invariant terms constructed by electromagnetic field tensor ( @xmath145 ) and quadratic invariant terms constructed by riemann tensor ( @xmath146 , @xmath147 and @xmath148 ) in action , we have obtained a new class of rotating spacetimes in various dimensions , with negative cosmological constant .
these solutions are asymptotically anti - de sitter and reduce to the solutions of gauss - bonnet - maxwell gravity when @xmath149 and reduce to those of einstein - born - infeld gravity as @xmath150 .
this class of solutions yields an @xmath0-dimensional spacetime with a longitudinal nonlinear and nonsingular magnetic field ( the only nonzero component of the vector potential is @xmath151 ) generated by a static magnetic source .
we have found that these solutions have no curvature singularity and no horizons and may be interpreted as a traversable wormhole near @xmath78 .
also , we found that the weak energy condition is not violated at the throat , which shows there is no exotic matter near the throat .
also we have generalized these solutions to the case of rotating spacetimes with a longitudinal magnetic field . for the rotating wormhole , when a rotational parameter is nonzero , the wormhole has a net electric charge density which is proportional to the magnitude of the rotational parameter given by @xmath152 . for static case , the electric field vanishes , and
therefore the wormhole has no net electric charge .
finally , we applied the counterterm method and calculated the conserved quantities of the solutions .
we have found that the conserved quantities do not depend on the born - infeld parameter @xmath153 .
p. candelas , g. t. horowitz , a. strominger , and e. witten , nucl .
b * 258 * , 46 ( 1985 ) ; m. b. greens , j. h. schwarz and e. witten , _ superstring theory _
( cambridge university press , cambridge , england , 1987 ) ; d. lust and s. theusen , _ lectures on string theory _
( springer , berlin , 1989 ) ; j. polchinski , _ string theory _ ( cambridge university press , cambridge , england , 1998 ) . | a new class of solutions which yields an @xmath0-dimensional spacetime with a longitudinal nonlinear magnetic field is introduced .
these spacetimes have no curvature singularity and no horizon , and the magnetic field is non singular in the whole spacetime . they may be interpreted as traversable wormholes which could be supported by matter not violating the weak energy conditions .
we generalize this class of solutions to the case of rotating solutions and show that the rotating wormhole solutions have a net electric charge which is proportional to the magnitude of the rotation parameter , while the static wormhole has no net electric charge .
finally , we use the counterterm method and compute the conserved quantities of these spacetimes . |
adiabatic expansion and photon number conservation have produced a cosmic microwave background ( cmb ) with a black - body temperature of @xmath5k @xcite that evolves with redshift @xmath6 as @xmath7 .
this temperature - redshift relation is an important test of the big - bang paradigm @xcite and of spatial homogeneity @xcite .
models like decaying vacuum energy density and gravitational ` adiabatic ' photon creation predict a more general scaling @xcite .
an imprint on the @xmath8 relation can be produced if the period of accelerated expansion is driven by a phase transition @xcite .
the non - conservation of the photon number density changes the temperature - redshift and the distance duality relations .
two functional forms have been considered in the literature @xmath0 @xcite and @xmath9 @xcite , with @xmath4 and @xmath10 constant parameters ; at low redshifts , the differences between both functional forms are small and only the first scaling is usually tested . in most scenarios , deviations of adiabatic evolution are associated with distortions of the cmb black - body spectrum @xcite and are strongly constrained by the current firas upper limit of @xcite .
a departure of the standard temperature - redshift relation would represent an important challenge to the current cosmological model and it represents a test of these alternative scenarios .
the earliest measurements on cmb black - body temperature evolution were obtained using the relative populations of atomic fine - structure levels which are excited by the background radiation @xcite . @xcite
measured @xmath11k using the rotational excitation of co molecules in quasar spectral lines , value compatible with @xmath12k , the black - body temperature expected at that redshift with the standard temperature - redshift relation .
using quasar spectral lines , the best constraint on deviations from adiabatic evolution at present is @xmath13 @xcite ; they also obtained a very stringent individual measurement of the cmb temperature : @xmath14k .
recently the planck collaboration obtained a much tighter constraint @xcite by including data on large scale structure to the cmb data but their measurement @xmath15 applies to models where the deviation from adiabatic evolution starts at the last scattering surface .
the temperature - redshift relation has also been probed using the thermal sunyaev - zeldovich effect ( sunyaev & zeldovich , 1970 ; hereafter tsz ) .
the tsz anisotropy induced by clusters of galaxies along the line of sight @xmath16 is @xmath17 .
the comptonization parameter is defined as @xmath18 , where @xmath19 is the cluster optical depth , @xmath20 are the electron density and temperature evaluated along the line of sight @xmath21 , @xmath22 the thomson cross section , @xmath23 the boltzmann constant , @xmath24 the electron mass and @xmath25 the speed of light .
this effect depends on the frequency of observation @xmath26 as @xmath27 , where @xmath28 is the planck constant and @xmath29 the cmb black - body temperature at the cluster location . in the non - relativistic limit , @xmath30 .
this spectral dependence is different from that of all known foregrounds , making the tsz effect an effective tool for detecting clusters as well as a potential probe of the redshift evolution of the background temperature .
if the universe evolves adiabatically then @xmath31 is independent of redshift .
we will test deviations from adiabatic evolution assuming the @xcite parametrization , that is , @xmath32 and @xmath33 .
@xcite proposed to constrain @xmath4 by measuring the zero cross frequency of clusters at different redshifts that , for adiabatic evolution , occurs at @xmath34 .
the measurement of the cross - over frequency is problematic since the tsz is inherently weak and could be dominated by uncertain systematics . as an alternative
, early studies fit the tsz signal at different frequencies to measure the function @xmath35 .
we shall denote this procedure the _
fit method_. @xcite suggested the _ ratio method _ , constraining @xmath4 by using the ratio of the tsz anisotropy at different frequencies , @xmath36 . by taking ratios , the dependence on the comptonization parameter
is removed and there is no need to account for model uncertainties on the gas density and temperature profiles . on the negative side ,
the analysis is more complicated since the distribution of the temperature anisotropy ratios is highly non - gaussian @xcite and the kinematic sunyaev - zeldovich ( sunyaev & zeldovich , 1972 ; hereafter ksz ) generated by cluster peculiar velocities is a component always present at the cluster location that can not be separated from the cosmological signal . for small number of clusters the effect needs to be taken into account while it is expected to be less important for large cluster samples .
@xcite found @xmath37 from the tsz measurement of the coma and a2163 clusters .
later , from the data on just 13 clusters , @xcite set up an upper limit of @xmath38 in the range @xmath39 at the 68% confidence level . in 2013
the planck collaboration released their first maps of the temperature fluctuations of the cmb sky using 15.5 months of data @xcite .
the low frequency instrument ( lfi ) produced three maps with frequencies 30 , 44 and 70 ghz and angular resolution from 33 to @xmath40 .
the high frequency instrument ( hfi ) was sensitive to a wider range of frequencies , from 100 to 857 ghz and scanned the cmb sky at much higher angular resolution , from 10 to @xmath41 .
due to its large frequency coverage , high resolution and low noise , the _ planck _ satellite is an optimal instrument for detecting the tsz distortion of the cmb spectrum induced by clusters . to clean the cmb data from foreground contributions , the planck collaboration also released maps of ( a ) thermal dust and residual cosmic infrared background emission , ( b ) synchrotron , free - free and spinning dust emission , ( c ) co contributions , that are most important at 100 , 217 and 353 ghz and ( d ) maps of dust temperature and opacity @xcite .
additional products required to analyze the data such as masks and noise maps were also released . using _ planck _
data and a sample of 813 clusters up to redshift @xmath42 , @xcite obtained @xmath43 .
their sample included x - ray and tsz selected clusters .
using clusters detected with the south pole telescope , @xcite measured @xmath44 , also compatible with adiabatic evolution .
in this article we will apply the techniques developed in @xcite to test the standard temperature - redshift relation .
our cluster catalog comprises @xmath45 clusters selected from the x - ray rosat data and with well measured properties ; of those , 481 will be used in this study .
the outline of this paper is as follows : in sec . 2 we describe our data and pipeline ; in sec .
3 we present our results and in sec .
4 we summarize our main conclusions .
we constrain the temperature - redshift evolution of the background temperature using a cluster catalog selected from rosat data and the _ planck _ nominal maps released in 2013 .
we will only use hfi data , originally released in healpix format with resolution @xmath46 @xcite .
our catalog contains 782 clusters , of which 623 are outside the planck pccs - sz - union mask @xcite ( see below ) .
they have been selected from three x - ray flux limited surveys : the rosat - eso flux limited x - ray catalog ( reflex , bhringer et al . 2004 ) , the extended brightest cluster sample ( ebcs , ebeling et al . 1998 and 2002 ) and the clusters in the zone of avoidance ( ciza , ebeling et al .
details of the combined catalog are given in @xcite .
all clusters have well measured positions , x - ray fluxes and luminosities in the [ 0.1,2.4]kev rosat band , spectroscopic redshifts and angular extents of the x - ray emitting region .
the catalog also lists the x - ray temperature derived from the @xmath47 relation of @xcite and the core radii @xmath48 and central electron density @xmath49 obtained by fitting a @xmath50 model to the rosat data .
we compute the radius at which the mean overdensity of the cluster is 500 times the critical density , @xmath51 , from the @xmath52 relation of @xcite .
we define @xmath53 as the mass enclosed within a sphere of radius @xmath51 and the angular size @xmath54 , where @xmath55 is the diameter angular distance of each cluster in the @xmath56cdm model . in fig .
[ fig1 ] we represent the redshift and mass distribution of the clusters in our catalog .
the clusters in our sample have masses in the interval @xmath57\times 10^{14}$]m@xmath58 and are located at redshifts @xmath59 , relatively lower compared with other catalogs .
the catalog used by @xcite contains x - ray and planck selected clusters . compared with our catalog ,
the mcxc sample @xcite includes clusters from sources that we have not considered such as noras @xcite , sgp @xcite and nep @xcite .
also , our catalog does not include clusters from x - ray pointed surveys which tend to find mainly low - mass systems but out to high redshift . for redshifts @xmath60 and masses @xmath61 all our clusters except ten are listed in the mcxc sample . in the same mass and redshift
range the planck sz catalog contains 555 clusters . of those ,
only 239 are at less than @xmath62 away from members of our sample .
an important difference with the latter catalog is that the planck collaboration estimates @xmath53 using the scaling relation @xmath63 of @xcite while we use the @xmath52 relation from @xcite and the resulting masses differ , on average , by 30% .
@xcite used a sample of 158 clusters with redshifts @xmath64 , selected by their tsz signal . when cluster candidates are identified using cmb data the adiabatic evolution is assumed , biasing cluster selection towards those candidates that mimic this behavior .
the south pole group verified that this effect was not significant .
nevertheless , due to their selection criteria , negative temperature fluctuations were more likely negative than positive and taking into account this second effect increased their error bars a 30% @xcite . therefore , it is important to verify the results of other groups using only x - ray selected clusters since they will not be affected by these biases . in table
[ table1 ] we present the mean redshift , angular scale , x - ray luminosity and mass averaged over the full cluster sample and in several subsamples selected according to x - ray luminosity mass and redshift . for better comparison we chose the redshift bins as in @xcite ; within each redshift bin , we select all clusters with @xmath65 or @xmath66 erg / s . in the table
we do not quote a redshift bin in the mass and x - ray luminosity subsamples when it coincides with the bin of the full sample .
.average properties of different cluster subsamples from the general catalog .
clusters have been selected by luminosity and mass within each redshift bin . [
cols="^,^,^,^,^,^ " , ] we can use the upper limit on @xmath4 given above to constrain the phenomenological model of @xcite using the temperature - redshift relation derived by @xcite . the constrain is given in terms of the effective equation of state of a hypothetical decaying dark energy model .
from the bounds on the deviation from adiabatic invariance we find @xmath67 , fully compatible with the concordance @xmath56cdm model .
this constrain represents a 20% improvement over the latest measurement of @xcite .
our estimates do not take into account the effect of subtracting the tsz effect from the lgmca map using the @xmath68 frequency dependence .
[ fig3]a shows that the average temperature anisotropy at the cluster locations has @xmath69k residual compared with the same measurement at random positions in the sky averaged over 100 realizations .
this difference is well within the errors and compatible with sample variance .
we checked that the average on the six redshift bins was random , oscillating around the mean , indicating that there was not such systematic effect .
nevertheless , if the redshift evolution has @xmath70 then there would exist a residual tsz at the cluster location and we need to consider the biases introduced if the @xmath71k difference was in fact systematic . since the lgmca map is a combination of wmap and planck data at different frequencies , when subtracting the same lgmca map to all planck nominal maps , the residual tsz anisotropy would shift the zero cross frequency but will not change the dependence on @xmath4 .
the magnitude and sign of the temperature shift is difficult to predict since it would depend on the cluster redshift , mass and extent and on the lgmca mixing matrices estimated for a set of input channels on a patch of data at a given wavelet scale .
we reanalyzed the data subtracting a fix @xmath72k from the measured temperature anisotropies of all the clusters and at all frequencies and obtained that the value of @xmath4 decreased by @xmath73 in the fit method and remained within the errors in the ratio method ; i.e. , we would be obtaining a value closer to @xmath74 by this amount .
if @xmath75 then the residual tsz at the cluster location would be positive and so would be the bias in @xmath4 .
therefore , by using the tsz frequency dependence of an adiabatically evolving universe , our method could be masking the effect of a non - adiabatic evolution .
taking into account this effect , our final constraint would be @xmath76 where the parenthesis indicates the systematic contribution . at the 2@xmath77 level , this result is both compatible with all the upper limits from tsz and spectral lines .
we have constrained the deviation of the adiabatic evolution of the cmb black - body temperature applying two different estimators to a sample of x - ray selected clusters and using foreground cleaned planck nominal maps . by not including clusters selected by their tsz signature
, we avoid biasing our sample to those clusters that are closest to adiabatic evolution . following @xcite we distributed our cluster in six bins of redshift .
the constrains using the ratio and fit methods were @xmath78 and @xmath1 , respectively ; the fit method producing statistically more significant results than the ratio method since the latter is very sensitive to small denominators .
the constrains are weakened if we add a hypothetical systematic effect due to the component separation method used to construct the lgmca map and become @xmath76 and @xmath79 , compatible with the results given in the literature at the @xmath80 confidence level .
compared with @xcite the errors in the ratio method are at least a factor of two worse . while our sample contains four times as many clusters , they are located at much lower redshift , where the statistic is less sensitive .
however , our implementation of the fit method provides the best constraints on @xmath4 to date using only cmb data .
for instance , as illustrated in fig [ fig7 ] , by using the lgmca reconstruction of the intrinsic cmb temperature anisotropies to remove the cosmological contribution from our maps , we have reduced the errors compared with @xcite and obtained the best bound to date .
the constrain is also interesting since our cluster sample has the lowest mean redshift .
we constrain deviations of adiabatic evolution within @xmath59 , well in the period of accelerated expansion .
our results are comparable with a similar analysis carried out by @xcite using a catalog of clusters at higher redshifts and fully complements their findings .
since our clusters have been selected using x - ray data , they are not affected by possible biases associated with selecting clusters using the tsz effect , providing a consistency check to previous results .
we continue to expand our x - ray cluster sample to probe higher redshifts , where systematic biases would be larger and could be estimated better .
@xcite has argued that low redshift energy injections that could produce a non - adiabatic evolution of the universe are strongly constrained by the upper limit set by firas .
nevertheless , and independently of theoretical expectations , measurements of the temperature - redshift relation of the cmb black - body temperature provide a strong consistency check of the current big - bang paradigm .
we thank the referee for his / her comments that helped to improve the manuscrip .
this work was done in the context of the fct / micinn cooperation grant cosmology and fundamental physics with the sunyaev - zeldovich effect aic10-d-000443 , with additional support from projects fis2012 - 30926 from the ministerio de educacin y ciencia , spain and ptdc / fis/111725/2009 from fct , portugal .
cm is also supported by a fct research professorship , contract reference if/00064/2012 , funded by fct / mctes ( portugal ) and poph / fse ( ec ) . | we constrain the deviation of adiabatic evolution of the universe using the data on the cosmic microwave background ( cmb ) temperature anisotropies measured by the _ planck _
satellite and a sample of 481 x - ray selected clusters with spectroscopically measured redshifts . to avoid antenna beam effects
, we bring all the maps to the same resolution .
we use a cmb template to subtract the cosmological signal while preserving the thermal sunyaev - zeldovich ( tsz ) anisotropies ; next , we remove galactic foreground emissions around each cluster and we mask out all known point sources . if the cmb black - body temperature scales with redshift as @xmath0 , we constrain deviations of adiabatic evolution to be @xmath1 , consistent with the temperature - redshift relation of the standard cosmological model . this result could suffer from a potential bias @xmath2 associated with the cmb template , that we quantify it to be @xmath3 and with the same sign than the measured value of @xmath4 , but is free from those biases associated with using tsz selected clusters ; it represents the best constraint to date of the temperature - redshift relation of the big - bang model using only cmb data , confirming previous results . |
compressed sensing is a recently emerged technique for signal sampling and data acquisition which enables to recover sparse signals from undersampled linear measurements @xmath0 where @xmath1 is a sampling matrix with @xmath2 , @xmath3 denotes an @xmath4-dimensional sparse signal , and @xmath5 denotes the additive noise .
the problem has been extensively studied and a variety of algorithms , e.g. the orthogonal matching pursuit ( omp ) algorithm @xcite , the basis pursuit ( bp ) method @xcite , the iterative reweighted @xmath6 and @xmath7 algorithms @xcite , and the sparse bayesian learning method @xcite were proposed . in many practical applications , in addition to the sparse structure , sparse signals may exhibit two - dimensional cluster patterns that can be utilized to enhance the recovery performance .
for example , the target of interest in the synthetic aperture radar / inverse synthetic aperture radar ( sar / isar ) images often demonstrates continuity in both the range and cross - range domains @xcite . in video surveillance , the foreground image exhibits a cluster pattern since the foreground objects ( humans , cars , text etc . )
generally occupy a small continuous region of the scene @xcite . besides these , block - sparsity is also present in temporal observations of a time - varying block - sparse signal whose support varies slowly over time @xcite .
analyses @xcite show that exploiting the inherent block - sparse structure not only leads to relaxed conditions for exact reconstruction , but also helps improve the recovery performance considerably .
a number of algorithms have been proposed for recovering block - sparse signals over the past few years , e.g. , block - omp @xcite , mixed @xmath8 norm - minimization @xcite , group lasso @xcite , model - based cosamp @xcite , and block - sparse bayesian learning @xcite .
these algorithms , however , require _ a priori _ knowledge of the block partition ( e.g. the number of blocks and location of each block ) such that the coefficients in each block are grouped together and enforced to share a common sparsity pattern . in practice ,
the prior information about the block partition of sparse signals is often unavailable , especially for two - dimensional signals since the block partition of a two - dimensional signal involves not only the location but also the shape of each block .
for example , foreground images have irregular and unpredictable cluster patterns which are very difficult to be estimated _ a priori_. to address this difficulty , a few sophisticated bayesian methods which do not need the knowledge of the block partition were developed . in @xcite ,
a `` spike - and - slab '' prior model was proposed , where by introducing dependencies among mixing weights , the prior model has the potential to encourage sparsity and promote a tree structure simultaneously .
this `` spike - and - slab '' prior model was later extended to accommodate block - sparse signals @xcite .
nevertheless , for the `` spike - and - slab '' prior introduced in @xcite , the posterior distribution can not be derived analytically , and a markov chain monte carlo ( mcmc ) sampling method has to be employed for bayesian inference . in @xcite ,
a graphical prior , also referred to as the `` boltzmann machine '' , is employed as a prior on the sparsity support in order to induce statistical dependencies between atoms .
with such a prior , the maximum a posterior ( map ) estimator requires an exhaustive search over all possible sparsity patterns . to overcome the intractability of the combinatorial search , a greedy method @xcite and a variational mean - field approximation method @xcite were developed to approximate the map . in @xcite , to cope with the unknown cluster pattern , an expanded model is employed by assuming that the original sparse signal is a superposition of a number of overlapping blocks , and the coefficients in each block share the same sparsity pattern .
conventional block sparse bayesian learning algorithms such as those in @xcite can then be applied to the expanded model . recently in @xcite , we proposed a pattern - coupled hierarchical gaussian prior model to exploit the unknown block - sparse structure . unlike the conventional hierarchical gaussian prior model @xcite where each coefficient is associated independently with a unique hyperparameter , the pattern - coupled prior for each coefficient not only
involves its own hyperparameter , but also its immediate neighboring hyperparameters .
this pattern - coupled hierarchical model is effective and flexible to capture any underlying block - sparse structures , without requiring the prior knowledge of the block partition .
numerical results show that the pattern - coupled sparse bayesian learning ( pc - sbl ) method renders competitive performance for block - sparse signal recovery . nevertheless , a major drawback of the method is that it requires computing an @xmath9 matrix inverse at each iteration , and thus has a cubic complexity in terms of the signal dimension .
this high computational cost prohibits its application to problems with even moderate dimensions .
also , @xcite only considers recovery of one - dimensional block - sparse signals . in this paper , we generalize the pattern - coupled hierarchical model to the two - dimensional ( 2-d ) scenario in order to leverage block - sparse patterns arising from 2-d sparse signals . to address the computational issue , we resort to the generalized approximate message passing ( gamp ) technique @xcite and develop a computationally efficient method .
specifically , the algorithm is developed within an expectation - maximization ( em ) framework , using the gamp to efficiently compute an approximation of the posterior distribution of hidden variables .
the hyperparameters associated with the hierarchical gaussian prior are learned by iteratively maximizing the q - function which is calculated based on the posterior approximation obtained from the gamp .
simulation results show that the proposed method presents superior recovery performance for block - sparse signals , meanwhile achieving a significant reduction in computational complexity .
the rest of the paper is organized as follows . in section [ sec :
model ] , we introduce a 2-d pattern coupled hierarchical gaussian framework to model the sparse prior and the pattern dependencies among the neighboring coefficients . in section [ sec : algorithm ] , a gamp - based em algorithm is developed to obtain the maximum a posterior ( map ) estimate of the hyperparameters , along with the posterior distribution of the sparse signal .
simulation results are provided in section [ sec : simulation ] , followed by concluding remarks in section [ sec : conclusions ] .
we consider the problem of recovering a two - dimensional block - sparse signal @xmath10 from compressed noisy measurements @xmath11 where @xmath12 denotes the compressed measurement vector , @xmath13 is a linear map : @xmath14 , with @xmath15 , and @xmath16 is an additive multivariate gaussian noise with zero mean and covariance matrix @xmath17 .
let @xmath18 , the linear map @xmath19 can generally be expressed as @xmath20 where @xmath1 denotes the measurement matrix . in the special case
where @xmath21 , then we have @xmath22 , in which @xmath23 stands for the kronecker product .
the above model ( [ data - model ] ) arises in image applications where signals are multi - dimensional in nature , or in the scenario where multiple snapshots of a time - varying sparse signal are available . in these applications ,
signals usually exhibit two - dimensional cluster patterns that can be utilized to improve the recovery accuracy .
to leverage the underlying block - sparse structures , we introduce a 2-d pattern - coupled gaussian prior model which is a generalization of our previous work @xcite . before proceeding , we provide a brief review of the conventional hierarchical gaussian prior model @xcite , and some of its extensions . for ease of exposition , we consider the prior model for the two - dimensional signal @xmath24 instead of its one - dimensional form @xmath3 .
let @xmath25 denote the @xmath26th entry of @xmath24 . in the conventional sparse bayesian learning framework @xcite ,
a two - layer hierarchical gaussian prior was employed to promote the sparsity of the solution . in the first layer ,
coefficients @xmath27 of @xmath24 are assigned a gaussian prior distribution @xmath28 where @xmath29 is a non - negative hyperparameter controlling the sparsity of the coefficient @xmath25 .
the second layer specifies gamma distributions as hyperpriors over the hyperparameters @xmath30 , i.e. @xmath31 as discussed in @xcite , for properly chosen @xmath32 and @xmath33 , this hyperprior allows the posterior mean of @xmath29 to become arbitrarily large . as a consequence
, the associated coefficient @xmath25 will be driven to zero , thus yielding a sparse solution .
this conventional hierarchical model , however , does not encourage structured - sparse solutions since the sparsity of each coefficient is determined by its own hyperparameter and the hyperparameters are independent of each other .
in @xcite , the above hierarchical model was generalized to deal with block - sparse signals , in which a group of coefficients sharing the same sparsity pattern are assigned a multivariate gaussian prior parameterized by a common hyperparameter .
nevertheless , this model requires the knowledge of the block partition to determine which coefficients should be grouped and assigned a common hyperparameter . to exploit the 2-d block - sparse structure , we utilize the fact that the sparsity patterns of neighboring coefficients are statistically dependent . to capture the pattern dependencies among neighboring coefficients , the gaussian prior for each coefficient
@xmath25 not only involves its own hyperparameter @xmath29 , but also its immediate neighbor hyperparameters .
specifically , a prior over @xmath24 is given by @xmath34 where @xmath35 in which @xmath36 denotes the neighborhood of the grid point @xmath26 , i.e. @xmath37 changes accordingly .
] , and @xmath38 $ ] is a parameter indicating the pattern relevance between the coefficient @xmath25 and its neighboring coefficients . clearly , this model is an extension of our previous prior model @xcite to the two - dimensional case .
when @xmath39 , the prior model ( [ pc - model ] ) reduces to the conventional sparse bayesian learning model . when @xmath40 , we see that the sparsity of each coefficient @xmath25 is not only controlled by the hyperparameter @xmath29 , but also by the neighboring hyperparameters @xmath41 .
the coefficient @xmath25 will be driven to zero if @xmath29 or any of its neighboring hyperparameters goes to infinity .
in other words , suppose @xmath29 approaches infinity , then not only its corresponding coefficient @xmath25 will be driven to zero , the neighboring coefficients @xmath42 will decrease to zero as well .
we see that the sparsity patterns of neighboring coefficients are related to each other through their shared hyperparameters . on the other hand , for any pair of neighboring coefficients ,
each of them has its own hyperparameters that are not shared by the other coefficient .
hence , no coefficients are pre - specified to share a common sparsity pattern , which enables the prior to provide flexibility to model any block - sparse structures .
following @xcite , we use gamma distributions as hyperpriors over the hyperparameters @xmath43 , i.e. @xmath44 where we set @xmath45 , and @xmath46 .
the choice of @xmath32 will be elaborated later in our paper .
also , the noise variance @xmath47 is assumed unknown , and to estimate this parameter , we place a gamma hyperprior over @xmath48 , i.e. @xmath49 where we set @xmath50 and @xmath51 .
we now proceed to perform bayesian inference for the proposed pattern - coupled hierarchical model .
the following model is considered since the linear map @xmath19 can be expressed as @xmath52 @xmath53 we first translate the prior for the two - dimensional signal @xmath24 to a prior for its one - dimensional form @xmath3 . from ( [ pc - model ] ) , the prior over @xmath3 can be expressed as @xmath54 where @xmath55 in which @xmath56 denotes the neighbors of the point @xmath26 on the two - dimensional grid , i.e. @xmath57 changes accordingly . ] .
the relation between @xmath58 and @xmath26 is given by @xmath59 , that is , @xmath60 , and @xmath61 , in which @xmath62 denotes the ceiling operator .
note that for notational convenience , we , with a slight abuse of notation , use @xmath63 to denote the @xmath58th entry of @xmath3 and @xmath64 to denote the hyperparameter associated with the coefficient @xmath63 .
also , let @xmath65 since its exact meaning remains unaltered . from ( [ alpha - prior ] ) , we have @xmath66 an expectation - maximization ( em ) algorithm can be developed for learning the sparse signal @xmath3 as well as the hyperparameters @xmath67 . in the em formulation ,
the signal @xmath3 is treated as hidden variables , and we iteratively maximize a lower bound on the posterior probability @xmath68 ( this lower bound is also referred to as the q - function ) . briefly speaking , the algorithm alternates between an e - step and a m - step . in the e - step ,
we need to compute the posterior distribution of @xmath3 conditioned on the observed data and the hyperparameters estimated from the @xmath69th iteration , i.e. @xmath70 it can be readily verified that the posterior @xmath71 follows a gaussian distribution with its mean and covariance matrix given respectively by @xmath72 where @xmath73 .
the q - function , i.e. @xmath74 $ ] , can then be computed , where the operator @xmath75 $ ] denotes the expectation with respect to the posterior distribution @xmath71 . in the m - step , we maximize the q - function with respect to the hyperparameters @xmath67 . it can be seen that the em algorithm , at each iteration , requires to update the posterior distribution @xmath71 , which involves computing an @xmath9 matrix inverse .
thus the em - based algorithm has a computational complexity of @xmath76 flops , and therefore is not suitable for many real - world applications involving large dimensions . in the following
, we will develop a computationally efficient algorithm by resorting to the generalized approximate message passing ( gamp ) technique @xcite .
gamp is a very - low - complexity bayesian iterative technique recently developed @xcite for obtaining an approximation of the posterior distribution @xmath71 .
it therefore can naturally be embedded within the em framework to replace the computation of the true posterior distribution . from gamp
s point of view , the hyperparameters @xmath67 are considered as known . the hyperparameters can be updated in the m - step based on the approximate posterior distribution of @xmath3 .
we now proceed to derive the gamp algorithm for the pattern - coupled gaussian hierarchical prior model .
gamp was developed in a message passing - based framework . by using central - limit - theorem approximations , message passing between variable nodes and factor nodes
can be greatly simplified , and the loopy belief propagation on the underlying factor graph can be efficiently performed . as noted in @xcite ,
the central - limit - theorem approximations become exact in the large - system limit under i.i.d .
zero - mean sub - gaussian @xmath77 . for notational convenience ,
let @xmath78 denote the hyperparameters .
firstly , gamp assumes posterior independence among hidden variables @xmath79 and approximates the true posterior distribution @xmath80 by @xmath81 where @xmath82 and @xmath83 are quantities iteratively updated during the iterative process of the gamp algorithm . here
, we have dropped their explicit dependence on the iteration number @xmath84 for simplicity . substituting ( [ x - prior ] ) into ( [ eqn-1 ] )
, it can be easily verified that the approximate posterior @xmath85 follows a gaussian distribution with its mean and variance given respectively as @xmath86 another approximation is made to the noiseless output @xmath87 , where @xmath88 denotes the @xmath89th row of @xmath77 .
gamp approximates the true marginal posterior @xmath90 by @xmath91 where @xmath92 and @xmath93 are quantities iteratively updated during the iterative process of the gamp algorithm .
again , here we dropped their explicit dependence on the iteration number @xmath84 . under the additive white gaussian noise assumption
, we have @xmath94 .
thus @xmath95 also follows a gaussian distribution with its mean and variance given by @xmath96 with the above approximations , we can now define the following two scalar functions : @xmath97 and @xmath98 that are used in the gamp algorithm .
the input scalar function @xmath97 is simply defined as the posterior mean @xmath99 @xcite , i.e. @xmath100 the scaled partial derivative of @xmath101 with respect to @xmath82 is the posterior variance @xmath102 , i.e. @xmath103 the output scalar function @xmath98 is related to the posterior mean @xmath104 as follows @xmath105 the partial derivative of @xmath106 is related to the posterior variance @xmath107 in the following way @xmath108 given the above definitions of @xmath97 and @xmath98 , the gamp algorithm tailored to the considered sparse signal estimation problem can now be summarized as follows ( details of the derivation of the gamp algorithm can be found in @xcite ) , in which @xmath109 denotes the @xmath110th entry of @xmath77 , @xmath111 and @xmath112 denote the posterior mean and variance of @xmath63 at iteration @xmath84 , respectively . [ cols="<",options="header " , ]
we developed a pattern - coupled sparse bayesian learning method for recovery of two - dimensional block - sparse signals whose cluster patterns are unknown _ a priori_. a two - dimensional pattern - coupled hierarchical gaussian prior model is introduced to characterize and exploit the pattern dependencies among neighboring coefficients .
the proposed pattern - coupled hierarchical model is effective and flexible to capture any underlying block - sparse structures , without requiring the prior knowledge of the block partition .
an expectation - maximization ( em ) strategy is employed to infer the maximum a posterior ( map ) estimate of the hyperparameters , along with the posterior distribution of the sparse signal .
additionally , the generalized approximate message passing ( gamp ) algorithm is embedded in the em framework to efficiently compute an approximation of the posterior distribution of hidden variables , which results in a significant reduction in computational complexity .
numerical results show that our proposed algorithm presents a substantial performance advantage over other existing state - of - the - art methods in image recovery .
d. wipf and s. nagarajan , `` iterative reweighted @xmath6 and @xmath7 methods for finding sparse solutions , '' _ ieee journals of selected topics in signal processing _ ,
vol . 4 , no . 2 , pp . 317329 , apr .
2010 .
v. cevher , a. sankaranarayanan , m. f. duarte , d. reddy , r. g. baraniuk , and r. chellappa , `` compressive sensing for background subtraction , '' in _
european conf .
comp . vision ( eccv ) _ , marseille , france , october 12 - 18 2008 .
z. zhang and b. d. rao , `` sparse signal recovery with temporally correlated source vectors using sparse bayesian learning , '' _ ieee journal of selected topics in signal processing _
, vol . 5 , no . 5 , pp . 912926 , sept .
2011 .
a. drmeau , c. herzet , and l. daudet , `` boltzmann machine and mean - field approximation for structured sparse decompositions , '' _ ieee trans .
signal processing _ , vol .
60 , no . 7 ,
34253438 , july 2012 .
z. zhang and b. d. rao , `` extension of sbl algorithms for the recovery of block sparse signals with intra - block correlation , '' _ ieee trans . signal processing _ , vol .
61 , no . 8 , pp . 20092015 , apr . 2013 .
g. warnell , d. reddy , and r. chellappa , `` adaptive rate compressive sensing for background subtraction , '' in _ ieee international conference on acoustics , speech and signal processing ( icassp ) _ , kyoto , japan , march 25 - 30 2012 .
j. p. vila and p. schniter , `` expectation - maximization bernoulli - gaussian approximate message passing , '' in _
45th asilomar conference on signals , symtems and computers _ , pacific grove , ca , usa , november 6 - 9 2011 . | we consider the problem of recovering two - dimensional ( 2-d ) block - sparse signals with _ unknown _
cluster patterns .
two - dimensional block - sparse patterns arise naturally in many practical applications such as foreground detection and inverse synthetic aperture radar imaging . to exploit the block - sparse structure ,
we introduce a 2-d pattern - coupled hierarchical gaussian prior model to characterize the statistical pattern dependencies among neighboring coefficients . unlike the conventional hierarchical gaussian prior model where
each coefficient is associated independently with a unique hyperparameter , the pattern - coupled prior for each coefficient not only involves its own hyperparameter , but also its immediate neighboring hyperparameters .
thus the sparsity patterns of neighboring coefficients are related to each other and the hierarchical model has the potential to encourage 2-d structured - sparse solutions . an expectation - maximization ( em ) strategy is employed to obtain the maximum a posterior ( map ) estimate of the hyperparameters , along with the posterior distribution of the sparse signal .
in addition , the generalized approximate message passing ( gamp ) algorithm is embedded into the em framework to efficiently compute an approximation of the posterior distribution of hidden variables , which results in a significant reduction in computational complexity .
numerical results are provided to illustrate the effectiveness of the proposed algorithm .
pattern - coupled sparse bayesian learning , block - sparse structure , expectation - maximization ( em ) , generalized approximate message passing ( gamp ) . |
the standard model ( sm ) of quarks and leptons serves as a bible of high - energy physics .
it even resembles the real bible , in the sense that it is mainly a set of `` stories '' , rules and wisdoms , which reflect both real life and human thinking .
the sm is also flexible , since it can easily be complemented with new findings or ideas , such as neutrino oscillations , supersymmetry , or new fundamental quarks and leptons , just like the first bible was once complemented with the new testament by christians . unlike the bible , the sm does not give a clue to its deeper meaning , nor does it reveal any connections between its many bits and pieces .
the sm is hence a quantitative success , but a qualitative mess .
what i will argue in this talk , however , is that the situation is not that unclear or hopeless .
rather , the sm is full of prophecies and hints to its deeper background , many of which can be understood after a deeper look into those sm features that have been put in entirely by hand .
i will make frequent references to the `` history book '' , in order to investigate if such ideas or situations have appeared before .
as you will see , there have been numerous situations in the past when paradoxes and problems like the ones in the sm turned out to come from _ compositeness _ of the `` fundamental '' particles of those days . to be more precise ,
i will discuss the following aspects of the sm : 0.1 cm - there are six quarks and six leptons in three families - most quarks and leptons are unstable - there are several _ ad hoc _ quantum numbers - some quarks , leptons and gauge bosons mix / oscillate - the heavy vector bosons are massive and unstable - the @xmath0 mixes with the photon 0.1 cm the discussion will converge toward more precise features of a preon model , and i will conclude by briefly describing a recent one , presented by jean - jacques dugne , johan hansson and myself @xcite .
the discussion presented here is , in fact , an offspring of our work with that particular model .
before going into some detail with preons and the sm , a few more general observations can be made .
first of all , the history book tells us that `` there is always a deeper layer of compositeness '' .
earlier layers have normally been suggested and/or discovered when the model in fashion became too complex , or when there were too many different models for the `` fundamental '' particles . such leaps to new levels of compositeness come every @xmath1 years , which means that time is ripe for preons !
nevertheless , compositeness has never been a main trend of high - energy research .
rather , an overwhelming majority of theorists has traditionally regarded the elementary particles of the day as _ the _ fundamental ones .
compositeness usually comes into fashion as a viable theory only if strongly supported by experiment , or if suggested by some well - known theorist . in the latter case ,
the interest fades away after some time , if not supported by observations .
examples are the quark model by gell - mann and zweig @xcite and the early preon models by harari @xcite and shupe @xcite ( `` rishons '' ) and by fritzsch and mandelbaum @xcite ( `` haplons '' ) .
the quark model gained a wider acceptance only after the pioneering deep - inelastic scattering data of the late 1960s and the early 1970s , and their interpretation in terms of partons .
there are no such supporting data for preons , and the interest in the first preon models was therefore gone a long time ago .
the concept of compositeness still appears now and then in experimental work , but mostly in routine searches for deviations from the sm predictions in events with large transferred momenta .
the lack of signals are normally quoted as a minimal preon energy scale of a few tev @xcite .
it is to be understood as an inverse length - scale in a hypothetical form factor of internal preon wave functions ( and not as a preon mass scale ) . in order to restrict this length - scale further
, one would seemingly need the high energies of the upcoming cern lhc facility , or at least high - statistics data from run ii of the fermilab tevatron .
however , this does not exclude that preons could reveal themselves in other ways , e.g. , at the tevatron , such as through the discovery of new , exotic quarks or leptons .
an earlier example was the discovery of the exotic @xmath2 hyperon and its importance for the subsequent quark model .
the most common argument in favour of preons @xcite is that there are too many quarks and leptons to let any sm enthusiast feel comfortable .
there is no obvious logical reason why there would be ( at least ) twelve fundamental particles .
the case for preons is strengthened by the fact that these particles fall into a nice pattern of three `` families '' .
the history book tells us that the existence of too many fundamental particles in the past always preceded the discovery of yet smaller and fewer building blocks ( not counting the era of `` earth , fire , air and water '' ) .
examples are : 0.1 cm - the existence of too many elements was explained by the atoms consisting of electrons and nuclei ; - the existence of many isotopes of one and the same element is a by - product of the compositeness of nuclei , containing protons and neutrons ; - the hundreds of hadrons that puzzled high - energy physicists a half - century ago reflect their quark structure .
the patterns among the early hadrons led the mind to the number three , and resulted in a model with three different quarks , with a quark - flavour @xmath3 symmetry .
0.1 cm a first conclusion is therefore that the many quarks and leptons reflect a preon substructure , and that the pattern of three families comes from the existence of three preons , with a preon - flavour @xmath3 symmetry .
it is noteworthy that the early preon models did not , in general , provide a more rational explanation of the quarks and leptons of those days .
typically , four different preons were used to explain the existence of the four particles of the lightest quark / lepton family , while the heavier ones were not properly understood .
a good preon model should hence explain _ all _ quarks and leptons on equal terms , at least in some sense .
most quarks and leptons are unstable , decaying into lighter ones , until only stable ( ? ) electrons , protons and neutrinos remain . in my view
there is a logical problem with unstable fundamental particles : how can nature s most fundamental objects decay into equally fundamental objects ? as to the best of my knowledge , this simple argument has not been debated earlier in the literature .
when discussing this with colleagues , i am often met by the attitude that nature is anyway so strange , so why bother ?
again the history book gives some moral support to the preon solution , because all decays of previous `` fundamental '' particles have mirrored their compositeness .
examples are : 0.1 cm - the decays of atoms were found to come from the decays of their nuclei ; - which were later blamed on the decays of their nucleons ; - which were later blamed on the decays of their quarks . 0.1 cm so , where will this sequence of explanations end ? in my opinion , not until we find a level with _ stable _ constituents
. a preon model must therefore either have absolutely stable preons , or rely on yet unstable and composite preons ( `` pre - preons '' ) .
if preons are stable , all quark and lepton decays are just regrouping of preons into systems with lighter quarks and leptons .
this strongly limits the number of possible preon models .
the pioneering preon models quoted above did not have this property .
the heavier two families either remained unexplained , or were assumed to be internal excitations of the lightest one , decaying to their own ground states .
summing up the prophecies so far , one can envisage the existence of three absolutely stable preons .
this makes _ net preon flavour _ a completely conserved quantity in all particle reactions .
a by - product is that in case two quarks or leptons would have identical net preon flavours , they will ( or must ) mix into new mass eigenstates .
this would be equivalent to , e.g. , @xmath4- and @xmath5-mixing in the quark model .
hence such a preon model provides a possibility to understand the various mixings of quarks , neutrinos and weak isospin eigenstates in the sm .
these aspects will be discussed later .
the sm contains a few quantum numbers that have been put in by hand , without a deeper understanding , and just in order to describe that some quark and lepton quantities are conserved or partially conserved .
these are quantum numbers that either seem absolutely conserved , such as baryon number , or are known to be conserved only in some reactions , e.g. , weak isospin .
lepton numbers form a grey - zone .
they seemed to be absolutely conserved until quite recently , when neutrino oscillations were discovered .
the history book again gives some clues .
a long time ago , the `` isotopic numbers '' were understood only after the discovery of the neutron and the compositeness of atomic nuclei . in the 1950s
hadronic isospin and strangeness / hypercharge were introduced to describe the observed approximate symmetries of hadronic decays and interactions .
these three quantum numbers turned out to come from three quark flavours .
since the number three pops up also in the sm , it is tempting to guess that lepton number conservation has to do with preon number conservation , i.e. , preon - flavour @xmath3 symmetry .
the disturbing neutrino oscillations / mixings can then be understood if two or more neutrinos have identical net preon flavour .
this will be discussed in the next section .
the connection between weak isospin and the number of preons is not as obvious , but will also be discussed below .
the sm prescribes that certain created quarks , leptons and gauge bosons mix into new eigenstates before being detected .
the history book tells about a few such situations in the past .
a classical example is that different isotopes , created in nuclear reactions , mix in nature in certain proportions , and are inseparable in normal chemical reactions .
chemical isotope mixing hence has its root in the compositeness of atomic nuclei .
a more modern case is the mixing / oscillation of the two neutral kaons @xmath6 and @xmath7 into the mass eigenstates @xmath8 and @xmath9 .
this comes about because of quark reactions inside the kaons , i.e. , the compositeness of hadrons .
both cases are examples of what happens when virtually different states have identical net quantum numbers relative the particular interaction used to `` detect '' them .
i will now discuss three similar mixings of `` fundamental '' particles in the sm , and their interpretations in terms of compositeness .
let us assume that the @xmath10 and @xmath11 quarks mix into the weak - interaction ( `` mass '' ) eigenstates @xmath12 and @xmath13 because they have identical net preon contents .
the question is then how this can be arranged in detail .
there are in principle two different solutions .
one is that the two quarks have identical preon contents , although with some internal differences between the detailed preon wave functions , e.g. , with two different internal spin structures .
the other one is to focus on the word `` net '' , meaning that some preon flavours cancel inside either quark , e.g. , because they contain preon - antipreon pairs .
it turns out that the first alternative fits the neutrino sector better ( see below ) .
i therefore suggest that a quark contains a preon - antipreon pair plus an additional preon . as an example , consider a situation where the two quarks have the following compositions : @xmath14 and @xmath15 , both with the net flavour of the @xmath16 preon .
if we assume that quark production in strong ( qcd ) processes always starts with pure preon states , these two quarks will subsequently mix into mass eigenstates before they decay , or take part in a weak process .
if looked upon as an oscillation , the mixing can be illustrated with the feynman - like diagram of fig .
2 . a preon - antipreon pair inside a quark can annihilate and produce another pair , thereby forming another quark . for technical reasons
i have used the convention of an antipreon ( @xmath16 ) for the common preon of @xmath10 and @xmath11 .
it is naturally not clear in detail how one preon pair converts to another .
the intermediate system can be one or more photons , one or more gluons , or maybe some new gauge bosons ( `` hypergluons '' ) .
neither does it seem realistic that one and the same phenomenon can explain _ all _ quark mixings of the so - called ckm matrix @xcite .
there is simply no third state made up of these three preons that can oscillate in a similar way to a @xmath10 or an @xmath11 .
on the other hand , the smaller ckm matrix elements differ by an order of magnitude from the cabibbo one , indicating that the detailed mechanism here is different from the one of fig . 2 for @xmath10 and @xmath11 .
i refer to @xcite for a preon - based discussion of other quark mixings than the @xmath17 one .
the preon charges can easily be chosen as to fit the charge of the two quarks .
obviously , @xmath18 must have charge @xmath19 , while the other two can be `` anything '' .
it is wise to choose the charges @xmath19 also for @xmath20 and @xmath21 for @xmath22 .
this will be obvious from the discussion in 6.2 below , although it already now seems attractive to mimic the three charges of the original flavour-@xmath3 quark model .
once we choose preon charges as multiples of @xmath23 , and prescribe that a quark contains three entities , it becomes almost necessary to build the integer - charged leptons as three - preon states . with the charges defined above
there are just three main ways to make neutrinos : @xmath24 , @xmath25 and @xmath26 .
the three preons build up a total spin @xmath27 , which means that there are several possible spin combinations . for simplicity ,
let us assume that two unequal preons prefer to form a total spin-@xmath28 `` dipreon '' pair , in the same fashion as quarks tend to pair up in diquarks in many situations @xcite .
such restrictions result in the following _ five _ neutrinos : @xmath29 , @xmath30 , @xmath31 .
@xmath32 , @xmath33 .
note the similarity with the five neutral baryons in the spin-@xmath27 nonet ( octet + singlet ) of the original quark model .
it can be seen that the three neutrinos @xmath34 , @xmath35 and @xmath36 indeed have identical net preon flavours , and hence can mix / oscillate into three new mass eigenstates .
such an oscillation can be seen as one of the three preons oscillating between the two spin-@xmath28 pairs that it can form with the other two preons .
figure 3 illustrates the situation for an oscillation between @xmath34 and @xmath35 . at this primitive stage of the discussion it is not possible to pinpoint the actual neutrinos that correspond to the different preon states .
three of the five states must naturally be connected to the three known neutrinos , while the other two must have masses in excess of half the @xmath0 mass . )
pairs ] it can be added that with these particular preons it is impossible to construct _ charged _ leptons that would oscillate or mix in the same way .
neither can there be decays of the type @xmath37 . in many preon models @xcite the weak gauge bosons
are supposed to be composite and built up by preon - antipreon pairs in total spin @xmath40 . with the preons described earlier
, it is tempting to define @xmath41 and @xmath42 .
the neutral sector is more complicated .
there are five neutral combinations of the three preons , and three of these have identical net preon numbers : @xmath43 , @xmath44 and @xmath45 .
they resemble the @xmath46 , @xmath47 and @xmath48 of the vector meson nonet , and are expected to mix . looking closer into the preon - flavour @xmath3 structure
, it seems as if some ( wave - function ) combinations correspond to the two neutral weak - isospin eigenstates , namely @xmath49 and @xmath50 .
these two then mix into two mass eigenstates , @xmath0 and @xmath51 ( a new and even heavier boson ) .
note that the mixing partner of the normal @xmath52 is _ not _ the photon , like in the sm .
the reason is that the weak interaction is not fundamental in models with composite @xmath52s and @xmath53s , and there can hence not be an electroweak unification ( and no higgs ! ) .
this aspect will be further discussed later .
the mixing of @xmath54 with @xmath55 can again can be illustrated in the fashion of a feynman diagram , as in fig .
4 . the intermediate state must now be neutral in both charge and colour , as well as have spin @xmath40 .
it could hence be one or more photons , two or three gluons , or a number of hypothetical `` hypergluons '' .
a closer look shows that the action in figs . 2 & 4 takes place through very similar preon subprocesses , as shown again in fig .
5 . and
@xmath39 , in the upper part of the figure , and the cabibbo mixing of @xmath10 and @xmath11 , in the lower part , come from the same basic preon processes , and are hence related ] if the mixing fraction of the @xmath54 and @xmath55 pairs depends _ only _ on the masses and/or electric charges of @xmath20 and @xmath22 , and not on , e.g. , the systems they sit in , or their total colour or spin , then one can show that there is a relation between the cabibbo and weinberg angles @xcite : @xmath56 with @xmath57 and @xmath58 @xcite the @xmath59 and the @xmath60 .
this is a fair agreement considering the rough assumptions .
in addition , if we simplify the situation further and assume that the mixings occur through one - photon intermediate states , then the preon - antipreon pairs occur in proportion to their squared charges , and we expect : @xmath61 which again is not far from reality .
most _ ad hoc _ sm assumptions that lack a deeper background have to do with the ugly fact that the weak interaction has massive gauge bosons . that is the excuse for introducing the higgs mechanism with all its inherent problems and unsatisfactory logic .
this is typical for many of the odd features of the sm .
they are introduced to cure and/or quantify earlier theoretical problems , rather than to explain the observationes from more basic principles .
ofter the cure causes side - effects , which need new cures , etc .
the situation with massive and unstable `` gauge bosons '' is not new in high - energy physics . in nuclear physics
it is often productive to regard the vector mesons , @xmath62 , @xmath47 , @xmath48 etc , as gauge bosons of the nuclear forces that keep nucleons in place .
they are good phenomenological couriers of qcd , by leaking colour - neutral quark - antiquark pairs between nucleons .
it seems as if spin-@xmath40 particles are better than scalar ones in faking true ( massless ) gauge bosons .
the history book therefore warns us that massive gauge bosons are not fundamental , but some kind of `` neutral '' leakage of more basic forces .
hence @xmath53 and @xmath52 might well be false gauge bosons that just leak the basic preon forces .
it would imply that there are _ nine _ different heavy vector bosons ( if we believe in preon - flavour @xmath3 ) .
hence six very heavy ones are yet to be discovered .
five of the bosons are neutral ( @xmath52s ) and four charged ( @xmath53s ) .
three of the @xmath52s have identical net preon flavour and would mix , since they are built up by @xmath43 , @xmath44 and @xmath45 .
the remaining two @xmath52s are mutual antiparticles , being @xmath63 and @xmath64 .
it remains to discuss why @xmath0 seems to mix with the photon , in a way that is well described by the electroweak sector of the sm and parametrized by the weinberg angle .
the history book again tells why , because the situation is not new . in the 1960s
it was discovered that the photon sometimes behaves like a hadron in interactions with nucleons .
it was suggested that its wave function has a hadronic component , consisting mainly of the @xmath62 meson , but with some fraction also of more massive spin-@xmath40 mesons at higher @xmath66 values .
this idea was dubbed vector meson dominance ( vmd ) @xcite .
it was developed in great detail and used to understand , or at least describe , a wealth of data .
it is still a viable model for the behaviour of virtual photons in medium-@xmath66 reactions with hadrons .
hence a photon is believed to couple directly to a @xmath62 , like in fig .
6 , when interacting with a hadron .
according to the vector meson dominance model ] in the early days this was taken as a fact , and the origin of the coupling remained unknown for a while . a physicist with a gift to look into the future
could then have come up with a radical idea : the photon and the @xmath62 are one and the same particle !
this would lay the ground for a new theory of _ electrostrong unification _
, where the @xmath67 mixing is parametrized by an electrostrong mixing angle .
after a while the substantial difference in mass between the two particles would have inspired someone to invent an electrostrong higgs boson . or rather , a whole set of mixing angles and higgs bosons , because the photon couples also to other vector mesons .
why was this model never invented ( except here ) ?
because soon after the birth of vmd , the quark model gained ground .
then it became evident that any spin-@xmath40 neutral meson couples to ( `` mixes '' with ) the photon just because its quarks are electrically charged .
at high - enough @xmath66 the photon sees the constituents , instead of the ( invisible ) neutral hadron .
the true explanation of the @xmath67 mixing is hence _
compositeness _ , as illustrated by fig
mixing comes about because @xmath62 is composite ] the conclusion for the electroweak sector and preons is then obvious .
the @xmath52 contains charged preons , and is hence destined to couple to a high-@xmath66 photon , as in fig .
mixing comes about because @xmath52 is composite ] qualitatively , the electroweak formalism is very similar to the old parametrization of the vdm model , i.e. , the propagators in a feynman - graph formalism look the same .
the @xmath52 and @xmath53s are just very good at faking true gauge bosons .
one can even understand why one and the same weinberg angle would reasonably well describe not only the @xmath65 mixing in the sm but also the @xmath68 mixing in a preon model .
this would come about if the conjectures leading to ( 1 ) and ( 2 ) were true , i.e. , that the preon - antipreon states inside the various @xmath52s mix via virtual photons , rather than gluons or hypergluons . however , the electroweak and electrostrong situation differ quantitatively , since no heavier ( @xmath51 ) boson has revealed itself experimentally as deviations from the sm predictions , e.g. , at the cern lep machine .
neither are there any traces of direct @xmath51 production at the fermilab tevatron .
the hypothetical @xmath51 should therefore be much heavier than the @xmath0 @xcite , unlike the vector - meson situation , where the @xmath47 is very close to @xmath62 in mass .
this need not be troublesome for preons , because the preon-@xmath3 wave functions inside @xmath52s might differ substantially from the quark-@xmath3 wave functions inside mesons .
it could be fruitful , however , to analyse the unconventional alternative that the @xmath0 and the next @xmath51 have almost _ equal _ masses .
maybe the @xmath0 wave function is dominated by the @xmath55 combination and the first @xmath51 by @xmath54 .
then @xmath69 annihilation would produce mainly the @xmath0 , for two reasons : ( i ) an annihilation to a photon would couple the photon four times as strongly to a @xmath44 pair than to an @xmath43 pair ; ( ii ) an annihilation of only the dipreons in the @xmath69 pair would result in a @xmath44 system ( see table 2 below ) , which in turn would prefer to make a @xmath0 . at present
it is unclear to me if a `` hidden '' @xmath51 , near the @xmath0 in mass , would be in conflict with the cern lep precision tests of the sm .
this discussion of crucial _ ad hoc _ features of the standard model has , step - by - step , led to strict and rather detailed requirements of a hypothetical preon model for quarks , leptons and heavy vector bosons .
as you might have guessed , they are all in line with the `` preon - trinity '' model presented earlier by dugne , hansson and myself @xcite . the three preons have the charges given in table 1 .
also listed are the dipreons , necessary for understanding , e.g. , neutrino oscillations .
when entered as anti - dipreons , the scheme reveals a nice supersymmetric structure . the best way to test the model would be to search for the new ( heavy ) particles in table 2 . since the top quark is most probably the one marked @xmath70 ,
there is some hope that the missing quarks and leptons can be produced at least at the fermilab tevatron .
i am now analysing the conjecture that all three quarks , @xmath71 , @xmath72 and @xmath70 , in table 2 are hidden in the tevatron data .
the analyses of their decay modes are rather complex , especially in the light of the trigger and tagging conditions of the tevatron experiments .
the @xmath73 quark is a mystery .
it appears near the @xmath10 and @xmath74 in table 2 , but it has charge @xmath75 , since both its preon and anti - dipreon have charge @xmath21 .
this made us suspect @xcite that the system is not bound , and hence not a real quark .
a final observation is that the heavy neutrino @xmath76 in table 2 can be produced in @xmath69 annihilation together with a _
normal _ antineutrino .
we expect the @xmath76 to be lighter than its related quark ( @xmath70 ) , since quarks are in general heavier than the corresponding leptons .
then the @xmath76 can , in principle , be produced at energies lower than around @xmath77 gev .
hence it might be visible in existing cern lep data .
i am analysing the situation in some detail .
all lep groups have reported that there are no statistically significant signatures of decay products from a heavy neutrino ( `` neutral excited lepton '' ) .
so maybe a careful event - by - event check is the only hope of finding a hint of a heavy neutrino at lep ( or hera ) . 01.cm
* acknowledgements * + + first of all i would like to thank the organisers for providing a wonderful atmosphere at schlo@xmath78 ringberg , and a most fantastic weather all over bavaria .
remember that this praise comes right from the heart of a resident of the swedish arctic , where the ice on the baltic broke up in early may , six weeks ago !
i am also grateful for much advice from my preon partners jean - jacques dugne and johan hansson . at an early stage this preon research was supported in part by the european commission under contract chrx - ct94 - 0450 , within the network the fundamental structure of matter " . | the standard model of quarks and leptons is , at first sight , nothing but a set of _ ad hoc _ rules , with no connections , and no clues to their true background . at a closer look
, however , there are many inherent prophecies that point in the same direction : _ compositeness _ in terms of three stable preons . |
the use of type ia supernovae ( hereafter sne ) as standardizable candles has led to surprising and intriguing results when it comes to the expansion rate of the universe , however , it is still a most unsatisfactory situation that the specifics of the progenitor scenarios that lead to the sn explosion are mostly unknown .
two general scenarios of how a white dwarf ( wd ) accretes matter to reach the chandrasekhar mass and thereafter explode are often discussed . in the single degenerate ( sd )
scenario , the wd accretes mass from a companion sub - giant or giant star , while the double degenerate ( dd ) scenario involves the merger of two wds .
the different progenitor scenarios are expected to involve different delay time distributions , i.e. , the time between the formation of the progenitor star and sn explosion ( e.g. , yungelson & livio 2000 ; greggio 2005 ) .
the rate of type ia sne should mimic the cosmic star formation rate ( sfr ) , albeit shifted to lower redshift by the delay time distribution set by the progenitor scenario . in dahlen et al .
( 2004 ) and strolger et al .
( 2004 ) , we used a sample of 23 type ia sne detected with @xmath0 acs in the goods fields to derive the sn rate ( snr ) to @xmath31.6 . using the observed rate and a simple one - component model characterized by a single delay - time , @xmath4 , we found with 95% confidence @xmath5 gyr .
the main observational feature driving @xmath4 to a high value , was the deficit of detected sne at @xmath6 . while this was the first attempt to derive the snr at @xmath7 to set constraints on the delay time , different approaches to address the question
have also been taken .
the single delay - time model offers a simple parametrization of the relation between sfr and snr ( madau et al .
1998 ; dahlen & fransson 1999 ) , however , it is plausible that there are different channels , characterized by different delay - times , contributing to the type ia snr , making the single - model an oversimplified description of the data .
mannucci et al .
( 2005 ) calculated the type ia snr as a function of stellar mass in galaxies , and found that overall rate could be described as the sum of a young or prompt component proportional to the ongoing sfr and an old component proportional to the stellar mass .
similar results are found by sullivan et al .
( 2006 ) , who use results from the supernova legacy survey ( snls ) to derive the type ia sn rate as a function of host galaxy sfr and mass .
their results also support a scenario where at least one component of the snr is closely related to the instant sfr .
additional observations that support that a two - component model is plausible includes , e.g. , branch & van den bergh ( 1993 ) , who found differences in the spectra of type ia sne found in early - type and late - type galaxies , respectively .
furthermore , a number of results suggest differences in peak magnitude distributions between sne occurring in different environments , in particular , under - luminous type ia sne are mostly found in early type galaxies , while the brightest type ia sne are preferentially found in late - type galaxies ( filippenko 1989 ; della valle & panagia 1992 ; hamuy et al .
1995 ; 2000 ; reindl et al .
if sne with different characteristics ( e.g. , peak magnitudes ) occur in different environments ( old vs. young galaxies ) , then it is conceivable that different explosion mechanisms are involved , e.g. , one channel that produces the faint sne in old galaxies , and another that produces the bright sne in young galaxies .
furthermore , scannapieco & bildsten ( 2005 ) show that a two - component model with a large contribution of prompt sne can help explain the fe content of the inter - cluster medium in clusters of galaxies , as well as the [ o / fe ] abundance ratios in the galaxy .
while the snr we derived in dahlen et al .
( 2004 ) is consistent with other measurements and models with a large prompt component at @xmath8 , there is an apparent discrepancy at @xmath9 between the long delay - time ( @xmath10 gyr ) derived from the one - component fit to the goods data ( strolger et al .
2004 ) and models supporting a large prompt component , e.g. , scannapieco & bildsten ( 2005 ; hereafter sb05 ) and neill et al .
( 2006 ; hereafter n06 ) . the model presented by mannucci et al .
( 2007 ; hereafter m07 ) , which includes a larger fraction of the old component and a different treatment of dust extinction , appears to fit data better . in this paper , we make use of data from the continued monitoring of the goods fields within the pans program to increase the temporal coverage of the survey by a factor @xmath11 , allowing us to better constrain the high redshift snr . in particular , we investigate the proposed decline in snr at @xmath2 .
we organize this paper as follows : in 2 we describe our data and in 3 we present our results .
a discussion follows in 4 , while we in 5 focus on the bimodality of type is sne . in
6 we briefly discuss results at lower redshifts .
finally , conclusions and a summary are given in 7 .
in this investigation , we use the original goods data , consisting of one reference image and four search images in each of goods - north and goods - south obtained during @xmath0 cycle 11 ( riess et al . 2004 ) , together with the pans continued monitoring of during @xmath0 cycles 12 - 13 .
this added nine search images in goods - north and five search images in goods - s ( riess et al .
2007 ) . in total
, we have therefore searched 22 fields , each consisting of 15 pointings in the f850lp filters using @xmath0 acs . with each pointing
being @xmath1211 sq arcmin , _ we have in total searched an area of @xmath12one square degree with an average time separation of 45 days . _ in compliment to the f850lp search filter , observations with the f775w filter ( and f606w filter in cycle 11 ) added contemporaneous color information at all epochs .
we note here that @xmath0 has a number of unique features making it more reliable than recent ground - based surveys for detecting and confirming high redshift ( @xmath7 ) sne .
this includes high signal - to - noise detections made possible through low background , in particular in the reddest filters , and a high spatial resolution which is important for distinguishing sne from variable nuclei in their host galaxies .
furthermore , during this campaign we conducted a rolling search where the same two fields were observed repeatedly every 40 - 50 days , allowing multiple , homogeneous measurements of the sn lightcurves without interruption ( due to weather ) as well as the construction of an extremely deep reference image for the latter epochs .
finally , @xmath0 with the g800l grism on acs wfc has the capability to achieve high signal - to - noise spectra to @xmath121 micron to identify the redshift and type of sn ia up to @xmath13 .
this ability provided reliable identifications for the majority of our sne ia at the highest - redshift bin of our survey ( riess et al .
2007 ) , a feature unmatched in other surveys at this redshift range .
details on the observations and reduction techniques used are given in giavalisco et al .
( 2004 ) and strolger et .
al ( 2004 ) .
a total of 56 type ia sne were discovered during the full program , of which 23 belong to the original goods set , while 33 were discovered in the follow up program .
the sne are classified as `` gold '' ( 33 sne ) , `` silver '' ( 9 sne ) , or `` bronze '' ( 14 sne ) depending on the confidence of the classification .
a gold classification requires either a high s / n spectrum of the sn ( available for 30 of 33 ) , or a spectroscopic redshift of the host galaxy together with three additional requirements : 1 ) the host is of elliptical / early type , 2 ) the peak rest - frame @xmath14 color is consistent with a sn ia , and 3 ) the light - curve shape is consistent with a sn ia ( 3 of 33 are based on these criteria ) .
if two of the three criteria above are met , then the sn is considered a silver type ia . finally ,
if only the light - curve shape is consistent with a type ia , then a bronze confidence level is assigned .
while the bronze classification implies that the sne are most probably type ia , we can not rule out that a few may be interlopers , i.e. , misidentified core collapse sne .
this classification is further discussed in strolger et al .
spectroscopic redshift of either the sn or the host galaxy is available for 48 of the sne . for the remaining eight , we rely on photometric redshifts . at high redshift ( @xmath7 ) ,
our sample consists of 23 type ia . of these , we have spectroscopic redshift of the sn or the host galaxy for 21 objects . divided by classification , the high redshift sample consists of 15 gold , 5 silver , and 3 bronze type ia sne .
the sne used in this investigation are described in detail in riess et al .
( 2004 ; 2007 ) and strolger et al .
( 2004 , 2008 in prep . ) . for a full catalog of goods / pans
sne with type classification and positions , visit the goods / pans catalog web page .
additional information on this sample , including redshifts , will be published in strolger et al .
( 2008 , in prep . ) .
we derive the snr as described in detail in dahlen et al .
in short , we use monte carlo simulations to estimate the predicted number of detected sn for a search with the same observational setup as goods / pans , e.g. , field - of - view , detection limits and temporal spacing between observations . input to the simulations is a given intrinsic sn rate in units number of sne per year per mpc@xmath15 . taking into account the sn peak magnitude luminosity function , k - corrections , light - curve shape , light - curve stretch and extinction of the sn light due to dust in the host galaxy ( see 4.2 for details ) , we calculate the expected number of detected sne in each redshift interval under consideration .
we thereafter adjust the input rate so that the expected number of detected sne matches the actual number of sne found in the search . to account for the relative large redshift errors for the sne with only photometric redshifts available
, we use the redshift probability distribution derived from the photometric redshift fitting when distributing these sne in bins .
therefore , a single sne may contribute with fractional parts in more than one bin . in table
[ tab1 ] and figure [ fig1 ] , we show our resulting rates in four redshift bins @xmath16 , @xmath17 , @xmath18 , and @xmath19 . in the figure
, we show our results as black dots . in table
[ tab1 ] , we also give the raw number of type ia sne in each bin together with the number of sne after taking into account the redshift distribution of the sne with photometric redshift only .
error - bars for our points in figure [ fig1 ] only include statistical errors , however , in table [ tab1 ] , we also give systematic errors .
these include uncertainties in sn classifications .
to estimate the maximum contamination from core collapse sne , we assume that all bronze type ia sne have been misclassified . to estimate the maximum number of sne that we may have excluded due to misclassification
, we assume that all bronze cc sne detected in the survey are instead type ia sne .
the systematic effects also include uncertainties in the type ia sn luminosity function , k - corrections and extinction corrections .
details on how the systematic errors are derived are given in dahlen et al .
( 2004 ) . to investigate how the adopted binning affects results
, we also calculate rates using a window function of width @xmath20 , which we shift from redshift bin @xmath21 to @xmath22 .
results are shown with the solid line in figure [ fig1 ] .
it is clear from figure [ fig1 ] that the steep decrease in the snr at @xmath2 found in dahlen et al .
( 2004 ) is still present in our extended data set .
we also note that the suggested peak in the ia rate at @xmath30.8 from the old data is now not as pronounced , rather the rate seems fairly flat over the range @xmath23 .
in figure [ fig1 ] , we also include a compilation of rates taken from the literature ; cappellaro et al .
( 1999 ) , hardin et al .
( 2000 ) , strolger ( 2003 ) , reiss et al .
( 2000 ) , blanc et al .
( 2004 ) , pain et al .
( 1996 ; 2002 ) , tonry et al . ( 2003 ) , neill et al .
( 2006 ) , barris & tonry ( 2006 ) , kuznetsova et al .
( 2007 ) and poznanski et al .
( 2007 ) . considering the relatively large errors on these measurements
, there is a fairly good agreement between rates determined at similar redshifts . to make a quantitative measurement of how well the data from different surveys agree
, we fit to a third order polynomial using all data points in figure [ fig1 ] .
this results in a reduced chi - square @xmath24=2.6 , reflecting some of the existing discrepancies found in the rates derived by different authors .
in particular , the rates from barris & tonry at @xmath25 are a factor @xmath122 higher than other measurements at comparable redshifts . also , our rate at @xmath26 is almost twice the rate measured recently by neill et al .
( 2006 ) using the snls data , and three times the rates of poznanski et al .
( 2007 ) at @xmath27 .
furthermore , the rate at @xmath2 presented in poznanski et al . is about twice the rate found here .
we discuss these apparent discrepancies in 5 and 6 .
the best - fitting one - component model to the goods data has a gaussian delay time distribution with parameters @xmath28 gyr and @xmath29 ( for a definition of the delay - time distributions , see strolger et al .
this fit is shown in figure [ fig1 ] as the dotted line and in figure [ fig2 ] as the thick solid line .
the fit is largely consistent with the one - sigma errors of a number of rate measurements over the wide redshift range .
however , as previously indicated , there are several constraints that disfavor exceptionally long delay - times .
most prominently , the increased snr in low - redshift star forming galaxies suggests that at least some sn ia progenitors prefer channels which are more closely connected to ongoing star formation .
although the strolger et al .
( 2004 ) one - component delay - time model can not be ruled out ( with high confidence ) by the rate measurements alone , the low - redshift observations seemingly contradicts this simple one - parameter model which would decouple the snr from the ongoing sfr .
also plotted in figure [ fig2 ] are the two - component models from sb05 ( thin solid line ) and n06 ( dashed line ) .
these have the major contribution from a component that is directly proportional to the ongoing sfr , and a smaller component proportional to the old stellar component ( i.e. , the amount of accumulated mass ) . while these models are fairly consistent with measurements at @xmath30 , they differ significantly at @xmath31 ( @xmath32 ) .
the low measured rate at @xmath33 is therefore difficult to reconcile with the two - component models of sb05 and n06 .
m07 calculate the expected snr from a two - component model using a different approach compared to sb05 and n06 .
m07 include in the total sfr also the sfr that is hidden to optical searches due to dust extinction , i.e. , contributions from faint ir galaxies ( firgs ) , luminous ir galaxies ( lirgs ) and ultra - luminous ir galaxies ( ulirgs ) .
taking into account that also a large fraction of the type ia sne should be hidden in these galaxies , they find an observable type ia snr that turns over at @xmath34 .
we show this model as the dotted line in figure [ fig2 ] .
this model shows a better agreement at the highest redshift point , while slightly under - predicting the rate at mid redshifts .
below , we further discuss these different models together with other effects that may affect the comparisons between observations and models , including statistical fluctuations , clustering variance , detection efficiency , dust extinction , and evolving wd explosion efficiency . adopting the two - component model rates from sb05 and n06
, we find that 16 and 10 sne should be detected in our highest redshift bin using the detection efficiency adopted here .
this is significantly more than the three sne detected . using statistical error , we find that the probability is @xmath35 of finding the predicted number from sb05 , and @xmath36 of finding the number predicted by n06 .
therefore , it seems unlikely that the difference is due to statistical fluctuations alone , especially compared to the rate in sb05 .
compared to n06 , the difference is not ruled out with high confidence , but the difference is high enough that alternative explanations have to be investigated . both sb05 and n06 models predict that @xmath3790% of the snr comes from the prompt component at @xmath31.4 .
the m07 model on the other hand , only assumes @xmath1250% , which together with the different assumptions about sfr and dust extinction leads to a lower number of expected high redshift sne .
thus , in the highest redshift bin , the m07 model predicts 6 sne which is consistent with our measurement within one sigma .
fitting the models to all data points using statistical errors results in reduced chi - square @xmath24=7.6 , 4.0 , and 2.1 for the sb05 , n06 , and m07 models , respectively .
again , this shows that deviations are fairly large compared to expected statistical variations , with exception for the m07 model , which is marginally consistent with data . clustering variance could affect the derived high redshift rates if there is a significant under - density in the observed fields .
using two widely separates field , goods - s and goods - n , should help to decrease these effects .
also , the star formation rate density measured in goods - s ( dahlen et al .
2007 ) to @xmath38 is consistent with measurements in the same redshift range obtained for other fields .
this implies that we should not be missing part of the prompt component of the snr due to an under - density of star forming galaxies .
since the number counts of galaxies in goods - n and goods - s are comparable ( goods - n slightly higher ) , we also do nt expect the star formation rate in goods - n to be under - dense and thus affecting the number of detected sne .
assumptions on the amount of dust extinction in the sn host galaxies also affects the derived sn rates . for example , a high extinction prior in the models results in a high derived sn rate , and vice verse . in this investigation
, we have followed the prescription in hatano et al .
( 1998 ) to calculate the extinction distribution .
we hereafter refer to this as model a. we note that the hatano et al .
model was derived for estimating extinction in local hosts and may therefore not be representative at high redshifts . to investigate the effect of assumed extinction distribution models
, we have also considered the models in riello & patat ( 2005 ) and n06 .
the models in riello & patat ( 2005 ) use the same mc technique as hatano et al .
( 1998 ) , but include a more detailed approach aimed at being applicable also to non - local hosts . in particular ,
they include a tail of high extinctions for highly inclined host galaxies . here
we follow the approach in n06 , who show that at low inclinations ( @xmath39 ) , the riello & patat models can be approximated by a positive valued gaussian ( described below ) .
for high inclinations , we assign a random extinction value according to e(b v)=-ln@xmath40 , where @xmath41 is a random number between 0 and 1 and @xmath42 describes how fast the high extinction tail falls of . here
we use @xmath42=5 at inclinations @xmath43 and @xmath42= 3 at @xmath44 . finally , to make the distribution of @xmath45 consistent with figure 2 in riello & patat ( 2005 ) , we adjust the model so that @xmath1235% of the distribution falls within @xmath46 .
we call the model based on the riello & patat distributions model b. in n06 , the distribution of e(b v ) is given by a positive valued gaussian with @xmath47=0.2 , @xmath48=3.1 and an extinction law following cardelli et al .
we include this parameterization as model c. the three different distributions are shown in figure [ fig3 ] .
the main differences between the distributions are that model c lacks the high extinction tail ( at @xmath49 ) and has significantly less power at the low extinction end , i.e , this model has @xmath1213% with @xmath46 , compared to @xmath1235% for both model a and model b. as a consequence , model c has higher power at intermediate absorptions ( @xmath50 ) .
model a and model b are more similar , both with a significant peak at the lowest absorptions , but with model a having higher power at the high extinction tail ( @xmath51 ) , whereas model b is dominating at intermediate absorptions . in the left panel of figure [ fig4 ]
, we show our rates using the nominal assumption about host galaxy extinction distribution , i.e. , model a , together with the alternative distributions derived from model b and model c. we also show results in the case where no dust extinction is included . assuming model b
, we expect a lower average extinction ( see figure [ fig3 ] ) and therefore lower derived rates .
we find that the model b produces rates that are @xmath5210% lower compared to model a. in particular , the rates in the highest redshift bin decreases by @xmath124% assuming model b. model c affects the derived rates in a different way so that the rates decrease at low redshift and increase at high redshift .
for the highest bin , we find that the derived rate increases by @xmath127% assuming model c. it is surprising to note that the resultant snr shows this little variation with assumed extinction model .
if fact , the only notable exception is for no assumed host galaxy extinction , most notably in the highest redshift bin at @xmath53 .
the low rate of sne ia in the high redshift bin is therefore a robust conclusion for the viable regime of assumed host galaxy dust models . besides the distribution of extinction within the host galaxies studied here
, a change in the mean extinction in host galaxies with redshift may also affect results .
we discuss this more in 5 .
the detection efficiency of a survey is a crucial ingredient when deriving sn rates using both mc simulations and the control time method . assuming that a survey reaches deeper than is the actual case ,
will underestimate derived rates . in the goods survey we characterized the detection efficiency with a function @xmath54 where @xmath55 is the detection magnitude ( difference between detection and reference epoch ) and
@xmath56 gives the magnitude where the detection efficiency drops to 50% .
the parameter @xmath57 gives the slope of the roll off in the detection efficiency .
for the goods part of the survey we used two methods to determine the detection efficiency ( strolger et al .
first , fake sn were added to the actual images that were being searched and the detection efficiency was derived from how many of the input sne were recovered as a function of magnitude . in the second method , we used mc simulations and automated detection algorithms to derive the efficiency . combining results we derived @xmath56=25.94 and @xmath57=0.38 ( see fig . 9 in strolger
et al . 2004 ) . in the pans part of the survey , the exposure time of the monitoring images
was reduced by 20% ( from 2000s to 1600s ) . however , in the goods part , we used single epoch reference frames in the subtraction , while we in the latter pans part of the survey added multiple templates to the reference images .
this increased their depth , adding less noise signal when subtracting the reference image from the detection image . to quantify this
, we use the acs exposure time calculator and calculate the s / n for a faint source ( @xmath58 ) making two different assumptions .
first , we calculate the signal for the goods exposure time ( 2000s ) and the associated noise for a single epoch . to get the total noise in the subtracted image
, we add back - ground , dark , and readout noise from detection image and reference image in quadrature to the source noise from the detection image .
this gives us the estimated s / n in the subtracted image .
second , for the pans search , we make the same calculation but with the signal given by a 1600s exposure and where we decrease the noise in the reference image by a factor sqrt(5 ) , since at least five images were added when creating the reference image .
we thereafter calculate the corresponding s / n in the subtracted image .
we find that the pans part of the survey results in similar ( slightly higher ) s / n compared to the goods part .
this suggests that we reach at least as deep in the pans part of the survey and should not be missing sne despite the somewhat shorter exposure time .
we can therefore use the same detection efficiency for the whole program . in figure [ fig5 ]
, we show the ( f775w - f850lp ) color as a function of detection magnitude in the f850lp filter for the full sample of sne ( type ia and cc ) . in the figure , we also plot the color - magnitude relation for a type ia sn observed near the peak on the lightcurve .
the relation shows that as the type ia becomes more distant , it becomes both fainter and redder .
the figure clearly illustrates that we are able to detect a number of sne down to the detection limit at @xmath59 .
however , the figure also shows that the majority of the faint objects with @xmath60 have colors f775w - [email protected] , which is too blue to be consistent with high redshift type ia sne . based on colors ( in combination with redshift and lightcurve fitting ) , we find that these objects are almost exclusively cc sne at @xmath8 .
to evaluate how the derived rates depend on a correct determination of the detection efficiency , we recalculate our rates after assuming that we reach the 50% detection efficiency 0.2 mag brighter than the limit derived in strolger et al .
( 2004 ) . as a test case
, we also give rates assuming a 0.5 mag brighter cutoff .
the right panel of figure [ fig4 ] shows how the derived rates change if we assume magnitude limits that are brighter than the nominal values .
black dots are nominal rates , while red and blue dots show the cases where we have assumed a brighter cutoff by 0.2 and 0.5 mag compared to @xmath56=25.94 .
since the highest redshift bin will contain mostly faint sne , we expect the effect of changing the efficiency to be largest here , which is consistent with what the figure shows .
the change in rate is , however , relatively small . adopting a 0.2 mag brighter cutoff ( @xmath56=25.74 ) only increases the rate by 13% in the highest redshift bin .
even with a detection limit 0.5 brighter ( @xmath56=25.44 ) , the trend of a strong decrease in the rates at @xmath2 remains .
the investigation above shows that the drop in rate of sne ia in the high redshift bin is robust for viable detection efficiencies .
additional support that we are not missing a number of high redshift sne due to overestimating the detection efficiency comes from kuznetsova et al .
they independently searched the goods cycle 11 data - set and four additional epochs in goods - n collected during spring - summer 2004 ( which are also included in our investigation ) .
no additional high redshift sne were found by kuznetsova et al .
, in fact , their @xmath2 sample includes the same three sne as our sample .
however , due to different selection criteria , only one of high redshift sne is included in the sample used by kuznetsova et al . to calculate the snr .
figure [ fig1 ] shows that the rate calculated by kuznetsova et al .
( 2007 ) decreases at high redshift similar to what we found here , although the error bars are larger so that the decrease is not as significant .
furhermore , rodney & tonry ( 2007 ) have also searched the goods fields without finding any additional sn candidates at @xmath61 1 . both the one - component and the two - component models rely on one strict assumption : that the wd explosion efficiency , i.e. , the fraction of all wds that explode as sn , stays at a fixed value at all redshifts . if this assumption is relaxed , the two - component model can be made consistent with a drop in the rate at @xmath2 by assuming that a decreasing fraction of wds explode as type ia sne at higher redshift .
the models by kobayashi et al .
( 1998 ) suggest that the metallicity must reach [ fe / h]@xmath62 in order for the accretion onto the wd progenitor to be efficient enough to result in a type ia explosion . with the overall metallicity decreasing at higher
redshift , kobayashi et al . predicted a cutoff in the type ia explosion at @xmath2 .
refined models by nomoto et al .
( 2000 ) suggests a higher cutoff at @xmath63 in spiral galaxies and at @xmath64 in elliptical galaxies .
recent models by kobayashi & nomoto ( 2008 ) , based on the single degenerate scenario , suggest that the type ia rate should be low in spiral galaxies at @xmath65 due to the lower metallicity in these systems .
this model leads to a scenario where the type ia rate peaks at @xmath66 and thereafter declines , consistent with what our results suggests .
the new measurements of the type ia snr to @xmath67 presented here are very similar to our previous results derived only from the first - year sample . based solely from the dearth of events at @xmath2 in this complete sample
, it would be difficult to conclude that there could be strong bimodality in the delay time distribution with a dominating prompt component .
instead , our investigation suggests that a major fraction of the type ia sne should have a long characteristic delay time in the order of @xmath68 gyr .
therefore , our results are difficult to reconcile with the models of sb05 and n06 , which find that @xmath3790% of the type ia rate at @xmath2 should be prompt . in more agreement with our results
is the bimodal model of mannucci et al .
( 2006 ; 2007 ) , which suggests that the old component contributes @xmath69% to the type ia sne events . until recently , only @xmath0 surveys could provide insight on the redshift regime critical to this conclusion .
the clear advantage has been the remarkable sensitivity and unparalleled resolution that @xmath0 + acs provides , thusly producing the deepest and most complete surveys for supernovae to date .
however , recent ground - based `` ultra - deep '' surveys are beginning to close the gap in surveyable depth . the subaru deep field ( sdf )
has collaterally produced a deep supernova survey that is comparable to goods ( albeit with much poorer spatial resolution due to atmospheric effects and yet only a one - epoch search ) , with a @xmath70-band limit of approximately @xmath71 mag ( ab ) .
the results from their initial sample ( poznanski et al .
2007 ) show a appreciable similarity to the goods first - year sample ( when corrected for differences in effective volume and control time of the surveys ) in the redshift distribution of sne ia , including the clear reduction in the number of sne ia above @xmath72 .
their analysis of sdf survey detection efficiency has likewise concluded that this reduction can not be explained as simply a loss in sensitivity to sne ia at these redshifts .
still , there are notable differences in the rates derived from these similar samples .
foremost , poznanski et al .
( 2007 ) measure of an appreciably higher ( factor @xmath122 ) type ia snr in the highest redshift bins ( at @xmath73 ) than is presented in this manuscript . however , of the three type ia sne in the highest redshift bin in poznanski et al .
, two have large uncertainties in both type determination and redshift .
this is reflected in the large error - bars given by poznanski et al . , which makes their rate consistent with the rate found here .
but despite the uncertainty of these few supernovae , it is encouraging the sdf supernova survey will continue to gather a more definitive measure of the snr at @xmath2 less affected by these individual uncertainties .
we conclude that at present , the observed sn rates at high redshift can not alone rule out a scenario where the delay time distribution has a large contribution from an old ( @xmath68 gyr ) population .
there are , however , a number of results that show strong support for a bimodal model with a significant fraction of the type ia events coming from a prompt component , as discussed in 1 .
it is therefore desirable to try to reconcile the observed rate with the bimodal model .
as shown in 4 and figure [ fig2 ] , the models by m07 gives a better fit to the data compared to the other bimodal models .
there are two main reasons for this better fit .
first , m07 assumes a significantly larger fraction from the old type ia progenitor component compared to the other models and second , m07 use different assumptions about the cosmic sfr and dust extinction .
when we in 4 calculated the best - fitting one - component model for our data , we used the same assumptions about dust extinction in the population of galaxies that produced the uv - luminosity ( from which the sfr was calculated ) and that produced the sne .
in m07 , a large fraction of the sfr is assumed to be hidden in dusty environments and will be missed in optical searches ( we refer to this as the ir part of the sfr ) .
the total sfr used by these authors therefore has a different shape compared to sfrs based on uv luminosities only ( uv sfr ) . at the same time , most of the sne occurring in these obscured galaxies will also be missed from sn searches . in order to compare our results with this model
, we use the sfr in m07 , taking into account the missing sfr as well as the expected fraction of missed type ia sne as given in m07 . making a fit to the data ,
we find again that the narrow gaussian model gives the best fit , but now with a characteristic delay time of @xmath74 gyr , compared to @xmath75 gyr for our original model . for a wide gaussian delay time model ,
our model gives @xmath76 gyr , while the m07 model gives @xmath77 gyr .
investigating these differences in some detail , we find that part of the differences is due to the hidden ir sfr , but also that a significant fraction of the change is due to different assumptions on the shape of the uv sfr .
if we use the ir sfr given by m07 in combination with the uv sfr adopted in this paper ( taken from giavalisco et al .
2004 ) , we find that even though we do include the missing fraction of the sfr and the snr , the delay times do not change considerably . for the best - fitting narrow gaussian model we find @xmath78 gyr ( compared to our original result @xmath75 gyr ) , while for the wide gaussian we find @xmath79 gyr ( @xmath76 gyr ) .
this highlights an additional problem for deriving the delay time function , as long as there are significant uncertainties in the evolution of the cosmic sfr , the derivation of the delay time function will be hampered .
we find that taking missing fraction of sfr and snr into account , according to the m07 model , leads to a slightly lower characteristic delay time , but that this result is somewhat dependent of the assumption of the shape of the sfr .
the characteristic delay - time we find is in the order of a few gyr , implying that a significant fraction of the snr should be delayed and therefore decoupled from the ongoing sfr .
we also note that in a recent paper , botticella et al .
( 2008 ) studied the evolution of both core collapse and type ia sne to @xmath3 0.20.3 . from the stronger evolution in the core collapse sn rate they found , they conclude that a significant fraction of the type ia progenitors has a delay time 23 gyr , consistent with what found here .
besides at high redshift , there are other notable differences between the goods and other results . in the rate measurement within the two bins spanning the range of approximately @xmath80 , poznanski et al .
( 2007 ) have found roughly half the type ia supernovae near @xmath34 as was seen in the goods , and essentially none at lower-@xmath81 . given the single search epoch and acute solid angle surveyed , it is not unlikely for the sdf survey to have a null result at lower redshifts .
however , the differences near @xmath82 seem appreciable .
it serves as another example of a growing problem ; the relatively large discrepancies in rate measurements by several authors in this redshift range ( see figure [ fig1 ] ) .
it is disappointing that there does not as yet appear to be a resolution in these measured discrepancies , as there could be great power in combining rate results to gain further insight on the modality of the sn ia delay - time distributions , and ultimately the progenitors mechanisms of type ia supernovae .
but until then , this analysis will remain necessarily limited .
here we present new measurements of the type ia snr to @xmath83 .
similar to our previous results based on a smaller sample , these observations show a decrease in the snr at redshifts @xmath84 , with a high significance .
the results are consistent with a characteristic delay time in the order of @xmath85 gyr .
recent two - component models for the type ia snr , with one dominating prompt and one less prominent delayed channel , seems to fit low redshift snr data well .
these models are also consistent with the findings that the snr is higher in galaxies with a higher star formation , and they may also explain the fe content of the inter - cluster medium in clusters of galaxies . however , these two - component models predicts rates at @xmath2 that deviates from the measured rates from this investigation . here
we have discussed possible solutions to this discrepancy and found : * it is unlikely that the difference between model predicted rates and observed rates is due to statistical fluctuations or cosmic variance . * it is also unlikely that the low rate we measure is due to an underestimate of the host galaxy dust extinction or an overestimate detection efficiency . * a bimodal model with a larger fraction of delayed type ia and that takes into account sfr hidden by dust results in a better fit to data . *
another possible scenario that would result in a decrease in the high redshift snr is if the wd explosion efficiency deceases at high redshift .
barris , b. j. & tonry , j. l. 2006 , apj , 637 , 427 blanc , g. , et al .
2004 , a&a , 423,881 botticella , m. t. , et al .
2008 , a&a , 479 , 49 branch , d. & van en bergh , s. 1993 , aj , 105 , 2231 cappellaro , e. , evans , r. , & turatto , m. 1999 , a&a , 351 , 459 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , apj , 345 , 245 dahlen , t. , & fransson , c. 1999 , a&a , 350 , 349 dahlen , t. , et al .
2004 , apj , 613 , 189 dahlen t. , mobasher , b. , dickinson , m. , ferguson , h. c. , giavalisco , m. , kretchmer , c. , ravindranath , s. 2007 , apj , 654 , 172 della valle , m. , & panagia , n. 1992 , aj , 104 , 696 filippenko , a. v. , 1989 .
pasp , 101 , 588 giavalisco , m. , et al .
2004 , apj , 600 , l93 greggio , l. , 2005 .
a&a , 441 , 1055 hamuy , m. , phillips , m. m. , maza , j. , suntzeff , n. b. , schommer , r. a. , avils , r. 1995 , aj , 109 , 1 hamuy , m. , trager , s. c. , pinto , p. a. , phillips , m. m. , schommer , r. a. , ivanov , v. , suntzeff , n. b. 2000 , aj , 120 , 1479 hardin , d. , et al . 2000 ,
a&a , 362 , 419 hatano , k. , branch , d. , & deaton , j. 1998 , apj , 502 , 177 kobayashi , c. & nomoto , k. 2008 , apj , submitted kobayashi , c. , tsujimoto , t. , nomoto , k. , hachisu , i. , & kato , m. 1998 , apj , 503 , l155 kuznetsova , n. , et al .
2007 , apj , in press madau , p. , della valle , m. , & panagia , n. 1998 , mnras , 297 , l17 mannucci , f. , della valle , m. , panagia , n. , cappellaro , e. , cresci , g. , maiolino , r. , petrosian , a. , & turatto , m. 2005 , a&a , 433 , 807 mannucci , f. , della valle , m. , & panagia , n. 2006 , mnras , 370 , 773 mannucci , f. , della valle , m. , & panagia , n. 2007 , mnras , 377 , 1229 ( m07 ) neill , j. d. , et al . 2006 , aj , 132 , 1126 ( n06 ) nomoto , k. , umeda , h. , kobayashi , c. , hachisu , i. , kato , m. , & tsujimoto , t. 2000 , in aip conf .
522 , cosmic explosions , eds .
s. holt & w. zhang ( college park : aip ) , 35 pain , r. , et al .
2002 , apj , 577 , 120 pain , r. , et al .
1996 , apj , 473 , 356 poznanski , d. , et al .
2007 , mnras , 382 , 1485 reindl , b. , tammann , g. a. , sandage , a. , saha , a. 2005 , apj , 624 , 532 reiss , d.j .
2000 , phd thesis , university of washington riello , m. & patat , f. 2005 , mnras , 362 , 671 riess , a.g . ,
2004 , apj , 607 , 665 riess , a.g . ,
2007 , apj , 659 , 98 rodney , s. a. & tonry , j. l. 2007 , baas , 39 , p. 807
scannapieco , e. & bildsten , l. 2005 , apj , 629 , l85 ( sb05 ) strolger , l. g. 2003 , phd thesis , university of michigan strolger , l .- g . , et al .
apj , 613 , 200 sullivan , m. , et al .
apj , 648 , 868 tonry , j. l. , et al .
2003 , apj , 594 , 1 yungelson , l. r. , & livio , m. 2000 , apj , 528 , 108 lcccccc @xmath86 & 0.47 & 0.80 & @xmath87 & @xmath88 & 8 & 8.8 + @xmath89 & 0.83 & 1.30 & @xmath90 & @xmath91 & 25 & 23.5 + @xmath92 & 1.21 & 1.32 & @xmath93 & @xmath94 & 20 & 20.2 + @xmath95 & 1.61 & 0.42 & @xmath96 & @xmath97 & 3 & 3.1 [ tab1 ] | we use the @xmath0 acs imaging of the two goods fields during cycles 11 , 12 , and 13 to derive the type ia supernova rate in four redshift intervals in the range @xmath1 . compared to our previous results based on cycle 11 observations only
, we have increased the coverage of the search by a factor 2.7 , _ making the total area searched equivalent to one square degree_. the sample now consists of 56 type ia supernovae .
the rates we derive are consistent with our results based on the cycle 11 observations
. in particular , the few detected supernovae at @xmath2 supports our previous result that there is a drop in the type ia supernova rate at high redshift , suggesting a long time delay between the formation of the progenitor star and the explosion of the supernova .
if described by a simple one - parameter model , we find a characteristic delay time of 23 gyr .
however , a number of recent results based on e.g. , low redshift supernova samples and supernova host galaxy properties suggest that the supernova delay time distribution is bimodal . in this model ,
a major fraction of the type ia supernova rate is prompt and follows the star formation rate , while a smaller fraction of the rate has a long delay time , making this channel proportional to mass . while our results are fully consistent with the bimodal model at low redshifts , the low rate we find at @xmath2 appears to contradict these results .
models that corrects for star formation hidden by dust may explain at least part of the differences . here we discuss this possibility together with other ways to reconcile data and models . |
as an approximation to control problems for critically - loaded stochastic networks , harrison ( in @xcite , see also @xcite ) has formulated a stochastic control problem in which the state process is driven by a multidimensional brownian motion along with an additive control that satisfies certain feasibility and nonnegativity constraints .
this control problem , that is , usually referred to as the _ brownian control problem _ ( bcp ) has been one of the key developments in the heavy traffic theory of controlled stochastic processing networks ( spn ) .
bcps can be regarded as formal scaling limits for a broad range of scheduling and sequencing control problems for multiclass queuing networks .
finding optimal ( or even near - optimal ) control policies for such networks which may have quite general non - markovian primitives , multiple server capabilities and rather complex routing geometry is in general prohibitive . in that regard ,
bcps that provide significantly more tractable approximate models are very useful . in this diffusion approximation approach to policy synthesis , one first finds an optimal ( or near - optimal ) control for the bcp which is then suitably interpreted to construct a scheduling policy for the underlying physical network . in recent years
there have been many works @xcite that consider specific network models for which the associated bcp is explicitly solvable ( i.e. , an optimal control process can be written as a known function of the driving brownian motions ) and , by suitably adapting the solution to the underlying network , construct control policies that are asymptotically ( in the heavy traffic limit ) optimal .
the paper @xcite also carries out a similar program for the crisscross network where the state
space is three dimensional , although an explicit solution for the bcp here is not available .
although now there are several papers which establish a rigorous connection between a network control problem and its associated bcp by exploiting the explicit form of the solution of the latter , a systematic theory which justifies the use of bcps as approximate models has been missing . in a recent work @xcite it was shown that for a large family of _ unitary networks _
( following terminology of @xcite , these are networks with a structure as described in section [ secsetup ] ) , with general interarrival and service times , probabilistic routing and an infinite horizon discounted linear holding cost , the cost associated with any admissible control policy for the network is asymptotically , in the heavy traffic limit , bounded below by the value function of the bcp .
this inequality , which provides a useful bound on the best achievable asymptotic performance for an admissible control policy , was a key step in developing a rigorous general theory relating bcps with spn in heavy traffic .
the current paper is devoted to the proof of the reverse inequality .
the network model is required to satisfy assumptions made in @xcite ( these are summarized above theorem [ ab937 ] ) .
in addition , we impose a nondegeneracy condition ( assumption [ non - deg ] ) , a condition on the underlying renewal processes regarding probabilities of deviations from the mean ( assumption [ ldp ] ) and regularity of a certain skorohod map ( assumption [ ab1012 ] ) ( see next paragraph for a discussion of these conditions ) . under these assumptions
we prove that the value function of the bcp is bounded below by the heavy traffic limit ( limsup ) of the value functions of the network control problem ( theorem [ main1016 ] ) . combining this with the result obtained in @xcite ( see theorem [ ab937 ] )
, we obtain the main result of the paper ( theorem [ maincorr ] ) .
this theorem says that , under broad conditions , the value function of the network control problem converges to that of the bcp .
this result provides , under general conditions , a rigorous basis for regarding bcps as approximate models for critically loaded stochastic networks .
conditions imposed in this paper allow for a wide range of spn models .
some such models , whose description is taken from @xcite , are discussed in detail in examples [ exammodel](a)(c ) .
we note that our approach does not require the bcp to be explicitly solvable and the result covers many settings where explicit solutions are unavailable . most of the conditions that we impose are quite standard and we only comment here on three of them : assumptions [ ab148 ] , [ g - column ] and [ ab1012 ] .
assumption [ ab148 ] says that each buffer is processed by at least one basic activity ( see remark [ rem - ht ] ) .
this condition , which was introduced in @xcite , is fundamental for our analysis .
in fact , @xcite has shown that without this assumption even the existence of a nonnegative workload matrix may fail .
assumption [ g - column ] is a natural condition on the geometry of the underlying network . roughly speaking
, it says that a nonzero control action leads to a nonzero state displacement .
assumption [ ab1012 ] is the third key requirement in this work .
it says that the skorohod problem associated with a certain reflection matrix @xmath0 [ see equation ( [ refmat ] ) for the definition of @xmath0 ] is well posed and the associated skorohod map is lipschitz continuous . as example
[ exammodel ] discusses , this condition holds for a broad family of networks ( including _ all _ multiclass open queuing networks , as well as a large family of parallel server networks and job - shop networks ) .
the papers @xcite noted earlier , that treat the setting of explicitly solvable bcp , do much more than establish convergence of value functions .
in particular , these works give an explicit implementable control policy for the underlying network that is asymptotically optimal in the heavy traffic limit . in the generality treated in the current work , giving explicit recipes ( e.g. , threshold type policies ) is unfeasible , however , the policy sequence constructed in section [ subconstruct ] suggests a general approach for building near asymptotically optimal policies for the network _ given _ a near optimal control for the bcp .
obtaining near optimal controls for the bcp in general requires numerical approaches ( see , e.g. , @xcite ) , discussion of which is beyond the scope of the current work .
we now briefly describe some of the ideas in the proof of the main result
theorem [ main1016 ] .
we begin by choosing , for an arbitrary @xmath2 , a suitable @xmath3-optimal control @xmath4 for the bcp and then , using @xmath4 , construct a sequence of control policies @xmath5 for the network model such that the ( suitably scaled ) cost associated with @xmath6 converges to that associated with @xmath4 , as @xmath7 .
this yields the desired reverse inequality .
one of the key difficulties is in the translation of a given control for the bcp to that for the physical network .
indeed , a ( near ) optimal control for the bcp can be a very general adapted process with rcll paths . without additional information on such a stochastic process
, it is not at all clear how one adapts and applies it to a given network model .
a control policy for the network needs to specify how each server distributes its effort among various job classes at any given time instant . by a series of approximations
we show that one can find a rather simple @xmath3-optimal control for the bcp , that is , easy to interpret and implement on a network control problem . as a first step , using pde characterization results for general singular control problems with state constraints from @xcite ( these , in particular , make use of the nondegeneracy assumption assumption [ non - deg ] ) , one can argue that a near - optimal control can be taken to be adapted to the driving brownian motion and be further assumed to have moments that are subexponential in the time variable ( see lemma [ cont - bcp ] ) . using results from @xcite , one can perturb this control so that it has continuous sample paths without significantly affecting the cost .
next , using ideas developed by kushner and martins @xcite in the context of a two - dimensional bcp , one can further approximate such a control by a process with a fixed ( nonrandom ) finite number of jumps that take values in a finite set . two main requirements ( in addition to the usual adaptedness condition ) for such a process to be an admissible control of a bcp ( see definition [ bcp ] ) are the nonnegativity constraints ( [ ab934 ] ) and state constraints ( [ ab440 ] ) .
it is relatively easy to construct a pure jump process that satisfies the first requirement of admissibility , namely , the nonnegativity constraints , however , the nondegenerate brownian motion in the dynamics rules out the satisfaction of the second requirement , that is , state constraints , without additional modifications .
this is where the regularity assumption on a certain skorohod map ( assumption [ ab1012 ] ) is used .
the pure jump control is modified in a manner such that in between successive jumps one uses the skorohod map to employ minimal control needed in order to respect state constraints .
regularity of the skorohod problem ensures that this modification does not change the associated cost much .
the skorohod map also plays a key role in the weak convergence arguments used to prove convergence of costs .
the above construction is the essential content of theorem [ main - bcp ] .
the @xmath3-optimal control that we use for the construction of the policy sequence requires two additional modifications [ see part ( iii ) of theorem [ main - bcp ] and below ( [ 111 ] ) ] which facilitate adapting such a control for the underlying physical network and in some weak convergence proofs , but we leave that discussion for later in section [ nearopt ] ( see remark [ on3 ] and above theorem [ newmain518 ] ) . using a near - optimal control @xmath4 of the form given in section [ nearopt ] ( cf .
theorem [ newmain518 ] ) , we then proceed to construct a sequence of policies @xmath8 for the underlying network .
the key relation that enables translation of @xmath4 into @xmath8 is ( [ y - defn ] ) using which one can loosely interpret @xmath9 as the asymptotic deviation , with suitable scaling , of @xmath10 from the nominal allocation @xmath11 ( see definition [ defn - ht ] for the definition of nominal allocation vector ) .
recall that @xmath4 is constructed by modifying , through a skorohod constraining mechanism , a pure jump process ( say , @xmath12 ) .
in particular , @xmath4 has sample paths that are , in general , discontinuous . on the other hand , note that an admissible policy @xmath6 is required to be a lipschitz function ( see remark [ rem921 ] ) .
this suggests the following construction for @xmath6 . over time periods ( say , @xmath13 ) of constancy of @xmath12
one should use the nominal allocation ( i.e. , @xmath14 ) , while jump - instants should be stretched into periods of length of order @xmath15 ( note that in the scaled network , time is accelerated by a factor of @xmath16 and so such periods translate to intervals of length @xmath17 in the scaled evolution and thus are negligible ) over which a nontrivial control action is employed as dictated by the jump vector ( see figure [ policyfig ] for a more complete description ) .
this is analogous to the idea of a discrete review policy proposed by harrison @xcite ( see also @xcite and references therein ) .
there are some obvious difficulties with the above prescription , for example , a nominal allocation corresponds to the average behavior of the system and for a given realization is feasible only when the buffers are nonempty .
thus , one needs to modify the above construction to incorporate idleness , that is , caused due to empty buffers .
the effect of such a modification is , of course , very similar to that of a skorohod constraining mechanism and it is tempting to hope that the deviation process corresponding to this modified policy converges to @xmath4 ( in an appropriate sense ) , as @xmath7 .
however , without further modifications , it is not obvious that the reflection terms that are produced from the idling periods under this policy are asymptotically consistent with those obtained from the skorohod constraining mechanism applied to ( the state process corresponding to ) @xmath12 .
the additional modification [ see ( [ ab543 ] ) ] that we make roughly says that jobs are processed from a given buffer over a small interval @xmath18 , only if at the beginning of this interval there are a `` sufficient '' number of jobs in the buffer .
this idea of safety stocks is not new and has been used in previous works ( see , e.g. , @xcite ) .
the modification , of course , introduces a somewhat nonintuitive idleness even when there are jobs that require processing .
however , the analysis of section [ secproof ] shows that this idleness does not significantly affect the asymptotic cost .
the above very rough sketch of construction of @xmath6 is made precise in section [ subconstruct ] .
the rest of the paper is devoted to showing that the cost associated with @xmath6 converges to that associated with @xmath4 .
it is unreasonable to expect convergence of controls ( e.g. , with the usual skorohod topology)in particular , note that @xmath6 has lipschitz paths for every @xmath15 while @xmath4 is a ( modification of ) a pure jump process
however , one finds that the convergence of costs holds .
this convergence proof , and the related weak convergence analysis , is carried out in sections [ secconvprf ] and [ weakcgce ] .
the paper is organized as follows .
section [ secsetup ] describes the network structure , all the associated stochastic processes and the heavy - traffic assumptions as well as the other assumptions of the paper .
the section also presents the spn control problem , that is , considered here , along with the main result of the paper ( theorem [ maincorr ] ) .
section [ nearopt ] constructs ( see theorem [ newmain518 ] ) a near - optimal control policy for the bcp which can be suitably adapted to the network control problem . in section [ secproof ] the near - optimal control policy from section [ nearopt ]
is used to obtain a sequence of admissible control policies for the scaled spn .
the main result of the section is theorem [ mainweak ] , which establishes weak convergence of various scaled processes . convergence of costs ( i.e. , theorem [ jrtoj ] ) is an immediate consequence of this weak convergence result .
theorem [ maincorr ] then follows on combining theorem [ jrtoj ] with results of @xcite ( stated as theorem [ ab937 ] in the current work ) .
finally , the collects proofs of some auxiliary results . the following notation will be used .
the space of reals ( nonnegative reals ) , positive ( nonnegative ) integers will be denoted by @xmath19 ( @xmath20 ) , @xmath21 ( @xmath22 ) , respectively . for @xmath23 and @xmath24
@xmath25 $ ] will denote the space of continuous functions from @xmath26 ( resp . @xmath27 $ ] ) to @xmath28 with the topology of uniform convergence on compacts ( resp .
uniform convergence ) . also , @xmath29 $ ] will denote the space of right continuous functions with left limits , from @xmath30 ( resp . @xmath27 $ ] ) to @xmath28 with the usual skorohod topology . for @xmath31 and @xmath32 , we write @xmath33 and @xmath34 , where for @xmath35 , @xmath36 all vector inequalities are to be interpreted component - wise .
we will call a function @xmath37 nonnegative if @xmath38 for all @xmath39 .
a function @xmath37 is called nondecreasing if it is nondecreasing in each component . all ( stochastic ) processes in this work will have sample paths that are right continuous and have left limits , and thus can be regarded as @xmath40-valued random variables with a suitable @xmath41 . for a polish space @xmath42 , @xmath43 will denote the corresponding borel sigma - field .
weak convergence of @xmath44 valued random variables @xmath45 to @xmath46 will be denoted as @xmath47 .
sequence of processes @xmath48 is tight if and only if the measures induced by @xmath45 s on @xmath49 form a tight sequence . a sequence of processes with paths in @xmath40 ( @xmath23 ) is called @xmath50-tight if it is tight in @xmath40 and any weak limit point of the sequence has paths in @xmath51 almost surely ( a.s . ) . for processes @xmath48 , @xmath46 defined on a common probability space , we say that @xmath45 converge to @xmath46 , uniformly on compact time intervals ( u.o.c . ) , in probability ( a.s . ) if for all @xmath52 , @xmath53 converges to zero in probability ( resp .
. to ease the notational burden , standard notation ( that follow @xcite ) for different processes are used ( e.g. , @xmath54 for queue - length , @xmath55 for idle time , @xmath56 for workload process etc . ) .
we also use standard notation , for example , @xmath57 , to denote fluid scaled , respectively , diffusion scaled , versions of various processes of interest [ see ( [ fl - scaled ] ) and ( [ diff - scaled ] ) ]
. all vectors will be column vectors .
an @xmath41-dimensional vector with all entries @xmath58 will be denoted by @xmath59 . for a vector @xmath60
, @xmath61 will denote the diagonal matrix such that the vector of its diagonal entries is @xmath60 .
@xmath62 will denote the transpose of a matrix @xmath63 .
also , @xmath64 will denote generic constants whose values may change from one proof to the next .
let @xmath65 be a probability space .
all the random variables associated with the network model described below are assumed to be defined on this probability space .
the expectation operation under @xmath66 will be denoted by @xmath67 .
we begin by introducing the family of stochastic processing network models that will be considered in this work .
we closely follow the terminology and notation used in @xcite .
the network has @xmath68 infinite capacity buffers ( to store @xmath68 many different classes of jobs ) and @xmath69 nonidentical servers for processing jobs .
arrivals of jobs , given in terms of suitable renewal processes , can be from outside the system and/or from the internal rerouting of jobs that have already been processed by some server . several different servers may process jobs from a particular buffer .
service from a given buffer @xmath70 by a given server @xmath71 is called an _ activity_. once a job starts being processed by an activity , it must complete its service with that activity , even if its service is interrupted for some time ( e.g. , for preemption by a job from another buffer ) .
when service of a partially completed job is resumed , it is resumed from the point of preemption that is , the job needs only the remaining service time from the server to get completed ( preemptive - resume policy ) .
also , an activity must complete service of any job that it started before starting another job from the same buffer .
an activity always selects the oldest job in the buffer that has not yet been served , when starting a new service [ i.e. , first in first out ( fifo ) within class ] .
there are @xmath72 activities [ at most one activity for a server - buffer pair @xmath73 , so that @xmath74 . here
the integers @xmath75 are strictly positive .
figure [ fig1-general ] gives a schematic for such a model .
buffers , @xmath72 activities , @xmath69 servers and probabilistic routing ( given by the matrix @xmath76 ) . ]
let @xmath77 , @xmath78 and @xmath79 .
the correspondence between the activities and buffers , and activities and servers are described by two matrices @xmath80 and @xmath81 respectively .
@xmath80 is an @xmath82 matrix with @xmath83 if the @xmath84th activity processes jobs from buffer @xmath70 , and @xmath85 otherwise .
the matrix @xmath81 is @xmath86 with @xmath87 if the @xmath71th server is associated with the @xmath84th activity , and @xmath88 otherwise .
each activity associates one buffer and one server , and so each column of @xmath80 has exactly one 1 ( and similarly , every column of @xmath81 has exactly one 1 ) .
we will further assume that each row of @xmath80 ( and @xmath81 ) has at least one 1 , that is , each buffer is processed by ( server is processing , resp . )
at least one activity .
for @xmath89 , let @xmath90 , if activity @xmath84 corresponds to the @xmath71th server processing class @xmath70 jobs .
let , for @xmath91 , @xmath92 and @xmath93 .
thus , for the @xmath71th server , @xmath94 denotes the set of activities that the server can perform , and @xmath95 represents the corresponding buffers from which the jobs can be processed .
we are interested in the study of networks that are nearly critically loaded .
mathematically , this is modeled by considering a sequence of networks @xmath96 that `` approach heavy traffic , '' as @xmath97 , in the sense of definition [ defn - ht ] below .
each network in the sequence has identical structure , except for the rate parameters that may depend on @xmath15 .
here @xmath98 , where @xmath99 is a countable set : @xmath100 with @xmath101 and @xmath102 , as @xmath103 .
one thinks of the physical network of interest as the @xmath15th network embedded in this sequence , for a fixed large value of @xmath15 . for notational simplicity , throughout the paper , we will write the limit along the sequence @xmath104 as @xmath105 simply as `` @xmath106 . ''
also , @xmath15 will always be taken to be an element of @xmath99 and , thus , hereafter the qualifier @xmath107 will not be stated explicitly .
the @xmath15th network @xmath108 is described as follows .
if the @xmath70th class ( @xmath109 ) has exogenous job arrivals , the interarrival times of such jobs are given by a sequence of nonnegative random variables @xmath110 that are i.i.d with mean and standard deviation @xmath111 respectively .
let , by relabeling if needed , the buffers with exogenous arrivals correspond to @xmath112 , where @xmath113 .
we set @xmath114 and @xmath115 , for @xmath116 .
service times for the @xmath84th type of activity ( for @xmath89 ) are given by a sequence of nonnegative random variables @xmath117 that are i.i.d .
with mean and standard deviation @xmath118 respectively .
we will assume that the above random variables are in fact strictly positive , that is , @xmath119 we will further impose the following uniform integrability condition : @xmath120 rerouting of jobs completed by the @xmath84th activity is specified by a sequence of @xmath121-dimensional vector @xmath122 , where @xmath123 .
for each @xmath124 and @xmath125 , @xmath126 if the @xmath127th completed job by activity @xmath84 gets rerouted to buffer @xmath70 , and takes the value zero otherwise , where @xmath128 represents jobs leaving the system .
it is assumed that for each fixed @xmath15 , @xmath129 , @xmath130 , are ( mutually ) independent sequences of i.i.d @xmath131 , where @xmath132 .
that , in particular , means , for @xmath133 , @xmath134 .
furthermore , for fixed @xmath135 , @xmath136 where @xmath137 is @xmath58 if @xmath138 and @xmath139 otherwise .
we also assume that , for each @xmath15 , the random variables @xmath140 next we introduce the primitive renewal processes , @xmath141 , that describe the state dynamics .
the process @xmath142 is the @xmath143-dimensional exogenous arrival process , that is , for each @xmath144 , @xmath145 is a renewal process which denotes the number of jobs that have arrived to buffer @xmath70 from outside the system over the interval @xmath146 $ ] . for class @xmath70
to which there are no exogenous arrivals ( i.e. , @xmath147 ) , we set @xmath148 for all @xmath149
. we will denote the process @xmath150 by @xmath151 .
for each activity @xmath89 , @xmath152 denotes the number of complete jobs that could be processed by activity @xmath84 in @xmath153 $ ] _ if the associated server worked continuously and exclusively on jobs from the associated buffer in @xmath146 $ ] and the buffer had an infinite reservoir of jobs_. the vector @xmath154 is denoted by @xmath155 .
more precisely , for @xmath156 , let @xmath157 we set @xmath158 .
then @xmath159 , @xmath160 are renewal processes given as follows . for @xmath161 , @xmath162
finally , we introduce the routing sequences .
let @xmath163 denote the number of jobs that are routed to the @xmath70th buffer , among the first @xmath127 jobs completed by activity @xmath84 .
thus , for @xmath164 , @xmath165 we will denote the @xmath68-dimensional sequence @xmath166 corresponding to routing of jobs completed by the @xmath84th activity by @xmath167 . also , @xmath168 will denote the @xmath82 matrix @xmath169 .
a _ scheduling policy _ or _ control _ for the @xmath15th spn is specified by a nonnegative , nondecreasing @xmath72-dimensional process @xmath170 . for any @xmath171
, @xmath172 represents the cumulative amount of time spent on the @xmath84th activity up to time @xmath173 . for a control @xmath6 to be admissible , it must satisfy additional properties which are specified below in definition [ t - adm - defn ] . for a given scheduling policy @xmath6 ,
the state processes of the network are the associated @xmath68-dimensional queue length process @xmath174 and the @xmath69-dimensional idle time process @xmath175 . for each @xmath149 , @xmath176
, @xmath177 represents the queue - length at the @xmath70th buffer at time @xmath173 ( including the jobs that are in service at that time ) , and for @xmath178 , @xmath179 is the total amount of time the @xmath71th server has idled up to time @xmath173 .
let @xmath180 be the @xmath68-dimensional vector of queue - lengths at time @xmath139 .
note that , for @xmath171 , @xmath181 is the total number of services completed by the @xmath84th activity up to time @xmath173 .
the total number of completed jobs ( by activity @xmath84 ) up to time @xmath173 that get rerouted to buffer @xmath70 equals @xmath182 . recalling the definition of matrices @xmath80 and @xmath81 , the state of the system at time @xmath183 can be described by the following equations : @xmath184 we now describe the main heavy traffic assumption@xcite .
we begin with a condition on the convergence of various parameters in the sequence of networks @xmath96 .
[ assum - limit - param ] there are @xmath185 , @xmath186 , @xmath187 such that @xmath188 , @xmath189 if and only if @xmath190 , and , as @xmath97 , @xmath191 \\[-8pt ] \nonumber \sigma ^{u , r}&\rightarrow&\sigma^{u},\qquad \sigma^{v , r}\rightarrow\sigma^{v } , \qquad\hat q^r { \doteq}\frac{q^r}{r } \rightarrow q.\end{aligned}\ ] ] the definition of _ heavy traffic _ , for the sequence @xmath96 , as introduced in @xcite ( also see @xcite ) , is as follows .
[ defn - ht ] define @xmath82 matrices @xmath192 , such that @xmath193 , for @xmath164 , and @xmath194 we say that the sequence @xmath96 approaches heavy traffic as @xmath195 if , in addition to assumption [ assum - limit - param ] , the following two conditions hold : there is a unique optimal solution @xmath196 to the following linear program ( lp ) : @xmath197 the pair @xmath196 satisfies @xmath198 [ assum - ht ] the sequence of networks @xmath96 approaches heavy traffic as @xmath199 .
[ rem - ht ] from assumption [ assum - ht ] , @xmath200 given in ( i ) of definition [ defn - ht ] is the unique @xmath72-dimensional nonnegative vector satisfying @xmath201 following @xcite , assume without loss of generality ( by relabeling activities , if necessary ) , that the first @xmath202 components of @xmath200 are strictly positive ( corresponding activities are referred to as _
basic _ ) and the rest are zero ( _ nonbasic _ activities ) . for later use
, we partition the following matrices and vectors in terms of _ basic _ and _ nonbasic _ components : @xmath203,\quad
t^r= \left [ \matrix { t^r_b\cr t^r_n } \right],\quad a = [ b : n ] , \quad r=[h : m],\ ] ] where @xmath6 is some control policy , @xmath204 is a @xmath205-dimensional vector of zeros , @xmath206 are @xmath207 , @xmath208 , @xmath209 and @xmath210 matrices , respectively .
the following assumption ( see @xcite ) says that for each buffer there is an associated basic activity .
[ ab148 ] for every @xmath176 , there is a @xmath89 such that @xmath211 and @xmath212 .
components of the vector @xmath200 defined above can be interpreted as the nominal allocation rates for the @xmath72 activities .
given a control policy @xmath6 , define the _ deviation process _
@xmath213 as the difference between @xmath6 and the nominal allocation : @xmath214 it follows from ( [ i - defn ] ) and ( [ xstar - defn ] ) that the idle - time process @xmath175 has the following representation : @xmath215 let @xmath216 .
next we define a @xmath217 matrix @xmath218 and @xmath219-dimensional process @xmath220
as follows : @xmath221,\qquad u^r(t ) { \doteq}k y^r(t),\qquad t \ge0,\ ] ] where @xmath222 denotes a @xmath223 identity matrix . note that , with @xmath224 as in ( [ partition ] ) , @xmath225,\qquad t \ge0 .
\label{flu1104}\ ] ] finally , we introduce the _ workload _ process @xmath226 which is defined as a certain linear transformation of the queue - length process and is of dimension no greater than of the latter .
more precisely , @xmath226 is an @xmath227-dimensional process ( @xmath228 , see @xcite ) defined as @xmath229 where @xmath230 is a @xmath231-dimensional matrix with rank @xmath227 and nonnegative entries , called the workload matrix
. we will not give a complete description of @xmath230 since that requires additional notation ; and we refer the reader to @xcite for details . the key fact that will be used in our analysis
is that there is a @xmath232 matrix @xmath233 with nonnegative entries ( see ( 3.11 ) and ( 3.12 ) in @xcite ) such that @xmath234 we will impose the following additional assumption on @xmath233 which says that each of its columns has at least one strictly positive entry .
the assumption is needed in the proof of lemma [ cont - bcp ] [ see ( [ cond612 ] ) ] .
[ g - column ] there exists a @xmath235 such that for every @xmath236 , @xmath237 .
we now introduce two types of scalings .
the first is the so - called fluid scaling , corresponding to a law of large numbers , and the second is the standard diffusion scaling , corresponding to a central limit theorem .
fluid scaled process : this is obtained from the original process by accelerating time by a factor of @xmath16 and scaling down space by the same factor .
the following fluid scaled processes will play a role in our analysis . for @xmath149 , @xmath238 here for @xmath239 , @xmath240 denotes its integer part , that is , the greatest integer bounded by @xmath241 . diffusion scaled process : this is obtained from the original process by accelerating time by a factor of @xmath16 and , after appropriate centering , scaling down space by @xmath15 .
some diffusion scaled processes that will be used are as follows . for @xmath149 , @xmath242)-\lfloor r^2 t\rfloor p ' ) } { r } , \nonumber \\[-8pt ] \\[-8pt ] \nonumber \hat
u^r(t ) & { \doteq } & r^{-1}u^r(r^2
t),\qquad \hat q^r(t){\doteq}r^{-1}q^r(r^2 t),\\ \hat w^r(t ) & { \doteq } & r^{-1}w^r(r^2 t),\qquad \hat y^r(t ) { \doteq}r^{-1}y^r(r^2 t).\nonumber\end{aligned}\ ] ] the processes @xmath243 are not centered , as one finds ( see lemma 3.3 of @xcite ) that , with any reasonable control policy , their fluid scaled versions converge to zero as @xmath106 . define for @xmath183 , @xmath244 \\[-8pt ] \nonumber & & { } - \sum_{j=1}^{{\mathbf{j } } } \hat{\bolds{\phi}}^{j , r}_i(\bar s^r_j(\bar t^r_j(t))),\qquad i \in\mathbb{i}.\end{aligned}\ ] ] recall @xmath245 and @xmath246 from assumption [ assum - limit - param ] . using ( [ q - defn1 ] ) , ( [ i - defn ] ) , ( [ xstar - defn ] ) and ( [ u - k - defn ] ) , one has the following relationships between the various scaled quantities defined above . for all @xmath183 , @xmath247 where @xmath248.\ ] ] also , using ( [ workload - defn ] ) , ( [ mrgk ] ) and ( [ q - relation ] ) , for all @xmath161 , @xmath249 + g\hat u^r(t).\hspace*{-30pt } \label{w - relation}\ ] ] the definition of admissible policies ( definition [ t - adm - defn ] ) , given below , incorporates appropriate nonanticipativity requirements and ensures feasibility by requiring that the associated queue - length and idle - time processes ( @xmath250 ) are nonnegative . for @xmath251
we define the multiparameter filtration generated by interarrival and service times and routing variables as @xmath252 \\[-8pt ] \nonumber & & \qquad = \sigma\{u_i^r(m'_i ) , v_j^r(n'_j ) , \phi_{i}^{j , r}(n'_{j})\dvtx m'_i\le m_i , n'_j\le n_j ; i \in\mathbb{i } , j \in\mathbb{j}\}.\hspace*{-20pt}\end{aligned}\ ] ] then @xmath253 is a multiparameter filtration with the following ( partial ) ordering : @xmath254 we refer the reader to section 2.8 of @xcite for basic definitions and properties of multiparameter filtrations , stopping times and martingales .
let @xmath255 for all @xmath256 , we define @xmath257 where @xmath258 denotes the vector of 1 s .
it will be convenient to allow for extra randomness , than that captured by @xmath259 , in formulating the class of admissible policies .
let @xmath260 be a @xmath261-field independent of @xmath262 .
for @xmath263 , let @xmath264 .
[ t - adm - defn ] for a fixed @xmath15 and @xmath265 , a scheduling policy @xmath266 is called admissible for @xmath108 with initial condition @xmath267 if for some @xmath260 independent of @xmath268 , the following conditions hold : @xmath269 is nondecreasing , nonnegative and satisfies @xmath270 for @xmath89 .
@xmath271 defined by ( [ i - defn ] ) is nondecreasing , nonnegative and satisfies @xmath272 for @xmath178 .
@xmath273 defined in ( [ q - defn1 ] ) is nonnegative for @xmath176 .
define for each @xmath274 , @xmath275 \\[-8pt ] \nonumber & { \doteq}&\bigl(e_i^r(r^2 t)+1\dvtx i \in\mathbb{i } ; s_j^r(t_j^r(r^2t))+1\dvtx j \in\mathbb{j}\bigr).\end{aligned}\ ] ] then , for each @xmath149 , @xmath276 define the filtration @xmath277 as @xmath278 \\[-8pt ] \nonumber & \!=\!&\sigma\bigl\{a \in{\mathcal{f}}^r\dvtx a \cap\{{\sigma } ^r_0(t)\le ( m , n)\ } \in\mathcal{f}^r((m , n ) ) , m \in{{\mathbb{n}}}^{{\mathbf{i } } } , n \in{{\mathbb{n}}}^{{\mathbf{j}}}\bigr\}.\hspace*{-35pt}\end{aligned}\ ] ] then @xmath279 denote by @xmath280 the collection of all admissible policies for @xmath108 with initial condition @xmath267 .
[ rem921 ] ( i ) and ( ii ) in definition [ t - adm - defn ] imply , in view of ( [ i - defn ] ) and properties of the matrix @xmath81 , that @xmath281 in particular , @xmath282 is a process with lipschitz continuous paths .
condition ( iv ) in definition [ t - adm - defn ] can be interpreted as a nonanticipativity condition .
proposition 2.8 and theorem 5.4 of @xcite give general sufficient conditions under which this property holds ( see also proposition [ tisadmisnew ] of the current work ) .
for the network @xmath108 , we consider an expected infinite horizon discounted ( linear ) holding cost associated with a scheduling policy @xmath6 and initial queue length vector @xmath267 : @xmath283 here , @xmath284 is the `` discount factor '' and @xmath285 , an @xmath68-dimensional vector with each component @xmath286 , is the vector of `` holding costs '' for the @xmath68 buffers . in the second term , @xmath287 is an @xmath219-dimensional vector .
the first @xmath69 block of @xmath288 corresponds to the idleness process @xmath55 , and , thus , the second term in the cost , in particular , captures the idleness cost .
the last @xmath289 components of @xmath288 correspond to the time spent on _ nonbasic _ activities .
thus , this formulation of the cost allows , in addition to the idleness cost , the user to put a penalty for using nonbasic activities .
the formulation of the cost function considered in our work goes back to the original work of harrison et al .
@xcite .
the _ scheduling control problem _ for @xmath108 is to find an admissible control policy @xmath6 that minimizes the cost @xmath290 .
the value function @xmath291 for this control problem is defined as @xmath292 the goal of this work is to characterize the limit of value functions @xmath291 as @xmath7 , as the value function of a suitable diffusion control problem . in order to see the form of the diffusion control problem
, we will like to send @xmath7 in ( [ q - relation ] ) . using the functional central limit theorem for renewal processes
, it is easily seen that , for all reasonable control policies ( see again lemma 3.3 of @xcite ) , when @xmath246 converges to some @xmath293 , @xmath294 defined in ( [ q - relation ] ) converges weakly to @xmath295 where @xmath296 here @xmath297 is the identity map and @xmath298 is a brownian motion with drift 0 and covariance matrix @xmath299 where @xmath300 is a @xmath301 diagonal matrix with diagonal entries @xmath302 , @xmath303 is a @xmath304 diagonal matrix with diagonal entries @xmath305 and @xmath306s are @xmath301 matrices with entries @xmath307 [ see ( [ sigma - phi ] ) ] .
although the process @xmath308 in ( [ q - relation ] ) , for a general policy sequence @xmath8 , need not converge , upon formally taking limit as @xmath195 , one is led to the following diffusion control problem .
[ bcp ] a @xmath72-dimensional adapted process @xmath4 , defined on some filtered probability space @xmath309 which supports an @xmath68-dimensional @xmath310-brownian motion @xmath298 with drift 0 and covariance matrix @xmath311 given by ( [ sigmadefn ] ) , is called an admissible control for the brownian control problem with the initial condition @xmath312 iff the following two properties hold @xmath313-a.s . : @xmath314 where @xmath315 and @xmath316 are as in ( [ 234b ] ) and ( [ theta - defn ] ) respectively .
we refer to @xmath317 as a system .
we denote the class of all such admissible controls by @xmath318 .
the brownian control problem is to @xmath319,\hspace*{-30pt}\ ] ] over all admissible controls @xmath320 .
define the value function @xmath321 recall our standing assumptions ( [ stpos1029 ] ) , ( [ ui1032 ] ) , ( [ ass1120 ] ) , assumptions [ assum - limit - param ] , [ assum - ht ] , [ ab148 ] and [ g - column ] .
the following is the main result of @xcite .
[ ab937 ] fix @xmath293 and for @xmath322 , @xmath323 such that @xmath324 as @xmath195 .
then @xmath325 the proof in @xcite is presented for the case where in the definition of @xmath326 [ see ( [ ab101 ] ) ] , @xmath280 is replaced by the smaller family @xmath327 which consists of all @xmath328 that satisfy ( iv ) of definition [ t - adm - defn ] with @xmath329 replaced by @xmath330 .
proof for the slightly more general setting considered in the current paper requires only minor modifications and , thus , we omit the details . for the main result of this work , we will need additional assumptions .
[ non - deg ] the matrix @xmath311 is positive definite .
we will make the following assumption on the probabilities of deviations from the mean for the underlying renewal processes .
similar conditions have been used in previous works on construction of asymptotically optimal control policies @xcite .
[ ldp ] there exists @xmath331 and , for each @xmath332 , some @xmath333 such that , for @xmath89 , @xmath176 , @xmath334 , @xmath335 , @xmath336 the third inequality above is a consequence of the first two , but we note it explicitly here for future use .
the assumption is clearly satisfied when @xmath337 and @xmath338 are poisson processes . for general renewal processes ,
such inequalities hold under suitable moment conditions on the interarrival and service time distributions . indeed , if for some @xmath339 @xmath340^{2{\bolds{\mathsf{m } } } } \bigr ] & < & \infty , \nonumber \\[-8pt ] \\[-8pt ] \nonumber { \mathbb{e}}\bigl [ \sup_{r } [ v_j^r(1)]^{2{\bolds{\mathsf{m}}}}\bigr ] & < & \infty \qquad\mbox{for all } i \in\mathbb{i } , j \in\mathbb{j},\end{aligned}\ ] ] then , from theorem 4 of @xcite , assumption [ ldp ] is satisfied .
we now introduce an assumption on the regularity properties of a certain skorohod map .
this map plays a crucial role in our analysis ; see proofs of theorems [ main - bcp ] and [ mainweak ] [ see in particular , ( [ 110 ] ) , ( [ ab302 ] ) , discussion below ( [ ab419 ] ) and the proof of theorem [ newmain518 ] ] .
let @xmath341 and @xmath342 [ ab948 ] given @xmath343 , we say @xmath344 solve the problem ( sp ) for @xmath345 if : ( i ) @xmath346 , ( ii ) @xmath347 , ( iii ) @xmath348 is nondecreasing and @xmath349 , ( iv ) @xmath350 for all @xmath183 , and(v ) @xmath351 for all @xmath352 . denoting by @xmath353 the set of @xmath343 such that there is a unique solution to the sp for @xmath354 , we define maps @xmath355 , @xmath356 as @xmath357 , @xmath358 if @xmath359 solve the @xmath360 for @xmath345 .
[ ab1012 ] @xmath361 and the maps @xmath362 are lipschitz , namely , there exists @xmath363 such that for all @xmath364 , @xmath365 we refer the reader to @xcite and @xcite for sufficient conditions under which the above regularity property of the skorohod map holds .
see also example [ exammodel ] below . for later use
we introduce the notation @xmath366 for @xmath367 .
since @xmath81 has nonnegative entries and @xmath368 for @xmath369 , we see from the definition of @xmath218 [ see ( [ u - k - defn ] ) ] that @xmath370 for rest of the paper , in addition to the assumptions listed above theorem [ ab937 ] , assumptions [ non - deg ] , [ ldp ] and [ ab1012 ] will be in force .
the main result of the paper is the following .
[ main1016 ] fix @xmath293 .
let for @xmath322 , @xmath371 be such that @xmath324 as @xmath195 .
then @xmath372 the theorem is an immediate consequence of theorem [ jrtoj ] below which is proved in section [ secproof ] . for @xmath2 ,
we say @xmath373 is @xmath3-optimal for the bcp with initial value @xmath374 if @xmath375 when clear from the context , we will omit the phrase `` for the bcp with initial value @xmath374 '' and merely say that @xmath376 is @xmath3-optimal .
[ jrtoj ] fix @xmath293 . for @xmath322 ,
let @xmath371 be such that @xmath324 as @xmath195 .
for every @xmath2 , there exists @xmath320 , which is @xmath3-optimal , and a sequence @xmath377 , @xmath334 such that @xmath378 combining theorems [ ab937 ] and [ main1016 ] , the following is immediate .
[ maincorr ] fix @xmath293 . for @xmath322 ,
let @xmath371 be such that @xmath324 as @xmath195 . then , as @xmath195 , @xmath379 .
assumptions made in this work can loosely be divided into two categories : assumptions on the underlying stochastic primitives , which include , in particular , the heavy traffic conditions ( assumptions [ assum - limit - param ] , [ assum - ht ] , [ non - deg ] and [ ldp ] ) , and assumptions made on the network structure ( assumptions [ ab148 ] , [ g - column ] and [ ab1012 ] ) . below we discuss the validity of these structural assumptions for some basic families of spn models .
[ exammodel ] the following examples have been described in detail in @xcite .
we will assume here , without loss of generality , that @xmath380 for all @xmath89 ( an activity @xmath84 for which @xmath381 can simply be deleted from the network description ) .
furthermore , for all three settings considered below , assumption [ ab148 ] can be made without loss of generality , since otherwise one can consider a reduced system obtained by omitting the buffers that are not processed by any basic activity .
assumption [ g - column ] states that the matrix @xmath233 can be chosen in a manner such that it has no columns that are identically zero . roughly speaking
, it says that a nonzero control action leads to a nonzero state displacement .
although this appears to be a very natural geometric condition and is trivially satisfied for networks in part ( a ) below , it is not clear that it holds always for examples in parts ( b ) and ( c ) below
. we will assume this condition to hold without further comment .
thus , in discussion below , we will focus only on assumption [ ab1012 ] . _ open multiclass queueing networks _ :
these correspond to a setting where each buffer is processed by exactly one activity and , consequently , there is a one - to - one correspondence between activities and buffers , that is , @xmath382 ( see left figure in figure [ fig - mult ] for an example ) . for such networks
, @xmath383 is an @xmath301-matrix of the form @xmath384 where @xmath76 is a nonnegative matrix with spectral radius less than 1 . in particular , @xmath383 is nonsingular , @xmath218 is a @xmath86 matrix with full row rank and one can take @xmath385 and @xmath386 . here
@xmath387 and from @xcite it is known that for such @xmath0 assumption [ ab1012 ] is satisfied . _ parallel server networks _ : for such spn , a buffer can be served by more than one activity , however , each job gets processed exactly once before leaving the system ( i.e. , there is no rerouting ) .
see right figure in figure [ fig - mult ] for an example .
in particular , @xmath388 and , hence , @xmath389 . in this case ,
@xmath390 , where @xmath391 , for @xmath392 . from assumption [ ab148 ] ( which
, as was noted above , can be made without loss of generality ) we have that @xmath393 and , thus , @xmath0 is a diagonal matrix with strictly positive diagonal entries .
assumption [ ab1012 ] is clearly satisfied for such matrices . ,
@xmath394 , @xmath395 . ]
_ job - shop networks _
: this subclass of networks combines features of both ( a ) and ( b ) : a buffer can be processed by more than one activity and jobs , once served , can get rerouted to another buffer for additional processing .
see figure [ fig - jshop ] for an example .
following specific examples considered in @xcite ( see also @xcite ) , we define job - shop networks as those which satisfy the following property : if for some @xmath396 , and @xmath176 , @xmath397 , then @xmath398 for all @xmath399 .
namely , jobs corresponding to any two activities that process the same buffer @xmath70 have an identical ( probabilistic ) routing structure , following their completion by the respective servers .
it is easily checked that in this case @xmath400 , where @xmath401 is as introduced in ( b ) and @xmath402 is an @xmath301-dimensional matrix with entries @xmath403 where @xmath89 is such that @xmath404 . under the condition that @xmath405 has spectral radius less than 1
, it follows from @xcite that assumption [ ab1012 ] is satisfied .
the rest of the paper is devoted to the proof of theorem [ jrtoj ] . toward that goal ,
in this section we construct near - optimal controls for the bcp with certain desirable features .
this construction is achieved in theorem [ newmain518 ] , which is the main result of this section . since an admissible control
is not required to be of bounded variation , the bcp is a somewhat nonstandard diffusion control problem and is difficult to analyze directly .
however , as shown in @xcite , under assumptions made in this paper , one can replace this control problem by an equivalent problem of singular control with state constraints ( scsc ) .
this control problem , also referred to as the equivalent workload formulation ( ewf ) of the bcp , is given below .
we begin by introducing the cost function , that is , optimized in this equivalent control problem .
_ effective cost function _ : recall the definition of the workload matrix @xmath230 introduced in ( [ workload - defn ] ) .
let @xmath406 .
for each @xmath407 , define @xmath408 since @xmath409 , the infimum is attained for all @xmath407 .
it is well known ( see theorem 2 of @xcite ) that one can take a continuous selection of the minimizer in the above linear program .
that is , there is a continuous map @xmath410 such that @xmath411 thus , in particular , @xmath412 is continuous .
one can check that @xmath413 satisfies linear lower and upper bounds . in order to see this , define @xmath414 then ( [ z - star - defn ] ) holds with @xmath415 replaced by @xmath416 . since @xmath417 and @xmath409 , we have from the above display that @xmath418 for some @xmath419 .
also , uniform continuity of @xmath416 on shows that @xmath420 \\[-8pt ] \eqntext{w_1 , w_2 \in\mathcal{w } , modulus , that is , a nondecreasing function from @xmath421 satisfying @xmath422 . inequalities ( [ cond511 ] ) and ( [ cond512 ] ) will be used in order to appeal to some results from @xcite ( see remark [ rem32 ] below ) . define @xmath423 an @xmath219-dimensional adapted process @xmath424 , defined on some filtered probability space @xmath425 which supports an @xmath68-dimensional @xmath426-brownian motion @xmath298 with drift 0 and covariance matrix @xmath311 defined in ( [ sigmadefn ] ) ,
is called an admissible control for the ewf with initial condition @xmath427 iff the following two properties hold @xmath428-a.s . :
@xmath429 \\[-8pt ] \nonumber & & \tilde w(t ) { \doteq}w + \lambda\theta t+ \lambda\tilde x(t ) + g \tilde
u(t ) \in \mathcal{w } \qquad \mbox{for all } t \ge0,\hspace*{-30pt}\end{aligned}\ ] ] where @xmath316 is as in ( [ theta - defn ] ) .
we denote the class of all such admissible controls by @xmath430 .
the control problem for the ewf is to @xmath431 over all admissible controls @xmath432 .
define the value function @xmath433 from theorem 2 of @xcite it follows that for all @xmath434 satisfying @xmath435 , @xmath436 the following lemma will be used in order to appeal to some results from @xcite . the proof is based on arguments in @xcite .
let @xmath437 .
[ intgood ] the cones @xmath438 and @xmath439 have nonempty interiors and @xmath440 . from @xcite ( see above corollary 7.4 therein ) it follows that @xmath441 has full row rank and so there is a @xmath442 matrix @xmath443 such that @xmath444 .
let @xmath445 and let @xmath446 be sufficiently small such that for all @xmath447 $ ] , @xmath448 .
let @xmath449 and @xmath450 ' \in{\mathbb{r}}^{{\mathbf{j}}}$ ] . then @xmath451 .
we will now argue that @xmath452 .
since the rows of @xmath230 are linearly independent , we can find an @xmath453 matrix @xmath454 such that @xmath455 . fix @xmath456 .
then , whenever @xmath457 , @xmath458 , we have @xmath459 and so @xmath460 .
this shows that @xmath461 .
next note that @xmath462 since @xmath463 and @xmath464 , we have that @xmath465 and so @xmath466 . since @xmath464 , we can find @xmath467 such that whenever @xmath468 is such that @xmath469 , @xmath470 .
now fix @xmath471 .
then , for any @xmath472 with @xmath473 and @xmath474 since @xmath475 , we have @xmath476 for all @xmath477 . also , @xmath478 for all @xmath479 .
thus , @xmath480 and , therefore , @xmath481 .
it follows that @xmath482 . finally , we show that @xmath483 . since @xmath484 , every row of @xmath485 must contain at least one strictly positive entry .
thus , @xmath486 . choose @xmath487 sufficiently small such that @xmath488 .
let @xmath489 ' \in{\mathbb{r}}^{{\mathbf{j}}}$ ] .
we now argue that @xmath490 .
note that by construction @xmath491 .
thus , we can find @xmath492 such that @xmath493 whenever @xmath494 satisfies @xmath495 .
also , since @xmath218 has full row rank ( see corollary 6.2 of @xcite ) , we can find a @xmath496 matrix @xmath497 such that @xmath498 .
thus , @xmath499 and , consequently , @xmath500 . the result follows .
note that the vector @xmath501 constructed in the proof of the lemma above has the property that @xmath502 and @xmath503 .
thus , we have shown the following : [ tset ] the set @xmath504 is nonempty .
the above result will be used in the construction of a suitable near optimal control policy for the bcp [ see below ( [ 111 ] ) ] .
[ rem32 ] we will make use of some results from @xcite and @xcite that concern a general family of singular control problems with state constraints . we note below some properties of the model studied in the current paper that ensure that the assumptions of @xcite and @xcite are satisfied : @xmath233 has full row rank .
this follows from the observation that @xmath505 and @xmath383 have full row ranks and , therefore , @xmath506 @xmath507 and @xmath438 has a nonempty interior ( see lemma [ intgood ] ) .
@xmath508 , @xmath509 for all @xmath510 and @xmath511 for all @xmath512 .
this is an immediate consequence of the fact that the entries of @xmath233 and @xmath230 are nonnegative [ see above ( [ mrgk ] ) ] .
since @xmath230 has full row rank and , by assumption [ non - deg ] , @xmath311 is positive definite , we have that @xmath513 is positive definite .
the above properties along with assumption [ g - column ] , ( [ cond511 ] ) and ( [ cond512 ] ) ensure that assumptions of @xcite and @xcite are satisfied in our setting .
in particular , assumption ( 2.1)(2.2 ) and ( 2.8)(2.10 ) of @xcite hold in view of properties ( b ) , ( c ) and ( d ) and equations ( [ cond511 ] ) and ( [ cond512 ] ) .
similarly , assumptions ( 1 ) , ( 5 ) and 2.2 of @xcite hold in our setting [ from property ( c ) , ( [ cond511 ] ) and assumption [ g - column ] , resp . ] .
henceforth , when appealing to results from @xcite and @xcite , we will not make an explicit reference to these conditions . recall @xmath514 and the map @xmath515 introduced above ( [ 234b ] ) and ( [ abn756 ] ) , respectively .
the following is a key step in the construction of a near - optimal control with desirable properties .
[ main - bcp ] fix @xmath293 . for each @xmath516
, there exists , given on some system @xmath517 , that is @xmath3-optimal and has the following properties : @xmath518 where @xmath519 is an adapted process with sample paths in @xmath520 satisfying the following : for some @xmath521 , @xmath522 , with @xmath523 and , a. @xmath524 for @xmath525 and @xmath526 for @xmath527 .
b. letting @xmath528 for @xmath529 and @xmath530 .
c. there is an i.i.d sequence of uniform ( over @xmath531 $ ] ) random variables @xmath532 , that is , independent of @xmath315 , and for each @xmath533 a measurable map @xmath534 \to \mathcal{s}_m^{\eta}$ ] , @xmath533 , such that the map @xmath535 is continuous , for a.e .
@xmath173 in @xmath536 $ ] , and @xmath537 where @xmath538 , @xmath539 .
proof of theorem [ main - bcp ] is given in section [ pfmainbcp ] .
[ on3 ] the above theorem provides an @xmath3-optimal control @xmath540 , which is the `` constrained''-version of a piecewise constant process @xmath541 .
the value of @xmath541 changes only at time - points that are integer multiples of @xmath316 and is constant for @xmath542 . also , the changes in ( the value of ) the process occur in jumps with sizes that are integer multiples of some @xmath543 and are bounded by @xmath63 .
the third property in the theorem plays an important role in the weak convergence proof [ theorem [ mainweak ] , see , e.g. , ( [ ab459n ] ) ] and says that the jump - sizes of this piecewise constant process are determined by the brownian motion @xmath298 sampled at discrete instants @xmath544 and the independent random variable @xmath545 ; furthermore , the dependence on @xmath298 is continuous .
the continuous dependence is ensured using a mollification argument [ see below ( [ bcp - main-18 ] ) ] that has previously been used in @xcite .
the following lemma is a straightforward consequence of the lipschitz property of the skorohod map , the linearity of the cost and the state dynamics .
proof is given in the .
[ ab210 ] there is a @xmath546 such that , if @xmath547 and @xmath548 defined on a common filtered probability space are such that @xmath549 where @xmath550 is defined by the right - hand side of ( [ ab440 ] ) by replacing @xmath4 there by @xmath551 , then @xmath552 define @xmath553 as @xmath554 note that @xmath555 is a lipschitz map : that is , for some @xmath556 , we have for @xmath557 , @xmath558 also , since @xmath559 , we have for @xmath560 , @xmath561 .
recall the set @xmath562 introduced in corollary [ tset ] .
fix @xmath2 , and a unit vector @xmath563 and define @xmath564 as @xmath565 let @xmath566 , @xmath567 be as in theorem [ main - bcp ] with @xmath568 .
let @xmath569 and define @xmath570 .
define control process @xmath571 with the corresponding state process @xmath572 [ defined by the right - hand side of ( [ ab440 ] ) ] by the following equations . for @xmath539 , @xmath573 and @xmath574 with the conventions that for @xmath575 , @xmath576 is replaced by @xmath26 and for @xmath577 , @xmath578 .
the control @xmath4 evolves in a similar manner to @xmath540 at all time points excepting @xmath579 , @xmath580 . since @xmath563 , for every @xmath581 and @xmath582 , @xmath583 .
this , along with the definition of the map @xmath515 ( see below assumption [ ab1012 ] ) , ensures that @xmath572 is nonnegative over the time intervals @xmath584 and @xmath585 , for @xmath586 . furthermore ,
( [ abn756 ] ) and the property @xmath587 ( see definition of @xmath562 in corollary [ tset ] ) ensure that @xmath588 is nondecreasing and nonnegative .
thus , the process defined by relations ( [ ab610 ] ) and ( [ ab616 ] ) is indeed an element of @xmath318 .
the strict positivity of @xmath572 at time instants @xmath589 , @xmath590 will be exploited in the weak convergence analysis of section [ secproof ] [ see equation ( [ ab1432 ] ) and also below ( [ 442half ] ) ] .
[ newmain518 ] the process @xmath4 defined above is @xmath3-optimal for the bcp with initial value @xmath374 .
since @xmath540 is @xmath591 optimal , in view of lemma [ ab210 ] , it suffices to show that @xmath592 for this we will introduce a collection of @xmath72-dimensional processes @xmath593 , @xmath594 , such that @xmath595 and @xmath596 .
these processes are only used in the current proof and do not appear elsewhere in this work .
define , recursively , for @xmath597 , processes @xmath598 with corresponding state processes @xmath599 , as follows : @xmath600 and for @xmath601 , @xmath149 , @xmath6021_{[n\theta , \infty)}(t),\end{aligned}\ ] ] where for all @xmath603 , @xmath604.\ ] ] note that for @xmath605 , @xmath606 where for all @xmath183 , @xmath607.\ ] ] using the lipschitz property of @xmath515 , it follows that , for @xmath161 , @xmath608 thus , for @xmath149 , @xmath609 the result follows on noting that @xmath595 and @xmath610 . throughout this section
we fix @xmath293 and @xmath611 .
we begin with some preparatory results .
let @xmath612 be the class of all @xmath72-dimensional adapted processes @xmath376 given on some filtered probability space such that @xmath613 is nondecreasing , @xmath614 and @xmath615 .
note that @xmath616 . also , a given @xmath617 is in @xmath318 if and only if ( [ ab440 ] ) is satisfied . given an adapted process @xmath618 , on some system @xmath517 , with sample paths in @xmath520 , and satisfying @xmath615 , we will denote the process @xmath376 , defined by @xmath619 as @xmath620 .
we claim that @xmath621 indeed , @xmath622 . also , @xmath623 .
since @xmath624 , @xmath625 is nondecreasing and @xmath626 . also , for @xmath343 , @xmath627 , which is a nonnegative and nondecreasing function since @xmath628 has these properties and the matrix @xmath629 c'\ ] ] has nonnegative entries . combining these observations , we see that the process @xmath630 satisfies ( [ ab440 ] ) and ( [ ab934 ] ) .
the claim follows .
next , from the lipschitz property of @xmath631 it follows that there is a @xmath632 such that , if @xmath633 , @xmath634 , then for all @xmath635 , @xmath636 in what follows , we will denote @xmath637 by @xmath638 .
[ stepwise - bcp ] let @xmath373 be a @xmath639-adapted process with a.s .
continuous paths .
suppose further that for some @xmath640 , @xmath641
< \infty\qquad \mbox{for all } t>0.\ ] ] then for any @xmath642 , there are @xmath643 , @xmath644 and a @xmath645 such that @xmath646 for some @xmath647 , that is , @xmath639-adapted and satisfies and of theorem [ main - bcp ] with @xmath523 and @xmath648<\varepsilon _ 1.\ ] ] the construction of @xmath540 proceeds by defining , successively , simpler approximations of @xmath376 , denoted as @xmath649 .
the process @xmath650 is given in terms of a sequence @xmath651 of @xmath72-dimensional processes , whereas the processes @xmath652 and @xmath653 are given in terms of one parameter families of @xmath72-dimensional processes @xmath654 , @xmath655 , respectively .
all the processes @xmath656 , @xmath657 , and @xmath651 , @xmath658 , @xmath659 are only used in this proof and do not appear elsewhere in the paper .
fix @xmath642 .
define @xmath660 .
note that for all @xmath661 $ ] , @xmath662 hence , by ( [ stepwise - bcp-1 ] ) , we have , for each @xmath127 , @xmath663 } \biggl|\frac{y_n(t)-y_n(t')}{t - t ' } \biggr|^m\biggr ] < \infty.\ ] ] note that for @xmath664 $ ] , @xmath665 and for @xmath666 $ ] , @xmath667 . since @xmath376 is continuous and @xmath668 , @xmath669 a.s . combining this with ( [ stepwise - bcp-1 ] ) and the estimate @xmath670
, we now have that , for some @xmath671 , @xmath672 satisfies @xmath673 with @xmath674 . also , @xmath675}\biggl|\frac{y^1(t)-y^1(t')}{t - t ' } \bigg|^m\biggr ] { \doteq}c_1 < \infty.\ ] ] note that @xmath676 given @xmath677 and @xmath523 , define @xmath678 fix @xmath679 large enough so that @xmath680 and set @xmath681 .
then , from ( [ step - pf-3 ] ) we have @xmath682 \nonumber\\ & & \qquad= { \mathbb{e}}\bigl[\max_{n=0,\ldots , p_0 - 1}\bigl\{\sup_{t\in[n\theta , ( n+1)\theta ) } \bigl|y^{(1)}(n\theta)-y^{(1)}(t)\bigr|^m\bigr\}\bigr ] \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & \qquad\leq \theta^m { \mathbb{e}}\biggl[\sup_{n=0,1,\ldots , p_0 - 1}\biggl\{\sup_{t , t'\in[n\theta , ( n+1)\theta ) } \frac{|y^{(1)}(t)-y^{(1)}(t')|}{|t - t'|}\biggr\}^m\biggr]\\ & & \qquad\leq \theta^m c_1 < \tilde\varepsilon_0.\nonumber\end{aligned}\ ] ] from ( [ step - pf-4 ] ) we have @xmath683 for @xmath684 , let @xmath685 denote the smallest integer upper bound for @xmath241 . for , let @xmath686 .
fix @xmath687 and , with convention @xmath688 , define for @xmath183 @xmath689 where for @xmath690 , @xmath691 denotes @xmath692 .
note that for @xmath693 , @xmath694 observing that for @xmath695 , @xmath696 , we have that @xmath697 \leq \bigl(p_0\eta\sqrt{{\mathbf{j}}}\bigr)^m \le\tilde\varepsilon_0 .
\label{ab1208}\ ] ] note that if @xmath582 satisfies @xmath698 , then @xmath699 for all @xmath700 and , consequently , for such @xmath84 , @xmath701 . combining this with the fact that @xmath81 has nonnegative entries
, we see that @xmath702 . from this observation , along with ( [ step - pf-6 ] ) , we have @xmath703 the process @xmath652 constructed above is constant on @xmath704 and the jumps @xmath705 take value in the lattice @xmath706 , for @xmath707 . also , @xmath708 for @xmath709 . for fixed @xmath710 ,
define @xmath711 then there exists @xmath712 such that , for all @xmath713 , @xmath714 \leq c_2\sum_{n=0}^{p_0}{\mathbb{e}}\bigl [ \bigl|\partial y^{(3)}(n)\bigr|^m i_{(|\partial y^{(3)}(n)|>m)}\bigr].\hspace*{-30pt}\ ] ] also , for some @xmath715 , we have from ( [ stepwise - bcp-1 ] ) , ( [ step - pf-2 ] ) , ( [ ab1206 ] ) and ( [ ab1208 ] ) that , for @xmath716 , @xmath717 \le c_3({\mathbb{e}}[|y|^m_{\infty , t } ] + 1 ) < \infty.\ ] ] fix @xmath713 such that the right - hand side of ( [ step - pf-75 ] ) is bounded by @xmath718 . setting @xmath719
, we now have that @xmath720 \leq \tilde\varepsilon_0 . \label{ab1221}\ ] ] also , @xmath721 combining ( [ stepwise - bcp-1 ] ) , ( [ step - pf-2 ] ) , ( [ ab1206 ] ) , ( [ ab1208 ] ) and ( [ ab1221 ] ) , we now have that @xmath722 satisfies ( [ ab102 ] ) as well as ( i ) and ( ii ) of theorem [ main - bcp ] .
this completes the proof .
[ cont - bcp ] for each @xmath723 there exists an @xmath724-optimal @xmath725 , which is @xmath726-adapted , continuous a.s . , and satisfies @xmath727 fix @xmath723 and let @xmath728 .
applying theorem 2.1(iv ) of @xcite , we have that @xmath729 , where the infimum is taken over all @xmath730-adapted controls @xmath288 .
hence , using ( [ mincost - equal ] ) , we conclude that there is an @xmath219-dimensional @xmath639-adapted process @xmath288 for which ( [ ewf-1 ] ) holds and @xmath731 from lemma 4.7 of @xcite and following the construction of proposition 3.3 of @xcite [ cf .
( 12 ) and ( 14 ) of that paper ] , we can assume without loss of generality that @xmath288 has continuous sample paths and for all @xmath732 , @xmath733 hence , using properties of the @xmath233 matrix ( see assumption [ g - column ] ) , we have that @xmath734 \\[-8pt ] \eqntext{\mbox{for all } m > 0.}\end{aligned}\ ] ] we will now use a construction given in the proof of theorem 1 of @xcite .
this construction shows that there is a @xmath496 matrix @xmath735 and a @xmath736 matrix @xmath737 such that letting @xmath738 we have that @xmath373 and @xmath739 we refer the reader to equations ( 35 ) and ( 36 ) of @xcite for definitions and constructions of these matrices . from ( [ bd622 ] ) we now have that @xmath376 is an @xmath724-optimal control , has continuous sample paths a.s . and
is @xmath740-adapted .
finally from ( [ cond615 ] ) , we have that for some @xmath741 , @xmath742 by combining ( [ 32half ] ) and ( [ ewf-1 ] ) , the fourth term on the right - hand side can be bounded above by @xmath743 for some @xmath744 .
the result then follows on using ( [ cond612 ] ) .
the following construction will be used in the proof of theorem [ main - bcp ] .
[ inter159 ] fix @xmath373 such that @xmath745 and ( [ ab127 ] ) holds .
for @xmath635 , let @xmath746 , @xmath747 , and @xmath748 .
then given @xmath723 , there exists @xmath749 such that @xmath750 and ( [ ab127 ] ) holds with @xmath376 replaced by @xmath751 . since @xmath745
, we have that @xmath752 where @xmath572 is the state process corresponding to @xmath376 and @xmath753 .
choose @xmath754 large enough so that @xmath755 using the lipschitz property of @xmath631 and @xmath756 ( see assumption [ ab1012 ] ) , we can find @xmath757 such that , for all @xmath635 , @xmath758 \\[-8pt ] \nonumber & & \qquad\le c_1 \sup_{t \le s \le t } t \ge t,\end{aligned}\ ] ] where @xmath759 is the state process corresponding to @xmath751 .
thus , for some @xmath712 , @xmath760 where @xmath761 . using ( [ ab127 ] )
, we can now choose @xmath762 large enough so that @xmath763 next , letting @xmath764 , we have from ( [ ab302 ] ) that for some @xmath715 , @xmath765 integration by parts now yields that for some @xmath766 , @xmath767 combining the estimates in ( [ lab16 ] ) , ( [ lab17 ] ) and ( [ lab18 ] ) , we now have that for all @xmath768 , @xmath769 finally , the fact that ( [ ab127 ] ) holds with @xmath376 replaced by @xmath751 is an immediate consequence of ( [ ab302 ] ) .
we can now complete the proof of theorem [ main - bcp ] .
proof of theorem [ main - bcp ] using lemma [ cont - bcp ] , one can find @xmath770 which is @xmath771-optimal , has continuous paths a.s .
, is @xmath726-adapted and satisfies ( [ ab127 ] ) . using lemma [ inter159 ] ,
we can find @xmath772 such that @xmath773 is @xmath774-optimal and ( [ ab127 ] ) holds with @xmath376 replaced by @xmath751
. we will apply theorem [ stepwise - bcp ] , with @xmath775 , @xmath376 replaced with @xmath751 , @xmath724 replaced by @xmath776 and denote the corresponding processes obtained from theorem [ stepwise - bcp ] , once again by @xmath540 and @xmath777 .
in particular , @xmath778 is such that @xmath779 , where @xmath12 is @xmath726-adapted , satisfies ( i ) and ( ii ) of theorem [ main - bcp ] , for some @xmath780 and @xmath781 , and ( [ ab102 ] ) holds with @xmath775 , @xmath376 replaced by @xmath751 and where @xmath782 is as in ( [ ab161 ] ) .
then @xmath783 \le\bar l { \mathbb{e}}\bigl[\bigl|y^t-\tilde y_0^{(1)}\bigr|_{\infty , t}\bigr ] \le\frac{\varepsilon}{5c_1}.\ ] ] thus , from lemma [ ab210 ] , @xmath540 is @xmath784-optimal .
processes @xmath785 as in the statement of theorem [ main - bcp ] will be constructed by modifying the processes @xmath785 above ( but denoted once more by the same symbols ) , by constructing successive approximations @xmath786 and @xmath787 .
these approximations are only used in the current proof and do not appear elsewhere in the paper .
consider @xmath788 such that @xmath789 .
let @xmath790 .
since @xmath791 as @xmath792 and @xmath541 is @xmath793-adapted , we have for each fixed @xmath794 , @xmath795 & \rightarrow & { \mathbb{p}}\bigl[\partial \tilde y^{(1)}_0(n ) = \varsigma | \mathcal{g}^{(n)}\bigr ] \nonumber \\[-8pt ] \\[-8pt ] \nonumber & = & 1_{\ { \partial \tilde y^{(1)}_0(n ) = \varsigma\}}\qquad \mbox{a.e . , as } \kappa\downarrow0.\end{aligned}\ ] ] note that for fixed @xmath796 and @xmath797 , @xmath798 = p^{\kappa } _ { n,\varsigma}(\mathcal{x}^{\kappa}(n)),\ ] ] for some measurable map @xmath799 $ ] satisfying @xmath800 using lemma [ appnov ] in the , we can construct , by suitably augmenting the filtered probability space , an adapted process @xmath801 such that @xmath802 , @xmath803 and @xmath804 satisfies @xmath805 = p^{\kappa } _ { n,\varsigma}(\mathcal{x}^{\kappa}(n)),\qquad 1 \le n \le p_0 , \varsigma\in\mathcal{s}_m^{\eta},\hspace*{-35pt}\ ] ] where @xmath806 .
note that if @xmath807 is a real bounded continuous map on @xmath808 , then by successive conditioning and ( [ ab401 ] ) , as @xmath792 , @xmath809 & \to & { \mathbb{e}}\biggl ( f(\tilde x ) \prod_{n=1}^{p_0 } 1_{\{\partial \tilde y_0^{(1)}(n ) = \varsigma_n\}}\biggr),\end{aligned}\ ] ] for any fixed @xmath810 .
thus , in particular , @xmath811 define @xmath812 . then @xmath813 and satisfies ( i ) and ( ii ) of theorem [ main - bcp ] . also , ( [ ab419 ] ) along with the lipschitz property of the skorohod map yields that @xmath814 as @xmath815 . fix @xmath816 sufficiently small so that @xmath817 is @xmath818-optimal and , suppressing @xmath816 , denote @xmath801 by @xmath12 , @xmath817 by @xmath4 and the generating kernels by @xmath819 .
finally , in order to ensure property ( iii ) in the theorem , we mollify the kernels @xmath819 as follows . for @xmath820 , and @xmath529 , define @xmath821 where @xmath822 is the density function of an @xmath68-dimensional normal random variable with mean 0 and variance @xmath823 .
note that the map @xmath824 is continuous for every @xmath825 and ( [ ab316 ] ) is satisfied with @xmath826 replaced with @xmath827 . from continuity of the maps @xmath828
, we can find measurable maps ( suppressing dependence on @xmath829 in notation ) @xmath534 \to\mathcal{s}_m^{\eta}$ ] , @xmath830 , such that @xmath831 is continuous , at every @xmath832 , for a.e .
@xmath173 in @xmath833 $ ] , and if @xmath834 is a uniform random variable on @xmath531 $ ] , then @xmath835 let @xmath836 be an i.i.d sequence of uniform random variables , that is , independent of @xmath298 .
now construct @xmath837 such that @xmath838 , @xmath839 and @xmath840 .
note that @xmath841 = \hat p^{\gamma}_{n,\varsigma}(\mathcal{x}^{\kappa}(n ) ) , \qquad n \ge1 , \varsigma\in\mathcal{s}_m^{\eta}.\vadjust{\goodbreak}\ ] ] since @xmath842 pointwise , as @xmath843 , we have for every real bounded map @xmath807 on @xmath808 , @xmath844 for every @xmath845 , as @xmath843 .
thus , @xmath846 , as @xmath843 .
let @xmath847 .
then @xmath848 and the above weak convergence and , once more , the lipschitz property of the skorohod map yields that @xmath849 as @xmath850 .
recall that @xmath4 is @xmath818-optimal .
we now choose @xmath829 sufficiently small so that @xmath851 is @xmath3-optimal . by construction ,
@xmath852 and @xmath853 satisfy all the properties stated in the theorem .
the goal of this section is to prove theorem [ jrtoj ] . fix @xmath854 and @xmath855 .
let @xmath320 be the @xmath856-optimal control introduced above theorem [ newmain518 ] .
fix @xmath857 , @xmath322 such that @xmath324 , as @xmath195 .
section [ subconstruct ] below gives the construction of the sequence of policies @xmath8 , @xmath858 , such that @xmath859 , yielding the proof of theorem [ jrtoj ] .
the latter convergence of costs is proved in section [ secconvprf ] . the main ingredient in this proof is theorem [ mainweak ] whose proof is given in section [ weakcgce ] . for the rest of this section @xmath4 as in theorem
[ newmain518 ] and parameters @xmath860 that specify @xmath861 shall be fixed .
in addition , let @xmath862 and @xmath863 be such that @xmath864 > m(\vartheta_{\mathrm{lip}}+1 ) \quad \mbox{and } \quad r_0 \theta > \rho,\ ] ] where @xmath865 is as in ( [ 110 ] ) .
@xmath866we will only specify a for @xmath867 and so henceforth , without loss of generality , we assume @xmath868 . in this section , since @xmath869 will be fixed , the superscript @xmath15 will frequently be suppressed from the notation .
the following additional notation will be used . for @xmath870 , define @xmath871 , @xmath872 , @xmath873 , @xmath874 and @xmath875 , where we set @xmath876 .
note @xmath877 , @xmath878 .
recall @xmath879 introduced in assumption [ ldp ] .
fix @xmath880 such that @xmath881 fix @xmath882 such that @xmath883 .
define @xmath884 also define @xmath885 as @xmath886 thus , if @xmath887 for some @xmath127 , @xmath888 and @xmath889 equals the left end point of the @xmath18-subinterval in which @xmath890 falls . otherwise , if @xmath891 for some @xmath127 , @xmath892 and @xmath893 .
recall the probability space @xmath65 introduced in section [ secsetup ] which supports all the random variables and stochastic processes introduced therein .
let @xmath894 be a sequence of uniform random variables on @xmath833 $ ] on this probability space ( constructed by augmenting the space if needed ) , independent of the @xmath261-field @xmath895 defined in ( [ fr - sigmafield - defn ] ) .
this sequence will be used in the construction of the control policy .
, of the @xmath71th server , that is , responsible for the @xmath896 activities ( @xmath897 ) .
the actual policy @xmath898 is given as a certain modification of @xmath899 that ensures feasibility and a strict positivity property . ]
the policy @xmath900 is constructed recursively over time intervals @xmath901 , as follows .
we will describe the effect of the policy on the @xmath71th server ( for each @xmath91 ) at every time - instant @xmath902 ( see figure [ policyfig ] ) .
recall the set @xmath94 introduced in section [ secsetup ] .
let for @xmath91 , @xmath903 be the ordered elements of @xmath94 ( the set of activities that the @xmath71th server can perform ) . _ step 1 _ : [ @xmath904 when @xmath905 .
let @xmath906 .
recall that @xmath907 .
write @xmath908 .
for @xmath909 , let @xmath910 note that , by our choice of @xmath911 , @xmath912 if @xmath913 is a basic activity . also , if @xmath913 is nonbasic , then @xmath914 , but since @xmath915 , we have @xmath916 .
thus , @xmath917 for every @xmath918 .
also , from assumption [ assum - ht ] [ see ( [ xstar - defn ] ) ] and recalling that @xmath919 , we get @xmath920 let @xmath921 and @xmath922 .
define , for each @xmath91 , and @xmath923 @xmath924 where for @xmath925 , @xmath634 , @xmath926 , @xmath927 and , by convention , @xmath928 , @xmath634 . since @xmath929 gives a partition of @xmath930 , the above defines the @xmath72-dimensional process @xmath931 for @xmath932 .
we set @xmath933 next , define @xmath934 for @xmath935 by the system of equations below : @xmath936 \\[-8pt ] \nonumber x^r(s ) & = & r \hat{x}^r(s / r^2),\qquad \hat x^r \mbox { defined by { ( \ref{xr - defn } ) } , } \\
t^r_j(s ) & = & \int_{[0 , s ] } 1_{\{q^r_{\sigma_1(j)}(u- ) > 0\ } } 1_{\ { q^r_{\sigma_1(j)}({\mathsf{p}}^r(u ) ) > \theta^r(u)\ } } \,dt^{r,(1)}_j(u),\qquad j \in\mathbb{j},\hspace*{-35pt } \nonumber\end{aligned}\ ] ] where ( as introduced in section [ secsetup ] ) , @xmath937 denotes the index of the buffer that the @xmath84th activity is associated with .
the above construction can be interpreted as follows .
the policy @xmath898 `` attempts '' to implement @xmath899 over @xmath938 , unless the corresponding buffer is empty ( in which case , it idles ) . over @xmath939
, @xmath898 has a similar behavior , but , in addition , it idles when the corresponding buffer does not have at least @xmath940many jobs at the beginning of the @xmath18-subintervals .
processing jobs only when there is a `` safety - stock '' at each buffer at the beginning of the @xmath18-subinterval ensures that with `` high probability '' either all the activities associated with the given buffer receive nominal time effort over the interval or all of them receive zero effort .
this , in particular , makes sure that the idling processes associated with the policy are consistent with the reflection terms for the skorohod map associated with the constraint matrix @xmath0 [ see ( [ ab1249 ] ) ] .
this property will be exploited in the weak convergence arguments of section [ secconvprf ] [ see , e.g. , arguments below ( [ ab1042 ] ) ] .
let @xmath175 be defined by ( [ i - defn ] ) . by construction , @xmath941 , @xmath942 and @xmath943 \\[-8pt ] \eqntext{\mbox{for } s\in\mathcal{i } ( 0 )
\mbox { and } t^r , i^r \mbox { are nondecreasing on } \mathcal{i}(0).}\end{aligned}\ ] ] this completes the construction of the policy and the associated processes on @xmath944 .
_ step 2 _ : [ @xmath904 for @xmath945 , for @xmath946 .
suppose now that the process @xmath947 has been defined for @xmath948 , where @xmath949 .
we now describe the construction over the interval @xmath950 .
let @xmath951 . recall that @xmath952 .
define @xmath953 where , as in section [ secsetup ] , @xmath954 , @xmath161 .
we will suppress @xmath816 and @xmath15 from the notation and write @xmath955 .
for each @xmath956 , define @xmath957 as before , ( [ ab1743 ] ) ( with @xmath958 replaced by @xmath959 ) is satisfied .
define , for each @xmath91 , and @xmath923 @xmath960 where for @xmath925 , @xmath634 , @xmath926 , @xmath961 are defined by the right - hand side of ( [ ab8a ] ) and ( [ ab8b ] ) respectively , with @xmath962 replaced by @xmath963 we set @xmath964 and define @xmath947 for @xmath945 by the system of equations in ( [ ab543 ] ) .
the above recursive procedure gives a construction for the process @xmath965 for all @xmath966 .
the policy constructed above clearly satisfies parts ( i ) , ( ii ) and ( iii ) of definition [ t - adm - defn ] .
in fact , it also satisfies part ( iv ) of the definition and , consequently , we have the following result . the proof is given in the .
[ tisadmisnew ] for all @xmath869 , @xmath858 . in this section
we prove theorem [ jrtoj ] by showing that , with @xmath967 as in section [ subconstruct ] and @xmath4 as introduced above theorem [ newmain518 ] , @xmath859 , as @xmath97 .
we begin with an elementary lemma .
[ invpr ] let @xmath968 , @xmath969 , @xmath970 be such that @xmath971 u.o.c .
let @xmath972 , @xmath969 , be such that @xmath973 as @xmath974 .
let @xmath975 be @xmath301 matrices such that @xmath976 as @xmath977 .
suppose @xmath978 and @xmath979 satisfy for all @xmath183 : @xmath980 , @xmath981 , @xmath982 } 1_{\{f^n_i(\gamma^n(s ) ) \le \varepsilon_n\}}\,ds , i \in\mathbb{i}$ ] and @xmath983 .
then @xmath984 is precompact in @xmath985 and any limit point @xmath986 satisfies for all @xmath149 , @xmath987 ; @xmath988 ; @xmath989}\tilde h(s ) \,ds$ ] , where @xmath990^{{\mathbf{i}}}$ ] is a measurable map such that @xmath991 } 1_{\{f_i(s ) > 0\ } } \tilde h_i(s ) \,ds = 0 \qquad\mbox{for all } t \ge0.\ ] ] the proof of the above lemma is given in the . recall the definitions of various scaled processes given in ( [ fl - scaled])([w - relation ] ) , and that @xmath992 for @xmath183 .
in addition , we define @xmath993 , @xmath994 , @xmath995 , @xmath183 .
[ tisadmis ] as @xmath195 , @xmath996 , u.o.c . in probability .
from the definition of @xmath997 [ see ( [ abnov8 ] ) ] , and the observation that the interval @xmath998 has length @xmath999 [ see ( [ ab1743 ] ) and ( [ ab8b ] ) ] , we have that over each interval @xmath1000 , that is , contained in @xmath1001 for some @xmath870 , the lebesgue measure of the time instants @xmath890 such that @xmath1002 equals @xmath1003 , @xmath1004 .
this is equivalent to the statement that @xmath1005 \\[-8pt ] \eqntext { \mbox{whenever } \bigl[m\delta , ( m+1)\delta\bigr ) \subset \mathcal{i}_2(n ) \mbox { for some } n \le p_0 .}\end{aligned}\ ] ] also , noting that for @xmath89 , @xmath1006 and @xmath1007 , we have that for some @xmath1008 , @xmath1009 \\[-8pt ] \eqntext{\mbox { whenever } \bigl[m\delta , ( m+1)\delta\bigr ) \subset \mathcal{i}(n ) \mbox { for some } n \le p_0 .}\end{aligned}\ ] ] fix @xmath89 and @xmath1010 such that @xmath1011 for some @xmath870 . then @xmath1012 using ( [ ab248 ] ) and ( [ ab248a ] ) and the fact that the number of @xmath18 intervals in @xmath1013 is bounded by @xmath1014
, we see that @xmath1015 thus , for some @xmath1016 , @xmath1017 also , since @xmath1018 for all @xmath183 , @xmath1019 , we get from ( [ xr - defn ] ) and standard estimates for renewal processes ( see , e.g. , lemma 3.5 of @xcite ) that @xmath1020 converges to @xmath139 u.o.c . in probability , as @xmath195 .
combining this with assumption [ assum - ht ] and ( [ ab1254 ] ) , we have @xmath1021 next , define @xmath1022 , @xmath1023 and @xmath1024\dvtx \bar q^r_{\sigma _ 1(j)}(s)=0 , \mbox{or } \bar q^r_{\sigma_1(j)}(\bar{\mathsf{p}}^r(s))\le \bar\theta^r(s ) \bigr\}\nonumber\\ & = & \bigl\{s \in[0 , t]\dvtx
\bar q^r_{\sigma_1(j)}(\bar{\mathsf{p}}^r(s))\le \bar\theta^r(s ) \bigr\ } \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & { } \cup\bigl\{s \in[0 , t]\dvtx \bar q^r_{\sigma_1(j)}(s)=0 , \bar q^r_{\sigma_1(j)}(\bar{\mathsf{p}}^r(s ) ) >
\bar\theta^r(s ) \bigr\ } \\ & { \doteq } & \mathcal{s}_j^{r , 1}(t ) \cup\mathcal{s}_j^{r , 2}(t).\nonumber\end{aligned}\ ] ] we will fix @xmath1025 for the rest of the proof and suppress @xmath173 from the notation when writing @xmath1026 , unless there is scope for confusion . using the above display and ( [ ab543 ] ) , we have that @xmath1027 } 1_{\mathcal{s}_j^r}(s ) \,d\bar t^{r,1}_j(s),\qquad j \in\mathbb{j } , t \ge0.\ ] ] using the fact that @xmath1028 along with ( [ q - relation ] ) and ( [ ab104 ] ) , we can write @xmath1029 where for @xmath89 , @xmath1030 } 1_{\mathcal{s}_j^{r,1}}(s)\ , d\bigl(\bar t^{r,1}_j(s ) - x^*_js\bigr ) \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & { } + \int_{[0 , t ] } 1_{\mathcal{s}_j^{r,2}}(s ) \,d\bar t^{r,1}_j(s)\end{aligned}\ ] ] and @xmath1031 } 1_{\mathcal{s}_j^{r,1}}(s ) \,ds = ( \operatorname{diag}(x^ * ) c'\bar\ell^r)_j(t),\ ] ] where @xmath1032 } 1_{\{\bar q^r_{i}(\bar{\mathsf{p}}^r(s))\le \bar\theta^r(s)\ } } \,ds,\qquad t \ge0 , i \in\mathbb{i}.\ ] ] since @xmath1033 [ see ( [ r - defn ] ) and ( [ refmat ] ) ] , we have @xmath1034 next , letting @xmath1035 , we have from the choice of @xmath1036 [ see above ( [ ab411 ] ) ] that @xmath1037 for some @xmath712 , where the next to last inequality makes use of assumption [ ldp ] .
recalling [ from ( [ lab23 ] ) ] that @xmath1038 , we get that @xmath1039 we note that the above convergence only requires that @xmath1040
. the property @xmath1041 will , however , be needed in the proof of proposition [ tight424 ] [ see ( [ ab632 ] ) ] .
next , using ( [ ab248 ] ) and ( [ ab248a ] ) , we have for some @xmath1042 , @xmath1043 } 1_{\mathcal{s}_j^{r,1 } } \,d\bigl(\bar t^{r,1}_j(s ) - x^*_js\bigr)\biggr| \le\frac{\varrho_1}{r } \to0 \qquad\mbox{as } r \to\infty,\ ] ] for all @xmath89 .
the above inequality follows from the fact that the integral can be written as the sum of integrals over @xmath1044-subintervals : when the subinterval is within some @xmath1045 , the integral is zero [ using the definition of @xmath889 in such intervals and ( [ ab248 ] ) ] and when the subinterval is within some @xmath1013 [ the number of such intervals is @xmath1046 which can be bounded by @xmath1047 for some @xmath1048 , the integral is bounded by @xmath1049 from ( [ ab248a ] ) . now , combining ( [ ab1254 ] ) , ( [ ab1230 ] ) and ( [ ab1240 ] )
, we have that , for each @xmath89 , u.o.c . in probability , as @xmath195 . from ( [ ab1249 ] ) , ( [ ab218n ] ) , lemma [ invpr ] and unique solvability of the skorohod problem for @xmath1050
, we now have that @xmath1051 , u.o.c . in probability , as @xmath195 .
thus , @xmath1052 converges to @xmath139 as well .
the result now follows on noting that @xmath1053 .
the following proposition gives a key estimate in the proof of theorem [ jrtoj ] .
[ tight424 ] for some @xmath1054 and @xmath1055 , @xmath1056 using standard moment estimates for renewal processes ( cf .
lemma 3.5 of @xcite ) , one can find @xmath757 such that @xmath1057 from ( [ 413half ] ) and ( [ ab1249 ] ) , we have @xmath1058 where @xmath1059 , @xmath1060 and @xmath1061 .
we rewrite the above display as @xmath1062 + \hat\zeta^r(t ) + r\hat h^r(t ) + d\hat\ell^r(t ) , \qquad t \ge0.\ ] ] from theorem 5.1 of @xcite , for some @xmath712 , @xmath1063 \\[-8pt ] \nonumber & & \qquad\le c_2 \biggl(\hat q^r + |\hat\zeta^r|_{\infty , t } + |\hat h^r|_{\infty , t } + { d_1r^{{\bolds{\mathsf{k}}}}}{r}\biggr).\end{aligned}\ ] ] also , from ( [ ab210n ] ) , ( [ ab1254 ] ) and ( [ ab1240 ] ) , for all @xmath183 and @xmath869 , @xmath1064 } 1_{\mathcal{s}_j^{r,2}}(s ) \,d\bar t^{r,1}_j(s ) \biggr| \le\varrho+ \varrho_1 . \label { ab612}\ ] ] next , using ( [ ab1230 ] ) , we get , for some @xmath715 , @xmath1065 } 1_{\mathcal{s } _
j^{r,2}}(s ) \,d\bar t^{r,1}_j(s ) \biggr)^{\upsilon\wedge2 } & \le & r^{\upsilon}c_3(t^3 + 1 ) r^{-\upsilon } \nonumber \\[-8pt ] \\[-8pt ] \nonumber & \le & c_3(t^3 + 1).\end{aligned}\ ] ] finally , for some @xmath1066 , for all @xmath1067 , @xmath1068 \\[-8pt ] \nonumber & & \qquad \le \frac{r^2t}{\delta^r}\biggl(\sum_{i \in\mathbb{i}}{\mathbb{p}}\biggl(a_i^r(\delta ^r ) \ge\frac{ar}{c_4}\biggr ) + \sum_{j \in\mathbb{j}}{\mathbb{p}}\biggl(s_j^r(\delta^r ) \ge\frac{ar}{c_4}\biggr)\biggr).\hspace*{-30pt}\end{aligned}\ ] ] using moment estimates for renewal process once more ( lemma 3.5 of @xcite ) , we can find @xmath1069 such that @xmath1070 thus , there is an @xmath1071 and @xmath1072 such that , for all @xmath1073 , @xmath1074 this shows that for some @xmath1075 and @xmath1076 , @xmath1077 the result now follows on using ( [ ab425 ] ) , ( [ ab612 ] ) , ( [ ab632 ] ) and ( [ ab629 ] ) in ( [ ab610n ] ) and observing that @xmath1078 . in preparation for the proof of theorem [ jrtoj ] , we introduce the following notation . for @xmath1079
, we define processes @xmath1080 with paths in @xmath1081 and @xmath1082 , respectively , as @xmath1083.$}\ ] ] we denote by @xmath1084 the processes with paths in @xmath1085 and @xmath1086 , respectively , defined by the right - hand side in the display above , by replacing @xmath127 by @xmath679 and @xmath1087 $ ] with @xmath1088 .
then @xmath1089 where @xmath1090 .
similarly , define the process @xmath1091 with paths in @xmath1092 .
recall @xmath1093 introduced in ( [ ab201 ] ) .
denote @xmath1094 and @xmath1095 , where we set @xmath1096 and @xmath1097 .
next , for @xmath1079 , define processes @xmath1098 with paths in @xmath1099 and @xmath1082 , respectively , as @xmath1100 also , define @xmath1101 by the first line of the above display by replacing @xmath316 by @xmath1102 .
then @xmath1103 , a.s .
, where @xmath1104 .
similarly , define the process @xmath1105 with paths in @xmath1092 .
also , let for @xmath533 , @xmath1106 , where @xmath519 is as above ( [ ab610 ] ) , and @xmath1107 .
then @xmath1108 define @xmath1109 and @xmath1110 , where @xmath1111 and @xmath1112 . then @xmath1113 and @xmath1114 .
next , let @xmath1115 then @xmath1116 let @xmath1117 note that @xmath1118 , @xmath334 and @xmath1119 are @xmath1120-valued random variables .
the following is the main step in the proof of theorem [ jrtoj ] .
[ mainweak ] as @xmath195 , @xmath1121 proof of the above theorem is given in the next subsection .
using theorem [ mainweak ] , the proof of theorem [ jrtoj ] is now completed as follows .
proof of theorem [ jrtoj ] from proposition and integration by parts , @xmath1122 & & \hspace*{18pt}{}+ { \mathbb{e}}\int_{[a^r(n)/r^2 , b^r(n)/r^2 ) } e^{-\gamma t } \bigl(h\cdot\hat q^r(t ) + \gamma p\cdot\hat
u^r(t)\bigr ) \,dt \biggr ] \\[-2pt ] & = & \sum_{n=0}^{p_0 } { \mathbb{e}}\int_{[b^r(n)/r^2 , a^r(n+1)/r^2 ) } e^{-\gamma t } \bigl(h\cdot\hat q^r(t ) + \gamma p\cdot\hat
u^r(t)\bigr ) \,dt + \varepsilon_r,\nonumber\vadjust{\goodbreak}\end{aligned}\ ] ] where , using proposition [ tight424 ] and the observation that @xmath1123 , we have that @xmath1124 as @xmath7 . from theorem [ mainweak ] , as @xmath7 , @xmath1125 .
combining this with proposition [ tight424 ] , we get for every @xmath1126 , @xmath1127 where , by convention , @xmath1128 when @xmath1129 .
next , for @xmath1130 , @xmath707 , @xmath1131 \\[-8pt ] \nonumber & = & \gamma p \cdot k { \mathbf{\mathpzc{z}}}^{r , n}(t - n\theta ) + \gamma p \cdot k\nu _ 0^{r , n}.\end{aligned}\ ] ] from theorem [ mainweak ] , as @xmath195 , @xmath1132 in @xmath1133 . combining this with ( [ ab1945 ] ) and proposition [ tight424 ] we now get similarly to ( [ ab753 ] ) , for @xmath1126 , @xmath1134 note that for @xmath1135 , @xmath1136 thus , the expression on the right - hand side of the above display equals @xmath1137 the result now follows on using this observation along with ( [ ab753 ] ) in ( [ ab804 ] ) . for @xmath89 , @xmath586 , and @xmath1138 ,
define @xmath1139\dvtx q^r_{\sigma_1(j)}(nr^2\theta+ rs , \omega ) = 0\}.\ ] ] from the definition of @xmath1140 , it follows that for some @xmath1141 , @xmath1142 as a consequence of this observation , we have the following result . the proof is given in section [ prop46 ] .
[ lemma45a ] for some @xmath1143 $ ] such that @xmath1144 as @xmath97 , we have @xmath1145 where @xmath1146 = \varnothing\}$ ] . for @xmath539 ,
let @xmath1147=(\bar\nu ^{r,0},\ldots,\bar\nu^{r , n})$ ] .
we define @xmath1148 , \nu_0^r[n ] , { \mathbf{\mathpzc{q}}}^r[n],\allowbreak { \mathbf{\mathpzc{z}}}^r[n]$ ] and their limiting analogues @xmath1149,\nu[n ] , \nu_0[n ] , { \mathbf{\mathpzc{q}}}[n ] , { \mathbf{\mathpzc{z}}}[n]$ ] in a similar fashion .
set @xmath1150 & = & ( \hat\zeta^r , \bar\nu^r[n ] , \nu^r[n ] , \nu _ 0^r[n ] , { \mathbf{\mathpzc{q}}}^r[n ] , { \mathbf{\mathpzc{z}}}^r[n]),\\ \mathcal{j}[n ] & = & ( \hat\zeta , \bar\nu[n ] , \nu[n ] , \nu_0[n ] , { \mathbf{\mathpzc{q}}}[n ] , { \mathbf{\mathpzc{z}}}[n]).\end{aligned}\ ] ] then @xmath1151 , \mathcal{j}[n]$ ] are @xmath1152$]-valued random variables , with @xmath1153 = \mathcal{d}^{{\mathbf{i } } } \times(\mathcal{s}_m^{\eta})^{\otimes(n+1 ) } \times\bigl({\mathbb{r}}^{{\mathbf{j}}(n+1)}\bigr ) \times\bigl({\mathbb{r}}^{{\mathbf{j}}(n+1)}\bigr ) \times\mathcal{d}_{\theta}^{(n+1){\mathbf{i } } } \times\mathcal{c}_{\theta}^{(n+1){\mathbf{j}}},\ ] ] where we follow the usual convention for @xmath575 . to prove the theorem , we need to show that @xmath1154\rightarrow\mathcal { j}[p_0]$ ] . in the lemma below
we will in fact show , recursively in @xmath127 , that @xmath1151\rightarrow\mathcal{j}[n]$ ] as @xmath195 , for each @xmath1155 , which will complete the proof of theorem [ mainweak ] .
[ riszero ] for each @xmath586 , @xmath1156\rightarrow \mathcal{j}[n]$ ] , as @xmath195 .
the proof will follow the following two steps : as @xmath195 , @xmath1157\rightarrow\mathcal{j}[0]$ ] .
suppose that @xmath1158\rightarrow\mathcal{j}[k]$ ] as @xmath97 for @xmath1159 , for some @xmath1160 .
then , as @xmath195 , @xmath1161\rightarrow\mathcal{j}[n+1]$ ] . consider ( i ) .
define scaled processes @xmath1162 , @xmath1163 .
processes @xmath1164 , @xmath1165 , @xmath1166 , @xmath1167 are defined similarly . by the functional central limit theorem for renewal processes and proposition [ tisadmis ]
, it follows that ( cf .
lemma 3.3 of @xcite ) @xmath1168 also , convergence of @xmath1169,\nu ^r[0])$ ] follows trivially since @xmath1170 = \bar \nu[0]=0 $ ] and @xmath1171 = \nu[0]=\varepsilon_0 y^*$ ] .
next , consider @xmath1172 = \hat y^r(\rho / r)$ ] . from the definition of the scaled processes defined above ( [ lab107 ] )
, we have that @xmath1173,\ ] ] and @xmath1174.\ ] ] also , observe that @xmath1175 can be written as @xmath1176 where , with @xmath1177 defined in ( [ lab107 ] ) , @xmath1178}1_{\check s_j^{r,0}}(s ) \,d \check t^{r,1}_j(s).\ ] ] next , note that , for a suitable @xmath757 , @xmath1179 also , @xmath1180 thus , from proposition [ lemma45a ] , @xmath1181 converges to @xmath139 in probability as @xmath195 , which shows that @xmath1182 \\[-8pt ] \eqntext { \mbox{in probability , as } r \to\infty.}\end{aligned}\ ] ] the above convergence is the key reason for introducing the modification of @xmath540 , through the vector @xmath1183 , described above theorem [ newmain518 ] .
next , standard moment bounds for renewal processes ( see , e.g. , lemma 3.5 of @xcite ) yield that @xmath1184 converges to zero in probability as @xmath195 . combining these observations
, we get from ( [ abnew558 ] ) and ( [ abnew558b ] ) that @xmath1185 converge , uniformly over @xmath1186 $ ] , in probability , to @xmath1187 .
in particular , this shows that @xmath1188 \\[-8pt ] \nonumber & & \quad \rightarrow\quad(q+\varepsilon_0ry^ * , \varepsilon_0y^ * ) = \bigl({\mathbf{\mathpzc{q}}}^{(0)}(0 ) , \nu_0^{(0)}\bigr).\end{aligned}\ ] ] finally , we prove the convergence of @xmath1189 to @xmath1190
. we will apply theorem 4.1 of @xcite . note that @xmath1191,\end{aligned}\ ] ] where @xmath1192 is a @xmath1193-valued random variable defined as @xmath1194 . from ( [ ab1042 ] ) and ( [ ab1122 ] ) @xmath1195 next , for @xmath1196 , write @xmath1197 where for @xmath89 , @xmath1198}\bigl(1 - 1_{\mathcal{s } _
d\bigl(x^*_js -\bar t^{r,1}_j(s ) \bigr ) + r \int_{[\rho / r , t]}1_{\mathcal{s}_j^{r,2}}(s ) \,d\bar
t^{r,1}_j(s),\\ \hat l^{r,0}_j(t ) & = & rx^*_j\int_{[\rho / r , t]}1_{\mathcal{s}_j^{r,1}}(s)\,ds,\end{aligned}\ ] ] with @xmath1199 defined in ( [ lab33 ] ) .
using calculations similar to those in the proof of proposition [ tisadmis ] [ see ( [ lab71 ] ) ] , we get @xmath1200 } 1_{\mathcal{s } _ j^{r,2 } } ( s ) \bar
t^{r,1}_j(s)\biggr| \to0\qquad \mbox{in probability , as } r \to\infty,\hspace*{-35pt}\ ] ] for all @xmath89 .
also , from ( [ ab248 ] ) it follows that @xmath1200 } \bigl(1 - 1_{\mathcal{s}_j^{r,1 } } ( s)\bigr)\ , d\bigl(x^*_js- \bar t^{r,1}_j(s)\bigr)\biggr| \le \frac{\delta^r}{r}.\ ] ] combining the above estimates , @xmath1201 also , @xmath1202 where for @xmath176 and @xmath1203 $ ] , @xmath1204 } 1_{\{\hat q^r_i(\bar { \mathsf{p}}^r(s ) ) \le r\bar\theta^r(s)\ } } \,ds \nonumber \\[-8pt ] \\[-8pt ] \nonumber & = & r \int_{[\rho / r , t ] } 1_{\{{\mathbf{\mathpzc{q}}}^{r,0}_i(\bar{\mathsf{p}}^r(s ) ) \le r\bar \theta^r(s)\ } } \,ds.\end{aligned}\ ] ] recall that @xmath1205 for @xmath1206 $ ] .
hence , setting @xmath1207 for @xmath1208 $ ] , we have from ( [ lab99 ] ) and ( [ ab441half ] ) @xmath1209.\ ] ] from ( [ ab430 ] ) and ( [ ab1211 ] ) we now have that , as @xmath199 , @xmath1210 in @xmath1193 .
using the definition of @xmath1211 , assumption [ ldp ] and elementary properties of renewal processes [ see similar arguments in ( [ lab71 ] ) and ( [ lab72 ] ) ] , we have that for some @xmath1212 , as @xmath195 , @xmath1213}|\hat q^r(\bar{\mathsf{p}}^r(s ) ) - \hat q^r(s)| > \varepsilon\bigr ) \le c_2\frac{r^2\theta}{\delta^r } \frac{1}{(\delta^r)^{{\bolds{\mathsf{m } } } } } = c_2 \frac{r^2\theta}{r^{{\bolds{\mathsf{k}}}({\bolds{\mathsf{m}}}+1 ) } } \to0.\ ] ] this shows that @xmath1214}|{\mathbf{\mathpzc{q}}}^{r,0}(s ) - { \mathbf{\mathpzc{q}}}^{r,0}(\bar{\mathsf{p}}^r(s))| \to0\quad \mbox{in probability , as } r \to\infty.\ ] ] using theorem 4.1 of @xcite along with ( [ ab1452n ] ) , ( [ ab444 ] ) , ( [ ab445 ] ) and ( [ ab426 ] ) , we now have that @xmath1215 as @xmath1216 valued random variables . since @xmath1217
we get from the above display that @xmath1218 as @xmath1219 . combining this with ( [ ab1042 ] ) and observations below ( [ ab1122 ] )
, we have @xmath1157 \rightarrow\mathcal{j}[0]$ ] , which completes the proof of ( i ) . we now prove ( ii ) .
we can write @xmath1220 & = & ( \mathcal { j}^r[n ] , ( \bar\nu ^{r , n+1 } , \nu^{r , n+1 } , \nu_0^{r , n+1 } , { \mathbf{\mathpzc{q}}}^{r , n+1 } , { \mathbf{\mathpzc{z}}}^{r , n+1}))\\ \mathcal{j}[n+1 ] & = & \bigl(\mathcal{j}^r[n],\bigl(\bar\nu^{(n+1 ) } , \nu^{(n+1 ) } , \nu_0^{(n+1 ) } , { \mathbf{\mathpzc{q}}}^{(n+1 ) } , { \mathbf{\mathpzc{z}}}^{(n+1)}\bigr)\bigr).\end{aligned}\ ] ] by assumption , @xmath1221 \rightarrow\mathcal{j}[n]\ ] ] and , thus , in particular , ( [ ab1122 ] ) holds .
this shows that @xmath1222 and as a consequence , using continuity properties of @xmath1223 , @xmath1224 \\[-8pt ] \nonumber & = & \bar\nu ^{(n+1)}.\end{aligned}\ ] ] in fact , this shows the joint convergence : @xmath1225 , \bar \nu ^{r , n+1 } ) \rightarrow(\mathcal{j}[n ] , \bar\nu^{n+1})$ ] .
in particular , we have @xmath1226 for the remaining proof , to keep the presentation simple , we will not explicitly note the joint convergence of all the processes being considered . from continuity of the map @xmath1140 , we now have that @xmath1227 next , we consider the weak convergence of @xmath1228 to @xmath1229 .
the proof is similar to the case @xmath1230 treated in the first part of the lemma [ cf .
below ( [ ab1122 ] ) ] and so only a sketch will be provided .
note that @xmath1231 weak convergence of @xmath1232 to @xmath1233 is a consequence of ( [ ab505 ] ) .
next , abusing notation introduced above ( [ ab1122 ] ) , define for @xmath1234 $ ] , @xmath1235 processes @xmath1236 are defined similarly to @xmath1175 . then , equations ( [ abnew558 ] ) and ( [ abnew558b ] ) are satisfied with these new definitions . hence ,
using arguments similar to the ones used in the proof of ( [ ab1042 ] ) ( in particular , making use of proposition [ lemma45a ] ) , we have that @xmath1237 converges in distribution to @xmath1238 as @xmath195 . combining the above observations
, we have , as @xmath1239 , @xmath1240 \\[-8pt ] \nonumber & & \quad\rightarrow\quad\bigl(\tilde q\bigl((n+1)\theta\bigr ) , \nu_0^{(n)}+ { \mathbf{\mathpzc{z}}}^{(n)}(\theta)+\nu^{(n+1)}\bigr)=\bigl({\mathbf{\mathpzc{q}}}^{(n+1)}(0 ) , \nu_0^{(n+1)}\bigr).\nonumber\hspace*{-30pt}\end{aligned}\ ] ] finally , we consider weak convergence of @xmath1241 to @xmath1242 .
similar to ( [ lab99 ] ) , we have @xmath1243,\ ] ] where @xmath1244 is a @xmath1193-valued random variable defined as @xmath1245 using ( [ ab604 ] ) and ( [ ab1122 ] ) , as @xmath195 , @xmath1246 weak convergence of @xmath1241 to @xmath1242 now follows exactly as below ( [ ab430 ] ) . combining the above weak convergence properties , we now have @xmath1161 \rightarrow\mathcal{j}[n+1]$ ] and the result follows
. we will only consider the case @xmath577 .
the general case is treated similarly .
let @xmath1247 . from assumption [ ldp ] , for each @xmath332 , one can find @xmath1248 such that , for @xmath176 , @xmath1249 , @xmath334 and @xmath1250 , @xmath1251\\[-8pt ] \qquad&\displaystyle{\mathbb{p}}\bigl(\bigl|\phi_i^{j , r}\bigl(s_j^r\bigl(t_j^{r,1}\bigl((k\,{+}\ , 1)\delta^r\bigr)\bigr)\bigr)\,{-}\,\phi_i^{j , r}(s_j^r(t_j^{r,1}(k\delta^r)))\,{-}\ , p^j_i\beta_j^r \tau_j^{r , k}\bigr|\,{\ge}\,\delta\delta^r\bigr)\hspace*{-30pt}&\nonumber\\ & \displaystyle\hspace*{-242pt}\,{\le}\ , \frac{c_1(\delta)}{r^{{\bolds{\mathsf{k}}}{\bolds{\mathsf{m}}}}},\hspace*{-30pt}&\nonumber\end{aligned}\ ] ] where @xmath1252 denote the union , over all @xmath1253 , of events on the left - hand side of the three displays in ( [ lab198 ] ) , by @xmath1254 .
then , the above estimates , along with ( [ lab23 ] ) , yield @xmath1255 define for @xmath1256 , @xmath1257 & & { } - c \bigl(s^r\bigl(t^{r,1}\bigl((k+1)\delta^r\bigr)\bigr)-
s^r(t^{r,1}(k\delta^r))\bigr ) \\[-2pt ] & & { } + \phi^r\bigl(s^r\bigl(t^{r,1}\bigl((k+1)\delta^r\bigr)\bigr)\bigr)- \phi^{r}(s^r(t^{r,1}(k\delta^r))).\end{aligned}\ ] ] note that , on the set @xmath1258 , we have for some @xmath1259 , @xmath1260 & & \qquad\ge q^r(k\delta^r ) + \alpha^r\delta^r - ( c - p')\operatorname{diag}(\beta^r)\biggl(x^*- \frac{\varepsilon_0y^*}{\rho } \biggr ) \delta^r - c_2\delta \delta^r{\mathbf{1}}_{{\mathbf{i } } } .\end{aligned}\ ] ] using assumption [ assum - limit - param ] , we now have that for some @xmath1261 , on the set @xmath1258 , for all @xmath869 , @xmath1262 recall that @xmath1263 .
fix @xmath1264 small enough so that for some @xmath723 and @xmath1265 , @xmath1266 then , for every @xmath1267 , @xmath1268 \\[-9.5pt ] \nonumber & \ge & q^r(k\delta^r ) + \varepsilon_1 { \mathbf{1}}_{{\mathbf{i } } } \delta ^r,\quad r \ge r_1.\hspace*{-30pt}\end{aligned}\ ] ] recall @xmath1036 introduced above ( [ ab411 ] ) .
let @xmath1269 be large enough so that @xmath1270
. then using ( [ ab1202 ] ) , we get that @xmath1271 \eqntext { \mbox{for all } k = m_0 , \ldots , m_r , r \ge r_1.}\end{aligned}\ ] ] next , let @xmath1272\}\\ & { \doteq } & g_k^r \cap b_k^r .\nonumber\vadjust{\goodbreak}\end{aligned}\ ] ] using estimates below ( [ ab1249 ] ) , we see that @xmath1273 also , from ( [ ab1202a ] ) , @xmath1274\biggr ) \nonumber \\[-10pt ] \\[-10pt ] \nonumber & = & \liminf_{r\to\infty } { \mathbb{p}}\biggl ( \bigcap_{k=0}^{m_r}(h_k^r)^c\biggr ) = 1.\vspace*{-2pt}\end{aligned}\ ] ] next , for @xmath1275 , @xmath1276 also , @xmath1277 \cap\biggl [ \bigcup_{k=0}^{m_r}h_k^r\biggr]\biggr)\nonumber\\[-3pt ] & & { } + { \mathbb{p}}\biggl ( \biggl[\bigcup_{k = m_0}^{m_r}(b_k^r \cap g^r_{m_0})\biggr ] \cap\biggl [ \bigcap_{k=0}^{m_r}(h_k^r)^c\biggr]\biggr)\\[-3pt ] & \le & { \mathbb{p}}\biggl ( \bigcup_{k=0}^{m_r}h_k^r\biggr)+{\mathbb{p}}\biggl ( \bigcup _ { k = m_0}^{m_r}(b_k^r \cap g^r_{k})\biggr),\nonumber\vspace*{-2pt}\end{aligned}\ ] ] where the last inequality is a consequence of the fact that on @xmath1278 , @xmath1279 for @xmath1280 . from ( [ ab1140 ] ) and ( [ ab1208n ] ) the above expression is seen to approach zero as @xmath195 . using this observation and ( [ ab565half ] ) in ( [ ab1235n ] ) , we now see that @xmath1281 as @xmath195 .
finally , recalling the definition of @xmath1282 , we have @xmath1283\bigr ) = { \mathbb{p}}\biggl(\bigcup_{k = m_0}^{m^r}b_k^r\biggr).\vspace*{-2pt}\ ] ] the proposition now follows on setting @xmath1284 .
[ appen ] [ appnov ] let @xmath1285 be a sequence of random variables , with values in a finite set @xmath99 , given on a probability space @xmath65 .
let @xmath260 be a sub-@xmath261 field of @xmath1286 .
suppose that @xmath1287 is a sequence of sub-@xmath261 fields of @xmath260 and @xmath1288 a sequence of @xmath1289-adapted , @xmath1290-valued random variables such that @xmath1291 where @xmath1292 $ ] are measurable maps such that @xmath1293 for all @xmath1294 , @xmath1295 .
then there is a sequence of @xmath99-valued random variables @xmath1296 defined on an augmentation of @xmath65 such that @xmath1297 where @xmath1298 . by suitably augmenting the space
, we can assume that the probability space @xmath65 supports an i.i.d .
sequence @xmath1299 of uniform @xmath531 $ ] random variables , independent of @xmath260 .
let , for @xmath1300 , @xmath1301 $ ] be measurable maps , such that for all @xmath1294 : @xmath1302 , @xmath1303 , @xmath1304 .
@xmath1305 , @xmath1306 , @xmath1307 , @xmath1295 .
the result follows on defining @xmath1309 from ( [ ab440 ] ) we have that for some @xmath757 , @xmath1310 thus , for some @xmath712 , @xmath1311 next , for @xmath183 and @xmath634 , @xmath1312 using assumption [ ab1012 ] and ( [ novab1 ] ) , we now have that for all @xmath183 , @xmath1313 this shows that , for some @xmath715 , @xmath1314 next , for some @xmath1315 , @xmath1316 } e^{-\gamma t } p \cdot d \tilde u^1(t ) - \int _ { [ 0,t ] } e^{-\gamma t } p \cdot d \tilde u^2(t ) \biggr|\nonumber\\[-2pt ] & & \qquad\le |p|
\biggl [ |\tilde u^1(0 ) - \tilde u^2(0)| + e^{-\gamma t } |\tilde u^1(t ) -
\tilde u^2(t)| \nonumber \\[-10pt ] \\[-10pt ] \nonumber & & \hspace*{78pt}\quad\qquad{}+ \gamma\int_{[0,t ] } |\tilde u^1(t ) - \tilde u^2(t)| \,dt \biggr ] \nonumber \\[-2pt ] & & \qquad\le c_4 |\tilde y^1 - \tilde y^2|_{\infty , t}.\nonumber\end{aligned}\ ] ] next , note that for @xmath1317 , @xmath634 , @xmath1318 } e^{-\gamma t } p \cdot d \tilde u^i(t ) \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & \qquad = \gamma\int_t^s e^{-\gamma t } p \cdot [ \tilde u^i(t ) - \tilde u^i(t ) ] \,dt + e^{-\gamma s } p \cdot [ \tilde u^i(s ) - \tilde u^i(t)].\end{aligned}\ ] ] also , using assumption [ ab1012 ] and ( [ ab934 ] ) , for some @xmath1319 , @xmath1320 which shows that , for @xmath634 , @xmath1321| \to0\qquad \mbox{as } s \to\infty.\ ] ] combining this observation with ( [ novab5 ] ) , we now have , on sending @xmath1322 , that for @xmath634 , @xmath1323 \,dt.\ ] ] thus , for some @xmath1072 , @xmath1324 - [ \tilde u^2(t ) - \tilde u^2(t)]| \,dt \\ & & \qquad \le c_6 \tilde{\mathbb{e}}\int_{(t,\infty ) } e^{-\gamma t } equality follows on using assumption [ ab1012 ] and ( [ novab25 ] ) . combining this with ( [ novab1 ] )
, we now have that @xmath1325 the result now follows on combining the above estimate with ( [ novab2 ] ) , ( [ novab4 ] ) and ( [ novab45 ] ) .
it is immediate from the construction that @xmath6 satisfies ( i)(iii ) of definition [ t - adm - defn ] .
we now verify that , with @xmath1326 , @xmath1327 the proof of ( [ admis1232 ] ) is similar to that of theorem 5.4 in @xcite , which shows that if a policy satisfies certain natural conditions ( see assumptions 5.1 , 5.2 , 5.3 therein ) , then it is admissible ( in the sense of definition [ t - adm - defn ] of the current paper ) .
the policy @xmath6 constructed in section [ subconstruct ] does not exactly satisfy conditions in section 5 of @xcite , but it has similar properties .
since most of the arguments are similar to @xcite , we only provide a sketch , emphasizing only the changes that are needed . for the convenience of the reader , we use similar notation as in @xcite .
also , we suppress the superscript @xmath15 from the notation .
recall from ( [ ab543 ] ) that @xmath1328 } 1_{\ { q_{\sigma_1(j)}(u ) > 0\ } } 1_{\{q_{\sigma_1(j)}({\mathsf{p}}(u ) ) > \theta(u)\ } } { \dot { t}^{(1)}_j}(u ) \,du,\\ \eqntext { j \in\mathbb{j } , t \ge0.}\end{aligned}\ ] ] in particular , assumption 5.1 of @xcite is satisfied . in view of ( [ stpos1029 ] ) , the integrand above has countably many points ( a.s . ) where the value of @xmath1329 changes from 0 to 1 ( or vice versa ) .
denote these points by @xmath1330 .
set @xmath1331 .
we refer to these points as the `` break - points '' of @xmath898 .
break - points are boundaries of the intervals of the form @xmath1332 or those of subintervals of length @xmath1333 or @xmath1334 for some @xmath1335 [ see ( [ abn532b ] ) , see also ( [ abn532a ] ) for @xmath1336 or @xmath1337 that are used to define the policy @xmath898 .
next , define @xmath1338 as the countable set of ( random ) `` event - points '' as defined in @xcite ( denoted there as @xmath1339 ) .
these are the points where either an arrival of a job or service completion of a job takes place anywhere in the network . combining the event - points and the break - points , we get the set of `` change - points '' of the policy @xmath898 denoted by @xmath1340 : @xmath1341 we will assume that the sequence @xmath1340 [ resp .
@xmath1342 , @xmath1343 is indexed such that @xmath1344 [ resp .
@xmath1345 , @xmath1346 is a strictly increasing sequence in @xmath1347 . as noted earlier , @xcite uses the notation @xmath1348 , instead of @xmath1349 , for event points .
we have made this change of notation since @xmath1350 here plays an identical role as that of event - points in the proof of @xcite .
in particular , it is easily seen that assumption 5.2 of @xcite holds with this new definition of @xmath1351
. we will next verify assumption 5.3 ( a nonanticipativity condition ) of @xcite in lemma [ appen - lemma ] below . for @xmath109 and @xmath1352 ,
let @xmath1353 .
thus , @xmath1354 is the residual ( exogenous ) arrival time at the @xmath70th buffer at time @xmath1355 , unless an arrival of the @xmath70th class occurred at time @xmath1344 , in which case it equals @xmath1356 .
similarly , for @xmath124 , @xmath1357 , define @xmath1358 .
next , write @xmath1359 , where @xmath1360 is right continuous . for @xmath109 ,
set @xmath1361 , and for @xmath1362 , @xmath1363 .
also , for @xmath1364 , and @xmath1365 , let @xmath1366 .
let @xmath1367 .
finally , define for @xmath1368 , @xmath1369 the definition of @xmath1370 above is similar to that in @xcite , with the exception of the sequence @xmath1371 .
this enlargement of the collection @xmath1370 is needed due to the randomization step , involving the sequence @xmath1372 , in the construction of the policy [ see ( [ ab201 ] ) ]
. in @xcite , part ( iv ) of the admissibility requirement ( for the smaller class of policies considered there ) was in fact shown with respect to a smaller filtration , namely , @xmath1373 . here , using the above enlargement
, we will show that part ( iv ) holds ( for the policy in section [ subconstruct ] ) with @xmath1374 .
in lemma [ appen - lemma ] below , we prove that @xmath1375 is a measurable function of @xmath1370 , for all @xmath1376 .
this shows that @xmath898 satisfies assumptions 5.15.3 of @xcite with the modified definition of @xmath1344 and @xmath1370 .
now part ( iv ) of the admissibility requirement [ i.e. , ( [ admis1232 ] ) ] follows exactly as the proof of theorem 5.4 of @xcite .
this completes the proof of the proposition .
[ appen - lemma
] @xmath1377 is a measurable function of @xmath1370 , for all @xmath1376 .
let for @xmath1378 , @xmath1379 denote the length of the @xmath41th break - point interval .
define @xmath1380 and @xmath1381 for @xmath1376 .
hence , @xmath1382 denotes the number of break - points that preceded the @xmath1347th change - point , and @xmath1383 is the `` last '' break - point before the @xmath1347th change - point @xmath1344 ( note that @xmath1384 ) for all @xmath1376 .
also , define for @xmath1376 , @xmath1385 as the `` residual '' time for the next break - point after @xmath1344 .
in particular , @xmath1386 implies that @xmath1344 itself is a break - point . by definition of @xmath898
[ see ( [ ab201 ] ) ] , it follows that @xmath1387 , @xmath1383 , @xmath1388 , and , hence , @xmath1389 are all measurable functions of @xmath1370 for @xmath1390 . summarizing this , we get @xmath1391 using notation from @xcite , let @xmath1392 be the set of all activities that are associated with the buffer @xmath70 and , for @xmath1393 , @xmath1394 be as defined by equation ( 5.2 ) of @xcite
. then @xmath1395 denotes all activities in @xmath1396 that are active at time @xmath183 , under @xmath898 . clearly , for @xmath1376 , @xmath1397 for @xmath176 , let @xmath1398 be the indicator function of the event that at the change - point @xmath1344 an arrival or service completion occurs at buffer @xmath70 .
more precisely , for @xmath176 and @xmath1368 , @xmath1399 from ( [ ch - pt-1 ] ) and ( [ chdef538 ] ) , it follows that @xmath1400 using ( [ ab201 ] ) and the construction below it , along with ( [ new545 ] ) , it is easily checked that @xmath1401 is a measurable function of @xmath1402 .
next , for @xmath89 , @xmath1403 from ( [ new545 ] ) and ( [ chdef538 ] ) , @xmath1404 and @xmath1405 are @xmath1370 measurable , thus , so is the first indicator in the above display . also ,
since @xmath1406 is either @xmath1344 or @xmath1407depending on whether @xmath1355 is in @xmath1408 or not and both @xmath1405 and @xmath1409 are @xmath1370 measurable , we see that the second indicator in ( [ abn617 ] ) is @xmath1370 measurable as well .
the lemma follows on combining the above observations .
since @xmath1410 is equicontinuous , pre - compactness of @xmath1411 is immediate .
suppose now that @xmath1412 converges ( in @xmath1413 ) , along some subsequence , to @xmath986 .
then @xmath987 for @xmath149 and @xmath1414 .
also , for suitable measurable maps @xmath1415 $ ] , @xmath176 , @xmath1416 } \tilde h_i(s ) \,ds , \qquad i \in\mathbb{i } , t\ge0.\ ] ] if @xmath1417 is a continuous map with compact support , then , along the above subsequence , @xmath1418 } \psi(f^n_i(s ) ) 1_{\{f^n_i(\gamma^n(s ) ) \le\varepsilon _ n\ } } \,ds \to \int_{[0 , t ] } \psi(f_i(s ) ) \tilde h_i(s ) \,ds,\qquad t \ge0 , i \in\mathbb{i}.\ ] ] suppose now that @xmath1419 for some @xmath332 .
then the left - hand side of the above display converges to @xmath139 and so for all @xmath1420 @xmath1421 } \psi(f_i(s ) ) \tilde h_i(s ) \,ds = 0 $ ] for such @xmath1422 .
since @xmath332 is arbitrary , we get @xmath1418 } 1_{\{f_i(s)=0\ } } \tilde h_i(s ) \,ds = h_i(t).\ ] ] the result follows .
we thank an anonymous referee for pointing us to the paper @xcite . | scheduling control problems for a family of unitary networks under heavy traffic with general interarrival and service times , probabilistic routing and an infinite horizon discounted linear holding cost are studied .
diffusion control problems , that have been proposed as approximate models for the study of these critically loaded controlled stochastic networks , can be regarded as formal scaling limits of such stochastic systems . however , to date , a rigorous limit theory that justifies the use of such approximations for a general family of controlled networks has been lacking .
it is shown that , under broad conditions , the value function of the suitably scaled network control problem converges to that of the associated diffusion control problem .
this scaling limit result , in addition to giving a precise mathematical basis for the above approximation approach , suggests a general strategy for constructing near optimal controls for the physical stochastic networks by solving the associated diffusion control problem . . . |
generalized parton distributions ( gpds ) @xcite provide a detailed description of the nucleon in terms of partonic degrees of freedom , which allows , e.g. , to address the three - dimensional distribution of quarks and gluons and the partonic decomposition of the nucleon spin @xcite .
their determination improves our understanding of non - perturbative qcd dynamics in general .
more specifically , they are a necessary ( though not sufficient ) input for the theoretical description of multiple - hard reactions in proton - proton collisions at lhc collider @xcite and for other applications reaching beyond the limit of collinear qcd .
information on gpds comes from many sides .
there exists already an impressive experimental data base for exclusive reactions , which is continuously improved .
a standard approach is either confrontation of model predictions @xcite or least - squares fit @xcite to this data .
the latter approach , formulated in mellin space , is analogous to the well - established global fitting framework of parton distribution functions ( pdfs ) .
complementary information on moments of gpds comes from lattice qcd @xcite .
however , compared to the situation for pdfs , extracting gpds from data is a much more intricate task and model ambiguities are much larger .
for instance , the theoretically cleanest process , used for the determination of gpds , is deeply virtual compton scattering ( dvcs ) , @xmath1 , which can be parametrized by twelve compton form factors ( cffs ) . at leading order in the ( inverse )
photon virtuality squared @xmath2 they are expressed as convolutions of gpds with perturbatively calculable coefficient functions . at leading order in the qcd coupling constant and for one of the cffs , @xmath0 , we have @xmath3 h ( x,\xi , t,{\cal q}^2)\ , , \label{eq : calh}\ ] ] where @xmath4 , @xmath5 is bjorken s scaling variable , @xmath6 is the momentum transfer squared , and @xmath7 is the gpd .
formulae such as ( [ eq : calh ] ) suggest to parametrize gpds at some input scale @xmath8 , to employ qcd evolution equations to determine the gpds at the desired scale @xmath9 , then to make the momentum fraction convolution to obtain cffs that are used to calculate measured cross - sections and asymmetries via known formulae @xcite , and , finally , to confront the result with experimental data by means of the least - squares method . due to the facts that gpds can not be fully constrained even by ideal data and that they depend at the input scale @xmath8 on three variables , the space of possible functions , although restricted by gpd constraints , is huge . as a result , the theoretical uncertainty induced by
the choice of the fitting model is much more serious than in the pdf case , where the model functions depend at the input scale only on one variable , namely , the longitudinal momentum fraction @xmath10 .
the fact that exclusive data is typically much less precise than inclusive one exacerbates the situation further .
in this paper we explore an alternative approach , in which _ neural networks _
are used in place of specific models .
this essentially eliminates the problem of model dependence and , as an additional advantage , facilitates a convenient method to propagate uncertainties from experimental measurements to the final result . in the context of nucleon structure studies ,
neural networks have already been successfully applied to extract pdfs by the nnpdf group @xcite and to parametrize electromagnetic form factors @xcite .
similarly , they have been employed for parametrization of spectral function of hadronic tau decays @xcite . in the case of gpds , because of the abovementioned reasons we expect that the advantages of this method should be even more pronounced . while our long - term goal are global fits of gpds , the data presently available allows only a less ambitious analysis , namely the fit of the dominant cff @xmath0 using neural networks @xcite ] .
this reduces the mathematical complexity of the problem ( increased also by gpd constraints ) , while still retaining the relevance for the study of hadron structure .
the imaginary part of the cff ( [ eq : calh ] ) is at leading order @xmath11 and so , although we consider only extraction of cffs , @xmath12 provides us with direct information about the shape of gpd @xmath7 on the cross - over line @xmath13 .
moreover , a dispersion relation " @xcite @xmath14 - { \cal c}(t,{\cal q}^2)\ , , \label{eq : recalh}\ ] ] can then be used as a sum rule to pin down this gpd outside of the kinematically accessible region .
extraction of cffs from dvcs data in a largely model - independent way was also performed in @xcite .
in contrast to the work presented here , these analyses are local in the sense that they extract values of cffs separately at each measured kinematic point .
we emphasize that in refs .
@xcite the systematic uncertainties , induced by other non - dominant cffs , were explored with model - dependent constraints for the four twist - two related cffs .
the paper is organized as follows . in section [ sec : nnmethod ] we describe salient features of the neural network approach . in section [ sec : fit ] we present fits to a well - defined subset of available dvcs data , so that we could study neural network method in a controlled situation , free of subtleties which surface in many fits involving data coming from a number of different experiments .
in particular , we used hermes measurements of two observables , beam spin asymmetry ( bsa ) and beam charge asymmetry ( bca ) in leptoproduction of a real proton ( of which dvcs is a subprocess ) @xcite .
this data belongs to a kinematic region where these two asymmetries are determined essentially by the imaginary and the real part of the compton form factor @xmath0 ( [ eq : calh ] ) , respectively , and where dependence of @xmath0 on @xmath9 is weak and , therefore , can be neglected for simplicity .
the neural network parameterization of cff @xmath0 is then used to make a prediction for another observable , namely , the beam charge - spin asymmetry in the kinematic region to be explored by the compass ii experiment @xcite . in section [ sec : comparison ]
we place particular emphasis on the determination of uncertainties .
to do so we fit a simple model using the standard least - squares method and compare the resulting uncertainties with those obtained by neural networks .
section [ sec : conclusion ] contains conclusions .
each blob symbolizes a neuron and thickness of arrows represents the strengths of weights @xmath15 .
( smaller blobs with no input are so - called _ biases _ they improve the network s properties @xcite . ) ] neural networks have been applied to a variety of tasks , including optimization problems .
one of the main attractions is their apparent ability to mimic the behavior of human brain : with their multi - connectedness and parallel processing of information they can be trained to perform relatively complex classification and pattern recognition tasks . to this end
many kinds of neural networks and corresponding `` learning '' algorithms have been developed , see , e.g. , @xcite . in our study , we used one of the most popular neural network types , known as _ multilayer perceptron_. schematically shown on fig .
[ fig : perceptron ] , it is a mathematical structure consisting of a number of interconnected `` neurons '' organized in several layers .
each neuron has several inputs and one output .
the value at the output is given as a function @xmath16 of a sum of values at inputs @xmath17 , each weighted by a a certain number @xmath15 .
consequently , a multilayer perceptron is defined by its architecture ( number of layers and number of neurons in each layer ) , by the values of its weights , and by the shape of its _ activation function _ @xmath18 .
again somewhat inspired by biological neuron properties , the activation function is often taken as a nonlinear function with saturating properties for large and small values of its argument .
we employed a very popular logistic sigmoid function @xmath19 for neurons in inner ( `` hidden '' ) layer(s ) , while for input and output layers we used the identity function @xmath20 . by iterating over the following steps the network
is then trained , i.e. , it `` learns '' how to describe a certain set of data points : 1 .
kinematic values ( two in our case : @xmath5 and @xmath6 ) of the first data point are presented to two input - layer neurons 2 .
these values are then propagated through the network according to weights and activation functions . in the first iteration , weights are set to some random value .
3 . as a result , the network produces some resulting values of output in its output - layer neurons . here
we have two : @xmath12 and @xmath21 . obviously ,
after the first iteration , these will be some random functions of the input kinematic values : @xmath22 and @xmath23 .
4 . using these values of cff(s ) ,
the observable(s ) corresponding to the first data point is ( are ) calculated and it is ( they are ) compared to actually measured value(s ) , with squared error used for building the standard @xmath24 function .
the obtained error is then used to modify the network : it is , possibly weighted by the inverse uncertainty of the experimental measurement , propagated backwards through the layers of the network and each weight is adjusted such that this error is decreased .
the concrete algorithm for this weight adjustment is discussed at the end of this section .
this procedure is then repeated with the next data point , until the whole training set is exhausted .
this sequence of steps ( called training _ epoch _ ) is repeated until the network is capable to describe experimental data with a sufficient accuracy the precise stopping criterion is specified below . from what is
said it should be clear that a neural network is nothing more but a complicated non - linear multi - parameter function and that the training procedure is equivalent to the least - squares fitting of this function to data .
the actual power of the approach stems from the following : * a neural network serves as _ unbiased _ interpolating function .
its complicated dependence on parameters enables it to approximate any smooth function with comparable ease can be arbitrary accurately approximated by a multilayer perceptron .
strictly spoken , one hidden layer is enough for this , but networks with more hidden layers can have fewer neurons and/or can be more efficiently trained . ] .
* updating the weights of a given neuron by back - propagation of error uses only values and gradients of activation functions in the immediate network neighborhood . in this sense
the training process is local and it is algorithmically efficient .
* this approach enables the following convenient method for propagation of experimental uncertainties ( and even their correlations ) into the final result : in our application a `` replica data set '' is interpolated .
it is obtained from original data by generating random artificial data points using gaussian probability distribution with a width defined by the error bar of experimental measurements . taking a large number @xmath25 of such replicas , the resulting family of trained neural networks @xmath26
defines a probability distribution of the represented cff @xmath27 and of any functional @xmath28 $ ] thereof .
thus , the mean value of such a functional and its variance are @xcite @xmath29 \big\rangle & = \frac{1}{n_{rep}}\sum_{k=1}^{n_{rep } } \mathcal{f}[\mathcal{h}^{(k)}]\ ; , \label{eq : funcprob } \\
\big(\delta \mathcal{f}[\mathcal{h}]\big)^2 & = \big\langle \mathcal{f}[\mathcal{h}]^2 \big\rangle - \big\langle \mathcal{f}[\mathcal{h } ] \big\rangle^2 \;. \label{eq : variance}\end{aligned}\ ] ] for details of this `` monte carlo '' procedure see refs .
@xcite and note that this is a general method of error propagation , which is not related to neural networks themselves and can be used also for error propagation in standard least - squares fitting of model functions to data .
the power of neural networks to approximate any continuous function implies also the danger of overfitting the data ( also known as _ overtraining _ ) .
namely , after a certain number of iterations the network will not only describe the general dependence of observables on kinematic variables , but will also adjust to the random fluctuations of data .
this unwanted behavior is prevented by the cross - validation procedure in which initial data is divided into two sets : _ training _ set and _ validation _ set .
then performance of the network is continuously checked on the validation set . since this set
is not used for training , after the onset of overfitting , the error of the network s description of validation sets will increase , see fig . [
fig : crossvalidation ] .
this is the moment at which training is stopped . for more informations on neural networks
we refer the reader to the vast literature on the subject , e.g. @xcite , while the phd thesis @xcite is a good reference for the particular approach used also in this work . in the remainder of this section
, we specify some details about the neural networks we used . in our study we employed the pybrain software library for neural networks @xcite
. this choice was motivated by ( i ) the great flexibility of this library , which allows to explore various types of neural networks , and ( ii ) it being written in the python programming language , such that it was easy to link with our already existing python code for dvcs analysis . like almost every other freely available neural network software ( at least to our knowledge )
, pybrain expects that the network output is directly compared to training data samples for error calculation . in our case , however , the network output must be transformed , i.e. , the dvcs observables must be calculated from @xmath12 and @xmath21 ( see step no . 4 in the training procedure described above ) , before the error can be evaluated . for this
, we had to make some modifications to the pybrain software .
concerning the network architecture , the number of neurons in the input and output layers were fixed by the problem at hand to be two , see fig .
[ fig : perceptron ] . to fix the number of hidden layers and their neurons
, one has to consider a trade - off between two requirements : the number of neurons in the hidden layer(s ) must be large enough to ensure sufficient expressive power of the network but must not be too large with respect to the number of available data points , otherwise the training will become difficult . in our case , it is already known from previous studies , e.g. @xcite , that cff @xmath0 is a reasonably well - behaved function .
thus , since we used a relatively small sample of few tens of data points , see next section , just one hidden layer with about ten neurons proved to be sufficient .
the specific results presented in this paper were obtained by training series of neural networks , with 13 neurons in the hidden layer each .
we checked that results do not change appreciably if a neuron is added or removed from the hidden layer , which is one of the standard tests in neural network applications .
there are various training algorithms available for updating the network weights , from a simple back - propagation algorithm @xcite to genetic algorithms like the one used by the nnpdf group @xcite . after some experimentation we ended up using the so called _ resilient back - propagation _ algorithm @xcite .
this algorithm is a modification of the usual back - propagation algorithm : only the signs of the partial derivatives of the error function , and not their magnitude are taken into account .
resilient back - propagation is known to be very reliable , fast and insensitive to parameters of the network and of the learning process .
more sophisticated algorithms are needed when correlations are included in the @xmath24 error function or in cases for which the connection between the network output and the experimental observables is very non - linear .
the latter would , e.g. be the case when the network output would represent gpds rather than cffs , so that the connection would involve convolutions with qcd evolution operators and hard - scattering coefficient functions .
let us finally note that besides perceptrons , there are attempts to apply other kinds of neural networks in nucleon structure studies , such as self - organizing maps @xcite .
for the first application , to assess the power of the neural network approach , we took just two sets of measurements of leptoproduction of a real photon by scattering leptons off unpolarized protons .
one set consisted of 18 measurements of the first sine harmonic @xmath30 of the beam spin asymmetry ( bsa ) ) and we disregard the difference of these two observables in this paper . ]
@xmath31 ( where @xmath32 is the azimuthal angle in the so - called trento convention ) , while in the other set there were 18 measurements of the first cosine harmonic @xmath33 of the beam charge asymmetry ( bca ) @xmath34 both sets cover identical kinematic regions @xmath35 as mentioned in the introduction , we assumed that qcd evolution effects can be neglected , i.e. , that cffs are independent of @xmath9 .
furthermore , it has been shown that the hypothesis of cff @xmath0 dominance leads to a successful description of this data @xcite and so we relied on this assumption ( with the understanding that it is only an approximation which should be given up , once sufficiently precise data is available ) .
thus , at present , just a single cff @xmath27 , or two real - valued functions @xmath22 and @xmath23 , are extracted from data by our neural networks . a comment about the relation between @xmath12 and @xmath21 is in order .
analytic properties of cff @xmath0 relate these two functions via a `` dispersion relation '' , which reads in twist - two approximation : @xmath36 cf
. eqs .
( [ eq : dvcslo ] ) and ( [ eq : recalh ] ) . as discussed in the introduction ,
this could be used to simplify the fitting function set from @xmath37 however , for two reasons we treat here the imaginary and real part of cff @xmath0 as independent quantities .
first , assumption of `` dispersion relation '' ( [ eq : dr ] ) by itself introduces a theoretical bias we want to avoid . in section [ sec : comparison ]
we will see that model - dependent least - squares fits that are based on the dispersion relation " ( [ eq : dr ] ) link , e.g. , assumptions about the low @xmath10 behaviour of @xmath12 to that of @xmath21 which might lead to rather misleading results . instead
, neural networks which do not make use of ( [ eq : dr ] ) give large uncertainties for both , which is probably more realistic .
+ second , to implement the dispersion relation " approach in a neural network framework one would need to either ( i ) disconnect topologically the network output representing @xmath38 from the input representing @xmath5 , or ( ii ) to create a separate neural network with one input and one output ( 11 ) for @xmath38 and train it together with the two - inputs - one - output ( 21 ) network representing @xmath22 . certainly , taking in future all available fixed target data into account , such a strategy might be feasible , and if so , would allow to get rid of the @xmath0 dominance hypothesis and might thus improve the phenomenological value of the final result .
( [ eq : bca ] ) and first sine harmonic of beam spin asymmetry @xmath30 ( [ eq : bsa ] ) resulting from neural network fit ( hatched areas ) , shown together with data @xcite , used for training .
two model fits , km09a ( solid ) and km09b ( dashed ) from @xcite are also shown for comparison . ] using the procedure described in section [ sec : nnmethod ] , we created 50 replicas from the set of 36 measured data points , which were randomly divided into a training set of 25 points and a validation set of 11 points .
then we trained one neural network on each training set , monitoring progress on the validation set , as in fig .
[ fig : crossvalidation ] .
training was performed either until the onset of overlearning , or for 2000 epochs , whichever came first .
we ended up with 50 neural networks , each representing cff @xmath0 .
most of the neural networks fit the original data set quite well ( 44 of our 50 neural networks have @xmath24 less than the number of data points , 36 ) , with few poorer fits as expected due to the stochastic nature of the set of data replicas . using this set of neural networks as a probability distribution in a functional space of all
@xmath0 one can predict any function of @xmath0 , together with its uncertainty , using ( [ eq : funcprob ] ) and ( [ eq : variance ] ) . as a consistency check we plot in fig .
[ fig : hermes09 ] the result for the very observables that are used for training of networks .
we observe that uncertainties of measurements have been correctly propagated into the corresponding uncertainties given by neural networks .
this is particularly visible in the first two panels of fig .
[ fig : hermes09 ] , where uncertainties of both data and neural networks increase with @xmath39 in the same fashion . error bands here and in other figures denote the uncertainty that corresponds to one standard deviation ( one sigma ) .
we checked for departures from a gaussian distribution of network results and found them to be small in the region where data is available .
this means that a one - sigma error band corresponds indeed to a 68% confidence level .
however , as observed also by the nnpdf group @xcite , in the extrapolation region , where there is no data , departures from gaussian distribution are significant , and 68% confidence level regions are generally smaller than the one - sigma regions we plot . ) in compass ii kinematics ( @xmath40 , @xmath41 ) ( hatched areas ) . two model fits , km09a ( solid ) and km09b ( dashed ) from @xcite are also shown for comparison . ] as an example of a proper prediction coming from our analysis we plot in fig .
[ fig : compass ] the beam charge - spin asymmetry ( bcsa ) , @xmath42 as a function of @xmath6 , for several kinematic points that are characteristic for the compass ii experiment ( where the muon is taken to be massless and the polarization is set equal to 0.8 ) .
this experiment was chosen because its kinematics overlaps with that of the hermes data used for neural network training .
hence these predictions represent partly interpolation and partly extrapolation of hermes data , thus testing this framework in a nontrivial way .
one of the main advantages of the neural network approach to hadron structure is that it offers a convenient way to assess the uncertainties of non - perturbative functions ( gpds , cffs , pdfs , ) .
let us now emphasize this point further by making a comparison with the common method in which one chooses some model and makes a least - squares fit of it to the data . to quantify this comparison
, we took a simple model of the cff @xmath0 and fitted it to the very same data to which neural networks were trained .
we adopted a version of the model used in @xcite for which the partonic decomposition of @xmath12 is : @xmath43\,.\end{aligned}\ ] ] we parametrized the gpds along the cross - over trajectory @xmath13 as : @xmath44 here @xmath45 is the normalization of pdf @xmath46 from pdf fits , @xmath47 is the skewness ratio at small @xmath10 ( the ratio of a gpd at some point on the cross - over trajectory and the corresponding pdf ) , @xmath48 is the `` regge trajectory '' , @xmath49 controls the large-@xmath10 behavior , and @xmath50 and @xmath51 control the @xmath6-dependence @xcite .
the parameters of the sea - quark gpd @xmath52 were taken to be as in @xcite ( @xmath53 , @xmath54 , @xmath55 , @xmath56 , @xmath57 , @xmath58 ) .
for the valence quark gpd @xmath59 , we also fixed @xmath60 , @xmath61 , and @xmath62 .
this left @xmath63 , @xmath64 and @xmath65 as free parameters , to be determined by the fit to the data .
we obtained @xmath21 from the dispersion relation " ( [ eq : dr ] ) , treating the subtraction constant @xmath66 as a fourth and final free parameter of the model ) to be @xmath6-independent , @xmath67 .
this is motivated a posteriori by the fact that such a simple model is good enough to fit the data . additionally , when we parametrize @xmath68 , the parameter @xmath69 tends to be strongly correlated with other parameters of the model , making error analysis , which is our main task , more difficult . ] .
fitting this model to the same 36 data points used for neural network fit , using the well - known minuit software package @xcite for minimization of the @xmath24 function , resulted in a good fit ( @xmath70 ) with parameter values @xmath71 the uncertainty of the resulting fit is commonly determined by the so - called hessian method .
the hessian matrix is given by the second derivatives of @xmath24 with respect to the model parameters @xmath72 at the minimum @xmath73 : @xmath74 the uncertainty @xmath75 of the function @xmath76 of these parameters , including the uncertainty of the parameters themselves , is given by the error propagation formula @xmath77 in a textbook approach to statistics , one sigma uncertainty is obtained by taking @xmath78 in this formula .
however , there are several problems with this simple procedure .
first , @xmath24 tends to vary very differently in different directions of parameter space , yielding a hessian with large variations in the eigenvalue spectrum which might then imply numerical difficulties . in our case , where eigenvalues vary over three orders of magnitude , the problem was not so severe .
however , in dvcs fits , involving more parameters , we have also observed much stronger variations . to improve the reliability of the hessian method
, we followed an iterative procedure @xcite which allows to find natural directions in parameter space and natural step sizes for the finite - difference calculation of the hessian matrix .
second , it is known that with global fits , i.e. , when one combines data from several different experiments , errors are not distributed in the expected way . often , reasoning strictly within textbook statistics , one would have to reject some data as incompatible either with itself ( i.e. , because its @xmath24 is too large ) or with other data sets ( i.e. , because likelihood curves poorly overlap ) .
this is mainly due to systematic errors which are not understood . without entering further into the discussion of this complex issue
, we can follow the cteq procedure @xcite , where @xmath79 in ( [ eq : errorprop ] ) is increased ad hoc to @xmath80 in order to accommodate all experimental measurements used for fitting .
the tolerance parameter @xmath81 is determined by calculating the average distance from the best fit along eigenvectors of the hessian matrix at which the model is still compatible with some particular experiment at some confidence level ( c.l . ) . in global pdf fits , with many data sets , values of @xmath82 are common , see e.g. @xcite for a recent review .
nevertheless , using this procedure , we arrived at a tolerance @xmath83 ( for 68% c.l . ; cteq use 90% ) .
we conclude that in our case with just two sets of data , coming from the same experiment , there are no statistical inconsistencies and we can actually use the textbook formula ( [ eq : errorprop ] ) with @xmath84 , which we did will have to be increased for a true global fit . ] .
we note in passing that we found that model parameters are significantly more constrained by the bca than by the bsa data . and @xmath85 ( ascending hatches ) from hermes bca and bsa @xcite data compared with three model fits , one of which with determined uncertainties ( descending hatches ) . [
fig : cff ] ] we can now compare the model fit ( [ eq : tdr3 ] ) , taking also into account uncertainties calculated by the procedure just described , with the neural network parameterization of the cff @xmath0 .
both findings are displayed in fig .
[ fig : cff ] , together with two models from @xcite , where @xmath12 ( upper panels ) and @xmath21 ( lower panels ) are separately plotted .
we notice several interesting features : * in the kinematic region of measured data ( this is roughly the middle vertical third of the left panels ) , there is a good agreement of values and uncertainties of neural networks ( ascending hatches ) and model fit ( descending hatches ) .
this shows that both fitting methods correctly _ interpolate _ the data , and that two , quite different , statistical methods used for determination of uncertainties are mutually consistent . * as one starts to _ extrapolate _ the fitted cff @xmath86 outside of the data region ( left and right thirds of left panels and the whole of the right panels ) , the two methods predict markedly different shapes and uncertainties for the cff @xmath86 .
all model fits show effects of theoretical bias by following the @xmath87 functional behavior at small @xmath10 and claiming a small uncertainty there , whereas in neural network parameterizations it is obvious that extrapolated functions are very unconstrained .
this is particularly visible in the right panels , illustrating the difficulty of a model - independent extrapolation towards @xmath88 , which is a limit of particular interest for hadron structure studies . *
the model fit and the corresponding uncertainties ( descending hatches ) show the effect of the dispersion relation integral constraint ( [ eq : dr ] ) @xmath12 is significantly constrained also outside of the data region .
this is actually a welcome feature of the dispersion relation fits , offering the opportunity to access nucleon structure in regions not directly accessible in the experiment .
however , its usefulness depends on the extent to which the functional forms used are firmly established . *
the model km09b ( dashed line ) , which was obtained in @xcite by including additional data , and , more importantly , by extending the model to include contributions of the cffs @xmath89 and @xmath90 , disagrees for @xmath21 with the other three fits .
this emphasizes that the assumed @xmath0 dominance might substantially affect the quality of our results .
based on hermes measurements of lepton scattering off unpolarized protons , a reaction which should be dominated by the cff @xmath91 , we performed the first neural network analysis of deeply virtual compton scattering data and obtained a neural network representation of the cff @xmath0 .
this was then used to make a prediction for the beam charge - spin asymmetry , measurable at compass ii .
the monte carlo method of propagation of experimental errors enabled us to determine also the uncertainty of the resulting cff , and we have found that ( in the kinematic region where data is available ) it agrees well with the uncertainty determined by model fits and standard statistical procedures . however , neural networks combined with monte carlo error propagation are a conceptually cleaner and practically simpler method .
the propagation of uncertainty from experimental data is intrinsic .
most importantly , this provides essentially a model - independent approach to determine cffs and gpds .
one could argue that the neural network approach , just because it is essentially model - independent , does nothing more than faithfully representing the information available in the experimental data by universal objects : cffs ( as presented here ) or gpds ( yet to be extracted by this approach ) .
this is valuable in itself but can also be considered as a stepping stone towards improved model - dependent studies , that in principle offer the advantage that our detailed theoretical and phenomenological understanding of qcd dynamics can be taken into account .
such a two - step procedure would be quite natural as models can be more directly compared to neural - network determined cffs or gpds than to actual measured observables .
finally , let us stress that we consider the presented neural network study only as a first step in the analysis of dvcs data .
the problem here is that the number of observables measured at a given kinematic point is in practice not sufficient to pin down all twelve cffs , some of which are kinematically suppressed .
thanks juan rojo for many valuable comments and zvonimir vlah for discussions on neural networks . d.m . and k.k .
are grateful to the institut fr theoretische physik , university of regensburg and to theory group of the nuclear science division at lawrence berkeley national laboratory for their warm hospitality .
this work was supported by the bmbf grant under the contract no .
06ry9191 , by eu fp7 grant hadronphysics2 , by dfg grant , contract no .
436 kro 113/11/0 - 1 and by croatian ministry of science , education and sport , contract no .
119 - 0982930 - 1016 .
d. mller , d. robaschik , b. geyer , f. m. dittes and j. hoeji , _ wave functions , evolution equations and evolution kernels from light - ray operators of qcd _ , _ fortschr .
* 42 * ( 1994 ) 101 [ http://arxiv.org/abs/hep-ph/9812448[hep-ph/9812448 ] ] .
a. v. radyushkin , _ scaling limit of deeply virtual compton scattering _ , _ phys .
* b380 * ( 1996 ) 417425 [ http://arxiv.org/abs/hep-ph/9604317[hep-ph/9604317 ] ] .
ji , _ deeply - virtual compton scattering _ , _ phys .
* d55 * ( 1997 ) 71147125 [ http://arxiv.org/abs/hep-ph/9609381 [ hep - ph/9609381 ] ] .
ji , _ gauge invariant decomposition of nucleon spin _ , _ phys .
* 78 * ( 1997 ) 610613 [ http://arxiv.org/abs/hep-ph/9603249[hep-ph/9603249 ] ] .
l. frankfurt , m. strikman and c. weiss , _ dijet production as a centrality trigger for @xmath92 collisions at cern lhc _ , _ phys.rev . _
* d69 * ( 2004 ) 114010 [ http://arxiv.org/abs/hep-ph/0311231[hep-ph/0311231 ] ] .
b. blok , y. dokshitzer , l. frankfurt and m. strikman , _ the four jet production at lhc and tevatron in qcd _ , _ phys .
* d83 * ( 2011 ) 071501 [ http://arxiv.org/abs/1009.2714[1009.2714 ] ] .
m. diehl and a. schfer , _ theoretical considerations on multiparton interactions in qcd _ , _ phys .
* b698 * ( 2011 ) 389402 [ http://arxiv.org/abs/1102.3081[1102.3081 ] ] .
s. v. goloskokov and p. kroll , _ the role of the quark and gluon gpds in hard vector - meson electroproduction _ , _ eur . phys . j. _
* c53 * ( 2008 ) 367384 [ http://arxiv.org/abs/0708.3569[0708.3569 ] ] .
k. kumeriki and d. mller , _ deeply virtual compton scattering at small @xmath5 and the access to the gpd h _ , _ nucl .
_ * b841 * ( 2010 ) 158 [ http://arxiv.org/abs/0904.0458[0904.0458 ] ] .
collaboration , j. d. bratt _ et .
_ , _ nucleon structure from mixed action calculations using 2 + 1 flavors of asqtad sea and domain wall valence fermions _ , _ phys .
rev . _ * d82 * ( 2010 ) 094502 [ http://arxiv.org/abs/1001.3620[1001.3620 ] ] .
a. v. belitsky , d. mller and a. kirchner , _ theory of deeply virtual compton scattering on the nucleon _ , _ nucl .
phys . _ * b629 * ( 2002 ) 323392 [ http://arxiv.org/abs/hep-ph/0112108[hep-ph/0112108 ] ] .
a. belitsky and d. mller , _ exclusive electroproduction revisited : treating kinematical effects _ , _
phys.rev . _ * d82 * ( 2010 ) 074010 [ http://arxiv.org/abs/1005.5209[1005.5209 ] ]
. collaboration , r. d. ball _ et .
al . _ , _ a determination of parton distributions with faithful uncertainty estimation _ , _ nucl .
* b809 * ( 2009 ) 163 [ http://arxiv.org/abs/0808.1231 [ 0808.1231 ] ] .
collaboration , r. d. ball , l. del debbio , s. forte , a. guffanti , j. i. latorre _ et .
_ , _ a first unbiased global nlo determination of parton distributions and their uncertainties _ ,
_ nucl.phys .
_ * b838 * ( 2010 ) 136206 [ http://arxiv.org/abs/1002.4407 [ 1002.4407 ] ] . r. d. ball _ et .
_ , _ impact of heavy quark masses on parton distributions and lhc phenomenology _ , http://arxiv.org/abs/1101.1300[1101.1300 ]
. k. m. graczyk , p. plonski and r. sulej , _ neural network parameterizations of electromagnetic nucleon form factors _ , _ jhep _ * 09 * ( 2010 ) 053 [ http://arxiv.org/abs/1006.0342[1006.0342 ] ] .
j. rojo and j. i. latorre , _ neural network parametrization of spectral functions from hadronic tau decays and determination of qcd vacuum condensates _ , _ jhep _ * 0401 * ( 2004 ) 055 [ http://arxiv.org/abs/hep-ph/0401047[hep-ph/0401047 ] ] .
s. forte , l. garrido , j. i. latorre and a. piccione , _ neural network parametrization of deep - inelastic structure functions _ , _ jhep _ * 05 * ( 2002 ) 062 [ http://arxiv.org/abs/hep-ph/0204232 [ hep - ph/0204232 ] ] .
collaboration , l. del debbio , s. forte , j. i. latorre , a. piccione and j. rojo , _ unbiased determination of the proton structure function f(2)**p with faithful uncertainty estimation _ , _ jhep _ * 0503 * ( 2005 ) 080 [ http://arxiv.org/abs/hep-ph/0501067[hep-ph/0501067 ] ]
. o. v. teryaev , _ analytic properties of hard exclusive amplitudes _ , http://arxiv.org/abs/hep-ph/0510031[hep-ph/0510031 ] .
k. kumeriki , d. mller and k. passek - kumeriki , _ towards a fitting procedure for deeply virtual compton scattering at next - to - leading order and beyond _ , _ nucl .
* b794 * ( 2008 ) 244323 [ http://arxiv.org/abs/hep-ph/0703179[hep-ph/0703179 ] ] .
m. diehl and d. y. ivanov , _
dispersion representations for hard exclusive processes _ , _ eur .
phys . j. _ * c52 * ( 2007 ) 919932 [ http://arxiv.org/abs/0707.0351[0707.0351 ] ]
. k. kumeriki , d. mller and k. passek - kumeriki , _ sum rules and dualities for generalized parton distributions : is there a holographic principle ? _ , _ eur .
phys . j. _ * c58 * ( 2008 ) 193215 [ http://arxiv.org/abs/0805.0152[0805.0152 ] ] .
m. guidal , _ a fitter code for deep virtual compton scattering and generalized parton distributions _ , _ eur .
phys . j. _ * a37 * ( 2008 ) 319332 [ http://arxiv.org/abs/0807.2355[0807.2355 ] ] .
m. guidal and h. moutarde , _ generalized parton distributions from deeply virtual compton scattering at hermes _ , _ eur .
j. _ * a42 * ( 2009 ) 7178 [ http://arxiv.org/abs/0905.1220[0905.1220 ] ]
. m. guidal , _ generalized parton distributions from deep virtual compton scattering at clas _ , _ phys .
* b689 * ( 2010 ) 156162 [ http://arxiv.org/abs/1003.0307[1003.0307 ] ] .
h. moutarde , _ extraction of the compton form factor h from dvcs measurements at jefferson lab _ , _ phys .
_ * d79 * ( 2009 ) 094021 [ http://arxiv.org/abs/0904.1648[0904.1648 ] ] .
collaboration , a.
airapetian _ et . al .
_ , _ separation of contributions from deeply virtual compton scattering and its interference with the bethe
heitler process in measurements on a hydrogen target _
, _ jhep _ * 11 * ( 2009 ) 083 [ http://arxiv.org/abs/0909.3587 [ 0909.3587 ] ] . collaboration , f. gautheron _ et .
_ , _ compass - ii proposal _ , _ cern - spsc-2010 - 014 _ ( 2010 ) spsc
[ http://wwwcompass.cern.ch / compass / proposal / compass - ii_proposal / compass% -ii_proposal.pdf[http://wwwcompass.cern.ch / compass / proposal / compass - ii_proposal / compass% -ii_proposal.pdf ] ] .
w. t. giele , s. a. keller and d. a. kosower , _ parton distribution function uncertainties _ , http://arxiv.org/abs/hep-ph/0104052 [ hep - ph/0104052 ] .
j. c. rojo , _ the neural network approach to parton distribution functions _ , http://arxiv.org/abs/hep-ph/0607122 [ hep - ph/0607122 ] .
phd thesis .
t. schaul , j. bayer , d. wierstra , y. sun , m. felder , f. sehnke , t. rckstie and j. schmidhuber , _ pybrain _ , _ the journal of machine learning research _ * 11 * ( 2010 ) 743746 .
h. honkanen , s. liuti , j. carnahan , y. loitiere and p. reynolds , _ new avenue to the parton distribution functions : self - organizing maps _ , _ phys.rev . _ * d79 * ( 2009 ) 034022 [ http://arxiv.org/abs/0810.2598[0810.2598 ] ] .
d. s. hwang and d. mller , _ implication of the overlap representation for modelling generalized parton distributions _ , _ phys .
lett . _ * b660 * ( 2008 ) 350359 [ http://arxiv.org/abs/0710.1567 [ 0710.1567 ] ] .
f. james and m. roos , _ minuit : a system for function minimization and analysis of the parameter errors and correlations _
* 10 * ( 1975 ) 343367 .
j. pumplin , d. r. stump and w. k. tung , _ multivariate fitting and the error matrix in global analysis of data _ , _ phys .
* d65 * ( 2001 ) 014011 [ http://arxiv.org/abs/hep-ph/0008191[hep-ph/0008191 ] ] .
j. pumplin , d. stump , j. huston , h. lai , p. m. nadolsky _ et .
_ , _ new generation of parton distributions with uncertainties from global qcd analysis _ , _ jhep _ * 0207 * ( 2002 ) 012 [ http://arxiv.org/abs/hep-ph/0201195[hep-ph/0201195 ] ] . a. de roeck and r. thorne , _ structure functions _ , http://arxiv.org/abs/1103.0555[1103.0555 ] . | we have generated a parametrization of the compton form factor ( cff ) @xmath0 based on data from deeply virtual compton scattering ( dvcs ) using neural networks .
this approach offers an essentially model - independent fitting procedure , which provides realistic uncertainties .
furthermore , it facilitates propagation of uncertainties from experimental data to cffs .
we assumed dominance of the cff @xmath0 and used hermes data on dvcs off unpolarized protons .
we predict the beam charge - spin asymmetry for a proton at the kinematics of the compass ii experiment . |
the possibility to exploit electron spin to manipulate currents in electronic microdevices has generated wide interest in the rapidly expanding field of spin electronics @xcite .
technically , the itinerant electron spin may be put to use if the spin - flip scattering , intermixing the spin - up and spin - down channels , occurs at a rate @xmath0 which is slow compared to the inverse mean free time @xmath1 which characterizes the dissipative processes in charge of electrical resistivity .
separation of electrons in two channels is a key element of the physics of a central spintronics phenomenon of giant magneto - resistance ( gmr ) of metallic ferromagnetic - nonferromagnetic metal multilayers .
the two - channel description of diffusive electrical transport from a ferromagnetic ( fm ) to a nonferromagnetic metal was given in @xcite and on a thermodynamic basis also in @xcite .
due to the exchange splitting of the fermi surface in a fm metal the fermi energy corresponds to different parts of the band structure for spin - up and spin - down electrons , and therefore the two types of electrons have different densities of state @xmath2 and conductions @xmath3 .
because of the conduction asymmetry @xmath4 an electrical current passing from a fm to a nonferromagnetic metal is polarized and injects magnetization .
this effect shows up as an additional contact resistance .
the size @xmath5 of the region of the spin accumulation is set by the equilibrium between spin injection through the interface and the spin - flipping processes in the bulk of the paramagnet .
the spin diffusion length @xmath5 relates to the mean free path @xmath6 as @xmath7 .
an analysis of the two - channel model was given by valet and fert @xcite on the basis of the boltzmann transport equation in the limit @xmath8 .
they have also applied the model to the geometry of the gmr with current perpendicular to the plane ( cpp - gmr ) .
a significant advantage of their approach is the ability to describe measurable quantities in detail in terms of parameters of the theory , which allows to extract the latter from experiment @xcite .
the formulas in @xcite , however , have been derived in a flawed assumption that while the conductivities @xmath3 of the majority and minority spins are different , the corresponding densities of states ( dos ) , @xmath2 , are equal . but the difference in @xmath2 is essential as it leads to a jump of the electric potential on an intermetallic interface , which also was not observed in the later work @xcite .
neglect of the difference in @xmath2 results in an incorrect variation of the electric potential with distance from an intermetallic interface , and changes , _
inter alia _ , the boundary contribution to resistance .
the aim of the present paper is to provide expressions for boundary resistances of common multilayered systems corrected to allow the measurements of potential drop to be performed at any distance from interfaces , not necessarily @xmath9 .
correlative results for an array of domain walls have been obtained in @xcite . in sec .
[ sec : basic - equations ] a review of the two - channel diffusive transport theory is given , whereas in sec .
[ sec : results ] we demonstrate our results for a single ferromagnetic - nonferromagnetic interface and for typical spin - valve structures . sec .
[ sec : conclusions ] summarizes the findings and provides a brief discussion .
finally , an appendix with details of our approach to calculation is presented at the end of the paper .
starting from the boltzmann equation , valet and fert @xcite showed that in the limit @xmath8 one can describe transport in the system introducing electrochemical potentials @xmath10 for the two spin channels @xmath11 as sums of chemical potentials ( accounting for the kinetic energy of electrons ) and the potential energy . for small deviations from equilibrium
the chemical potential is given by the non - equilibrium electron density @xmath12 divided by the density of states @xmath13 on the fermi surface : @xmath14 here @xmath15 is the electron charge , @xmath16 is the electric potential , and the condition @xmath17 is fulfilled owing to electroneutrality .
we now briefly review the equations of two - channel diffusive transport to establish notations .
electron current @xmath18 in the channel @xmath19 is a sum of a diffusion component and a contribution from the electron drift under the influence of an electric field @xmath20:@xmath21 where we used the einstein relation @xmath22 between the conductivity @xmath23 and the diffusion constant @xmath24 . in the bulk
the excess electron density @xmath25 is zero , the electrochemical potentials for the two channels are equal : @xmath26 , and the current polarization is @xmath27 ( [ eq : ca ] ) .
relaxation of non - equilibrium polarization of current near a boundary between two materials due to spin - flip processes is described by the continuity equations @xmath28 it is convenient to introduce the divergence @xmath29 between the two electrochemical potentials .
it is non - zero only in the presence of spin accumulation : from ( [ eq : muupdown ] ) one finds that it is related to @xmath25 via @xmath30 substituting ( [ eq : musdef ] ) into ( [ eq : j ] ) and ( [ eq : dj ] ) we find that @xmath31 satisfies the diffusion equation @xmath32 where the spin diffusion length@xmath33 equation ( [ eq : mus ] ) by means of relation ( [ eq : musdef ] ) describes the exponential decay of non - equilibrium spin density away from a boundary . in order to find currents @xmath18 in each channel and the electric potential @xmath16 we introduce @xmath34 where @xmath35 is the net conductivity .
differentiating @xmath36 yields @xmath37 , where @xmath38 is the net current being conserved throughout the sample ( @xmath39 ) .
the spatial distribution of @xmath36 is found from @xmath40 , hence in one dimension @xmath41 the spin current @xmath42 may then be found from the known distribution of @xmath31:@xmath43 from the definition ( [ eq : muupdown ] ) we find that the spatial distribution of @xmath16 is given by @xmath44 where we have introduced the asymmetry of the density of states@xmath45 equations ( [ eq : musdef ] ) , ( [ eq : mus ] ) , ( [ eq : js ] ) , and ( [ eq :- eu ] ) completely describe electron transport inside a layer . at distances @xmath9 from boundaries , when the divergence @xmath31 decays to zero , the two electrochemical potentials @xmath46 as well as the electric potential ( [ eq :- eu ] ) collapse to @xmath36 . i.e. @xmath36 is the value of the electrochemical potential that would be realized were there not non - equilibrium distribution of spin .
however , if measurements are done closer than @xmath5 to an interface , one has to take into account that the non - equilibrium divergence of @xmath47 has not yet subsided . to find charge and current distribution through the whole multilayer sample , eqs .
( [ eq : musdef ] ) , ( [ eq : mus ] ) , ( [ eq : js ] ) , and ( [ eq :- eu ] ) have to be supplemented by boundary conditions . in the absence of an interface resistance and spin - flip scattering ,
the appropriate boundary conditions are the continuity of @xmath47 and @xmath48 @xcite .
the continuity of @xmath47 shows that @xmath36 and @xmath16 are discontinuous at intermetallic interfaces with the jumps@xmath49 } _
{ 0}={\textstyle \frac}{1}{2}{[{\alpha}]}\mu _ { s},\qquad -e{[u]}={\textstyle \frac}{1}{2}{[{\beta}]}\mu _ { s},\label{eq : difmu0}\ ] ] where @xmath50}$ ] is the jump in the conduction asymmetry ( [ eq : ca ] ) and @xmath51}$ ] is the jump in the asymmetry of the density of states .
the discontinuity in @xmath36 appears in transport measurements farther than @xmath5 from an interface as a boundary contribution to resistances@xmath52 } _ { 0}/ej={[{\alpha}]}\mu _ { s}/2ej.\label{eq : rinfty}\ ] ] discontinuities of the electrochemical potentials and the spin current may be introduced @xcite to account for interface resistance and spin - flip scattering .
one then has to allow for the jumps of @xmath47 in ( [ eq : difmu0 ] ) , which affects the answer ( [ eq : rinfty ] ) .
but the inclusion of new parameters for interface resistance and spin - flip scattering renders the formulas very cumbersome , not changing the principal point of the paper .
therefore , we have delegated the discussion of the interface resistance and spin - flip scattering to appendix .
equation ( [ eq :- eu ] ) presents the spatial variation of the electric potential , provided @xmath36 and @xmath31 for the problem in question are found . as revealed by its second term , if @xmath53 , spin accumulation near an intermetallic interface causes @xmath16 to collapse to its bulk asymptotic value @xmath54 on a spin diffusion length scale .
hence , the form of the boundary contribution to resistance has to be modified dependent on how close to the boundary the measurements take place . anyhow , since the jump @xmath55}$ ] at an interface as given by the second of eqs .
( [ eq : difmu0 ] ) differs , generally speaking , from the jump of @xmath56 } _ { 0}$ ] , if measured in the immediate vicinity to the interface , instead of ( [ eq : rinfty ] ) the interface resistance would rather be @xmath57}/j={[{\beta}]}\mu _
{ s}/2ej.\label{eq : r0}\ ] ] note that the jumps @xmath55}$ ] and @xmath56 } _ { 0}$ ] not only differ in absolute value , but may in principle have opposite signs . at intermediate distances from an interface
the boundary resistance must be obtained from ( [ eq :- eu ] ) : @xmath58 by a complete solution of the problem for the whole multilayer structure .
the solution is straightforward in the simplest case of an isolated boundary at the origin between two different materials with constants denoted by the two subscripts 1 and 2 .
we here have decided to illustrate our claim with more general than simply a ferromagnetic - nonferromagnetic interface example at an expense of only slightly more complex expressions . equating spin currents ( [ eq : js ] ) we find @xmath31 at the origin and
reestablish the valet - fert expression for the boundary resistance:@xmath59 in this expression and below we use the shorthand notation@xmath60 for arbitrary @xmath61 , @xmath62 the resistance between the two points is @xmath63 where the first two terms are the series resistances of the two materials over the lengths @xmath64 and @xmath65 respectively , and the interface contribution is @xmath66 this expression exponentially decays from the value @xmath67 very close to the boundary to @xmath68 farther than @xmath5 for it . note that if one of the materials is a nonferromagnetic metal , then the corresponding exponential term in ( [ eq:18 ] ) , which describes decay of the electric potential in it , is zero .
this is because in a normal metal the electric potential ( [ eq :- eu ] ) falls onto the asymptote @xmath36 right after the interface , not exponentially with the distance from it .
position dependence of the electric potential @xmath69 and of the electrochemical potentials @xmath36 , @xmath70 , @xmath71 , @xmath31 for an isolated boundary between a ferromagnetic ( @xmath72 , @xmath73 ) and a normal metal ( @xmath74 ) is plotted in fig .
[ cap : sketch - of - spatial ] .
note that because @xmath75 is taken negative , the jumps @xmath55}$ ] and @xmath56 } _
{ 0}$ ] have opposite signs . allowing for interface resistance and spin - flips complicates the formulas ( [ eq : isolated ] ) , ( [ eq:18 ] ) , and ( [ eq:20 ] ) without changing the qualitative conclusion of the difference between @xmath76 and @xmath68 .
expressions for boundary resistances in this case are provided in appendix . in the end , we cite formulas for a simplest spin - electronic device a trilayer ` spin - valve ' : a ( thin ) layer of paramagnetic metal of thickness @xmath77 sandwiched between two ferromagnetic layers with parallel or antiparallel orientations of the magnetic momenta .
we consider both the cases with and without supplying normal metal electrodes ( see fig .
[ cap : multi ] ) . in the first case
we assume the two ferromagnetic layers to have equal thickness @xmath78 . in case of a symmetric trilayer without the supplying electrodes ( the uppermost structure in fig .
[ eq : isolated ] ) the valet - fert boundary resistances of the two interfaces are @xmath79 where @xmath80 . in the limit of large @xmath81
this expression turns into ( [ eq : isolated ] ) for an isolated boundary .
when the thickness of the paramagnetic layer @xmath82 , we obtain that @xmath68 vanishes as @xmath83 . in what follows
we omit the series contributions to resistance like the first two term in ( [ eq:17 ] ) . for any structure in fig .
[ cap : multi ] these contributions are trivial .
we thus simply cite the boundary part of the total resistance of a symmetric trilayer between two probe points @xmath84 and @xmath85:@xmath86 for the measurement points very close to the boundaries @xmath87 and @xmath88 ( [ eq:22 ] ) reduces to @xmath89 for an antisymmetric trilayer without the supplying electrodes ( the second structure from the top in fig .
[ cap : multi ] ) , when the right ferromagnetic layer has the opposite conduction asymmetry @xmath90 , the valet - fert boundary resistances are @xmath91 when @xmath82 the resistance tends to a finite value @xmath92 , thus justifying the term ` spin - valve ' .
the boundary contribution to resistance between two arbitrary points @xmath84 and @xmath93 is@xmath94 and tends to @xmath95 when @xmath87 and @xmath88 .
for a symmetric five - layer with two supplying normal - metal electrodes ( the third structure from the top in fig .
[ cap : multi ] ) the boundary resistances of the two external and the two internal interfaces are respectively@xmath96 for an antisymmetric five - layer ( the lowermost structure in fig .
[ cap : multi ] ) @xmath97 the boundary contribution to resistance between two arbitrary points @xmath84 and @xmath98 is for both symmetric and antisymmetric five - layered structures simply the total boundary resistance of all the interfaces @xmath99 since in the normal - metal supplying electrodes the electric potential coincides with its bulk asymptote @xmath54 everywhere after the interface .
in this brief study we focused on the effects of the density of states asymmetry in the diffusive transport through multilayered ferromagnetic - nonferromagnetic metal structures . whereas conduction asymmetry brings about spin accumulation in a layer of thickness of the order of the spin diffusion length , and determines the overall drop of the electric potential in this layer ( apart from the trivial series resistance contributions ) , the dos asymmetry results in a jump of the electric potential directly at an intermetallic interface .
in a standard valet - fert theory only the boundary contributions to resistance @xmath68 from the overall potential drop in the spin accumulation layer are calculated .
these boundary resistances emerge only when measurements are done farther than the spin diffusion length from interfaces . however , the series resistances that build up over the distance @xmath100 between the measurement points turn out to be larger than the boundary resistance @xmath68 . to identify the part of the resistance from the current conversion process on the boundary it is advantageous to attach probes closer to the interface .
but in that case it should be remembered that because of the asymmetry of the density of states the interface contribution to the resistance differs from @xmath68 .
the authors are indebted to prof .
lev p. gorkov for innumerable enlightening discussions and guidance through the work , and acknowledge support from darpa through the naval research laboratory grant no .
n00173 - 00 - 1 - 6005 .
in this appendix we expound the general approach we used to calculate the spatial distribution of potentials in a multilayered structure .
the diffusion equation ( [ eq : mus ] ) in an interval @xmath101 in one dimension has a solution@xmath102 that falls off exponentially on a length scale @xmath5 from the values @xmath103 and @xmath104 at the boundaries @xmath105 and @xmath106 .
first consider the boundaries with no interface resistance or spin - flip scattering . in this case
the boundary conditions of continuity of the electrochemical potentials and of the currents apply . the form ( [ eq : musx ] ) for a solution in each layer with respective @xmath5 relies on the values of @xmath31 on the boundaries , which thus need to be put equal on the two sides of each interface to satisfy the boundary condition of the continuity of @xmath107 .
the second boundary condition for the electrochemical potentials leads to the expression ( [ eq : difmu0 ] ) for the jump of @xmath36 , which together with solution ( [ eq : mu0sol ] ) in each layer with respective @xmath108 yields answer for the spatial distribution ( [ eq : mu0sol ] ) of @xmath36 . substituting ( [ eq : musx ] ) into ( [ eq : js ] )
, we obtain spin currents at the boundaries:@xmath109 where @xmath110 in case of a semi - infinite interval , when either @xmath111 or @xmath112 , @xmath31 on the corresponding end tends to zero and @xmath113 , @xmath114 . equating ( [ eq : jsajsb ] ) on both sides of each interface , we obtain an algebraic system of equations on the interface values of @xmath31
. once found , these values then give boundary resistances ( [ eq : rinfty ] ) and ( [ eq : r0 ] ) . in this way we derived eqs .
( [ eq : r3s ] ) , ( [ eq : r3a ] ) , ( [ eq : r5se ] ) , ( [ eq : r5si ] ) , and ( [ eq : r5ae ] ) , ( [ eq : r5ai ] ) .
interface resistance as well as spin - flip scattering at an interface may be taken into account by appropriate boundary conditions @xcite .
non - zero surface resistance leads to discontinuity of electrochemical potential proportional to the current in the respective channel and spin - flip scattering induces discontinuity of the spin current proportional to the average value of @xmath31:@xmath49 } _ { { \rho}}=er_{{\rho}}{\{j\}}_{{\rho}},\qquad { [ j]}_{s}=\frac{1}{er_{\mathrm{sf}}}{\{\mu\ } } _ { s},\label{eq : hardbc}\ ] ] where @xmath115 denotes an average of a discontinuous quantity between the values on the two sides of a boundary .
the surface resistances @xmath116 in each channel may be parametrized by the total resistance @xmath117 and the surface resistance asymmetry @xmath118 solution ( [ eq : musx ] ) of eq .
( [ eq : mus ] ) inside a layer then obviously still holds , if @xmath103 and @xmath104 are understood as values of @xmath31 on the internal side of each boundary . substituting this into expression ( [ eq : js ] ) yields the values of the spin current on the internal side of each boundary , which may be combined with the boundary conditions ( [ eq : hardbc ] ) to give an algebraic equation on the interface values of @xmath31 .
however , it is easier to rewrite eqs .
( [ eq : musx ] ) and ( [ eq : jsajsb ] ) in terms of the average values @xmath119 and @xmath120 to produce:@xmath121 in analogy with ( [ eq : jsajsb ] ) . here
parameters of interface resistance @xmath122 , @xmath123 and of interface spin - flip scattering @xmath124 now enter only through the renormalized constants @xmath125 , @xmath126 , and @xmath127:@xmath128 the formal analogy between ( [ eq:36 ] ) and ( [ eq : jsajsb ] ) allows to reuse the system of algebraic equations on the interface values of @xmath31 with the substitution of @xmath119 as variables and ( [ eq:37])-([eq:39 ] ) as parameters .
however , the expressions for the jumps of @xmath36 and @xmath16 at the boundaries are more complex than ( [ eq : difmu0]):@xmath129 } _ { 0 } & = & er\left((1+{\{\alpha\ } } \gamma ) j+({\{\alpha\ } } + \gamma ) { \{j\}}_{s}\right)+{\textstyle \frac}{1}{2}{[\alpha ] } { \{\mu\ } } _ { s},\\ -e{[u ] } & = & er\left((1+{\{\beta\ } } \gamma ) j+({\{\beta\ } } + \gamma ) { \{j\}}_{s}\right)+{\textstyle \frac}{1}{2}{[\beta ] } { \{\mu\ } } _ { s}.\end{aligned}\ ] ] hence the boundary resistances ( [ eq : rinfty ] ) , ( [ eq : r0 ] ) will also be different .
as an example and to give the reader a feel of the complexity of the formulas that arise when surface resistance and spin - flips are taken into account we cite our answers for a problem considered earlier in the text with simple boundary conditions of continuity , i.e. a single boundary at the origin between two materials .
@xmath130}^{2}+r\left({\lambda}_{1}({\alpha}_{2}+\gamma ) ^{2}+{\lambda}_{2}({\alpha}_{1}+\gamma ) ^{2}\right)+2r_{\mathrm{sf}}^{-1}(\gamma + { \{{\alpha}\}})^{2}}{{\lambda}_{1}+{\lambda}_{2}+{\lambda}_{1}{\lambda}_{2}r+r_{\mathrm{sf}}^{-1}(2+({\lambda}_{1}+{\lambda}_{2})r/2)},\\ r_{i0 } & = & r(1-\gamma ^{2})+\frac{{[{\alpha}]}{[{\beta}]}+r\left({\lambda}_{1}({\alpha}_{2}+\gamma ) ( { \beta}_{2}+\gamma ) + { \lambda}_{2}({\alpha}_{1}+\gamma ) ( { \beta}_{1}+\gamma ) \right)+2r_{\mathrm{sf}}^{-1}(\gamma + { \{{\alpha}\}})(\gamma + { \{{\beta}\}})}{{\lambda}_{1}+{\lambda}_{2}+{\lambda}_{1}{\lambda}_{2}r+r_{\mathrm{sf}}^{-1}(2+({\lambda}_{1}+{\lambda}_{2})r/2)}.\end{aligned}\ ] ] where @xmath132 } + r\left({\lambda}_{1}({\alpha}_{2}+\gamma ) + ( { \{{\alpha}\}}+\gamma ) /4r_{\mathrm{sf}}\right)}{{\lambda}_{1}+{\lambda}_{2}+{\lambda}_{1}{\lambda}_{2}r+r_{\mathrm{sf}}^{-1}(2+({\lambda}_{1}+{\lambda}_{2})r/2)},\\ \mu _ { s}(-0 ) & = & 2ej\frac{{[{\alpha } ] } -r\left({\lambda}_{2}({\alpha}_{1}+\gamma ) + ( { \{{\alpha}\}}+\gamma ) /4r_{\mathrm{sf}}\right)}{{\lambda}_{1}+{\lambda}_{2}+{\lambda}_{1}{\lambda}_{2}r+r_{\mathrm{sf}}^{-1}(2+({\lambda}_{1}+{\lambda}_{2})r/2)}\end{aligned}\ ] ] are the values of @xmath31 on the two sides of the boundary . 10 _ semiconductor spintronics and quantum computation _ , edited by d. d. awschalom , d. loss and n. samarth ( springer - verlag , berlin , 2002 ) .
g. a. prinz , science * 282 * , 1660 ( 1998 ) .
j. f. gregg , i. petej , e. jouguelet , and c. dennis , j. phys .
phys . * 35 * , r121 ( 2003 ) .
p. c. van son , h. van kempen and p. wyder , phys .
* 58 * , 2271 ( 1987 ) .
m. johnson and r. h. silsbee , phys .
b * 35 * , 4959 ( 1987 ) ; phys .
lett . * 60 * , 377 ( 1988 ) . t. valet and a. fert , phys .
b * 48 * , 7099 ( 1993 ) . a. fert and s
lee , phys .
b * 53 * , 6554 ( 1996 ) .
s. dubois , l. piraux , j. m. george , k. ounadjela , j. l. duvail , a. fert , phys .
b * 60 * , 477 ( 1999 ) .
w. p. pratt , jr .
lee , j. m. slaughter , r. loloee , p. a. schroeder , and j. bass , phys .
* 66 * , 3060 ( 1991 ) ; s. f. lee , w. p. pratt , jr .
, r. loloee , p. a. schroeder , and j. bass , phys . rev .
b * 46 * , 548 ( 1992 ) .
l. piraux , s. dubois , and a. fert , j. magn .
magn . mater . *
159 * , l287 ( 1996 ) .
m. dzero , l. p. gorkov , a. k. zvezdin and k. a. zvezdin , phys . rev .
b * 67 * , 100402(r ) ( 2003 ) . | asymmetry of conduction of electrons in the spin majority and minority bands of a ferromagnetic metal is well known to produce spin accumulation at the boundary with a normal metal when a current is injected through the interface .
however , little emphasis has been put on the density of states ( dos ) asymmetry in diffusive ferromagnetic - nonferromagnetic multilayers .
we found that if dos and conduction asymmetry differ , the electric potential in a ferromagnet falls off to its bulk value exponentially on a scale of the spin diffusion length .
therefore the boundary contribution to resistance dramatically depends on whether the potential difference is measured close to the interface or farther than the spin diffusion length from it .
this result is not altered by taking surface resistance or spin - flips on the intermetallic boundary into account .
explicit answers for common multilayered spin - valve structures are given .
pacs number(s ) : 72.25.pn , 72.25.ba , 73.40.cg , 73.40.jn |
a variety @xmath6 of universal algebras is _ subtractive _ in the sense of a.ursini @xcite , if the theory of @xmath6 contains a binary term @xmath7 ( called a _ subtraction term _ ) and a nullary term @xmath8 , satisfying the identities @xmath9 and @xmath10 .
an example of a subtractive variety is the variety of ( additive ) groups , for which @xmath11 . more generally , any maltsev variety @xcite , whose theory contains a nullary term , is subtractive .
an example of a subtractive variety , which is not a maltsev variety , is the variety of implication algebras @xcite .
a _ subtraction algebra _ is a triple @xmath12 , where @xmath13 is a set , `` @xmath14 '' is a binary operation on @xmath13 , and @xmath8 is a nullary operation on @xmath13 , satisfying the axioms @xmath15 and @xmath16 .
a _ group _ can be defined as a subtraction algebra with the additional axiom @xmath17 , and an _ abelian group _ can be defined as a subtraction algebra with the stronger axiom @xmath18 in both cases , the group addition @xmath19 is defined by the equality @xmath20 , while the unary operation of taking the inverse @xmath21 of an element @xmath22 , is defined by the equality @xmath23 . note that , to require the axiom ( [ commutativity of subtraction ] ) , is the same as to require that the binary operation @xmath24 is a homomorphism of subtraction algebras .
thus , an abelian group is a subtraction algebra whose algebraic structure is at the same time an internal subtraction structure in the category @xmath25 of all subtraction algebras .
further , any internal subtraction structure on an object @xmath13 in @xmath25 , coincides with the underlying subtraction structure of @xmath13 ( see @xcite , @xcite , and `` subtraction '' is used only when @xmath13 is a group ; in @xcite ( and also in most cases in @xcite too ) , `` @xmath26 '' is written as `` @xmath27 '' , and accordingly , a `` ( proto)subtraction '' is called a `` ( proto)devision '' . ] ) , and so an internal subtraction algebra in @xmath25 is always an internal abelian group .
the variety @xmath25 is clearly a subtractive variety .
further , the algebraic theory of @xmath25 contains a unique nullary term ; equivalently , @xmath25 is a _ pointed category _ , i.e. the terminal object in @xmath25 coincides with the initial object .
the notion of a subtractive category , introduced in @xcite ( see also @xcite ) , extends the notion of a pointed subtractive variety to abstract pointed categories a pointed variety is subtractive in the sense of ursini , if and only if it is a subtractive category .
the purpose of this paper is to show that for a finitely complete pointed category , with binary sums , and such that each morphism @xmath2 in @xmath0 can be presented as a composite @xmath3 , where @xmath4 is a monomorphism and @xmath5 is an epimorphism , we have the following result : @xmath0 is additive if and only if both @xmath0 and its dual @xmath1 are subtractive ( theorem [ thm : the main theorem ] ) .
this implies that a barr exact @xcite pointed category @xmath0 , having binary sums , is an abelian category , if and only if both @xmath0 and @xmath1 are subtractive .
in particular , for varieties of universal algebras ( which are always barr exact and have binary sums ) , we obtain : a pointed subtractive variety @xmath0 is abelian ( i.e. @xmath0 is the variety of modules over a fixed ring ) if and only if @xmath1 is also subtractive .
let @xmath0 be a pointed category with binary products .
@xmath0 is said to be subtractive , if for any object @xmath13 in @xmath0 , any subobject @xmath28 of @xmath29 satisfies the following condition : if there exist morphisms @xmath30 , such that both diagrams @xmath31 ^ -{r } & & & r\ar[d]^-{r } \\a\ar[r]_-{(1_a,1_a)}\ar[ur]^-{f_1 } & a{\times}a & & a\ar[r]_-{(1_a,0)}\ar[ur]^-{f_2 } & a{\times}a}}\ ] ] commute , then there exists a morphism @xmath32 , such that the diagram @xmath33 ^ -{r } \\a\ar[r]_-{(0,1_a)}\ar[ur]^-{f_3 } & a{\times}a}}\ ] ] commutes .
the above condition on the subobject @xmath34 can be reformulated , using generalized elements , as follows : if @xmath35 and @xmath36 for all @xmath37 , then @xmath38 for all @xmath37 . as we know from @xcite , if @xmath0 has finite limits , then @xmath0 is subtractive if and only if the following condition is satisfied for any object @xmath13 in @xmath0 , and any subobject @xmath34 of @xmath29 : for all @xmath37,@xmath39 the condition ( [ implication i ] ) on @xmath34 states precisely that @xmath34 is ( regarded as a binary relation on @xmath13 ) _ @xmath40-closed _ in the sense of @xcite , where @xmath40 is the extended matrix @xmath41 of terms in the algebraic theory @xmath42 of pointed sets @xmath8 is the unique constant of @xmath42 , while @xmath22 is an arbitrary variable of @xmath42 .
note that this matrix directly arises from the system of equations @xmath43 which defines an operation of subtraction .
in the present paper we write such matrices mostly in the transposed form @xmath44 _ @xmath34 is @xmath40-closed _ means that for any _ interpretation _ @xmath45 of @xmath40 , where @xmath37 and @xmath8 is the base point of @xmath13 , if the first two rows of @xmath46 belong to @xmath34 ( considered as generalized elements of @xmath47 ) , then also the third row ( i.e. the row below the horizontal line ) belongs to @xmath34 .
thus , @xmath34 is @xmath40-closed if and only if the implication ( [ implication i ] ) is satisfied for all @xmath37 , as already said above .
recall from @xcite , that if we take @xmath48 then , again , a ( finitely complete pointed ) category with @xmath40-closed relations is the same as a ( finitely complete ) subtractive category . if we further modify @xmath40 by adding to it columns of the form @xmath49 then again ( by proposition 1.7 of @xcite )
, a subtractive category is the same as a pointed category in which every @xmath50-ary relation @xmath51 is @xmath40-closed , where now @xmath40 is the new matrix , and @xmath50 denotes the number of columns of @xmath40 .
a _ semi - abelian category _ in the sense of g.janelidze , l.mrki , and w.tholen @xcite , is a pointed category @xmath0 satisfying the following conditions : as shown in @xcite , a pointed variety of universal algebras is semi - abelian if and only if it is _ classically ideal determined _ in the sense of a.ursini @xcite ( this is the same as _ bit speciale _ in the sense of @xcite ) .
the following equation for categories was obtained in @xcite : @xmath52 this equation says : a category @xmath0 is an abelian category if and only if both @xmath0 and its dual category @xmath1 are semi - abelian . from theorem
[ thm : the main theorem ] below , it follows that , more generally , for any barr exact category with binary sums , we have ( see corollary [ cor : abelian categories ] , and see also remark [ rem : strongly unital categories ] ) : @xmath53 recall that a barr exact category @xmath0 is an abelian category if and only if it is additive , i.e. if and only if @xmath0 is enriched in the category of abelian groups each object in @xmath0 is equipped with an internal abelian group structure , so that each morphism in @xmath0 is a homomorphism of internal abelian groups . more generally , we have : [ lem : additivity via subtraction algebras ] a category with finite products is additive if and only if each object in it is equipped with an internal subtraction structure , so that each morphism is a homomorphism of internal subtraction algebras .
this follows immediately from the definition of an additive category , and the known fact that if the operation of subtraction of a subtraction algebra is homomorphic , then the subtraction algebra becomes an abelian group . if @xmath0 is additive , then it is subtractive .
indeed , then for any pair of commutative diagrams ( [ dia : subtractivity a ] ) , we can form the third one ( [ dia : subtractivity b ] ) , by taking in it @xmath55 , where `` @xmath14 '' denotes the operation of subtraction on @xmath56 , induced by the operation of subtraction of the abelian group structure of @xmath34 .
we will then have @xmath57 so the diagram ( [ dia : subtractivity b ] ) indeed commutes . on the other hand , if @xmath0 is additive , then so is @xmath1 , and hence @xmath1 is subtractive .
this proves the first part of the theorem .
assume now that both @xmath0 and @xmath1 are subtractive . according to lemma [ lem : additivity via subtraction algebras ] , to show that @xmath0 is additive , it suffices to show that each object @xmath13 in @xmath0 is equipped with an internal subtraction structure , @xmath12 , so that each morphism @xmath58 in @xmath0 is a homomorphism of ( internal ) subtraction algebras .
let @xmath13 be an object in @xmath0 .
then we can form a commutative diagram @xmath59 ^ -{r=(r_1,r_2 ) } & \\
\ar[ur]^-{q={\fontsize{8}{9.2}\selectfont\left(\begin{array}{c}q_1\\#2\end{array}\right)}}\ar[rr]_-{{\fontsize{8}{9.2}\selectfont\left(\begin{array}{cc}1_a & 1_a\\#3 & 0\\\end{array}\right ) } } & & a{\times}a\\}\ ] ] where @xmath60 is a monomorphism and @xmath61 is an epimorphism . this commutative diagram yields the following two commutative diagrams : @xmath62 ^ -{r_1}&\\ a+a \ar[u]^-{q}\ar[r]_-{{\fontsize{8}{9.2}\selectfont\left(\begin{array}{c}1_a\\#2\end{array}\right ) } } & a\\ } { \;\;\;}{\;\;\;}{\;\;\;}\xymatrix@=3pc { r\ar[dr]^-{r_2 } & \\
a+a \ar[u]^-{q}\ar[r]_-{{\fontsize{8}{9.2}\selectfont\left(\begin{array}{c}1_a\\#2\end{array}\right ) } } & a\\}\ ] ] since @xmath1 is subtractive , there exists a morphism @xmath63 such that the diagram @xmath64 ^ -{s } & \\ a+a \ar[u]^-{q}\ar[r]_-{{\fontsize{8}{9.2}\selectfont\left(\begin{array}{c}0\\#2\end{array}\right ) } } & a\\}\ ] ] commutes .
next , we regard @xmath34 as the subobject @xmath28 of @xmath47 . then @xmath63 becomes a partial binary operation on @xmath13 .
commutativity of the two diagrams @xmath65 ^ -{r}\\ a \ar[ur]^-{q_1}\ar[r]_-{(1_a,1_a ) } & a{\times}a\\ } { \;\;\;}{\;\;\;}{\;\;\;}\xymatrix@=3pc { & r\ar[d]^-{r}\\ a \ar[ur]^-{q_2}\ar[r]_-{(1_a,0 ) } & a{\times}a\\}\ ] ] and the fact that @xmath0 is subtractive , yield that there exists a morphism @xmath66 such that the diagram @xmath65 ^ -{r}\\ a \ar[ur]^-{p}\ar[r]_-{(0,1_a ) } & a{\times}a\\}\ ] ] commutes . from the commutativity of the last four diagrams we obtain : for all @xmath37 , @xmath67 and , further , @xmath68 and @xmath69 .
we now show @xmath70 .
consider the ternary relation @xmath71 defined as follows : @xmath72 since @xmath0 is subtractive , @xmath73 is closed with respect to the matrix @xmath74 for each @xmath75 we then have : @xmath76 thus , for all @xmath75 , @xmath77 , which shows @xmath70 .
so @xmath7 is a full binary operation on @xmath13 , and since we already know that it satisfies the identities @xmath68 and @xmath69 , we get that @xmath78 is a subtraction algebra .
we have thus shown * that the morphism @xmath60 is an isomorphism ; this implies that the morphism @xmath79_-{{\fontsize{8}{9.2}\selectfont\left(\begin{array}{cc}1_a & 1_a\\#3 & 0\\\end{array}\right ) } } & & a{\times}a\\}\ ] ] is an epimorphism ; * that there exists a subtraction algebra @xmath80 ; note that the subtraction axioms for the operation @xmath81 state precisely that the diagram @xmath82 ^ -{{\fontsize{8}{9.2}\selectfont\left(\begin{array}{c}0\\#2\end{array}\right)}}\ar[rr]_-{{\fontsize{8}{9.2}\selectfont\left(\begin{array}{cc}1_a & 1_a\\#3 & 0\\\end{array}\right ) } } & & a{\times}a\ar[r]_-{- } & a\\}\ ] ] commutes .
now let @xmath83 be another subtraction algebra , obtained from an object @xmath84 , like @xmath80 is obtained from @xmath13 .
let @xmath2 be any morphism @xmath58 . in the diagram @xmath85_-{f+f}\ar[rr]^-{{\fontsize{8}{9.2}\selectfont\left(\begin{array}{cc}1_a & 1_a\\#3 & 0\\\end{array}\right ) }
} & & a{\times}a\ar[d]_-{f{\times}f}\ar[r]^-{- } & a\ar[d]^-{f}\\ a'+a'\ar[rr]_-{{\fontsize{8}{9.2}\selectfont\left(\begin{array}{cc}1_{a ' } & 1_{a'}\\#3 & 0'\\\end{array}\right ) } } & & a'{\times}a'\ar[r]_-{- ' } & a'\\}\ ] ] the left inner rectangle commutes and the outer rectangle commutes .
since the first morphism in the top row is an epimorphism , this implies that the right inner rectangle commutes . we thus obtain that @xmath2 is a homomorphism @xmath86 of subtraction algebras .
this concludes the proof .
an ablian category is the same as an additive barr exact category . on the other hand , a barr exact category , being a regular category ,
has finite limits and any morphism @xmath2 in it can be decomposed @xmath3 into a monomorphism @xmath4 and an epimorphism @xmath5 ( moreover , any morphism can be decomposed into a monomorphism and a regular epimorphism ) . after these observations
, it is only left to apply theorem [ thm : the main theorem ] .
[ rem : strongly unital categories ]
i do not know if theorem [ thm : the main theorem ] remains true if @xmath0 in it does not have the property ( * ) .
however , it does remain true if in addition we replace `` subtractive '' in the theorem with `` strongly unital '' in the sense of d. bourn @xcite ; this is an immediate consequence of the following known facts ( see @xcite and @xcite ) : * a finitely complete category is additive if and only if it is subtractive and enriched in the category of commutative monoids , * a category @xmath0 with binary sums and products is enriched in the category of commutative monoids if and only if both @xmath0 and its dual @xmath1 are unital in the sense of d. bourn @xcite , * a finitely complete category is strongly unital if and only if it is subtractive and unital . | the notion of a subtractive category , recently introduced by the author , is a `` categorical version '' of the notion of a ( pointed ) subtractive variety of universal algebras , due to a.ursini .
we show that a subtractive variety @xmath0 , whose theory contains a unique constant , is abelian ( i.e. @xmath0 is the variety of modules over a fixed ring ) , if and only if the dual category @xmath1 of @xmath0 , is subtractive .
more generally , we show that @xmath0 is additive if and only if both @xmath0 and @xmath1 are subtractive , where @xmath0 is an arbitrary finitely complete pointed category , with binary sums , and such that each morphism @xmath2 in @xmath0 can be presented as a composite @xmath3 , where @xmath4 is a monomorphism and @xmath5 is an epimorphism . |
recent work on parton distributions functions ( pdf s ) in the nucleon has focussed on probing the sea and gluon distribution at small @xmath4 .
the valence quark distributions has been thought to be relatively well understood .
however , the precise knowledge of the @xmath1 and @xmath0 quark distribution at high @xmath4 is very important at collider energies in searches for signals for new physics at high @xmath12 .
in addition , the value of @xmath5 as @xmath13 is of theoretical interest .
recently , a proposed cteq toy model @xcite included the possibility of an additional contribution to the @xmath1 quark distribution ( beyond @xmath14 ) as an explanation for both the initial hera high @xmath12 anomaly @xcite , and for the jet excess at high-@xmath15 at cdf @xcite . in this communication
we conclude that a re - analysis of data from nmc and slac leads to a great improvement in our knowledge of the pdf s at large @xmath4 .
information about valence quarks originates from proton and neutron structure function data .
the @xmath1 valence quark distribution at high @xmath4 is relatively well constrained by the proton structure function @xmath16 . however , the neutron structure function @xmath17 , which is sensitive to the @xmath0 valence quark at high @xmath4 , is actually extracted from deuteron data .
therefore , there is an uncertainty in the @xmath0 valence quark distribution from the corrections for nuclear binding effects in the deuteron . in past extractions of @xmath17 from deuteron data ,
only fermi motion corrections were considered , and other binding effects were assumed to be negligible .
recently , the corrections for nuclear binding effects in the deuteron , @xmath18 , have been extracted empirically from fits to the nuclear dependence of electron scattering data from slac experiments e139/140 @xcite .
the empirical extraction uses a model proposed by frankfurt and strikman @xcite , in which all binding effects in the deuteron and heavy nuclear targets are assumed to scale with the nuclear density .
the correction extracted in this empirical way is also in agreement ( for @xmath19 ) with recent purely theoretical calculations @xcite of nuclear binding effects in the deuteron .
the total correction for nuclear binding in the deuteron is about 4% at @xmath20 ( shown in fig .
[ fig : f2dp](a ) ) , and in a direction which is opposite to what is expected from the previous models which only included the fermi motion effects .
the ratio @xmath2 is directly related to @xmath5 . in leading order qcd , @xmath21 at high @xmath4 .
we perform a nlo analysis on the precise nmc @xmath2 data @xcite to extract @xmath5 as a function of @xmath4 .
we extract the ratio @xmath22 by applying the nuclear binding correction @xmath18 to the @xmath2 data . as shown in fig .
[ fig : f2dp](b ) , the standard pdf s do not describe the extracted @xmath22 .
since the @xmath1 distribution is relatively well constrained , we find a correction term to @xmath5 in the standard pdf s ( as a function of @xmath4 ) by only varying the @xmath0 distribution to fit the data .
the correction term is parametrized as a simple quadratic form , @xmath23 for the mrs(r2 ) pdf , where the corrected @xmath5 ratio is @xmath24 . based on this correction
, we obtain a mrs(r2)-modified pdf as shown in fig [ fig : dou](a ) .
the corrections to other pdf s such as cteq3 m is similar .
the nmc data , when corrected for nuclear binding effects in the deuteron , clearly indicate that @xmath5 in the standard pdf s @xcite is significantly underestimated at high @xmath4 as shown in fig .
[ fig : dou ] .
fig.[fig : dou](a ) also shows that the modified @xmath5 ratio approaches @xmath25 as @xmath26 , in agreement with a qcd prediction @xcite . in contract , if the deuteron data is only corrected for fermi motion effects ( as was mistakenly done in the past ) both the @xmath5 from data and the @xmath5 in the standard pdf s fits approach @xmath27 as @xmath13 .
information ( which is not affected by the corrections for nuclear effects in the deuteron ) on @xmath5 can be extracted from @xmath7 production data in hadron colliders .
fig.[fig : dou](b ) shows that the predicted @xmath7 asymmetry calculated with the dyrad nlo qcd program using our modified pdf is in much better agreement with recent cdf data @xcite at large rapidity than standard pdf s .
when the modified pdf at @xmath12=@xmath28 is evolved to @xmath12=@xmath29 using the dglap nlo equations , we find that the modified @xmath0 distribution at @xmath30 is increased by about 40 % in comparison to the standard @xmath0 distribution .
the modified pdf s have a significant impact on the prediction of the cross sections in the hera high @xmath12 region because the charged current scattering with positrons is on @xmath0 quarks only .
it also impacts neutral current scattering because of the large coupling of the @xmath31 to @xmath0 quarks at high @xmath12 .
the modified pdf s also lead to an increase of 10% in the qcd prediction for the production rate of very high @xmath32 jets in hadron colliders .
since all the standard pdf s , including our modified version , are fit to data with @xmath4 less than 0.75 , we now investigate the validity of the modified mrs(r2 ) at very high @xmath4 by comparing to @xmath16 data at slac .
although the slac data at very high @xmath4 are at reasonable values of @xmath12 @xmath33 , there are in a region in which non - perturbative effects such as target mass and higher twist are very large .
we use the georgi - politzer calculation @xcite for the target mass corrections ( tm ) .
these involve using the scaling variable @xmath34 instead of @xmath4 .
since a complete calculation of higher twist effects is not available , the very low @xmath12 data is used to obtain information on the size of these terms .
we use two approaches in our investigation of the higher twist contributions : an empirical method , and the renormalon model . in the empirical approach ,
the higher twist contribution is evaluated by adding a term @xmath35 to the perturbative qcd ( pqcd ) prediction of the structure function ( including target mass effects ) .
the @xmath4 dependence of the higher twist coefficients @xmath36 is fitted to the global dis @xmath3 ( slac , bcdms , and nmc ) data @xcite in the kinematic region ( @xmath37 , @xmath38 ) with the following form , @xmath39 . here
@xmath40 is a floating factor to investigate possible @xmath4 dependent corrections to our modified pdf .
a functional form , @xmath41 for @xmath36 is used in the higher twist fit to estimate the size of the higher twist terms above @xmath42 .
the slac and bcdms data are normalized to the nmc data . in the case of the bcdms data ,
a systematic error shift @xmath43 ( in standard deviation units ) is allowed to account for the correlated point - to - point systematic errors .
the empirical higher twist fits with the modified nlo mrs(r2 ) pqcd prediction with tm have been performed simultaneously on the proton and deuteron @xmath3 data with 11 free parameters ( 2 relative normalizations and 3 parameters for @xmath44 per target and the bcdms @xmath43 ) .
the results of the fit are given in table [ ht - fit ] .
.results of the higher twists fits ( with pqcd+tm ) to the dis f2 data [ cols="^,^,^,^,^,^",options="header " , ] we find that empirical higher twist fit describes the data well ( @xmath45 ) .
the size of the higher twist contributions in the proton and deuteron are similar .
the magnitude is almost half of those extracted in previous analysis of slac / bcdms data @xcite .
this is because that analysis was based on ( ) @xmath46 , while the mrs(r2 ) pdf uses ( ) @xmath47 , which is close to the current world average . in the renormalon model approach @xcite ,
the model predicts the complete @xmath4 dependence of the higher twist contributions to @xmath3 , @xmath48 , and @xmath11 , with only two unknown parameters @xmath49 and @xmath50 .
we extract the @xmath49 and @xmath50 parameters , which determine the overall level of the @xmath51 and @xmath52 terms by fitting to the global data set for @xmath3 and @xmath53 .
the values of @xmath49 and @xmath50 for the proton and deuteron are same in this model .
the @xmath4 dependence of @xmath48 differs from that of @xmath3 but is same as that of @xmath11 within a power correction of @xmath51 .
our fits can also be used to estimate the size of the higher twist effects in @xmath11 ( e.g. the gls sum rule ) .
the higher twist fit in this approach has employed the same procedure as the empirical method .
[ fig : disht ] shows that the model yields description of @xmath4 dependence of higher twist terms in both @xmath3 and @xmath9 with just the two free parameters ( @xmath54 ) .
the ccfr neutrino data @xcite is shown for comparison though it is not used in the fit .
the extracted values of @xmath49 are @xmath55 and @xmath56 , for proton and deuteron , respectively .
the contribution of @xmath50 is found to be negligible .
we find that the floating factor @xmath40 for the deuteron deviates from 1 and is also bigger than that for the proton , unless the modified mrs(r2 ) pdf is used .
this reflects our earlier conclusion that the standard @xmath0 distribution is underestimated at high @xmath4 region .
as expected , the extracted @xmath49 value is half of the previous estimated value @xcite of @xmath49 based on slac / bcdms ( ( ) @xmath46 ) analysis .
since both of these approaches yield a reasonable description for the higher twist effects , we proceed to compare the predictions of the modified pdf s ( including target mass and renormalon higher twist corrections ) to the slac proton @xmath3 data at very high @xmath4 ( @xmath57 ) .
there is a wealth of slac data @xcite in the region up to @xmath58 and intermediate @xmath12 @xmath33 .
previous pdf fits have not used these data .
we use the estimate of the higher twist effects from the models , based on the data ( below @xmath19 ) described above . note that the data for @xmath14 is in the dis region , and the data for @xmath59 is the resonance region .
it is worthwhile to investigate the resonance region also because from duality arguments @xcite it is expected that the average behaviour of the resonances and elastic peak should follows the dis scaling limit curve .
[ fig : highx ] shows the ratio of the slac data at very high @xmath4 to the predictions of the modified mrs(r2 ) . with the inclusion of target mass and the renormalon
higher twist effects , the very high @xmath4 data from slac is remarkably well described by the modified mrs(r2 ) up to @xmath58 . the good description of the data by the modified mrs(r2 ) is also achieved using the empirical estimate ( @xmath35 ) of higher twist effects as shown ( dashed line ) in fig . [
fig : highx](c ) .
the data at @xmath59 and @xmath60 is somewhat lower than the predictions of the modified mrs(r2 ) but relatively the large @xmath12 data ( @xmath61 ) show a good agreement as shown in fig .
[ fig : highx_highq2](b ) .
this could be due to the missing elastic contribution as expected from duality arguments .
[ fig : highx_highq2](b ) also shows that the cteq toy model ( with an additional 0.5% component of @xmath1 quarks beyond @xmath14 ) overestimates the slac data by a factor of three at @xmath62 ( dis region ) . from these comparison ,
we find that the slac @xmath3 data do not support the cteq toy model which proposed an additional @xmath1 quark contribution at high @xmath4 as an explanation of the initial hera high @xmath12 anomaly and the cdf high-@xmath15 jet excess . 0.3 cm
we find that nuclear binding effects in the deuteron play a significant role in our understanding of @xmath5 at high @xmath4 .
the modified pdf s with our @xmath5 correction are in good agreement with the prediction of qcd at @xmath6 , and with the cdf @xmath7 asymmetry data . with the inclusion of target mass and higher twist corrections ,
the modified pdf s describe all dis data not only up to the very highest @xmath4 , but also all the way down to @xmath63 .
note that when the analysis is repeated in nnlo @xcite , we find that the additional nnlo contributions to @xmath9 account for most of the higher twist effects extracted in the nlo fit .
the analysis in nnlo indicates that the highest twist corrections to the gls sum rule are very small ( the fractional contribution to the pqcd gls sum rule is @xmath64 ) .
99 s. kuhlmann , h.l .
lai , w.k . | we extract the ratio of the down ( @xmath0 ) and up ( @xmath1 ) parton distribution functions ( pdf s ) from the ratio of nmc deuteron and proton structure function @xmath2 , using corrections for nuclear binding effects in the deuteron , which are extracted from the nuclear dependence of slac @xmath3 data .
significant corrections to the @xmath0 quark distribution in standard pdf s are required , especially at high @xmath4 .
the corrected @xmath5 ratio is in agreement with the qcd prediction of 0.2 at @xmath6 .
the predictions for the recent cdf @xmath7 asymmetry data using pdf s with the corrected @xmath5 ratio give much better agreement at large rapidity . using the updated @xmath5 ratio and the most recent world average for @xmath8
, we perform a nlo global fit to all dis data for @xmath3 and @xmath9 , and estimate the size of the higher twist contributions using both a renormalon model and an empirical model .
we find that with the updated value of @xmath8 , the magnitude of the higher twist terms is half the value of previous analysis . with the inclusion of target mass and higher twist corrections , the standard nlo pdf s with the updated @xmath5 ratio describe the slac @xmath3 data up to @xmath10 .
when the analysis is repeated in nnlo , we find that the additional nnlo contributions to @xmath9 account for most of the higher twist effects extracted in the nlo fit . the analysis in nnlo indicates that the higher twist effects in @xmath9 , @xmath3 and @xmath11 ( e.g. gls sum rule ) are very small .
-0.8 cm 17.cm -0.5 cm # 1#2#3 _ nucl . phys . _
* # 1 * , # 2 ( # 3 ) # 1#2#3 _ proc . cam .
phil .
soc . _
* # 1 * , # 2 ( # 3 ) # 1#2#3 _ phys .
lett .
_ * # 1 * , # 2 ( # 3 ) # 1#2#3 _ phys .
rep .
_ * # 1 * , # 2 ( # 3 ) # 1#2#3 _ phys .
rev . _
* # 1 * , # 2 ( # 3 ) # 1#2#3 _ phys .
rev .
lett . _ * # 1 * , # 2 ( # 3 ) # 1#2#3 _ proc .
roy .
soc . _ * # 1 * , # 2 ( # 3 ) # 1#2#3 _ prog .
th .
phys . _ * # 1 * , # 2 ( # 3 ) # 1#2#3 _ rev .
mod .
phys .
_ * # 1 * , # 2 ( # 3 ) # 1#2#3 _ rep .
prog .
phys . _
* # 1 * , # 2 ( # 3 ) # 1#2#3 _ zeit .
phys . _ * # 1 * , # 2 ( # 3 ) # 1#2#3 _ eur .
phys .
jour . _ * # 1 * , # 2 ( # 3 ) -.3 cm 0.5 in psfig.tex ur-1518 + er-40685 - 919 + jun . 10 , 1998 u. k. yang and a. bodek department of physics and astronomy , university of rochester , rochester , ny 14627 + |
negative ions ( anions ) were first discovered in the interstellar medium ( ism ) by @xcite , who detected the linear hydrocarbon anion c@xmath0h@xmath1 in .
this was followed by c@xmath0h@xmath1 and c@xmath3h@xmath1 detections in the protostellar core l1527 by @xcite and @xcite , respectively , which hinted of a role for anions in the chemistry of star - formation .
@xcite performed the first dedicated survey for c@xmath0h@xmath1 in twenty - four molecular sources , but detected the anion in only two star - forming clouds , with anion - to - neutral ratios on the order of a few percent , similar to previously - observed values .
interstellar anion detections were confined to the taurus molecular cloud complex until the recent discoveries of c@xmath0h@xmath1 in the lupus , cepheus and auriga star - forming regions @xcite , which proved that anions are widespread in the local ism , and not an artifact of any particular physical or chemical conditions in the taurus region .
the possible importance of anions in interstellar chemistry was first discussed by @xcite , and prospects for the detection of molecular anions using radio astronomy were examined by @xcite .
@xcite argued that large interstellar molecules ( including carbon - chain - bearing species ) can undergo rapid radiative electron attachment as exemplified by the laboratory experiments of @xcite potentially resulting in significant anion abundances in dense molecular clouds .
based on this idea , models for anion chemistry have been successful in reproducing the observed abundances of c@xmath0h@xmath1 and c@xmath8h@xmath1 in tmc-1 , irc+10216 , l1527 and l1512 @xcite .
however , there is a major discrepancy between the modeled and observed c@xmath3h@xmath1 anion - to - neutral ratio ( see * ? ? ?
* ) , and the lack of anions in pdrs @xcite is at variance with the model predictions of @xcite .
clearly , our understanding of molecular anion chemistry is incomplete .
nevertheless , as demonstrated by @xcite , anion measurements have the potential to offer insight into interstellar cloud properties due to their reactivity @xcite , and consequently , their sensitivity to the abundances of gas - phase electrons and atomic c , h and o , and to physical conditions such as cloud density and molecular depletion .
observations of anions in a variety of interstellar environments will be key to a complete understanding of their chemistry . given the relatively small number of molecular anion detections in the ism to date , we set out to address the question of just how widespread anions are in interstellar clouds , and to ascertain the behavior of the anion - to - neutral ratio over a range of cloud types and ages . in this article ,
results are presented from our survey for c@xmath0h@xmath1 and c@xmath0h in a sample of seven carbon - chain - rich interstellar clouds , protostars and prestellar cores .
detections of c@xmath0h@xmath1 in two of the sources ( l1251a and l1512 ) were previously reported by @xcite , and here we report the final results for all seven sources , as well as the first ( tentative ) detection of c@xmath3h@xmath1 in tmc-1 .
lcclcc l1172 smm & 21:02:22.1&+67:54:48 & star - forming & 440&1 + l1251a & 22:30:40.4&+75:13:46 & star - forming & 300&2 + l1389 ( cb17 ) smm1&04:04:36.6&+56:56:00 & protostellar & 250&3 + l1495b & 04:15:41.8&+28:47:46 & quiescent & 140&4 + l1512 & 05:04:07.1&+32:43:09 & star - forming & 140&4 + tmc-1c & 04:41:35.6&+26:00:21 & star - forming & 140&4 + tmc-1 cp & 04:41:41.9&+25:41:27 & quiescent & 140&4 due to the close chemical relationship between polyynes and cyanopolyynes ( see _ e.g. _ @xcite ) , carbon - chain - bearing species such as c@xmath3h and c@xmath0h are expected to be abundant in dense molecular clouds , close to the peak of hc@xmath9n emission .
target sources with strong hc@xmath9n emission lines were selected from an hc@xmath9n @xmath10 mapping survey of twenty nearby molecular clouds , prestellar cores and young , low - mass protostars obtained using the onsala space observatory ( oso ) 20-m telescope between 2005 and 2012 .
some of these data were presented by @xcite and @xcite .
maps of the integrated hc@xmath9n @xmath10 emission intensity towards six of our seven chosen sources ( l1172 , l1251a , l1389 , l1495b , l1512 and tmc-1c ) are shown in figure [ fig : maps ] .
these observations were obtained using the oso with a beam size of @xmath11 and beam efficiency of @xmath12 .
a @xmath13 map spacing was used , which was resampled to a @xmath14 per pixel grid using bilinear interpolation . for tmc-1 ,
hc@xmath9n maps were published by @xcite and @xcite . to search for c@xmath0h and c@xmath0h@xmath1 , we targeted the strongest hc@xmath9n peak positions in our maps of l1389 , l1495b , l1512 and tmc-1c .
for l1172 we targeted the sub - mm core cataloged by @xcite , which fell within 15@xmath15 of the hc@xmath9n peak .
our chosen l1251a position does not coincide with the main hc@xmath9n peak because it was based on an earlier , lower - sensitivity map than that shown in figure [ fig : maps ] .
for tmc-1 , we targeted the cyanopolyyne peak ( cp ) where @xcite previously attempted to detect c@xmath3h@xmath1 .
the adopted coordinates for our anion survey targets are listed in table [ tab : sources ] .
our l1172 and l1512 c@xmath0h@xmath1 positions overlap the prestellar cores identified by @xcite and @xcite , respectively . for l1389 , the hc@xmath9n emission maps out a very compact clump , coincident with the center of the bok globule cb17 @xcite .
the hc@xmath9n peak also matches closely the location of strongest microwave / sub - mm continuum emission observed by @xcite and denoted l1389 smm1 ( the position of which is indicated in the bottom - left panel of figure [ fig : maps ] by the cross inside the dashed black circle of the gbt beam ) .
as shown in figure [ fig : maps ] , our targeted l1251a position is @xmath16 south - east of the center of the protostar l1251
irs3 @xcite , although the outer envelope of this protostar probably intersects the gbt beam to some extent , as discussed by @xcite . from dust continuum emission
, @xcite obtained a number density for l1389 smm1 of @xmath17 @xmath18 .
@xcite derived a moderately high co depletion factor ( @xmath19 ) , and inferred a relatively evolved chemical age of @xmath20 myr for this collapsing core .
the core was recently found to contain a very young , low - luminosity protostar ( referred to as cb17 mms ; @xcite ) , @xmath21 ne of our targeted gbt beam center , as well a more evolved class 0/i protostar @xmath22 nw .
given that the radii of typical low - mass protostellar envelopes are @xmath23 au ( _ e.g. _ @xcite ) , at a distance of 250 pc , our targeted l1389 beam is likely to be dominated by protostellar envelope matter .
l1495b , on the other hand , is a quiescent molecular cloud with no known protostars nearby .
it appears to be chemically young , with an apparently low level of co depletion consistent with less - evolved interstellar clouds @xcite . in tmc-1c
, @xcite identified an anti - correlation between the spatial distributions of hc@xmath9n and c@xmath24o , at least partly attributable to co depletion .
thus , our observations of tmc-1c ( at the hc@xmath9n peak ) probably sample more chemically - evolved , depleted gas than l1495b .
no protostars are known to be present in our observed tmc-1c beam , but several sub - mm sources are located in the surrounding cloud ( indicated by white crosses in the lower - right panel of figure [ fig : maps ] ) , showing this to be a region of active star formation .
south - west of tmc-1c lies the source tmc-1 ( cp ) , which is a well - known chemically - young dark cloud , with only a modest degree of depletion @xcite . in table
[ tab : sources ] , we provide a basic categorization of our target sources in light of their properties described above : ` quiescent ' refers to those cloud cores that are chemically young , show no evidence for active star - formation within the telescope beam ( such as outflows or compact sub - mm / ir emission ) , and do not appear to be undergoing collapse ; ` star - forming ' is used for clouds showing nearby active star - formation and a greater degree of chemical evolution ; finally , ` protostellar ' describes our l1389 position , which contains the low - luminosity protostar cb17 mms .
observations of emission lines of c@xmath0h ( @xmath25 ) , c@xmath0h@xmath1 ( @xmath10 and @xmath26 ) and hc@xmath9n ( @xmath27 and @xmath28 ) were carried out between 2010 april and 2011 january using the ka receiver of the nrao 100-meter robert c. byrd green bank telescope ( gbt ) .
the gbt spectrometer was used with a bandwidth of 50 mhz and @xmath29 khz channels ( corresponding to a velocity spacing of @xmath30 ) .
four spectral windows were used , which allowed simultaneous observation of the c@xmath0h and c@xmath0h@xmath1 lines . for the compact , spatially - isolated source l1512 , beam switching ( with @xmath31 throw ) was used , and for all other targets , frequency switching was used . pointing was checked every one to two hours and was typically accurate to within @xmath32 . in the middle of the observed frequency range ( 28 ghz ) ,
the telescope beam fwhm was @xmath33 and the main beam efficiency was 0.90 .
total system temperatures were typically in the range @xmath34 k and the zenith opacity was @xmath35 .
intensity calibration was performed using beam - switched observations of the compact radio source ngc7027 .
measured antenna temperatures were subsequently corrected for opacity , spillover , ohmic loss , blockage efficiency and beam efficiency , then averaged using standard gbtidl routines .
we obtained observations of c@xmath3h ( @xmath36 , @xmath37 , @xmath38 ) and c@xmath3h@xmath1 ( @xmath39 ) between 2011 december and 2012 february at around 19 ghz using the gbt k - band focal - plane array ( kfpa ) .
these observations were obtained with channel spacing 12.2 khz , beam efficiency 0.92 , beam size 39@xmath15 and zenith opacity @xmath40 .
additional single - pointing hc@xmath9n @xmath10 spectra of our sources were obtained using the oso at our adopted target positions .
a list of all the observed transitions is given in table [ tab : trans ] .
llcccc c@xmath3h & @xmath36 , @xmath37 , @xmath38 & 19054.4760&1&gbt&38 + c@xmath3h@xmath1 & @xmath41 & 18619.7580&2,3&gbt&39 + c@xmath0h@xmath1 & @xmath42 & 27537.1302&4&gbt&26 + c@xmath0h@xmath1 & @xmath43 & 30290.8133&4&gbt&24 + c@xmath0h & @xmath44 & 29109.6437&5&gbt&25 + c@xmath0h & @xmath45 & 29109.6855&5&gbt&25 + c@xmath0h & @xmath46 & 29112.7087&5&gbt&25 + c@xmath0h & @xmath47 & 29112.7503&5&gbt&25 + hc@xmath9n & @xmath48 & 27292.9018&6&gbt&27 + hc@xmath9n & @xmath49 & 27294.0758&6&gbt&27 + hc@xmath9n & @xmath50 & 27294.2927&6&gbt&27 + hc@xmath9n & @xmath51 & 27294.3451&6&gbt&27 + hc@xmath9n & @xmath52 & 27296.2334&6&gbt&27 + hc@xmath9n & @xmath53 & 36390.8861&6&gbt&20 + hc@xmath9n & @xmath54 & 36392.2358&6&gbt&20 + hc@xmath9n & @xmath55 & 36392.3293&6&gbt&20 + hc@xmath9n & @xmath56 & 36392.3630&6&gbt&20 + hc@xmath9n & @xmath57 & 36394.1765&6&gbt&20 + hc@xmath9n & @xmath58 & 90978.9838&6&oso&42 + hc@xmath9n & @xmath59 & 90978.9948&6&oso&42 + hc@xmath9n & @xmath60 & 90979.0024&6&oso&42 + + + lccccc l1172 & 6.7 ( 0.8 ) & 41.1 ( 1.6 ) & 2.4 ( 0.3 ) & 7.1 ( 0.3 ) & 3.3 ( 0.5 ) + l1251a & 6.5 ( 1.0 ) & 43.6 ( 2.1 ) & 2.3 ( 0.4 ) & 7.6 ( 0.4 ) & 3.0 ( 0.6 ) + l1389&5.9 ( 0.8 ) & 27.1 ( 1.3 ) & 2.1 ( 0.3 ) & 4.7 ( 0.2 ) & 4.4 ( 0.8 ) + l1495b & 9.6 ( 1.0 ) & 141.6 ( 2.2)&3.4 ( 0.4 ) & 24.6 ( 0.4)&1.4 ( 0.2 ) + l1512 & 4.3 ( 0.4 ) & 26.3 ( 0.9 ) & 1.5 ( 0.1 ) & 4.6 ( 0.2 ) & 3.3 ( 0.4 ) + tmc-1c & 13.6 ( 1.1)&88.1 ( 1.7 ) & 4.8 ( 0.4 ) & 15.3 ( 0.3)&3.1 ( 0.3 ) + tmc-1
cp & 41.6 ( 5.0)&332.4 ( 9.0)&14.7 ( 1.8)&57.8 ( 1.6)&2.5 ( 0.4 ) .
the c@xmath0h@xmath1 anion and its neutral counterpart c@xmath0h were detected in all of the surveyed sources .
this is the first time anions have been detected in l1172 , l1389 , l1495b and tmc-1c .
the observed c@xmath0h@xmath1 and c@xmath0h spectra are shown in figure [ fig : spectra ] .
spectra of the @xmath10 and @xmath61 transitions of c@xmath0h@xmath1 were averaged in velocity space to improve the signal - to - noise ratio ; integrated intensities of the averaged spectra are given in table [ tab : results ] .
due to partial blending of the hyperfine components of the c@xmath0h ( @xmath62 ) lines , the integrated intensity summed over both components is given in table [ tab : results ] .
given the weakness of these lines , they can safely be assumed to be optically thin , and equation 2 of @xcite was used to derive column densities for c@xmath0h and c@xmath0h@xmath1 ( given in table [ tab : results ] ) .
spectroscopic data for the transitions of interest ( table [ tab : trans ] ) were obtained from the cologne database for molecular astronomy @xcite , and an excitation temperature of 10 k was assumed .
error estimates were derived using 500 monte carlo noise replications for each measurement . in tmc-1 ,
c@xmath0h excitation temperatures of 5.2 k and 6.7 k were measured by @xcite and @xcite , respectively .
such sub - thermal excitation might be expected given the large einstein @xmath63 coefficients for rotational transitions of this molecule , which are a direct consequence of its large ( 5.5 d ) dipole moment .
however , the value obtained by @xcite may be subject to uncertainty due to the effects of telescope beam dilution , and the @xcite value may not be applicable to our observations because it was obtained using the @xmath64 transition , which has an einstein @xmath63 coefficient about four times greater than that of the @xmath25 transition we observed .
the larger einstein @xmath63 results in an increased likelihood of sub - thermal excitation of the @xmath65 level .
if the c@xmath0h excitation temperature is as low as 5 k , then the calculated column densities for c@xmath0h and c@xmath0h@xmath1 will be about 36% smaller than the values in table [ tab : results ] . due to the similar moments of inertia of c@xmath0h and c@xmath0h@xmath1
, their rotational levels occur at similar energies , so that the calculated anion - to - neutral column density ratios ( c@xmath0h@xmath1/c@xmath0h ) are insensitive to small uncertainties in excitation , provided both molecules share a common excitation temperature .
the mean c@xmath0h@xmath1/c@xmath0h ratio for the seven sources is 3.0% , with a standard deviation of 0.92% .
our new measurements for c@xmath0h and c@xmath0h@xmath1 in l1251a and l1512 supersede those presented by @xcite .
although the column densities and anion - to - neutral ratios measured in the present study match those of @xcite within the stated errors , the newer values benefit from improved calibration and analysis methods and have smaller uncertainties .
the difference between our tmc-1 ( cp ) c@xmath0h@xmath1/c@xmath0h value of @xmath66% and the value of @xmath67% obtained by @xcite is just beyond the range of the one - sigma error bars , and could be the result of statistical noise in the spectra . alternatively , because @xcite derived their value from a different set of transitions , non - lte excitation effects in c@xmath0h and/or c@xmath0h@xmath1 could be partially responsible for the discrepancy .
the observed c@xmath3h@xmath1 and c@xmath3h spectra of tmc-1 ( cp ) are shown in the bottom - right panel of figure [ fig : spectra ] .
the @xmath68 line of c@xmath3h@xmath1 is detected in a single channel with a peak antenna temperature of 3.4 mk , which corresponds to 4.8@xmath4 ( where @xmath4 is the rms noise of the baseline - subtracted spectrum ) . using a gaussian fit to this c@xmath3h@xmath1 line with 1000 monte carlo noise replications ,
the central velocity was found to be @xmath69 kms@xmath70 with a fwhm of @xmath71 kms@xmath70 . within the errors ,
these parameters match those derived for hc@xmath9n ( given in table [ tab : hc3n ] ) , which adds confidence to our c@xmath3h@xmath1 detection .
the integrated line intensity is @xmath72
mkkms@xmath70 , which corresponds to a column density of @xmath73 @xmath74 ( assuming an excitation temperature of 10 k ) .
the neutral c@xmath3h line has an integrated intensity of @xmath75 mkkms@xmath70 , which requires @xmath76 @xmath74 .
the corresponding anion - to - neutral ratio c@xmath3h@xmath1/c@xmath3h is @xmath77 .
this value is consistent with the upper limit of @xmath78 measured by @xcite , and our c@xmath3h column density also closely matches their value of @xmath79 @xmath74 . both @xmath80 and @xmath81 are relatively insensitive to the adopted excitation temperature , and vary by only 12% over the range @xmath82 k. lcccc l1172 & 2.7 & 2.74 & 0.49&7.5 ( 2.3 ) + l1251a & 3.3 & -3.97&0.31&2.1 ( 0.4 ) + l1389 & 1.5 & -4.75&0.32&5.2 ( 1.3 ) + l1495b & 8.2 & 7.62 & 0.19&1.1 ( 0.3 ) + l1512 & 4.2 & 7.11 & 0.10&2.6 ( 0.9 ) + tmc-1c & 6.4 & 5.37 & 0.18&1.1 ( 0.4 ) + tmc-1 cp & 19.5&5.75 & 0.32&1.0 ( 0.2 ) our hc@xmath9n maps ( figure [ fig : maps ] ) show the presence of one or more compact molecular condensations in each source , and highlight a clumpy structure in this carbon - chain - rich gas .
the peaks in hc@xmath9n emission tend to coincide approximately with the locations of known prestellar cores and protostars , the positions of which are denoted with crosses and asterisks , respectively . as a result of the collapse and infall processes occurring in these objects , their densities are expected to be greater than the surrounding gas .
this should result in increased collisional excitation of the @xmath83 level of hc@xmath9n , which has a critical density of @xmath84 @xmath18 @xcite .
thus , the intensity variations shown in our hc@xmath9n oso maps represent a combination of variations in gas density and total hc@xmath9n column .
spectra of three rotational transitions of hc@xmath9n for each source , observed at the locations targeted by our gbt anion survey , are shown in figure [ fig : hc3n ] .
these span a range of upper - state energy levels @xmath85 cm@xmath70 ( @xmath87 k ) . in order to derive gas densities and velocities ,
the hc@xmath9n spectra were subject to fitting using the radex radiative transfer code @xcite .
collisional transition rates were taken from the leiden atomic and molecular databasemoldata ] @xcite , which tabulates scaled versions of the original data of @xcite .
hyperfine structure ( hfs ) was accounted for by assuming lte population of the hyperfine levels in each @xmath88 state . the primary collision partner number density ( @xmath89 ) ,
molecular column density ( @xmath90 ) , doppler line fwhm ( @xmath91 ) and doppler velocity ( @xmath92 ) were optimized for the spectra of each source using the mpfit least - squares algorithm @xcite .
the best - fitting parameters are given in table [ tab : hc3n ] .
the kinetic temperature was fixed at 10 k , then varied by @xmath93 k to produce the quoted error estimates on @xmath89 .
for all sources , a temperature of 10 k produced a good fit to all five hfs components ( the two weakest hfs components are not shown in figure [ fig : hc3n ] ) .
the hc@xmath9n doppler velocities are shown with vertical dashed lines in figure [ fig : spectra ] and closely match the central velocities of the detected hydrocarbons and anions . in several cases ,
the hc@xmath9n line widths are very narrow ; for l1512 ,
@xmath94 kms@xmath70 , which is only slightly greater than the 10 k thermal line - width of 0.07 kms@xmath70 , indicating an unusual lack of turbulence , flows or shears in the central @xmath95 of this source .
the c@xmath0h@xmath1/c@xmath0h ratio is plotted as a function of @xmath89 in figure [ fig : anr ] , and shows a positive correlation .
there is considerable scatter and uncertainty on @xmath89 , but consistent with previous studies ( @xcite , @xcite , @xcite and @xcite , respectively ) , we find the lowest density in tmc-1 , highest densities in in l1172 and l1389 , and an intermediate density in l1512 .
our calculated @xmath89 values for l1172 , l1389 and l1512 are , however , systematically lower than previously - derived values by up to two orders of magnitude .
this discrepancy is likely indicative of a problem in our method , which can be attributed to the effects of different degrees of beam dilution affecting the different hc@xmath9n line frequencies we observed . given the larger @xmath11 beam size for the hc@xmath9n @xmath10 line observed with the oso 20-m compared to the @xmath96 gbt beam for the @xmath27 line , it is theoretically possible that the oso data suffers up to an additional factor of 2.6 in beam dilution compared with that of the gbt , which could result in sufficient reduction of the derived @xmath89 values to account for the observed discrepancies .
figure [ fig : maps ] ( and figure 7 of @xcite ) shows that the hc@xmath9n emission from all our observed sources is distributed over an area larger than the oso beam , which suggests that beam dilution may not be so severe .
however , given the lack of spatial information on the extent of the @xmath10 emission on size scales less than the oso beam size , it is impossible to quantify the importance of beam dilution in our radex calculations , so the @xmath89 values given in table [ tab : hc3n ] should be treated with caution .
more accurate estimates for @xmath89 could be obtained from hc@xmath9n @xmath10 observations using a telescope with a smaller beam ( such as the iram 30-m ) , which would be a better match to that of our gbt observations .
h@xmath1 anion - to - neutral ratio _ vs.
_ h@xmath97 number density ( @xmath89 ) .
dashed line shows linear least - squares fit . [
fig : anr ] ] since the 2006 discovery of c@xmath0h@xmath1 in tmc-1 , there have been seven reported detections of this anion in various parts of the taurus - auriga molecular cloud complex ( including our latest detections in l1495b , l1512 and tmc-1c ) .
thus far , there have been only two detections of interstellar anions outside of this small region @xcite , which highlights the importance of our new discoveries of c@xmath0h@xmath1 in the vicinity of l1172 smm ( a prestellar core in cepheus ) and l1389 smm1 ( a prestellar / protostellar core inside bok globule cb17 in camelopardalis ) .
these new detections confirm that anions are indeed widespread in star - forming regions outside of taurus , and that c@xmath0h@xmath1 appears to be ubiquitous in dense interstellar clouds wherever its parent neutral is present .
@xcite detected an outflow in l1172 that @xcite identified as originating from a protostar ( l1172 smm1 ) that lies about 30@xmath15 south of our targeted position .
they also detected a dense core ( presumed prestellar ) , about 50@xmath15 se of our target , with @xmath98 @xmath74 .
if this column density also applies to our observed position , the c@xmath0h@xmath1 abundance is @xmath99 , which is a factor of a few less than in tmc-1 , but an order of magnitude greater than in l1512 , l1521f and l1544 ( see * ? ? ? * and references therein ) , suggesting a relatively high absolute anion abundance in l1172 . by utilizing the available physical and chemical information on the interstellar clouds
for which c@xmath0h@xmath1 has so far been detected ( as summarized for our observed targets in section [ sec : targets ] ) , it is possible to examine the c@xmath0h@xmath1/c@xmath0h ratio with respect to each cloud s chemical and dynamical evolutionary state .
the evolutionary states of the observed clouds are given in table [ tab : sources ] , and are shown by the different symbols on the plot in figure [ fig : anr ] .
there is a clear division in the anion - to - neutral ratio between young , ` quiescent ' and older , ` star - forming ' cores : for the quiescent , chemically - young cores l1495b and tmc-1 ( cp ) , c@xmath0h@xmath1/c@xmath0h @xmath100% , and for the star - forming cores / protostars , c@xmath0h@xmath1/c@xmath0h @xmath101% .
l1495b is the most chemically - young and quiescent of our sources @xcite , with no evidence for nearby star - formation , and has the lowest c@xmath0h@xmath1/c@xmath0h ratio in our sample ( @xmath102% ) .
the very low - luminosity , low - mass protostar ( vello ) l1389
smm1 , on the other hand , evidently contains among the most evolved matter in our sample ( having passed through the quiescent and star - forming stages ) , and has the highest anion - to - neutral ratio ( c@xmath0h@xmath1/c@xmath0h = @xmath103% ) .
this value is similar to the value of @xmath104% found in l1521f by @xcite , which also contains a vello @xcite . comparing our results with other sources from the literature ,
l1527 @xcite contains a somewhat more - evolved class 0/i protostar , and this source has a significantly larger c@xmath0h@xmath1/c@xmath0h ratio of @xmath105% .
thus , we confirm the apparent relationship identified by @xcite between the evolutionary state of interstellar matter and its c@xmath0h@xmath1 anion - to - neutral ratio .
the quiescent cloud lupus-1a observed by @xcite also fits the trend ( with c@xmath0h@xmath1/c@xmath0h = @xmath106% ) , as does the very low - luminosity protostar chameleon mms1 ( with its upper limit of c@xmath0h@xmath1/c@xmath0h @xmath107% ; @xcite ) .
the dense prestellar core l1544 may deviate , however , with a relatively low ratio of @xmath108% @xcite .
the observed c@xmath0h@xmath1/c@xmath0h trend and its correlation with @xmath89 shown in figure [ fig : anr ] can be understood in the context of the theoretical study of @xcite , who identified the effects of depletion on the c@xmath0h@xmath1 anion - to - neutral ratio , the degree of which is related to both the density and chemical age of the cloud .
depletion occurs as atoms and molecules collide with and stick to dust grains , and thus proceeds faster in denser media .
increased depletion affects the anion - to - neutral ratio in two ways : ( 1 ) the free electron abundance goes up so that more c@xmath0h@xmath1 is produced by radiative electron attachment and ( 2 ) the atomic o and h abundances go down , thus reducing the anion destruction rate .
our c@xmath0h@xmath1 observations agree with this theory , particularly for the quiescent cores and the protostars , which are believed to lie at opposite ends of the interstellar chemical / dynamic evolutionary path .
measured anion - to - neutral ratios , however , are not sufficiently accurate to draw a clear distinction among the ( moderately evolved ) prestellar cores and star - forming gas clouds of l1172 , l1251a , l1512 and tmc-1c .
these clouds likely contain gas that spans a range of densities and degrees of chemical evolution .
more accurate measurements of density , depletion and the c@xmath0h@xmath1/c@xmath0h ratio at high spatial resolution will be required in order to confirm the utility of the anion - to - neutral ratio as a measure of the evolutionary state of interstellar clouds .
our detection of c@xmath3h@xmath1 constitutes the smallest reported column density for this molecule in any source to - date .
consistent with the trend described above for c@xmath0h@xmath1 , the c@xmath3h@xmath1 anion - to - neutral ratio of @xmath77 in tmc-1 ( cp ) is about nine times smaller than observed in l1527 by @xcite .
theory regarding c@xmath3h@xmath1 chemistry is less well understood than for c@xmath0h@xmath1 , but it seems plausible to again ascribe the lower value in tmc-1 ( and the moderately low value of @xmath109 in lupus-1a ; @xcite ) , to chemical effects resulting from the lower densities and younger evolutionary states of these sources compared with l1527 .
the very low observed c@xmath3h@xmath1 anion - to - neutral ratios compared with c@xmath0h@xmath1 are at variance with chemical models that consider the formation of c@xmath3h@xmath1 to be by radiative electron attachment to c@xmath3h at the rate calculated by @xcite .
recent ab - initio calculations by v. kokoouline and colleagues ( private communication , 2012 ) show that the c@xmath3h radiative attachment rate may be several orders of magnitude less than previously thought , and based on observations of the c@xmath3h@xmath1/c@xmath3h ratio in l1527 , @xcite calculated a c@xmath3h radiative attachment rate a factor of 122 less than the theoretical value of @xcite . however , using the laboratory data of @xcite , @xcite identified that a significant pathway to smaller hydrocarbon anions is _ via _ reactions of larger anions with atomic oxygen .
thus , an important source of c@xmath3h@xmath1 is the reaction @xmath110 as a consequence of this ( and the analogous reaction that forms c@xmath7h@xmath1 from c@xmath0h@xmath1 + o ) , using the model of @xcite at @xmath111 @xmath18 ( applicable to tmc-1 ) , it is possible to set the radiative attachment rates for both c@xmath3h and c@xmath7h to zero and still produce an absolute c@xmath3h@xmath1 abundance and c@xmath3h@xmath1/c@xmath3h ratio that are within a factor of a few of the observed values .
it is therefore plausible that radiative electron attachment is not the dominant route to the formation of c@xmath3h@xmath1 in molecular clouds .
we conducted a search for the carbon - chain anion c@xmath0h@xmath1 in seven nearby molecular clouds selected based on their strong hc@xmath9n emission .
c@xmath0h@xmath1 and c@xmath0h were detected in all sources , with a mean anion - to - neutral ratio of 3.0% ( standard deviation 0.92% ) .
combined with the four previous c@xmath0h@xmath1 detections in other sources , we deduce that anions are ubiquitous in the ism wherever sufficient c@xmath0h is present to be easily detectable . from the combined sample of eleven known interstellar sources with c@xmath0h@xmath1 detections , we confirm the trend identified by @xcite for anion - to - neutral ratios to be smaller in younger , less dense environments and larger in older , denser environments .
this is consistent with the theory of @xcite that anion - to - neutral ratios are sensitive to the depletion of atomic o and h , and to the electron density , both of which become larger in denser , more chemically - evolved cores .
molecular hydrogen number densities were derived for each source based on radiative transfer fits to observed hc@xmath9n spectra ; a positive correlation is observed between c@xmath0h@xmath1/c@xmath0h and @xmath89 .
we report the first detection of c@xmath3h@xmath1 in tmc-1 , and derive an anion - to - neutral ratio c@xmath3h@xmath1/c@xmath3h @xmath112 ( @xmath6% ) .
this is the smallest anion - to - neutral ratio yet observed and confirms a problem with the theoretical value for the c@xmath3h electron attachment rate , which is apparently several orders of magnitude too large .
the observed c@xmath3h@xmath1 may instead be formed by reactions of the larger hydrocarbon anions ( c@xmath0h@xmath1 and c@xmath7h@xmath1 ) with atomic oxygen .
given the widespread presence of anions in space , a combination of further dedicated laboratory , observational and theoretical studies will be required in order to better understand their behavior and thus their full importance in astrochemistry .
measurements of radiative electron attachment rates for hydrocarbons , and branching ratios for reactions of anions with atomic oxygen will be of particular importance in this regard .
this research was supported by the nasa exobiology program and the nasa astrobiology institute through the goddard center for astrobiology .
we gratefully acknowledge the assistance of tom millar and catherine walsh during the 2010 onsala hc@xmath9n observing run .
jvb thanks john richer for support .
agndez , m. , fonfra , j. p. , cernicharo , j. , pardo , j. r. , gulin , m. 2008 , a&a , 479 , 493 bell , m. b. , feldman , p. a. , watson , j. k. g. et al . 1999 ,
apj , 518 , 740 bourke , t. l. , myers , p. c. , evans , n. j. et al .
2006 , apj , 649 , l37 brnken , s. , gupta , h. , gottlieb , c. a. , mccarthy , m. c. , thaddeus , p. 2007 , apj , 664 , l43 buckle , j. v. , rodgers , s. d. , wirstrm , e. s. et al .
2006 , faraday discussions , 133 , 63 clemens , d. p. barvainis , d. p. 1988 ,
apjss , 68 , 257 cordiner , m. a. , millar , t. j. , walsh , c. et al .
2008 , iaus , 251 , 157 cordiner , m. a. , charnley , s. b. , buckle , j. v. , walsh , c. , millar , t. j. 2011 , apj , 730 , l18 cordiner , m. a. , charnley , s. b. , wirstrm , e. s. , smith , r. g. 2012 , apj , 744 , 131 cordiner , m. a. , charnley , s. b. 2012 , apj , 749 , 120 chen , x. , arce , h. , dunham , m. et al .
2012 , apj , 751 , 89 dalgarno , a. , mccray , r. a. 1973 , apj , 181 , 95 de zafra , r. l. 1971 , apj , 170 , 165 di francesco , j. , johnstone , d. , kirk , h. , mackenzie , t. , ledwosinska , e. 2008 , apjs , 175 , 227 eichelberger , b. , snow , t. p. , barckholtz , c. , bierbaum , v. m. 2007 , apj , 667 , 1283 elias , j. h. 1978 , apj , 224 , 857 federman , s. r. , huntress , w. t. , jr . ,
prasad , s. s. , 1990 , apj , 354 , 504 gottlieb , c. a. , gottlieb , e. w. , thaddeus , p. , kawamura , h. 1983 , apj , 275 , 916 gupta , h. brnken , s. , tamassia , f. et al .
2007 , apj , 655 , l57 gupta , h. , gottlieb , c. a. , mccarthy , m. c. , thaddeus , p. 2009 , apj , 691 , 1494 green , s. , chapman , s. 1978 , apjs , 37 , 169 harada , n. , herbst , e. 2008 , apj , 685 , 272 herbst , e. 1981 , nature , 289 , 656 herbst , e. , osamura , y. 2008 , apj , 679 , 1670 hirahara , y. , suzuki , h. , yamamoto , s. 1992 , apj , 394 , 539 hirota , t. , hiroyuki , m. , yamamoto , s. 2004 , apj , 617 , 399 jrgensen , j. k. , schier , f. l. , van dishoeck , e. f. 2002 , a&a , 389 , 908 kirk , j. m. , ward - thompson , d. , andr , p. 2005 , mnras , 360 , 1506 kun , m. , prusti , t. 1993 , a&a , 272 , 235 launhardt , r. , nutter , d. , ward - thompson , d. et al .
2010 , apjs , 188 , 139 lee , j. , lee , h. , shinn , j. et al .
2010 , apj , 709 , 74 lis , d. c. , roueff , e. , gerin , m. et al .
2002 , apj 571 , l55 markwardt , c. b. 2008 , astronomical data analysis software and systems xviii , quebec , canada , asp conference series , vol . 411 , eds .
d. bohlender , p. dowler & d. durand ( astronomical society of the pacific : san francisco ) , p. 251
- 254 mccarthy , m. c. , thaddeus , p. 2005
, j. chem . phys . , 122 , 174308 mccarthy , m. c. , gottlieb , c. a. , gupta , h. c. , thaddeus , p. 2006 , , 652 , l141 mccarthy , m. c. , thaddeus , p. 2008
, j. chem .
phys . , 129 , 054314 millar , t. j. , herbst , e. 1994 , a&a , 288 , 561 millar , t. j. , walsh , c. , cordiner , m. a. , n chuimn , r. , herbst , e. 2007 , apjl , 662 , l87 mller , h. s. p. , schlder , f. , stutzki , j. and winnewisser , g. 2005 , j. mol .
, 742 , 215 myers , p. c. , heyer , m. , snell , r. , goldsmith , p. 1988
, apj , 324 , 907 pavlyuchenkov , y. , wiebe , d. , launhardt , r. , henning , t. 2006 , apj , 645 , 1212 pratap , p. , dickens , j. e. , snell , r. l. et al .
1997 , apj , 486 , 862 remijan , a. j. , hollis , j. m. , lovas , f. j. et al .
2007 , apjl , 664 , l47 sarre , p. j. 1980 , j. chim .
phys . , 77 , 769 schier , f. l. , van der tak , f. f. s. , van dishoeck , e. f. , black , j. h. 2005 , a&a , 432 , 369 sakai , n. , sakai , t. , osamura , y. , yamamoto , s. 2007 , apjl , 667 , l71 sakai , n. , shiino , t. , hirota , t. , sakai , t. , yamamoto , s. 2010 , apj , 718 , l49 thorwirth , s. , mller , h. s. p. , winnewisser , g. 2000 , j. mol .
spectrosc . , 204 , 133 van der tak , f. f. s. , black , j. h. , schier , f. l. , jansen , d. j. , van dishoeck , e. f. 2007 , a&a , 468 , 627 visser , a. e. , richer , j. s. , chandler , c. j. 2002 , apj , 124 , 2756 ward - thompson , d. , scott , p. f. , hills , r. e. , andr , p. 1994
mnras , 268 , 276 woodin , r. , foster , m. s. , & beauchamp , j. l. 1980 , j. chem .
phys . , 72 , 4223 | results are presented from a survey for molecular anions in seven nearby galactic star - forming cores and molecular clouds .
the hydrocarbon anion c@xmath0h@xmath1 is detected in all seven target sources , including four sources where no anions have been previously detected : l1172 , l1389 , l1495b and tmc-1c .
the c@xmath0h@xmath1/c@xmath0h column density ratio is @xmath2% in every source , with a mean value of 3.0% ( and standard deviation 0.92% ) .
combined with previous detections , our results show that anions are ubiquitous in dense clouds wherever c@xmath0h is present .
the c@xmath0h@xmath1/c@xmath0h ratio is found to show a positive correlation with molecular hydrogen number density , and with the apparent age of the cloud .
we also report the first detection of c@xmath3h@xmath1 in tmc-1 ( at 4.8@xmath4 confidence ) , and derive an anion - to - neutral ratio c@xmath3h@xmath1/c@xmath3h @xmath5 ( @xmath6% ) .
such a low value compared with c@xmath0h@xmath1 highlights the need for a revised radiative electron attachment rate for c@xmath3h .
chemical model calculations show that the observed c@xmath3h@xmath1 could be produced as a result of reactions of oxygen atoms with c@xmath7h@xmath1 and c@xmath0h@xmath1 . |
the study of instabilities in disks has a long history , following the seminal work of toomre ( 1964 ) , with a considerable literature on many aspects of it .
however , relatively little attention has been given to the most massive agglomerations that can form by the fragmentation of galactic gas disks .
these most massive agglomerations are of interest because they may be the precursors of the most massive star clusters known , the globular clusters .
globular clusters were until recently viewed as exclusively old objects , and cluster formation models were therefore based on ideas about early stages of galaxy formation .
this view changed with the realization that elliptical galaxies often contain two populations of globular clusters that appear to have different origins , suggested to be a ` primordial ' population and a ` merger ' population ( ashman & zepf 1992 ) .
the existence of a merger population was confirmed by the hubble space telescope with the discovery of massive young clusters in merging galaxies ( ashman & zepf 1998 ; schweizer 1998 ) .
a modern theory of globular cluster formation should therefore address why globular cluster formation is common in some environments such as merging and high - redshift galaxies but not in others such as the present milky way .
another question to be addressed is why some regions of galaxies are more favorable for the formation of massive clusters than others .
for example , the inner part of the milky way galaxy contains young clusters with masses up to several times @xmath0 , including some quite close to the central black hole ( krabbe et al .
1995 ; figer 2008 ) , that are an order of magnitude more massive than the typical open clusters found elsewhere in the galaxy . the giant molecular clouds ( gmcs ) in which open clusters currently form in the outer milky way have masses of order @xmath1 , but they produce clusters with masses of only several times @xmath2 .
the formation of globular clusters with masses of @xmath3 therefore requires either much more massive super - gmcs ( harris & pudritz 1994 ) or a much higher star formation efficiency . in reality ,
a combination of these effects is probably involved .
several scenarios have been proposed for the formation of massive super - gmcs , involving for example a hot primordial plasma ( fall & rees 1985 ) , collisions between normal gmcs ( harris & pudritz 1994 ) , or confinement by high pressures in mergers ( ashman & zepf 2001 ) .
relatively little attention has been given to the possible role of disks in the formation of globular clusters , with the exception of larson ( 1988 , 1996 ) , but this possibility now seems worth further investigation given the evidence for massive cluster formation in rotationally supported disks in merging galaxies ( ashman & zepf 1998 ) .
in this @xmath4 we study the gravitational instability of galactic gas disks , with special attention to the most massive clumps that can form in them .
we also consider which galaxies or environments will produce the largest unstable clumps and are therefore most favorable for massive cluster formation .
we start by reviewing the theory of instabilities in disks in 2 , and continue with its application to the formation of globular - type clusters in 3 .
we discuss the most promising environments for globular - cluster formation in 3.1 , and in 4 we summarize the main implications of our work .
for illustration we consider the simplest possible case : a uniformly rotating thin sheet or disk .
linear stability analysis of such a system yields the dispersion relation for small perturbations in surface density ( binney & tremaine 1987 ) @xmath5 where @xmath6 is the sound speed , @xmath7 is the surface density , and @xmath8 is the angular rotation speed . to see the implications of equation [ dispersion ] , we first consider the limit of a non - rotating sheet with @xmath9 . in this case
the sheet is unstable ( @xmath10 ) for wavelengths @xmath11 larger than the jeans length , @xmath12 .
thus pressure can only stabilize the sheet at small scales .
another limiting case is a pressureless sheet with @xmath13 .
equation [ dispersion ] then shows that only perturbations with @xmath14 are unstable , so that rotation can only stabilize the sheet at large scales .
the existence of a maximum length scale not stabilized by rotation was first derived and discussed by toomre ( 1964 ) .
clearly neither pressure nor rotation can by itself stabilize the sheet , and there is a range of unstable length scales limited on small scales by thermal pressure ( at the jeans length @xmath15 ) and on large scales by rotation ( at the critical length set by rotation , @xmath16 ) .
all intermediate length scales are unstable , but the most rapidly growing mode has a wavelength @xmath17 .
only a combination of pressure and rotation can stabilize the sheet , and this happens when the range of unstable wavelengths shrinks to zero ; according to equation [ dispersion ] this occurs when @xmath18 , which is equivalent to @xmath19 , the classical toomre ( 1964 ) criterion for the stability of a uniformly rotating disk .
such stability analyses apply only to fluid disks that are well described by a simple equation of state ( eos ) , such as an isothermal or a polytropic eos .
unfortunately , the real interstellar medium in galaxies is highly complex and not well described by a simple equation of state .
the complex dynamics and thermodynamics of the real interstellar medium produce large fluctuations and sharp transitions in its properties , as happens for example in molecular clouds , that change the jeans mass by orders of magnitude and are not well described by a simple eos .
a medium with complex structure and dynamics may therefore have no well - defined jeans length , and there may be no real lower limit on the sizes of the self - gravitating structures that can form until the much smaller thermal jeans scale is reached in molecular cloud cores .
this thermal jeans scale may be the next smaller scale that has any clear physical basis and where a relatively simple eos may again apply ; stability analyses give length and mass scales in molecular cloud cores of around 0.1 pc and 1 @xmath20 . the maximum length scale set by rotation @xmath16
is however still a meaningful quantity even when the ism is not well described by a simple eos , because @xmath16 depends only on the surface density and angular frequency of the disk and not on its small - scale physics .
if a disk has a well - defined average surface density , it should have a well - defined maximum length scale for instabilities .
this rotational maximum scale may then be the most relevant length scale for the large - scale dynamics of a disk with complex small - scale physics . in the milky way , @xmath16 is typically of the order of 1 kpc . at intermediate scales between 0.1 pc and 1 kpc
there may not be any preferred scale , and it has been suggested by larson ( 1979 , 1981 ) and others ( ballesteros - paredes et al .
2007 and references therein ) that the structure and dynamics of the ism on these intermediate scales may be roughly self - similar and described by power laws , as in a turbulent cascade . since the maximum unstable length scale set by rotation is the only clearly meaningful length scale for a gas disk with complex physics , we focus on the characteristic mass associated with this scale .
the resulting maximum unstable mass , calculated as @xmath21 , is @xmath22 for typical parameters in the milky way disk , equation [ maxcl ] gives a maximum unstable mass of the order of a few times @xmath1 , similar to the scale of giant cloud complexes in our galaxy .
equation [ maxcl ] gives a mass scale similar to that derived previously by many authors ( balbus 1988 ; elmegreen 1994 , 2002 ; kim & ostriker 2001 ) on the assumption that it is the jeans mass @xmath23 that determines the masses of giant cloud complexes , where the jeans mass is calculated assuming that on large enough scales the ism can be modeled as an isothermal fluid in which random cloud motions take the place of thermal motions .
this mass scale is again about @xmath24 , and it is similar to the rotational maximum mass because the analysis assumes that disks are marginally stable ( @xmath25 ) , implying that the jeans scale and the rotational scale are similar . in the more general case of non - uniform rotation , equations [ dispersion ] and [ maxcl ]
are easily generalized by replacing @xmath26 by the epicyclic frequency @xmath27 , which is given by @xmath28 .
since the epicyclic frequency @xmath27 varies only between @xmath8 and @xmath26 for centrally concentrated disks , we will continue to use @xmath8 because it has a more simple and intuitive meaning . for a review of results of more general gravitational stability analyses see larson ( 1985 ) .
the effect of rotation on the properties of the massive clumps that form in disks is illustrated by the simulations of nuclear gas disks by escala ( 2007 ) .
these simulations study the evolution of a gas disk with a mass of @xmath29 and a diameter of 600 pc , varying only the constant rotation velocity of the disk and studying its effect on the disk s evolution .
this was done by varying a fixed external potential in such a way that the disk is always rotationally supported , but with a range of rotation velocities .
here we compare the development of the instabilities in two disks with rotational velocities @xmath30 of 110 and 321 @xmath31 .
further details of the simulations are given by escala ( 2007 ) .
figure @xmath32 illustrates a representative stage in the evolution of the two disks and shows the density distribution in the disk midplane after 0.4 orbital periods .
the disk with a rotational velocity @xmath30 of 110 @xmath31 is shown in figure @xmath32a and the disk with @xmath30 = 321 @xmath31 in figure @xmath32b .
the plotted region in the x - y plane is 0.5kpc on a side .
both disks develop a complex clumpy structure , and the figures illustrate how rotation affects the growing instabilities . in the run with lower rotation ( fig .
@xmath32a ) the clumps are denser and more numerous , while in the more rapidly rotating disk ( fig .
@xmath32b ) the clumps are less marked and filaments are more prominent , reflecting the disruption of some clumps by rotation .
this is an expected result since figure @xmath32b has a smaller @xmath33 , so that rotation can here disrupt clumps that are able to survive in figure @xmath32a ; this explains why the clumps are denser and more numerous in the run with less rotation .
the gas clumps in figure @xmath32a have typical masses of several times @xmath1 , in rough agreement with the maximum clump mass of @xmath34 predicted by equation [ maxcl ] . for the disk with higher rotation ( fig .
@xmath32b ) , the clumps have smaller typical masses of several times @xmath35 , also in agreement with the value of @xmath36 predicted by equation [ maxcl ] . these values
are valid for the early evolution of the disk before much of its gas has been consumed by star formation . at later times , as @xmath37 decreases , less massive clumps are formed .
the condition for rotational support of a homogeneous uniformly rotating gas sheet against the gravity produced by the total mass enclosed within any radius @xmath38 , @xmath39 , can be written @xmath40 where @xmath41 is the ratio of the gas mass to the total enclosed mass . for a uniformly rotating ,
rotationally supported gas sheet , the maximum clump mass predicted by equation [ maxcl ] can be written @xmath42 equation [ eq4 ] shows that the formation of massive clusters requires not only a large gas content , as in the milky way , but also a large gas fraction ( large @xmath43 ) ; the formation of massive clusters is thus strongly favored in regions where the gas constitutes a large fraction of the mass . in the local universe
this condition is satisfied in ulirgs , which are typically characterized by an ongoing burst of star formation .
recent observations have shown that massive clusters are in fact currently forming in these systems ( ashman & zepf 1998 ) . in these starburst systems ,
not all of the star formation is confined to a single well - defined disk , especially during the early stages of a galaxy merger , as in the antennae system .
however , the maximum mass derived above is set basically by rotation via the balance between centrifugal and self - gravitational forces , and it is not strongly sensitive to the geometry ; equation [ maxcl ] should therefore remain approximately valid within geometrical factors . for a typical nuclear starburst disk with a mass of a few times @xmath44 and a gas fraction @xmath45 of 1/5 ( downes & solomon 1998 ) ,
we predict the formation of clumps with masses up to @xmath46 .
the formation efficiency of massive clusters is very uncertain , but even with an efficiency of only 1% we predict the formation of clusters with masses up to @xmath1 , characteristic of the most massive globular clusters . in normal disk galaxies like the milky way ,
the total gas mass is similar but the gas fraction is only about 1/30 of the total mass .
we then predict clumps with masses of order a few times @xmath1 , as mentioned in 2 , that are not massive enough to form globular clusters .
however , conditions for massive cluster formation become much more favorable in the protogalactic gas disks that are expected to exist during the early stages of disk galaxy formation , where the gas is expected to constitute 1/5 or more of the total mass ( @xmath47 ) .
such protogalactic disks , like for example the clump cluster and chain galaxies observed at high redshift ( elmegreen & elmegreen 2005 ) , would then also be very favorable sites for massive cluster formation .
thus we see from equation [ eq4 ] that there are two favorable environments for the formation of massive clusters : ( 1 ) nuclear gas disks in starburst systems , and ( 2 ) protogalactic disks .
this expectation agrees with the scenario for globular cluster formation proposed by ashman & zepf ( 1992 ) , whereby globular clusters form either in the early stages of galaxy formation or in galaxy mergers and interactions .
this scenario has been successfully tested against several observed properties of globular cluster systems , such as the typically bimodal age and metallicity distributions of globular clusters in elliptical galaxies and the presence of young globular clusters in merger remnants .
moreover it has been recently found that the oldest dwarf galaxies have the highest globular cluster frequencies ( peng et al .
2008 ) , which is expected since systems that formed earlier would probably have had higher gas fractions and therefore they were more favorable sites for cluster formation . the formulation presented in this _
letter _ also allows us to predict which present environments are most favorable for forming massive clusters , given the total gas mass and the gas fraction in galaxies , quantities that can easily be extracted from observations of many systems .
figure [ f1 ] shows the total gas mass of a sample of galaxies plotted against the gas fraction @xmath41 , where the black dots are nuclear starburst disks , the open circles are spirals , and the triangles are dwarf irregular galaxies . the data for spiral galaxies are from kent ( 1987 ) , giraud ( 1998 ) , and helfer et al .
( 2003 ) . for dwarf irregular galaxies the data are from c^ ot ' e et al .
( 2000 ) , van den bosch et al .
( 2001 ) , and for starburst galaxies they are from downes and solomon ( 1998 ) .
the parallel dotted lines show the maximum clump mass predicted by equation [ eq4 ] , with values varying from @xmath0 to @xmath46 ; the solid line is for a maximum clump mass of @xmath48 .
we expect globular clusters to form in clumps with masses above @xmath48 , so that galaxies above the solid line in figure [ f1 ] should be good candidates for massive cluster formation . if massive clusters form with an efficiency of 1% in gmcs ( lada & lada 2003 ) , this corresponds to cluster masses of @xmath35 or more , comparable to globular cluster masses .
figure [ f1 ] shows that with one exception , only galaxies classified as nuclear starbursts lie on or above the @xmath48 line , in agreement with the fact that only starburst galaxies show ongoing massive cluster formation .
the one exception is the spiral galaxy ngc 2403 , which has many giant hii regions and is forming massive stellar clusters like starbursts ( drissen et al 1999 ) , in agreement with our prediction .
figure [ f1 ] also shows that dwarf irregular galaxies should not form massive clusters even though most of their baryonic mass is in gas .
this is because the more important quantity is the gas fraction , which is only between 1% and 10% in these systems because they are dark - matter dominated . in spiral galaxies , although the total amount of gas is comparable to that in starburst galaxies , the gas is more widely distributed and the gas fraction in the star - forming region is therefore smaller , so we do not expect massive cluster formation in most spirals , in agreement with observations .
the position of the milky way in figure [ f1 ] is denoted by mw and the position of the large magellanic cloud is denoted by lmc . both the milky way and the lmc are predicted to have maximum clump masses of several times @xmath1 , which for a cluster formation efficiency of 1% is consistent with the fact that they are both currently forming clusters with masses up to several times @xmath0 ( see figer 2008 for the milky way and larson 1988 for lmc ) .
however , given the uncertainty in the cluster formation efficiency , what is more significant is the prediction that the milky way and the lmc should form clusters two orders of magnitude less massive than the most massive clusters that form in starbursts , as is observed .
in this paper we have studied gravitational instabilities in disks , with special attention to the most massive clumps that form because they are expected to be the progenitors of globular - type clusters .
the maximum unstable mass is set by rotation and depends only on the surface density and orbital frequency of the disk , unlike other mass scales such as the jeans mass that depend on the complex gas physics .
the maximum unstable mass is therefore a well - defined quantity even if the interstellar medium is not well described by a simple equation of state .
the maximum clump mass can be expressed in terms of the total gas mass and the gas fraction in a galaxy , and this formulation makes it clear that environments with a high gas fraction are the most promising places to form massive clusters . using data from the literature , we predict that massive globular - like clusters can form in nuclear starburst disks and protogalactic disks but not in typical spiral galaxies , in agreement with observations .
the scenario proposed here relates massive clusters to the largest scale in galaxies not stabilized by rotation , which is the only scale intermediate between stars and galaxies that has a clear physical basis .
there is no well - defined ` jeans mass ' on these intermediate scales , and the next smaller scale that has any clear physical basis is the thermal jeans scale in molecular clouds , which is related to the masses of individual stars .
the next larger physical scale is that of the galaxy itself , so we predict three physically well - motivated scales corresponding to stars , massive stellar clusters , and galaxies .
another application of studies of the stability of nuclear gas disks , not discussed here , is to their likely role in feeding supermassive central black holes in galaxies ( escala 2007 ) .
this requires the outward transfer of angular momentum , and if gravity is the most important force involved , the same mass concentrations that form massive clusters may also play an important role in the outward transfer of angular momentum and the growth of a central black hole .
we plan to investigate this problem further in ongoing work .
ashman , k.m .
, zepf , s.e .
1992 , , 394 , 50 ashman , k.m . , zepf , s.e .
1998 , globular cluster systems .
cambridge university press , cambridge ashman , k.m . , zepf , s.e .
2001 , aj , 122 , 1888 balbus , s. a. 1988 , apj , 324 , 60 ballesteros - paredes , j. , et al . 2007 .
protostars and planets v , university of arizona press , tucson .
binney , j. , tremaine , s. 1987 , galactic dynamics .
princeton university press , princeton c^ ot , s. , et al .
2000 , aj , 120 , 3027 downes , d. , solomon , p. m. 1998 , , 507 , 615 drissen , l. , et al .
1999 , , 117 , 1249 elmegreen , b. g. 1994 , apj , 433 , 39 elmegreen , b. g. 2002 , apj , 577 , 206 escala , a. 2007 , apj , 671 , 1264 fall , s. m. , & rees , m. j. 1985 , apj , 298 , 18 figer , d. f. 2008 arxiv : astro - ph/08014178 giraud , e. 1998 , aj , 116 , 1125 harris , w. , pudritz , r. 1994 , , 429 , 177 helfer , t.t . , et al .
2003 , apjs , 145 , 259 kent , s.m .
1987 , aj , 93 , 816 kim , w .- t . , ostriker , e.c . 2001 , apj , 559 , 70 krabbe , a. , et al .
1995 , apj , 447l , 95 lada , c.j .
, lada , l.a .
2003 , , 41 , 57 larson , r.b .
1981 , mnras , 194 , 809 larson , r.b . 1985 , mnras , 214 , 379 larson , r.b . 1988 .
proceedings of the 126th iau symposium , kluwer academic publishers .
larson , r.b .
formation of the galactic halo ... inside and out , asp conference series , vol .
92 , 1996 .
peng , e.w . , jordan , a. , cote , p. 2008
, arxiv : astro - ph/08030330 schweizer , f. 1998 .
galaxies : interactions and induced star formation , ed .
r. c. kennicutt et al .
, springer : berlin .
toomre , a. 1964 , apj , 139 , 1217 van den bosch , f.c . 2001 ,
mnras , 326 , 1205 | we study gravitational instabilities in disks , with special attention to the most massive clumps that form because they are expected to be the progenitors of globular - type clusters .
the maximum unstable mass is set by rotation and depends only on the surface density and orbital frequency of the disk .
we propose that the formation of massive clusters is related to this largest scale in galaxies not stabilized by rotation .
using data from the literature , we predict that globular - like clusters can form in nuclear starburst disks and protogalactic disks but not in typical spiral galaxies , in agreement with observations . |
in spite of the remarkable progress that has proved the cosmological origin of grbs , there remains a number of extremely important issues which are still not resolved .
some of the main issues dealt with in this work are the following .
what is the the cosmological evolution of the source population of gamma - ray bursts ( grbs ) ?
does it evolve as the star formation rate ( hereafter the sf rate ) or does it has its own specific evolution ?
( ii ) what is the luminosity distribution ( or function ) of grbs ? ( iii ) what is the total rate of grbs in the universe ?
there are two competing approaches for such studies : ( i ) the `` statisitical '' one using large samples of poorly localized grbs , and ( ii ) the `` individual '' one , using the small sample of optically followed - up grbs , where we have additional information per event including the redshift and the intrinsic luminosity .
the main data array for statistical studies was supplied by the burst and transient source experiment ( batse ) ( fishman et al .
1989 ) onboard the _ compton gamma - ray observatory _
( _ cgro _ ) .
the batse sample is a few times larger than the yield of all other experiments that have detected grbs .
it includes 2702 events in its final form .
fitting the batse data to various cosmological / evolutionary models has been the subject of many studies since the start of the batse operation in 1991 . for references to early works on fitting the batse brightness distribution of grbs to cosmological models , see bulik ( 1999 ) . for a number of results setting an upper limit to the width of the luminosity function of grbs ,
see , e.g. , hakkila et al .
then loredo & wasserman ( 1998 ) demonstrated that the 3rd batse catalog does not constrain the luminosity function . later work using the larger sample of the 4th batse catalog gave few constraints .
krumholtz , thorsett , & harrison ( 1998 ) fitting the batse sample at peak fluxes @xmath5 0.42 photons s@xmath1 @xmath6 found that both a non - evolving grb source population and sf evolutionary models fit the data even using standard candle grbs .
wijers et al .
( 1998 ) demonstrated the same for the sf model .
totani ( 1999 ) showed that the sf model does not fit the data using the standard candle assumption .
panchenko ( 1999 ) using the sf model accounting for the time delay due to binary system evolution estimated the minimal width of the luminosity function as being two orders of magnitude in luminosity .
finally , the recent work of porciany & madau ( 2001 ) deals with a larger sample including the batse non - triggered bursts of kommers et al .
( 2000 ) extending the peak flux down to 0.18 photons s@xmath1 @xmath6 .
they found that the data fit can not distinguish between different variants of the grb source evolution at large redshifts .
however , they did not check how sensitive the fit is to the evolution at _ low _ redshifts .
the general impression arising from fitting the batse data using cosmological models was that this approach had little future .
indeed , the `` statistical '' approach demonstrated an agreement between the data and a wide set of models .
the very few constraints obtained were trivial .
the main reason for the poor progress so far is the insufficient depth of the batse sample , i.e. , the too narrow brightness range .
the brightest burst has a peak photon flux of 160 photons s@xmath1 @xmath6 .
bursts useful for a usual least @xmath7 fit are , however , at @xmath8 photons s@xmath1 @xmath6 .
the batse trigger threshold is at 0.2 photons s@xmath1 @xmath6 but the bursts near the threshold are difficult to use because of poorly known threshold effects .
all works cited above used the peak flux range above 0.4 or even above 1 photon s@xmath1 @xmath6 , which is then narrower than two orders of magnitude .
loredo and wasserman ( 1998 ) showed that this is a too narrow range for obtaining constraints from the fits .
the `` individual '' approach gave a wealth of important data . by itself
this approach is , however , still unable to resolve the issues stated above .
besides having a too poor statistics ( presently we have only 17 grbs with known redshifts ) , the approach is subjected to very strong selection biases .
nevertheless , this small sample tells us that the luminosity function is at least 2.5 orders of magnitude wide and that it extends up to @xmath9 erg s@xmath1 .
after most previous work in this area was completed , the following progress concerning the data accumulation has taken place : - batse obtained additional data until the deorbiting of _ cgro _ in june 2000
. - searches for non - triggered bursts were performed by schmidt ( 1999 ) , kommers et al .
( 2000 ) , and stern et al .
( 2000 , 2001 ) . in the latter work ,
the statistics of useful grb events was increased by a factor 1.7 and the threshold effect was measured .
thus , the useful fitting range was extended down to 0.1 photons s@xmath1 @xmath6 .
- a sample of grbs with known intrinsic luminosities ( presently 17 events ) appeared due to the optical afterglow observations . in this work
, we use these advances including the `` individual '' grb data .
in addition , we include the brightest peak flux interval , which is statistically poor , into the maximum likelihood fit and find that this peak flux interval is very informative .
thus we extend the fitting brightness range to 3 orders of magnitude .
this allow us to obtain a number of conclusive results . in
2 - 4 , we describe the set of fitted data , the fitting models including the cosmology , the source evolution , and the luminosity function of grbs , and finally the the fitting procedure . in 5 , we present results of the fits and show that the scenario of a non - evolving population of grbs does not fit the data .
instead , we demonstrate that the grb - population should decline approximately as fast as the star formation rate .
we also determine the approximate shape of the luminosity function and give an estimate for the lower limit of the total rate of grbs as being 3000 grbs per year in the visible universe ( that is up to a reasonably large redshift ) .
probing various cosmological and evolutionary models we fitted the sample of 3255 batse grbs longer than 1 s found by stern et al .
( 2000 , 2001 ) in the off - line scan of the batse continuous daily records in 1.024 s time resolution .
this sample , which is selected from the catalog of stern & tikhomirova , is essentially uniform and has a corresponding efficiency matrix ( measured by a test burst method ) , which is needed when fitting the weak end of the log @xmath10 - log @xmath11 distribution .
( hereafter the term log @xmath10 - log @xmath11 distribution means the _ differential _ distribution of grbs vs. the logarithm of the _ peak photon flux _ ) .
we excluded short bursts ( consisting of one 1.024 s bin ) from the analysis for two reasons : ( i ) short and long bursts could be separate phenomena , and ( ii ) the sample is incomplete regarding short bursts as they have a lower detection efficiency and a wrong brightness estimate in 1.024 s time resolution . by excluding one - bin events , we make our sample more homogeneous .
the brightness distribution of the grbs in this sample was fitted using a hypothetical brightness distribution folded with the detection efficiency matrix described in stern et al .
this matrix was obtained using a sample of @xmath12 11,000 artificial test bursts , which were superimposed on the batse continuous records and then passed through the same procedure of search and processing as real grbs .
the efficiency matrix is approximately given by @xmath13 , \ ] ] where @xmath14 is the expected and @xmath15 the measured count rate in units of counts s@xmath1 @xmath6 ; @xmath16 = @xmath17)^\nu$ ] is the efficiency function with fitted parameters @xmath18 = 0.097 counts s@xmath1 @xmath6 , @xmath19 = 2.34 ; the log - normal factor describes the relative error of the measured count rate , @xmath20 = @xmath21 , and the selection bias is crudely expressed as @xmath22 = @xmath23 . in order to constrain the intrinsic luminosity function ( hereafter the luminosity function or the lf )
, we used the sample of gamma - ray bursts with measured redshifts .
we can not infer the lf from this sample as it is subjected to strong selection biases .
this is demonstrated in section 3.3 .
the redshift data , however , give us a useful piece of information , i.e. , the existence of very intrinsically bright grbs .
three of the intrinsically brigthest bursts are : grb990123 , grb991216 , and grb000131 ( named by dates ) , with redshifts 1.6 ( djorgovski et al .
1999 ) , 1.02 ( vreeswijk et al . 1999 ) , and 4.5 ( andersen et al .
2000 ) , respectively , and with batse peak fluxes 16.4 , 67.5 , and 6.3 photons s@xmath1 @xmath6 in the 50 - 300 kev range , respectively ( estimated using the batse catalog ) . if they were emitted at @xmath24 , their peak fluxes taking into account the `` k - correction '' ( i.e. , the correction due to the spectral redshift effects on a fixed spectral band of the detector ) would be 45 , 69 , and 84 photons s@xmath1 @xmath6 , respectively , assuming the cosmological parameters ( @xmath25 .
hereafter , we use the photon peak flux at redshift @xmath26=1 , @xmath27 , as a measure of the intrinsic brightness .
we , furthermore , use the intrinsic brightness interval of these three events , @xmath28 photons s@xmath1 @xmath6 , to constrain the lf when fitting the batse log @xmath10 - log @xmath11 distribution .
the choice of the three brightest events for this purpose is somewhat arbitrary .
we can not use a much wider brightness interval because of a brightness dependent selection bias . on the other hand
, we can neglect such problems for the narrow brightness range of these three events .
the data we fitted are presented in figure 1 . in order to impose a proper constraint on the lf
we must estimate the sampling function for strong grbs . with the sampling function
we mean the probability that a burst will be detected , localized , its afterglow observed and its redshift measured .
this function evolves with time .
it was zero before 1997 .
then this function was limited by the field - of - view of the two beppo - sax wide field cameras , @xmath29 of the sky , as this was the main instrument supplying precise coordinates of grbs during 1997 - 1998 . in 1999 and 2000 , many precise localizations were made by other systems , with most of them being made by the interplanetary network _ ulysses_/_konus_/_near _
( see the ipn home page ) .
this means that in this period the sampling function became larger and for very bright events it could , in principle , approach unity as all instruments of the ipn had a 4@xmath30 field - of - view .
actually it should be considerably less as in the same period , three very strong batse events ( triggers 7301 , 7491 , 7595 ) were not localized .
for one strong batse event ( trigger 7954 , grb000115 ) an x - ray transient was found , but no optical transient . with this background ,
let us take a conservatively high estimate of the sampling function , @xmath31 for grbs with @xmath32 40 photons s@xmath1 @xmath6 , and the conservatively low estimate of the rate of these grbs , @xmath33 yr@xmath1 , in the visible universe .
we tried two sets of cosmological parameters , the flat matter - dominated universe , commonly used in most of previous works : ( @xmath34 ) = ( 1,0 ) and the vacuum - dominated cosmology , which is supported by recent data ( @xmath34 ) = ( 0.3 , 0.7 ) ( see , e.g. , lukash , 2000 ) .
hereafter , these two models are denoted as m - models and @xmath35-models , respectively .
the distribution of grbs over redshift for a non - evolving ( ne ) population for @xmath36 is @xmath37 the photon number `` luminosity '' distance is defined by @xmath38 where @xmath39 is the hubble constant , which is assumed to be 75 km s@xmath1 mpc@xmath1 , when estimating the luminosities of grbs . for the standard luminosity distance , the factor
@xmath40 is replaced with @xmath41 .
another important component of the model is the evolution of the population of grb sources .
we probed four cases : a non - evolving population and three evolution functions correlated with the history of star formation following porciani & madau ( 2001 ) .
the declineing phase of the star formation ( sf ) rate at @xmath42 1.5 is a relatively well measured function of @xmath26 .
its history at @xmath43 2 is , however , controversial .
this issue is discussed in porciani & madau ( 2001 ) giving the relevant references .
below we reproduce three versions of the sf evolution suggested in that work : @xmath44 i.e. , decreasing sf at @xmath2 , @xmath45 i.e. , roughly constant sf at @xmath46 , @xmath47 i.e. , increasing sf at large @xmath26 . models will from now on be denoted as ne , m ; sf1,@xmath35 ; and so on , where m and @xmath35 denotes the two types of cosmologies .
the next step is the generation of the standard candle log @xmath10 - log @xmath11 distributions . at this step
, we introduced an additional broadening of the observed brightness distribution due to the k - correction which depends on the type of grb spectrum .
for this purpose , we obtained the log @xmath10 - log @xmath11 distributions using monte - carlo simulations .
to each simulated grb we prescribed one of 54 spectra of bright batse bursts parametrized by the band expression ( band et al .
all these template spectra were assumed to be emitted at @xmath26= 1 .
then we sampled the @xmath26 of the burst and the corresponding k - correction for the 50 - 300 kev band was applied to the apparent brightness of the simulated grb .
the resulting log @xmath10 - log @xmath11 distributions are shown in figure 2 .
if we neglect the k - correction which depends on the grb spectrum , then each distribution in figure 2 is a direct reflection of the corresponding redshift distributions of grbs .
the standard candle luminosity ( before applying the k - correction ) corresponds to a peak photon flux of 1 photon s@xmath1 @xmath6 in the 50 - 300 kev band at @xmath26=1 .
if we vary this value when fitting the data , none of the 8 models ( two cosmological cases , four evolutionary cases ) would still fit the observed log @xmath10 - log @xmath11 distribution of 3255 long batse grbs ( shown by the crosses in figure 2 ) . the fact that the evolution of the sf1 type with the standard candle luminosity function can not fit data was shown by totani ( 1999 ) and lloyd & petrosian ( 1999 ) .
the ne model gives the smallest deviation in this case , but still the value of @xmath7 is unacceptable ( 61 for 27 degrees of freedom ) .
however , as it will be shown below , the discrepancy in the case of the ne model can not be compensated by any hypothesis of the luminosity function .
the slopes of the three log @xmath10 - log @xmath11 distributions for an evolving sf rate are close to the euclidean slope -3/2 starting from a peak flux @xmath11 @xmath48 photon s@xmath1 @xmath6 , which at this standard candle brightness corresponds to @xmath24 .
the agreement with the euclidean slope is an accidental coincidence due to the superposition of cosmological and evolutionary effects .
as was stated above , the luminosity distribution of events with known @xmath26 can not be used as a base for the lf model in the whole brightness range .
this fact is clear from figure 3 where we present log @xmath10 - log @xmath11 distributions that would give the sample of 17 grbs with known absolute luminosities for the ne , sf2 and sf3 evolutionary models .
all models give a striking disagreement with the data .
the only way to reduce this disagreement is to assume an unreasonably sharp increase of grbs at large redshifts which , in turn , will contradict the redshift data .
therefore the shape of a hypothetical lf remains arbitrary ( except the constraints imposed on the brightest end of the lf ) . in order to get a handle on the lf of grbs
, we tried different types of functions that describe common shapes of wide distributions in nature : the log - normal distribution ( lgn ) , a truncated power law ( tpl ) , a power law with an exponential cutoff ( plexp ) , and a broken power law ( bpl ) .
lgn : @xmath49 , with three free parameters @xmath50 , @xmath51 , and @xmath52 .
tpl : @xmath53 for @xmath54 and 0 outside this interval .
free parameters are @xmath55 , @xmath56 , and @xmath52 . in some fits
, @xmath57 was fixed to @xmath58 leaving 3 free parameters .
plexp : @xmath59 , with three free paremeters @xmath55 , @xmath50 , and @xmath52 .
bpl : @xmath53 for @xmath54 , @xmath60 for @xmath61 and @xmath62 outside the [ @xmath63 interval .
free parameters are @xmath55 , @xmath64 , @xmath57 , @xmath50 , and @xmath52 , while @xmath65 was fixed to a value above the maximum observed grb brightness .
we also considered a smoothed version of the broken power law : sbpl : @xmath66 for technical convenience , we measure the intrinsic brightness as peak count rate or peak photon flux , @xmath27 , in the 50 - 300 kev range produced by a grb at a distance corresponding to @xmath24 .
the absolute peak luminosity of the grbs is related to @xmath27 as @xmath67 erg s@xmath1 assuming isotropic emission .
below we present the main results on the lf both in @xmath27 and in absolute luminosity units .
we used the forward folding method when fitting the observed distribution of grbs , i.e. , the hypothetical brightness distribution was convolved with the efficiency matrix ( 1 ) and fitted to the observed distribution of grbs over peak count rate ( crosses in figure 2 ) .
this distribution was represented by 29 data points spaced by 0.1 in log @xmath11 in the interval 0.067 - 50 photons s@xmath1 @xmath6 in the 50 - 300 kev range . during 9.1 years , there are 3 grbs detected by batse that are brighter ( the rightmost cross in figure 2 and one grb at 160 photons s@xmath1@xmath6 , which is not shown ) .
we treat this brightness range separately , estimating the likelihood function of the fit for each peak flux interval . for the main interval , this is the standard @xmath7 probability function . for the tail of the brightness distribution ,
the likelihood is the poisson probability to sample not more then 3 events brighter than 50 photons s@xmath1 @xmath6 at the given number of such events for the full observation period , @xmath68 , predicted by the model .
the final likelihood finction is the product of these two factors . instead of the usual @xmath7 minimization procedure
, we explore the parameter space sampling @xmath69 random points .
this is sufficient to find the minimum of @xmath7 with a good accuracy while at the same time investigating the @xmath7 `` topography '' of the parameter space region .
this method does not work if the `` valley '' of the minimum has a very small volume in parameter space .
for our cases , the minima are wide and smooth enough .
the maximum likelihood point for the whole sample of points in parameter space represents the _ unconstrained _ fit , where the requirement of the redshift data , @xmath70 yr@xmath1 , was ignored .
the subsample selected including this requirement represents the _ constrained _ fit .
the effective number of degrees of freedom ( dof ) in the second case is smaller by unity as compared to the unconstrained fit .
we present results of both the constrained and the unconstrained fits in order to demonstrate the role of the redshift data and the possible effects of the uncertainty in the estimate of the rate of intrinsically strong grbs .
the best fit parameters are presented in tables 1 and 2 .
table 1 includes fits using various lf models and selected cosmological models .
table 2 summarizes fits using a broken power law lf for all cosmological models and provides data to compare their relevance using a bayesian approach .
the best unconstrained fit using the ne models has @xmath7 = 36.8 for 25 degrees of freedom ( table 1 , truncated power law , tpl ) , which is marginally acceptable if we ignore the tail of the brightness distribution . for the tail ,
the model predicts @xmath68 = 11.5 , while the real number is 3 yr@xmath1 and the corresponding maximum likelihood drops down to 1.3@xmath71 ( see table 1 ) .
when we impose the redshift data constraint , @xmath70 yr@xmath1 , the maximum likelihood factor that we can obtain using the ne,@xmath35 model is 3.2@xmath72 , which is for the case of a broken power law lf ( see table 2 ) . for the ne , m model , the results are even worse .
the likelihood factor is not yet a rejection factor for the ne model , because we can not exclude some bias or contamination which would increase the @xmath7 . to reject the ne hypothesis
, we should demonstrate a good fit for other equally simple and reasonable models .
indeed , data fits with the sf models are much better ( see tables 1 and 2 ) .
their maximum likelihood factor is about 0.02 .
this is the case when we can apply the bayesian approach .
the estimate of the rejection level for models with nonevolving grb source population is the ratio of the maximum ne likelihood factor ( ne,@xmath35 model in table 2 ) to that for sf models ( e.g , sf1,@xmath35 in table 2 ) , which is @xmath73 . figures 4 , 5 , and 6 demonstrate the differences between fits using ne and sf models . as seen in figure 5 , the high brightness slope of the log @xmath10 - log @xmath11 distribution for the broken power law ne model is too flat and it can not be made considerably steeper by modifying the lf .
note that the ne standard candle log @xmath10 - log @xmath11 distribution with @xmath27 = 1 photon s@xmath1 @xmath6 is already flatter than the observed log @xmath10 - log @xmath11 distribution .
then if the lf is extended to @xmath12 100 photons s@xmath1 @xmath6 , the model tail will be considerably flatter than the observed tail independently of how the lf is extended .
looking at the integral distribution of real grbs in figure 6 , one notices that it declines faster than an euclidean distribution .
however , this is still not a statistically significant fact .
the probability of such a deviation from the -3/2 slope by chance is 0.1 , which is relatively large ( integral distributions are known to produce an illusion of statistically significant features from fluctuations ) .
most probably we are just dealing with a moderate fluctuation .
nevertheless , the observed slope could really be steeper than the euclidean one if the decline of grbs is steep enough .
more data are required to clarify this issue .
the rejection of the ne models is significant even without the redshift data .
the unconstrained rejection factor , taken as the ratio of the unconstrained likelihoods of the ne,@xmath35 and sf1,@xmath35 models is @xmath74 ( see table 2 ) .
this means that the result is not very sensitive to the value of the constraining @xmath75 .
if we , e.g. , overestimate the rate of intrinsically strong grbs by a factor 3 ( suppose that it is a fluctuation ) and that @xmath76 , then the rejection factor is 4@xmath71 . while the ne models are rejected using the low redshift behavior , one does not obtain any preference for a certain kind of sf evolution at large redshifts .
different sf models give similar likelihood results ( see the tables ) except for the sf3 scenario , which gives a slightly worse fit .
furthermore , the data do not allow us to distinguish between matter - dominated and vacuum - dominated cosmologies .
once the ne models have been rejected at a significant level , we now concentrate on the sf models , i.e. , models with evolution of the grb source population .
there are two clear features of the lf that we see for all sf models : a near power law interval at the lower brightness range and a break or turnover towards the bright end of the distribution . attempting to replace this construction of the lf by a log - normal lf
gives a decrease of the maximum likelihood by 2 orders of magnitude ( see , e.g. , the two sf1 models in table 1 ) . if we study the @xmath7 topography for the broken power law lf , we find a power law fragment at least 1.7 orders of magnitudes wide and there is no upper limit on its width .
it can be arbitrarily extended to lower brightnesses ( see fig .
indeed , one can see from figure 7 that the @xmath7 distribution reaches its asympthotics : by extending the lf further to lower brightnesses does not affect the model in the range of the data points .
on the other hand , the slope of this power law is surprisingly well constrained , especially for the sf1 model ( fig .
7 ) , and it is sligtly sensitive to the choosen sf behavior at large redshifts . the shape of the @xmath7 minimum changes , however .
larger sf rates at high redshifts allow flatter slopes @xmath55 ( see fig .
the @xmath77 plots for a matter - dominated cosmology , which are not shown , are very close to those shown in figure 8 .
they are just slightly narrower and more symmetric .
a break or a turnover is necessary at a high significance level .
its removal increases @xmath7 by @xmath78 for sf1 and @xmath79 19 - 25 for sf2 models ( see table 2 ) .
the properties of the break are , however , less certain than the parameters of the power law fragment .
the @xmath7 minimum is non - parabolic , asymmetric and relatively wide as seen in figure 9 .
all we can say is that some turnover in the power law lf is required at an intrinsic brightness of about 10 photons s@xmath1 @xmath6 at @xmath24 , or @xmath80 erg s@xmath1 for isotropic emission .
the fitted position of the break is slightly sensitive to the cosmological model but the difference is within the statistical errors .
we , however , can not distinguish between a power law break and an exponential cutoff of the lf . both give a good @xmath7 and the same maximum likelihood ( compare tables 1 and 2 ) .
summarizing , we can claim that the behavior of the lf below @xmath81 photons s@xmath1@xmath6 or @xmath82erg s@xmath1 ( in the 50 - 300 kev range ) is close to the power law @xmath83 ( or , more exactly , the power law index @xmath84 can vary from @xmath85 to @xmath86 depending on the model ) .
then it breaks to a steeper slope . a smooth break ( i.e. , the parametrization sbpl in section 3.3 )
also gives a good result .
the likelihood is 0.016 for the sf2,@xmath35 model when the difference in the slopes is large enough , i.e. , @xmath87 .
figure 10 shows a set of the best fit lfs for different models .
the reason why the best fit lfs is more or less well defined is clear when comparing the lf curves with the batse log @xmath10 - log @xmath11 distribution shown by the crosses .
the latter should be a convolution of the former with the standard candle curves in figure 2 .
for the sf1 and sf2 models , the standard candle curves are `` narrower '' then the batse log @xmath10 - log @xmath11 distribution .
therefore the lf should roughly correspond to the main features of the observed log @xmath10 - log @xmath11 distribution : a power law with a turnover . in the case of the sf1 model with narrower redshift distribution ,
the lf is closer to the observed brightness distribution ( see fig.10 ) .
it is natural that the required turnover of the lf is sharper than that of the log @xmath10 - log @xmath11 distribution with the sharp lf turnover being smoothed by the convolution .
the last column of table 2 shows the lower limits on the total rate of grbs in the visible universe .
these limits correspond to @xmath88 , i.e. , to @xmath89 of a @xmath7 distribution with 25 dof .
note that their values are obtained using an abrupt cutoff of the lf at the dim end .
for the sf2,@xmath35 model , the highest allowed cut off is at @xmath90 erg s@xmath1 .
a more realistic smoother cut off would give a higher lower limit .
we suggest that the value 3000 grbs per year gives a realistic estimate of the minimum grb rate .
this estimate is for _ long _ grbs only .
the result depends on the sf model in a natural way predicting a larger result for a higher sf rate at large redshifts . for the sf2,@xmath35 model of the grb source evolution ,
the highest minimum comoving grb rate at large redshifts is @xmath91 yr@xmath1 gpc@xmath92 . at the present epoch
, it becomes 0.13 yr@xmath1 gpc@xmath92 .
this lower limit coincides with the value claimed by porciani & madau ( 2001 ) as the estimate derived from a fit of the data of kommers et al .
it is , however , very difficult to compare results as the slopes and the ranges of the lf are different .
in fact , the shape of the lf derived by porciany and madau has the same parametrization as one of our models , a power law with an exponential cut off , but , neverheless , contradict our results predicting a different slope and a finite estimate for the low brightness cutoff of the lf .
the latter could be a consequence of the different behavior of the log @xmath10 - log @xmath11 distribution near the threshold in kommers et al .
( 2000 ) .
our estimate is close to the result of schmidt ( 2000 ) which is @xmath12 0.2 yr@xmath1 gpc@xmath92 at @xmath93 derived assuming that the present grb rate is 10 times less than at @xmath94 ( in our case the decline is a factor 23 ) . on the basis of this estimate
, we can discuss the possible association of grb980425 and the supernova sn 1998bw ( bloom et al .
, 1999 , see also lamb , 1999 for a discussion ) .
schmidt ( 2000 ) estimates the probability of events such as grb980425 if the association is real as being as low as @xmath95 yr@xmath1 .
indeed , the distance to sn 1998bw is @xmath12 40 mpc .
the corresponding sampling volume is 2.7@xmath96 gps@xmath97 .
our estimate of the grbs rate 0.13 yr@xmath1 gpc@xmath92 with the cutoff @xmath98 erg s@xmath1 .
the 50 - 300 kev peak luminosity of grb980425 is 3@xmath99 erg s@xmath1 ( our rough estimate ) .
if the luminosity function extends down to this luminosity with the determined slope @xmath100 , then the rate of grbs above this luminosity will be @xmath101 times larger , i.e. , 2.6 yr@xmath1 gpc@xmath92 at @xmath93 and a corresponding rate of 0.7@xmath71 yr@xmath1 in the sampling volume .
furthermore , one should take into account the probability that such events will be localized with the accuracy of a few arcseconds which reduces the estimate by at least an extra order of magnitude .
thus the probabliity of occurance and good localization of an event such as grb980425 within @xmath102 mpc with the power law behavior of lf down to @xmath103 erg s@xmath1 is below @xmath104 per year .
such a small probability can hardly be compensated by a break in the lf below the observational cut off .
the break must be so steep that it will affect the observed log @xmath10 - log @xmath11 distribution near the batse threshold .
this is a strong argument that grb980425 might represent a different phenomenon that does not overlap with classic grbs in its luminosity function ( but at the same time being indistinguishable from typical grbs in its general appearence ) .
a more probable possibility is that we are dealing with an accidental coincidence .
the source population of grbs sharply declines from @xmath105 towards the present epoch .
this fact is established at the confidence level @xmath106 . the measured decline of the star formation rate being used as
the evolution hypothesis for the source population of grbs fits the data satisfactorily .
this is what is expected according to prevailing view of grbs as being the product of stellar evolution .
such a scenario was already successfully applied for the description of the observed log @xmath10 - log @xmath11 evolution , e.g. , by wijers et al .
( 1998 ) , panchenko ( 1999 ) , kommers et al .
( 2000 ) , schmidt ( 2000 ) , and porciani & madau ( 2001 ) .
however , nobody has quantitatively demonstrated that an evolution similar to the sf rate at some range of @xmath26 is _ necessary _ to describe data .
we can mention just the work of schmidt ( 2000 ) where the sf rate hypothesis fits the bright tail of the log @xmath10 - log @xmath11 distribution better ( by a visual impression ) than the ne hypothesis .
however , no quantitative comparison of different scenarios has been done .
we have here demonstrated that the decline of the grb population consistent with the sf decline is a _ necessary _ requirement .
the main issue is whether the grbs are associated with the collapse of massive stars ( collapsars ) or the merging of neutron stars ( mergers ) . on the small scale ,
these scenarios differ regarding the expected correlations of grbs with star formation regions in galaxies : collapsar grbs should be well correlated , merger grbs should show no correlation with the star formation .
this fact stimulated searches for such correlations using optical grb afterglows ( see , e.g. , the review of van paradijs , koveliotou , & wijers , 2000 ) . on the cosmological scale , such correlations should exist in both cases .
however , in the merger scenario the occurance of grbs will be described by the sf rate convolved with a delay function .
if the latter extends to a few billions years , then the decline of the grb population will be considerably flatter . for estimates of the delay function for mergers ,
see portegies zwart & yungelson ( 1998 ) and panchenko et al .
( 1999 ) .
qualitatively , our results favor the collapsar scenario .
a very interesting situation occurs if the very steep , steeper than -3/2 , slope of the log @xmath10 - log @xmath11 tail as shown in figure 5 will persist to larger @xmath11 ( _ ulysses _ data reduced by atteia , boer & hurley , 1999 indicate that this can be the case ) .
then one will have to accept that the grb progenitors are very massive collapsars whose population can decline faster than the general sf rate .
but , at present , we do not have sufficient statistical arguments to rule out mergers .
we believe that the data can give tight constraints on the delay function for the merger scenario and the latter can be challenged by such constraints . however , we leave such estimates for future studies for the reason that in order to obtain solid conclusions , it is worth to incorporate the _ ulysses _ data , which would provide at least a doubling of the statistics of the brightest grbs .
our results concerning the luminosity function of grbs confirm the conclusion of loredo & wasserman ( 1998 ) that the width of the luminosity function is not constrained by the batse data being wider than two orders of magnitude .
the shape of the lf tried in the most of previous works was a truncated power law or a lognormal distribution which satisfied earlier data in a narrower brightness range .
schmidt ( 2000 ) used a broken power law hypothesis similar to one of our models .
the position of the break is consistent with our results .
the interpretation of the shape of the luminosity function is beyond the scope of this work . in principle , such type of distributions - broken or exponentially cut power laws are common in nature
. then the break implies some physical limit such as a finite energy source .
we believe that the outlined shape of the lf might be a useful clue in the development of physical models for the grb emission .
this research made use of data obtained through the heasarc online service provided by nasa / gsfc .
this work was supported by the swedish natural science research council , the royal swedish academy of science , the wenner - gren foundation for scientific research , and the russian foundation for basic research grant 00 - 02 - 16135 .
we thank the anonymous referee for useful suggestions .
llllllll ne,@xmath35&tpl&36.8,39.8&11.5&1.3@xmath10710@xmath108 & -0.37@xmath109 & & 2.8 , 10.7 + ne,@xmath35&lgn&44.3,42.5&16.6&1.2@xmath10710@xmath110 & & 1.9&0.032 , 0.36 + ne,@xmath35&plexp&54.0 , 49.7 & 18.7&2@xmath10710@xmath111&-0.21@xmath112 & & 15 , 21 + sf1,@xmath35&lgn&44.5,39.8&6.0&@xmath113&&2.0 & 0.032 , 0.12 + sf1,@xmath35&plexp&31.2,30.7&7.0&0.022&[email protected] & & 14 , 28 + sf2,@xmath35&plexp&34.2,31.7&6.7&0.016&-0.36@xmath115 & & 11 , 33 + sf2,m&plexp&32.7,31.4&6.9&0.018&-0.36@xmath116 & & 12 , 26 + [ fits ] lllllllll ne , m&39.1&16.8 & 6@xmath10710@xmath117&@xmath118&[email protected]&0.23 , 3.3 & 7.&2800 + ne.,@xmath35&40.4&14.5 & 3.2@xmath72&0.87@xmath71 & -0.3@xmath119&0.43 , 5.0&5.&3000 + sf1,m & 29.0&7.2&0.015&0.043&[email protected] & 2.8 , 9.2 & 33 & 1800 + sf1,@xmath35&29.4&6.7&0.023&0.050&[email protected] & 2.9 , 16 .
& 28 & 2000 + sf2,m & 30.2&6.8&0.018&0.031&[email protected] & 2.9 , 15.&25 & 2700 + sf2,@xmath35&30.3&6.0&0.021&0.028&[email protected] & 3.4 , 21.&19 & 3000 + sf3,m & 30.9&7.2&0.014 & 0.20&[email protected]&2.9 , 11.&21&4400 + sf3,@xmath35&30.7&6.55&0.014 & 0.17&-0.34@xmath120&4.2 , 21.&16 & 4500 + [ fits ] | the source population of gamma - ray bursts ( grbs ) declines towards the present epoch being consistent with the measured decline of the star formation rate .
we show this using the brightness distribution of 3255 long batse grbs found in an off - line scan of the batse continuous 1.024 s count rate records .
the significance of this conclusion is enhanced by the detection of three grbs with known redshifts brighter than 10@xmath0 erg s@xmath1 during the last two years .
this is an argument in favor of the generally believed idea that grbs are strongly correlated with the star production , at least on cosmological time scales , and favors the association of long grbs with collapses of supermassive stars .
however , we still can not rule out neutron star mergers if the typical delay time for binary system evolution is relatively short .
if we assume a steep decline of the grb population at @xmath2 , then their luminosity function can be clearly outlined .
the luminosity function is close to a power law , @xmath3 , for low luminosities over at least 1.7 orders of magnitude
. then the luminosity function breaks to a steeper slope or to an exponential decline around @xmath4 erg s@xmath1 in the 50 - 300 kev range assuming isotropic emission . |
distributed control of multi - agent systems has been receiving a great deal of attention in the recent literature due to its broad applications in a number of areas ; e.g. , see @xcite-@xcite .
the leader - follower tracking problem represents a particular class of distributed control problems which is concerned with the design of control protocols for each agent based only on the information from the nearest neighbours , with the aim to guarantee that states of all followers converge to that of a dynamic leader ; e.g , see @xcite .
this paper considers the leader - following tracking problem for a class of discrete - time linear multi - agent systems with a high - dimensional leader and undirected communications between followers .
closely related work includes @xcite@xcite .
these references exemplify a common trend in the existing literature , which has a significant focus on consensus of discrete - time multi - agent systems with first or second - order integrator dynamics .
in contrast , consensus tracking for discrete - time multi - agent systems with identical linear higher - order node dynamics was studied in @xcite .
the work in @xcite covers the results on consensus tracking for multi - agent systems consisting of first- and second - order integrator dynamics as special cases , respectively .
also , in the majority of the existing papers on this topic , including some of the previously mentioned references , the leader and all the followers are assumed to have identical dynamics models .
this assumption allows one to directly analyze dynamics of the tracking error arising in the corresponding multi - agent networks consisting of closed - loop agent systems . on the contrary , the case where dynamics of the leader and those of the followers have different models ( e.g. ,
are described by state - space equations of different order ) has not received as much attention . in this paper
, we focus on the case where dynamics of the leader are more complex than those of the followers .
therefore , the existing theoretical approaches for analyzing leader - following tracking problems which have been developed for networks of identical agents can not be directly applied in this case .
furthermore , the state of the leader which evolves independently of the followers is not directly measurable by all of the followers .
it is assumed that only a partial information about the state of the leader can be sensed by a small group of followers which are subject to uncertainty .
thus , the multi - agent system under consideration is more general and contains some other commonly studied classes of leader - following multi - agent systems such as , e.g. , the multi - agent systems in @xcite , as special cases .
the control goal here is to design a tracking protocol for each agent such that the closed - loop system of agents achieves a desired level of @xmath0 leader - following tracking performance .
note that the proposed problem of distributed @xmath0 leader - following tracking for multi - agent systems with a high - dimensional leader is meaningful in a number of practical applications such as the design of distributed sensor networks @xcite where dynamics of the leader are more complex than those of the followers .
the fact that the followers can only sense partial information about the state of the leader prompts us to introduce a dynamic protocol where a local controller together with a neighbor - based state observer is designed for each follower . in the present framework ,
the state observer embedded in the followers performs the task of estimating the unmeasurable states of the leader in a distributed way . under the assumptions that the leader is detectable and the communication topology contains
a directed spanning tree , we propose a procedure for the design of a tracking protocol which involves a solution to a modified algebraic riccati equation .
the analysis of distributed @xmath0 leader - following tracking performance of this protocol when applied to a multi - agent system with a high - dimensional leader is then presented . the protocol design to achieve a pre - specified level of @xmath0 leader - following tracking performance for a system of agents whose dimension are different form that of the leader is the main contribution of this paper .
we note that dynamic protocols have been considered in a number of recent papers .
for example , a dynamic protocol for synchronization of multiagent systems has been proposed recently in @xcite .
similar to this paper , the analysis in @xcite is based on the reduction of the problem to the analysis of @xmath1 decoupled systems .
however , in contrast to our paper , robust performance issues are not considered in @xcite .
it is also worth noting that there is another possible way to solve tacking problems under partial information about the leader , by using the distributed output regulation theory and internal model principle @xcite .
the remainder of this paper is organized as follows . in section
ii , some preliminaries from the graph theory and the problem formulation are given .
in section iii , the main results are presented . a numerical example and simulations to illustrate our theoretical analysis are provided in section iv .
section v concludes the paper . [ [ notation ] ] notation + + + + + + + + let @xmath2 and @xmath3 be the sets of @xmath4 real matrices and complex matrices , respectively .
let @xmath5 and @xmath6 be , respectively , the sets of natural numbers and the @xmath7-dimensional real square summable functions . if not explicitly stated , all matrices are assumed to have compatible dimensions .
the superscripts @xmath8 and @xmath9 denote the transpose and the hermitian adjoint of a matrix , respectively .
a matrix @xmath10 is a unitary matrix if @xmath11 .
@xmath12 represents a diagonal matrix with @xmath13 , @xmath14 on its diagonal .
the notation @xmath15 denotes the vector whose elements are equal to @xmath16 .
let @xmath17 and @xmath18 be the @xmath4 zero and identity matrices , respectively .
a square matrix is said to be schur stable if the magnitude of all of its eigenvalues is less than @xmath16 .
the symbols @xmath19 and @xmath20 denote , respectively , the kronecker product and the euclidian norm .
let @xmath21 be a directed graph with a set of nodes @xmath22 , a set of directed edges @xmath23 , and a weighted adjacency matrix @xmath24_{n\times n}$ ] .
we will use the simplified notation @xmath25 for the graph when this causes no confusion . a directed edge @xmath26 is associated with an ordered pair of nodes @xmath27 , where @xmath28 and @xmath29 are called the parent and child nodes , respectively , and @xmath30 if and only if @xmath31 .
furthermore , self - loops are not allowed , i.e. , @xmath32 for all @xmath33 . a directed path from node @xmath29 to @xmath28 is an ordered sequence of edges , @xmath34 , with distinct nodes @xmath35 .
a directed tree is a directed graph such that ( a ) its every node @xmath36 @xmath37 , except for the root node @xmath38 , has exactly one parent , and ( b ) there exists a unique directed path from @xmath38 to each node @xmath36 @xmath37 .
a directed spanning tree of @xmath25 is a directed tree that has the same node set @xmath39 and whose edge set is a subset of @xmath40 .
@xmath25 will reduce to an undirected graph if and only @xmath41 , for all @xmath42 . a matrix @xmath43_{n\times n}$ ] is called a row - stochastic matrix associated with the graph @xmath25 , if the following properties hold : @xmath44 ; @xmath45 if @xmath46 , and @xmath47 otherwise ; and @xmath48 , for all @xmath33 .
[ lemmarowsto ] for any row - stochastic matrix @xmath49 associated with the graph @xmath25 , 1 is an eigenvalue of @xmath49 , and all other eigenvalues of @xmath49 lie in the open unit disk .
furthermore , 1 is a simple eigenvalue of @xmath49 if and only if the graph @xmath25 contains a directed spanning tree .
consider a group of @xmath1 agents indexed by @xmath50 . without loss of generality
, it is assumed that the agent labeled @xmath16 is the leader , whose dynamics are governed by the following equations @xmath51 the vector @xmath52 represents the state of the leader at time instant @xmath53 , @xmath54 and @xmath7 are two positive integers .
this vector is assumed to be partitioned as @xmath55 , where @xmath56 , @xmath14 .
accordingly , the state matrix @xmath57 $ ] is in @xmath58 , and admits a compatible partition of the following form : @xmath59 where @xmath60 , @xmath61 .
the vector @xmath62 represents the output information about the leader that is made available to the followers for sensing at time instant @xmath63 .
however , as will be seen later , it will be assumed that only a partial information about this vector is sensed by some of the followers .
therefore , without loss of generality it is assumed that @xmath64 , i.e. , @xmath65 , for @xmath66 .
the above model for the leader reflects the common situation where the leader of the group plays the role of a command generator providing reference states to be tracked by the followers .
therefore , it is natural to assume that the state of the leader evolves in accordance with its intrinsic nominal model and is not influenced by the followers .
the dynamics of the followers , labeled as @xmath67 , @xmath68 , are described by the state equations @xmath69 where @xmath70 and @xmath71 are constant matrices , @xmath72 is the control input to be designed , and @xmath73 is a disturbance input .
as mentioned , we assume that some of the followers are able to sense the output of the leader . in a general form
, the output @xmath74 of the leader sensed by the follower @xmath67 , @xmath75 , is expressed as @xmath76 where @xmath77 , and @xmath78 if and only if the leader is a neighbor of follower @xmath67 .
the matrix @xmath79 characterizes the protocol for information exchange in the network , and @xmath80 .
this reflects the situation where only a partial information about the leader is made available to selected followers .
clearly , if the agent @xmath67 is not connected to the leader , then @xmath81 .
the present framework has several features that distinguish it from similar problems considered in the literature .
firstly , while we assume that the state matrix @xmath82 is known to the followers , though the initial condition @xmath83 is unknown , the leader and the followers have significantly different models in that the states of the leader and the followers have different dimensions , and are also governed by different state matrices .
secondly , the followers employ different output matrices @xmath84 ; the scaling parameters @xmath85 may reflect differences in the strength of the leader s signal received by the followers positioned at a different distance from the leader .
finally , dynamics of the followers are subject to uncertain perturbations .
as is well known , within the @xmath86 framework , such perturbations are often associated with unmodeled uncertain dynamics , and can be used to account for imperfections in the followers models . in regard to communications between the followers , define the information output of follower @xmath67 to be @xmath87 where @xmath79 is the matrix from equation ( [ outputinformation ] ) .
we will assume that each follower @xmath67 can only use for control the information outputs of its neighbours relative its own output , @xmath88 . as mentioned previously
, the matrix @xmath89 characterizes the protocol for information exchange in the network .
the matrix @xmath89 is in general a rectangular matrix ; this allows for the vectors @xmath88 exchanged between the neighboring followers @xmath67 and @xmath90 to have a lower dimension than their states .
note that all agents are assumed to use the same matrix @xmath89 . as discussed in @xcite , in certain applications ,
using a common communication matrix by all followers is not a significant limitation . for the notational convenience ,
let is replaced with @xmath91 , and then using this modified matrix instead of @xmath92 in the subsequent derivations . ]
@xmath93 and introduce the relative information available to follower @xmath67 , @xmath75 , given by @xmath94,\ ! \nonumber \\
& = & \frac{1}{\kappa}\!\bigg[\sum_{j=2}^{n}\!a_{ij}e(x_{i}(k){-}x_{j}(k)){+}a_{i1}e\left(x_{i}(k){-}\theta_{n}(k)\right)\!\bigg],\ ! \quad\end{aligned}\ ] ] where @xmath95 , @xmath96 , and @xmath97 is a given positive constant .
the factor @xmath98 on the right hand side of ( [ relativeinformation ] ) is a weighting factor which scales the actual communication weights between follower @xmath67 and its neighbors into positive scalars within the interval @xmath99 , for each @xmath68 .
note that such scaling technique is commonly used in consensus problems for discrete - time multi - agent systems @xcite . before closing this section ,
we state the standing assumptions about the structure of the system communication topology .
the first assumption is concerned with interactions between the followers , while the second assumption describes the communication topology between the leader and the rest of the network .
[ assumptionundirected ] the adjacency matrix of graph @xmath25 , @xmath100_{n\times n}$ ] has the property that for all @xmath101 , @xmath41 .
note that assumption [ assumptionundirected ] indicates that the subgraph describing the communication topology between the followers is undirected .
however , this subgraph is not required to be connected in the present framework .
[ assumptionspanningtree ] the communication topology graph @xmath25 contains a directed spanning tree with the leader node being its root .
note that assumption [ assumptionspanningtree ] is not restrictive . for example , it holds when the subgraph describing the communication topology between the followers is connected , and also at least one follower senses the output of the leader .
more generally , when the communication topology between the followers consists of @xmath102 separate connected components , assumption [ assumptionspanningtree ] will be satisfied if each component of the graph includes a node which directly senses the output of the leader .
the control problem in this paper is to design a distributed protocol @xmath103 , @xmath68 , to enable the closed - loop multi - agent system ( [ followerdynamics ] ) , equipped with this protocol , to achieve a prescribed level of @xmath0 leader - following tracking performance .
the mathematical definition of the @xmath0 leader - following tracking performance index will be given later .
to guarantee the @xmath0 leader - following consensus tracking performance , the following observer - based dynamic distributed tracking protocol is proposed for each follower @xmath67 , @xmath68 .
the protocol consists of two parts : * the neighbor - based local controller : @xmath104 where @xmath105 , @xmath106 , @xmath61 , are defined in ( [ partition ] ) and @xmath107 is given in ( [ relativeinformation ] ) . *
distributed state estimator : @xmath108 the gain matrix @xmath109 is the design parameter of the protocol which will be defined later . combining equations ( [ followerdynamics ] ) , ( [ localcontroller ] ) and ( [ distributedestimator ] ) yields the closed - loop system describing dynamics of each follower governed by the proposed protocol : @xmath110 where @xmath111 is the state of the closed - loop system , @xmath68 , and @xmath112 @xmath113 .
then , it is easy to see that the difference between the state of the leader and the state of the @xmath67-th closed - loop system , @xmath114 , satisfies the following equation @xmath115 @xmath116 . to characterize performance of the proposed tracking protocol ,
define the performance variable @xmath117 where @xmath118 where @xmath119 is a given performance output matrix , @xmath68 . using the expression for @xmath120 given in ( [ relativeinformation ] )
, we have @xmath121\!\right\}\rho(k ) \\ & \quad \quad \quad \quad \;\;+ ( i_{n-1}\otimes \hat{b}_{\omega})\omega(k ) , \\ & e(k+1)=\left(i_{n-1}\otimes c\right)\rho(k+1 ) , \end{aligned } \right.\ ] ] where @xmath122 denotes the @xmath123 matrix defined as @xmath124_{(n-1)\times ( n-1)}$ ] with @xmath125 , @xmath126 , @xmath127 and @xmath128 , @xmath129 and @xmath130 are the constants defined in ( [ relativeinformation ] ) , @xmath131 where @xmath89 is the matrix from equation ( [ outputinformation ] ) .
denote by @xmath132 the transfer function matrix of the system ( [ errordynamics-2 ] ) from disturbance input @xmath133 to the performance output @xmath134 .
we are now in a position to formulate the leader - following @xmath86 tracking problem under consideration in this paper .
[ definition1 ] the multi - agent system consisting of the leader ( [ leaderdynamics ] ) and the followers ( [ followerdynamics ] ) and equipped with the protocol ( [ localcontroller ] ) is said to solve the distributed @xmath0 leader - following tracking problem with performance index @xmath135 , if the following two conditions hold : 1 .
the multi - agent system described by ( [ leaderdynamics ] ) and ( [ followerdynamics ] ) with @xmath136 , @xmath68 , achieves consensus in the sense of @xmath137 , where @xmath138 is the state of the closed loop system defined in ( [ closed - loop - state ] ) , @xmath68 .
2 . the @xmath0 norm of @xmath132 satisfies the following condition : @xmath139 .
in this section , the main theoretical results are presented .
since the leader has no neighbors , the matrix @xmath49 associated with communication topology @xmath25 , has the following structure @xmath140 and is a row - stochastic matrix ; @xmath129 is the constant defined in ( [ relativeinformation ] ) . by assumption
[ assumptionundirected ] , the block @xmath122 in ( [ row - stochasticd ] ) is symmetric .
let @xmath141 , @xmath142 , be the eigenvalues of @xmath122 .
it then follows from lemma [ lemmarowsto ] and assumption [ assumptionspanningtree ] that @xmath143 , for all @xmath142 .
[ maintheorem1 ] suppose the communication graph @xmath25 satisfies assumptions [ assumptionundirected ] and [ assumptionspanningtree ] .
then , for a given @xmath135 , the distributed @xmath0 leader - following tracking problem stated in definition [ definition1 ] admits a solution if and only if the following @xmath144 systems are simultaneously internally stable and have the @xmath0 norm less than @xmath145 : @xmath146 where @xmath142 .
[ remarkbelowtheorem1 ] theorem [ maintheorem1 ] shows that the distributed @xmath0 tracking problem for the networked agent system ( [ errordynamics-2 ] ) can been converted into a collection of @xmath0 control problems for a group of uncoupled systems ( [ maintheorem - eq1 ] ) , each having the same dimension .
thus , the complexity of the design reduces significantly . note also that the effect of topology on the distributed @xmath0 leader - following tracking performance is characterized by the eigenvalues of the @xmath122 .
although theorem [ maintheorem1 ] gives necessary and sufficient conditions for the distributed @xmath0 leader - following tracking problem to admit a solution , it does not explain how the feedback gain matrix @xmath147 should be selected in order to obtain such a solution .
the following theorem shows that this issue can be addressed using tools from the @xmath0 control theory , based on the result of theorem [ maintheorem1 ] .
[ maintheorem2 ] suppose that assumptions [ assumptionundirected ] and [ assumptionspanningtree ] hold , and let @xmath148 .
given a constant @xmath135 , suppose there exist real matrices @xmath149 , @xmath150 and a positive scalar @xmath151 such that @xmath152 and @xmath153 then the protocol ( [ localcontroller ] ) augmented with the distributed state estimator ( [ distributedestimator ] ) , with the feedback gain matrix @xmath147 , defined as @xmath154 , solves the @xmath0 leader - following tracking problem for the multi - agent system described by ( [ leaderdynamics ] ) and ( [ followerdynamics ] ) , with a disturbance attenuation level @xmath145 .
theorem [ maintheorem2 ] provides sufficient conditions for solvability of the distributed @xmath0 leader - following tracking problem for the multi - agent system described by ( [ leaderdynamics ] ) and ( [ followerdynamics ] ) .
it is not hard to see that a necessary condition for this tracking problem to have a solution is that the matrix pair @xmath155 must be detectable .
note that performance of the proposed @xmath86 tracking protocol is determined by the matrix @xmath119 in ( [ errordynamics-2 ] ) which defines the performance variable @xmath134 . in the special case where the performance variable of interest is the tracking error @xmath156 , the conditions of theorem [ maintheorem2 ] are simplified by letting @xmath157 .
consider a leader whose dynamics are described by equation ( [ leaderdynamics ] ) , with @xmath158 furthermore , let @xmath159 , @xmath160 , and @xmath161 .
then the corresponding matrices @xmath162 and @xmath163 are @xmath164 , @xmath165 .
clearly , the pair @xmath166 is detectable . to illustrate theorem [ maintheorem2 ] ,
let us consider a network of agents of the form ( [ followerdynamics ] ) connected over the communication graph @xmath25 shown in fig .
the adjacency matrix of this graph is @xmath167 . ]
it is easy to check that assumption [ assumptionspanningtree ] holds .
thus , the neighbor - based protocol consisting of the local controller of the form ( [ localcontroller ] ) and the distributed state estimator of the form ( [ distributedestimator ] ) can be designed by solving the conditions in theorem [ maintheorem2 ] . to design the protocol in this example ,
the parameter @xmath97 in ( [ relativeinformation ] ) is set to be equal @xmath168 .
calculations show that the eigenvalues of @xmath122 defined in ( [ row - stochasticd ] ) are @xmath169 , @xmath170 , @xmath171 , @xmath172 . then , @xmath173 .
also , the output matrix @xmath174 and performance level @xmath175 were chosen in this example . solving the linear matrix inequality ( [ maintheorem2firstlmi ] ) with @xmath176
gives that @xmath177 and @xmath178 . to illustrate properties of this protocol , we simulated the closed loop system without disturbances and also with disturbance inputs of the form @xmath179 , where @xmath180 , @xmath181 , and @xmath182 to illustrate asymptotic convergence of the tracking agents in the absence of disturbances , the corresponding state trajectories of the closed - loop multi - agent system ( [ followerdynamics ] ) with a high - dimensional leader ( [ leaderdynamics ] ) , are shown in figs .
[ figure2][figure4 ] .
let @xmath183 be the square of the norm of the consensus tracking error vector for the multi - agent system , where @xmath184 , @xmath185 .
[ figure5 ] indicates that the proposed distributed dynamic tracking protocol indeed ensures consensus tracking in the absence of disturbances , i.e. , when @xmath136 , @xmath180 .
next , the @xmath0 consensus tracking under disturbances is considered .
under zero initial conditions , the ` energy trajectories ' @xmath186 and @xmath187 were computed as functions of the evolution time @xmath188 and were plotted in fig .
[ figure9 ] .
it can be seen from fig .
[ figure9 ] that the proposed distributed dynamic tracking protocol indeed ensures the set level of disturbance attenuation .
the distributed @xmath0 leader - following tracking problem for a class of discrete - time multi - agent systems with a high - dimensional active leader has been investigated in this paper . in the presented framework ,
the outputs of the leader are only sensed by some informed followers .
a new kind of dynamic tracking protocol consisting of a local controller and a distributed state estimator has been constructed and employed to solve such a coordination problem . using tools from the @xmath0 control theory , it has been proved that distributed @xmath0 leader - following tracking can be ensured if the underlying topology graph contains a directed spanning tree with the leader being its root while the communication topology among the followers is undirected .
future work will focus on solving the distributed tracking problem for multi - agent systems with a high - dimensional leader and time - varying topologies as well as leader - following tracking for multi - agent systems with nonlinear dynamics .
g. wen would like to thank prof .
yiguang hong for the inspiring discussions and helpful suggestions .
l. ballard , y. cao , and w. ren , `` distributed discrete - time coupled harmonic oscillators with application to synchronized motion coordination , '' _ iet control theory & applications _ , vol .
5 , pp . 806816 , 2010 .
z. li , z. duan , and g. chen , `` consensus of discrete - time linear multi - agent systems with observer - type protocols , '' _ discrete and continuous dynamical systems - series b _ , vol .
2 , pp . 489505 , 2011 .
z , zhou , h. fang , and y. hong , `` distributed estimation for time - varying target in noisy environment , '' _ proceedings of the 10th world congress on intelligent control and automation _ , beijing , china , pp . 43414346 , 2012 .
c. de souza and l. xie , `` on the discrete - time bounded real lemma with application in the characterization of static state feedback @xmath0 controllers , '' _ systems & control letters _ , vol . 18 , no .
1 , pp . 61 - 71 , 1992 . | this paper considers the distributed @xmath0 leader - following tracking problem for a class of discrete - time multi - agent systems with a high - dimensional dynamic leader .
it is assumed that output information about the leader is only available to designated followers , and the dynamics of the followers are subject to perturbations .
to achieve distributed @xmath0 leader - following tracking , a new class of control protocols is proposed which is based on the feedback from the nearest neighbors as well as a distributed state estimator . under the assumptions that dynamics of the leader are detectable and the communication topology contains a directed spanning tree , sufficient conditions are obtained that enable all followers to track the leader while achieving a desired @xmath0 leader - following tracking performance .
numerical simulations illustrate the effectiveness of the theoretical analysis . |
public opinion expressed in results of elections and opinion polls have been studied widely using traditional statistics .
an alternative approach is information theory , which can be applied to probabilistic data .
electoral data can be easily transformed from percentages to probabilities .
thus the use of information theory to investigate public opinion about political parties and the conduct of governments and its opposition is obvious , but has not been carried out so far .
such an application , the only one to our knowledge , is an analytical approach to interpret the public s high job approval rating for president clinton @xcite .
this rating has been high and nearly constant between january 1998 and february 1999 , despite the well known unfavorable conditions for the us president in that period .
such a high rating could be explained partially , but is still considered unusual for several reasons @xcite . the political situation in greece in the recent three years is completely different than the united states of the years 1998 - 1999 .
however , an interesting ( parallel ) question arises : greek prime minister karamanlis enjoyed a high job approval ( 2004 - 2007 ) , although his party new democracy approval by the public was just @xmath0 higher than the opposition party pasok , headed by george papandreou .
we note that we used statistical data from a specific greek opinion polls company ( metron analysis ) in the period 2004 - 2007 , stopping just three months before the latest parliament elections in greece ( september 2007 ) .
it is of interest to try to clarify the above striking fact by extending the usual statistical treatment to shannon s information theory @xcite .
information theory was used for the first time in telecommunications in the late 40 s .
our aim is to investigate the possibility to extract some general , qualitative conclusions from typical opinion polls in greece , employing the tools of information and complexity theories . as we mentioned above
, our inspiration comes from a similar study in the united states .
although the political systems and the conditions in the usa and greece are very different , our work leads to the same mathematical model .
it is seen that information - theoretic methods can be used to extend the results of usual statistics , which illuminate certain statistical data of public opinion .
information theory can proceed further towards an interpretation , in some sense , of statistical processes .
the use of the logarithm in the definition of information entropy smooths small differences in statistical data from various companies and yields the same qualitative conclusions .
this illustrates the strength of information theory to give quantitative ( numerical ) answers to qualitative questions .
specifically information entropy @xmath1 , corresponding to a probability distribution @xmath2 of @xmath3 events occuring with probabilities @xmath4 , respectivelly , can be defined as @xmath5 @xmath1 is an information theoretic quantity which takes into account all the moments of a probability distribution and can be considered , in a sense , superior to traditional statistics employing the well - known quantities of average value and variance . @xmath1 in relation ( [ eq : eq1 ] ) is measured in bits ( if the base of logarithm is 2 ) , nats natural units of information ( if the base is e ) and hartleys ( if the base is 10 ) . in the present paper
the base is 10 , for the sake of comparison with @xcite .
however , one case can be transformed to the other one , by multiplying with just a constant . definition ( [ eq : eq1 ] ) represents the average information content of an event , which occurs with a specific probability distribution @xmath6 .
the use of the logarithm is justified because in such a way @xmath1 obeys certain mathematical and intuitive properties expected from a quantity related to information content of a probability function .
specifically , @xmath1 is positive and the joint information content of two simultaneous independent events translate to the addition of the corresponding information measures of each event e.t.c . for more properties and a pedagogical description see @xcite .
@xmath1 is maximum for an equiprobable or uniform probability distribution @xmath7 , i.e. @xmath8 .
@xmath1 is minimum when one of the @xmath4 s is 1 ( @xmath9 ) and all the other @xmath4 s are 0 , i.e. @xmath10 , under the convention that @xmath11 . in this case ,
one of the outcomes is certain , while all the other ones are impossible to occur .
@xmath1 represents a measure of information content of a probabilistic event , i.e. the average number of `` yes '' or `` no '' questions needed to specify the event ( in the case of bits ) .
@xmath1 is reciprocal to the degree of surprise of an event , i.e. the least probable event has the most information and vice versa .
we give a simple example in order to understand the meaning of relation ( [ eq : eq1 ] ) .
let us ask to a certain number of people the following question : _ is c. karamanlis suitable for the position of prime minister of greece ?
_ we receive answers with percentages and corresponding probabilities @xmath12 ( yes ) , @xmath13 ( no ) , @xmath14 ( something else ) . a direct application of ( [ eq : eq1 ] ) for the normalized probability distribution @xmath15 , where @xmath16 ( @xmath17 ) gives the information content in hartleys of that set of probabilities . in the case of a uniform ( equiprobable )
distribution i.e. @xmath18 , relation ( [ eq : eq1 ] ) gives @xmath19 .
this is the maximum information entropy with uniform probability distribution ( @xmath17 ) .
this can be interpreted as a distribution of complete ignorance ( unbiased ) in the sense that a specific answer does not contain more information than any other one .
a case of maximum entropy @xmath1 corresponds to a minimum amount of information @xmath20 about our question .
thus information @xmath20 , is reciprocal with @xmath1 @xmath21 the above convention agrees with our intuition , i.e. the information content of an event corresponding to a probability distribution can be quantified by the magnitude of our surprise after the event has occurred or how unpredictable is the outcome . the case of equiprobable distribution for @xmath22 , i.e. @xmath23 occurred in the recent general parliament elections in italy ( april 2006 ) .
there were two large coalition of parties and one of the coalitions won with a slight difference in votes , about 40,000 -while the number of votes was about 40,000,000 .
thus , with real results @xmath24 versus @xmath25 , we can consider with a very satisfactory approximation that @xmath26 .
the application of information theory in this case gives @xmath27 bit ( base of the logarithm equals 2 ) .
this fact is completely equivalent with throwing a fair coin ( equal probability for the two results heads - tails ) or with the question yes - no ( equiprobable ) which coalition will win .
that means that @xmath28 gives @xmath29 and the minimum information can be interpreted as a complete homogenization of the public opinion about the two coalitions . in other words ,
the results of elections in italy correspond to the random throw of a fair coin i.e. a complete lack of knowledge of the voters .
our observation does not intend to depreciate the process of elections , the culmination of democracy , but it is an extreme case with maximum possible information entropy @xmath30 .
there are other measures of information such as onicescu s information energy @xmath31 @xcite and fisher s information @xmath32 @xcite .
shannon s information is a global measure , while fisher s is a local one i.e. @xmath1 does not depend on the ordering of the probabilities @xmath6 , while @xmath32 does depend , due to the existence of the derivative of the distribution in its definition .
their definitions are given below together with appropriate comments .
it is stressed that all are based on the same probability distributions @xmath6 as @xmath1 .
landsberg s definition of disorder @xmath33 @xcite is @xmath34 and order @xmath35 disorder @xmath33 is a normalized disorder ( @xmath36 ) .
@xmath37 ( zero disorder , @xmath38 ) corresponds to complete order @xmath39 and @xmath40 ( complete disorder , @xmath28 ) corresponds to zero order @xmath41 .
@xmath42 enable us to study the organization of data , described probabilistically .
the next important step is the _ statistical complexity _ @xmath43 defined by shiner - davison - landsberg ( sdl ) @xcite , where @xmath44 is the strength of disorder and @xmath45 is the strength of order . in the present work we consider the simple case @xmath46 and @xmath47 .
another measure of _ complexity _ is @xmath48 according to lopez ruiz - mancini - calbet ( lmc ) @xcite . here
@xmath49 is the so - called _ disequilibrium _ ( or distance from equilibrium ) defined as @xmath50 sdl complexity @xmath51 describes correctly the two extreme cases of complete order and complete disorder , where we expect intuitively zero complexity or organization of the data .
an example taken from the physical world is illuminating .
a perfect crystal ( complete order ) has @xmath52 and the same holds for a gas ( complete disorder ) where @xmath52 as well .
thus ( perfect ) crystals and gases are not interesting , lacking complexity or organization .
this is given by @xmath51 and agrees with intuition .
instead , for the information entropy we have @xmath38 for crystals and @xmath28 for gases , which is not satisfactory . thus extending from physics , @xmath33 , @xmath53 , @xmath51 and @xmath54
enable us to study quantitatively the ( organized ) complexity of probabilistic data of opinion polls and elections .
an other very important information measure is fisher information @xmath32 @xcite .
recently , there is a revival of interest for fisher information , culminating in two books @xcite and @xcite , defined as @xmath55 for a continuous probability distribution @xmath56 , which is modified accordingly in the present work for discrete probability distributions .
specifically , for a discrete probability distribution @xmath6 employed in the present work , relation ( [ eq : eq8 ] ) becomes @xmath57 thus the treatment of high job approval of clinton in @xcite , will be repeated for the case of the greek prime minister constantinos karamanlis and the greek political scene in the recent three years ( 2004 - 2007 ) and extended in the present paper using new quantities e.g. @xmath33 , @xmath53 , @xmath51 , @xmath54 and @xmath32 as functions of time .
we used statistical data for the public opinion coming from the greek opinion polls company _ _ metron analysis__. specifically , we focused our interest on the following three questions , presented in table [ tab : tab1 ] . | a general methodology to study public opinion inspired from information and complexity theories is outlined .
it is based on probabilistic data extracted from opinion polls .
it gives a quantitative information - theoretic explanation of high job approval of greek prime minister mr .
constantinos karamanlis ( 2004 - 2007 ) , while the same time series of polls conducted by the company metron analysis showed that his party new democracy ( abbr .
nd ) was slightly higher than the opposition party of pasok -party leader mr .
george papandreou .
it is seen that the same mathematical model applies to the case of the popularity of president clinton between january 1998 and february 1999 , according to a previous study , although the present work extends the investigation to concepts as complexity and fisher information , quantifying the organization of public opinion data . |
the genetic architecture of biological organisms shows remarkable robustness against both structural and environmental perturbations @xcite .
for example , quantitative models indicate that the functionality of the _ drosophila _ segment polarity gene network is extremely insensitive to variations in initial conditions @xcite and robust against architectural modifications @xcite .
( also see @xcite . ) gene knock - out studies on yeast have shown that almost 40% of the genes on chromosome v have either negligible or no effects on the growth rate @xcite .
certain cellular networks , such as the _ e. coli _ chemotaxis network , are also known to be very robust to variations in biochemical parameters @xcite .
the genetic regulatory networks that control the developmental dynamics buffer perturbations and maintain a stable phenotype .
that is why phenotypic variation within most species is quite small , despite the organisms being exposed to a wide range of environmental and genetic perturbations @xcite .
it has been proposed that genetic robustness evolved through stabilizing selection for a phenotypic optimum @xcite .
@xcite showed that this in fact can be true by modeling a developmental process within an evolutionary scenario , in which the genetic interaction sequence represents the development , and the stationary configuration of the gene network represents the phenotype .
his results indicate that the genetic robustness of a population of model genetic regulatory networks can gradually increase through stabilizing selection , in which deviations from the `` optimal '' stationary state ( phenotype ) are considered deleterious @xcite . in this paper , we focus on the effects of evolution of genetic robustness on the dynamics of gene regulatory networks in general .
first , we examine the relationship between genetic robustness and the dynamical character of the networks . by dynamical character ,
we mean the stability of the expression states of the network against small perturbations , or noise .
models that are employed to study dynamics of gene networks , such as random boolean networks ( rbn ) @xcite or some variants of random threshold networks ( rtn ) @xcite have been known to undergo a phase transition at low connectivities @xcite , giving rise to change in dynamical behavior : on average , small perturbations will percolate through the network above the threshold connectivity ( chaotic phase ) , whereas they stay confined to a part of the network below the threshold ( ordered phase ) . intuitively , one might expect to find that robustness to mutations ( which are permanent structural changes , not dynamic perturbations ) is related to the dynamical behavior of the system , therefore , ordered gene regulatory networks should be genetically more robust than the chaotic ones . however , this is not necessarily true . here , we show that the relation between the dynamical character of a genetic regulatory network and its mutational robustness can be quite the opposite .
in fact , even earlier studies provide a clue on this issue : for gene networks that have undergone selection ( i.e. , evolved ) , mutational robustness is known to increase with increasing connectivity @xcite . on the other hand
, chaoticity has been shown to increase with increasing connectivity in random gene regulatory networks @xcite .
these facts seem to contradict the intuitive interpretation of robustness since they suggest that chaotic networks can be mutationally robust . although this inference is proven true in this paper , such an assessment can not be based on the previous studies of the dynamics of random networks @xcite , as the evolved networks ( with high mutational robustness and connectivity ) mentioned above @xcite have undergone selection .
the selection process could potentially tune a network to exhibit a different dynamical character than its random ancestors .
therefore , the dynamical character of the evolved networks should be studied independently and then compared with their random counterparts . here , we study the dynamics of the evolved networks numerically and show that selection for an optimal phenotype indeed has only a minor effect on their global dynamical behavior .
this result indicates that the evolution of mutational robustness can not be understood in terms of simple dynamical measures .
we also provide statistics on robustness to _ noise _ , which is the ability of a network to reach its `` optimal '' steady state after a perturbation to the gene - expression _
trajectory_. computer simulations indicate that mutational robustness is correlated to robustness of the gene - expression trajectory to small perturbations ( noise ) , at least for short trajectories .
this result is supported by recent studies @xcite . for perturbations of arbitrary magnitude
, the basin size of the steady - state attractor provides a better measure .
our analysis shows that basin sizes of densely connected networks ( which are highly chaotic ) have a broad distribution , and therefore such networks can have very large attractor basins @xcite .
the yeast cell - cycle network has been reported to have similar properties @xcite .
although chaoticity is just a side - effect of high connectivity , and not directly selected for during evolution , it appears that this intrinsic property of chaotic networks can be useful in terms of robustness to noise .
when all of these measures are taken into account , chaotic dynamics does not seem be an obstacle for the networks with high connectivity since they are more robust to both mutations and noise than the ones that are sparsely connected .
the organization of the rest of this paper is as follows .
we describe the model in sec .
[ sec : model ] and explain its implementation in simulations in sec .
[ sec : methods ] .
we give the results in sec .
[ sec : results ] and discuss their implications in sec [ sec : discussion ] .
we use the model introduced by @xcite , which has also been used with some modifications by other researchers @xcite . each individual
is represented by a regulatory gene network consisting of @xmath0 genes .
the expression level of each gene , @xmath1 can be either @xmath2 or @xmath3 , meaning that the gene is expressed or not , respectively .
the expression states change in time according to regulatory interactions between the genes .
the time development of the expression states ( i.e. , the dynamical trajectory taken by the network ) represents a developmental pathway .
this deterministic , discrete - time dynamics of the development is given by a set of nonlinear difference equations , @xmath4 where sgn is the sign function and @xmath5 is the strength of the influence of gene @xmath6 on gene @xmath7 .
nonzero elements @xmath5 of the @xmath8 matrix @xmath9 are independent random numbers drawn from a gaussian distribution with zero mean and unit variance
. the diagonal elements of * @xmath9 * are allowed to be nonzero , corresponding to self - regulation .
the mean number of nonzero elements in @xmath9 is controlled by the connectivity density , @xmath10 , which is the probability that any given @xmath5 is nonzero .
thus , the mean degree of the network is @xmath11 to clarify , @xmath12 represents the _ developmental _ time , through which genetic interactions occur .
it is different from the _ evolutionary _ time , @xmath13 , which will be explained in the next section .
the dynamics given by eq . can display a wide variety of features . for a specified initial state @xmath14 ,
the network reaches an attractor ( either a fixed point or a limit cycle ) after a transient period .
transient time , number of attractors , attractor periods , etc . can differ depending on the connectivity of the network , and from one realization of @xmath9 to another .
the fitness of an individual is defined by whether it can reach a developmental equilibrium , i.e. , a fixed point , which is a fixed gene - expression pattern , @xmath15 , in a `` reasonable '' transient time .
( it has been shown that selection for developmental stability is sufficient for evolution of mutational robustness @xcite , i.e. , deviations from @xmath15 do not have to be deleterious , as long as the gene - expression configuration reaches a fixed point .
however , we shall adopt the criterion used by @xcite for compatibility . )
further details of the model are explained in the next section .
we studied populations of random networks with @xmath16 .
we use these relatively small networks to be able to enumarate all network states exhaustively for a large set of realizations . in the simulations ,
each network was first assigned a random interaction matrix @xmath9 and an initial state @xmath14 .
@xmath9 was generated as follows . for each @xmath5 ,
a random number uniformly distributed on @xmath17 was generated , and @xmath5 was set to zero if the random number was greater than the connectivity density , @xmath10 .
otherwise , @xmath5 was assigned a random number drawn from a gaussian distribution with zero mean and unit variance .
then , each `` gene '' of the initial configuration , @xmath18 , was assigned either @xmath3 or @xmath2 at random , each with probability 1/2 .
after @xmath9 and @xmath14 were created , the developmental dynamics were started , and the network s stability was evaluated . if the system reached a fixed point , @xmath19 in @xmath20 time steps , then it was considered _ viable _ and kept .
otherwise it was considered unstable , both @xmath9 and @xmath14 were discarded , and the process was repeated until a viable network was generated .
( a viable network can have several fixed points and/or limit cycles in addition to @xmath21 . ) for each viable network , its fixed point , @xmath15 , was regarded as the `` optimal '' gene - expression state of the system .
this is the only modification we made to the model used by @xcite : we accept any @xmath21 as long as it can be reached within @xmath20 time steps from @xmath14 , whereas @xcite generated networks with preassigned random @xmath14 and @xmath15 .
, between @xmath14 and @xmath21 to assess mutational robustness of these networks @xcite .
we ignore this fact since producing networks with a predetermined @xmath22 is computationally not very feasible .
the probability distribution of @xmath22 for the networks we generate follows approximately a gaussian with a mean between 2.8 and 4.5 , increasing with @xmath23 .
see supplementary fig . s2 . ] in order to generate a collection of more robust networks , a mutation - selection process was simulated for each viable network as follows .
first , a clan of @xmath24 identical copies of each network was generated .
for each member of the clan , a four - step process was performed for @xmath25 generations : 1 .
recombination : each pair of the @xmath0 rows of consecutive matrices in the clan were swapped with probability 1/2 . since
the networks were already shuffled in step 4 ( see below ) , there was no need to pick random pairs .
2 . mutation : each nonzero @xmath5 was replaced with probability @xmath26 by a new random number drawn from the same standard gaussian distribution .
thus , on average , one matrix element was changed per matrix per generation .
fitness evaluation : each network was run starting from the original initial condition @xmath14 . if the network reached a fixed point , @xmath27 within @xmath28 developmental time steps , then its fitness was calculated using @xmath29 where @xmath30 denotes the normalized hamming distance between @xmath31 and @xmath32 , @xmath33 denotes the inverse of the strength of selection , @xmath21 is the optimal gene - expression state , which is the final gene - expression state of the original network that `` founded '' the clan .
we used @xmath34 .
if the network could not reach a fixed point , it was assigned the minimum nonzero fitness value , @xmath35 4 .
selection / asexual reproduction : the fitness of each network was normalized to the fitness value of the most fit network in the clan .
then a network was chosen at random and duplicated into the descendant clan with probability equal to its normalized fitness .
this process was repeated until the size of the descendant clan reached @xmath36 .
then the old clan was discarded , and the descendant clan was kept as the next generation .
this process allows multiple copies ( offspring ) of the same network to appear in the descendant clan , while some networks may not be propagated to the next generation due to genetic drift . at the end of the @xmath25 generation ( evolutionary time ) selection ,
any unstable networks were removed from the evolved clan .
some steps of the process above may be unnecessary , and in fact , the results do not depend strongly on model details @xcite .
nevertheless , the entire procedure of @xcite was retained for compatibility . the mutational robustness , @xmath37 of a network was assessed as follows .
first , a nonzero @xmath5 was picked at random and replaced by a new random number with the same standard gaussian distribution .
then , the developmental dynamics were started , and it was checked whether the system reached the same stationary state , @xmath21 , within @xmath28 time steps .
this process was repeated @xmath38 times , starting from the original matrix .
the robustness of the original network before evolution was defined as the fraction of singly - mutated networks that reached @xmath21 . for the evolved networks
, we picked one sample network at random from the clan and used the same procedure to assess its mutational robustness . ) of 10000 sample networks before ( filled bars ) and after ( empty bars ) evolution with @xmath16 and @xmath39 .
before : @xmath40 @xmath41 .
after : @xmath42 @xmath43 .
the evolved distribution was calculated by sampling one network from each of 10000 evolved clans .
the mean indegree , @xmath44 , rather than the connectivity density , @xmath10 , is the parameter that controls the behavior of the system @xcite .
[ sec : methods ] for other model parameters.,scaledwidth=50.0% ]
the stabilizing selection described above increases the robustness of a model population of gene networks against mutations @xcite .
figure [ fig : symmetry - and - robustness ] shows that a population with a very large initial variation in robustness evolved an increased ability to absorb mutations after @xmath25 generations of stabilizing selection . however , it is not very clear what kind of a reorganization in the state spaces of these networks occur during the evolution .
we measured the changes in several system parameters to answer this question .
it has been analytically shown that the rtns undergo a phase transition from order to chaos with increasing @xmath23 @xcite .
obviously , the state space of an rtn is finite .
therefore , the expression states have to display periodicity in development after at most @xmath45 steps .
thus , chaos here is not a long - term aperiodic behavior ; rather it corresponds to a dynamical regime in which small perturbations percolate through the gene network @xcite .
we quantify the dynamical character of a network by comparing the time development of two configurations , @xmath46 and @xmath47 , that differ by one gene : we measure the mean number of different genes in time step @xmath48 , @xmath49 , averaged over all possible @xmath46 and @xmath47 pairs with @xmath50 .
this is known as damage spreading or damage propagation .
this incompatibility vanishes for larger @xmath23 as this detail in the update rule does not have any significance when almost all nodes have inputs .
nevertheless , we use the term `` damage spreading '' instead of `` damage - spreading rate '' when referring to @xmath51 to avoid a confusion . ]
. the dynamical behavior of random networks can be quantified analytically using damage - spreading analysis @xcite . however
, this analytical approach may not be applied to the _ evolved _ networks as the selection process can tune a network to behave dynamically very differently than its ancestor .
therefore , we calculated @xmath49 numerically for both viable and evolved networks , averaging ensembles of 10000 networks exhaustively over all possible configuration pairs , @xmath46 and @xmath47 , for each @xmath23 .
as seen in fig .
[ fig : matrixdistr - and - damage ] , the @xmath49 increases monotonically with increasing connectivity , indicating that highly connected networks are more chaotic on average .
the evolved networks are slightly more ordered ( on average ) than their viable ancestors .
these results indicate that the dynamical behavior of the evolved networks is not much different than that of the viable networks from which they are descended . , after a one - bit perturbation before ( viable ) and after evolution ( evolved ) for networks with @xmath16 , measured exhaustively for all possible state pairs , averaged over 10000 samples each .
the error bars represent one standard deviation ( not standard error ) .
evolved networks ( triangles ) are slightly more ordered compared to their viable ancestors .
( the differences are statistically significant . )
@xmath49 is always larger than unity due to our specific update rule .
see the discussion in the text for details .
the lines connecting the symbols are guides to the eye .
( probability distributions for @xmath49 for certain values of @xmath23 are given in supplementary figs .
s3 and s4 . )
[ fig : matrixdistr - and - damage ] ] to @xmath48 .
arrows indicate the direction of flow .
each state drains into an attractor ( pentagons ) through transient states .
( a ) state space of a sample network with @xmath52 and @xmath53 .
there are eight basins , each having a fixed - point attractor . the principal basin ( at the top )
contains the initial ( diamond shaped node ) and final states .
note that the basins are quite symmetric , and transients are very short as the system behaves less chaotically when the connectivity of the network is low .
( b ) basin of a sample network with @xmath52 and @xmath54 .
the principal basin occupies a large portion of the state space .
the basin on the left crosses the symmetry plane with a 2-cycle ( note the symmetry of the branches ) .
the other basins have mirror images on the other side of the state space ( not shown ) .
the principal basin is significantly larger than the others .
the high connectivity makes the network behave more chaotic , creating basins with broadly distributed sizes and branch lengths , and significantly longer transients .
see text for details
. generated using graphviz @xcite .
[ fig : state - space - layout],title="fig : " ] to @xmath48 .
arrows indicate the direction of flow .
each state drains into an attractor ( pentagons ) through transient states .
( a ) state space of a sample network with @xmath52 and @xmath53 .
there are eight basins , each having a fixed - point attractor . the principal basin ( at the top )
contains the initial ( diamond shaped node ) and final states .
note that the basins are quite symmetric , and transients are very short as the system behaves less chaotically when the connectivity of the network is low .
( b ) basin of a sample network with @xmath52 and @xmath54 .
the principal basin occupies a large portion of the state space .
the basin on the left crosses the symmetry plane with a 2-cycle ( note the symmetry of the branches ) .
the other basins have mirror images on the other side of the state space ( not shown ) .
the principal basin is significantly larger than the others .
the high connectivity makes the network behave more chaotic , creating basins with broadly distributed sizes and branch lengths , and significantly longer transients .
see text for details .
generated using graphviz @xcite .
[ fig : state - space - layout],title="fig : " ] we also analyzed the state spaces of random , viable , and evolved networks @xcite .
figures [ fig : stats1 ] and [ fig : stats2 ] show these statistics for networks with @xmath16 and connectivities @xmath551 , 5 , and 10 . due to the up - down symmetry of the system ,
the state space is divided into two parts , where the dynamics on one side is the mirror image of the other .
therefore , the basin size of a fixed - point attractor can not exceed half the size of the state space , @xmath56 .
limit cycles , however , can cross the symmetry plane that divides the state space ( fig . [
fig : state - space - layout](b ) ) .
therefore , their basins can contain up to @xmath57 states .
typically , sparsely connected gene networks have many attractors ( fixed points and limit cycles ) and , consequently , smaller basin sizes on average , as depicted in fig .
[ fig : state - space - layout](a ) .
increasing connectivity brings state spaces with fewer attractors ( figs .
[ fig : stats1](a ) , ( b ) , and ( c ) ) , longer attractor ( limit cycle ) periods ( figs . [ fig : stats1](d ) , ( e ) , and ( f ) ) , and broadly distributed basin sizes ( figs , [ fig : stats1](g ) , ( h ) , and ( i ) ) . densely connected networks can contain basins that occupy a large portion , occasionally even all , of the state space . the basins of such networks tend to have broadly distributed branch lengths as seen in fig .
[ fig : state - space - layout](b ) , and their states also tend to have fewer precursors ( figures [ fig : stats2](a ) , ( b ) , and ( c ) ) .
( all states , @xmath58 , that go to state @xmath31 at time step @xmath12 + 1 are precursors of @xmath31 . ) therefore , mean transient time ( total number of steps from a state to the attractor ) on such networks are typically larger than on networks with lower connectivity ( figs .
[ fig : stats2](d ) , ( e ) , and ( f ) ) .
these changes in the state - space characteristics are consistent with the damage - spreading measurements .
the distributions for the evolved networks are shifted toward the distributions of more ordered networks of lower connectivity .
this means evolved networks are slightly more ordered , as the damage - spreading measurements shown in fig .
[ fig : matrixdistr - and - damage](b ) suggest .
for example , the distribution of the number of attractors ( figs . [ fig : stats1](a ) , ( b ) and ( c ) ) for the networks with low connectivity ( @xmath59 has a longer tail compared to the networks with @xmath60 and 10 . similarly , the distributions for the evolved networks have slightly longer tails compared to those of viables of the same connectivity .
the same effect can be seen in the attractor - period ( figs .
[ fig : stats1](d ) , ( e ) , and ( f ) ) , precursor ( figs . [ fig : stats2](a ) , ( b ) , and ( c ) ) , and transient - time distributions ( figs . [
fig : stats2](d ) , ( e ) , and ( f ) ) , as well .
the most significant changes are seen in the transient - time distributions , indicating that basins with relatively shorter branches are preferred by selection .
the basin - size distributions of viable and evolved networks , however , do not display much difference as they virtually overlap .
and @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
each curve represents an average over 20,000 realizations for the `` random '' networks , and 10,000 realizations for the `` viable '' and `` evolved '' ones . for the evolved networks , we did not use clan averages to avoid a bias
instead , we picked one sample network from each evolved clan .
error bars were calculated by grouping the data : each data set was divided into groups of 1000 samples and the average distribution for each group was calculated .
then , the error calculations were performed on the new set of averaged distributions .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) to reduce the noise .
( see supplementary fig .
s1 for histograms with linear bins . )
the basin size does not include the size ( period ) of the attractor . the evolved attractor - count and attractor - period distributions
are shifted toward those of viable networks with lower @xmath23 , indicating that evolved networks display a slightly more ordered character .
the basin - size distributions for the evolved networks seem to overlap with the ones for the viable networks except for the largest basins , which become less probable after selection .
the lines connecting the symbols are guides to the eye.[fig : stats1],title="fig : " ] and @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
each curve represents an average over 20,000 realizations for the `` random '' networks , and 10,000 realizations for the `` viable '' and `` evolved '' ones .
for the evolved networks , we did not use clan averages to avoid a bias .
instead , we picked one sample network from each evolved clan .
error bars were calculated by grouping the data : each data set was divided into groups of 1000 samples and the average distribution for each group was calculated . then , the error calculations were performed on the new set of averaged distributions .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) to reduce the noise .
( see supplementary fig .
s1 for histograms with linear bins . )
the basin size does not include the size ( period ) of the attractor . the evolved attractor - count and attractor - period distributions
are shifted toward those of viable networks with lower @xmath23 , indicating that evolved networks display a slightly more ordered character .
the basin - size distributions for the evolved networks seem to overlap with the ones for the viable networks except for the largest basins , which become less probable after selection .
the lines connecting the symbols are guides to the eye.[fig : stats1],title="fig : " ] and @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
each curve represents an average over 20,000 realizations for the `` random '' networks , and 10,000 realizations for the `` viable '' and `` evolved '' ones . for the evolved networks , we did not use clan averages to avoid a bias
instead , we picked one sample network from each evolved clan .
error bars were calculated by grouping the data : each data set was divided into groups of 1000 samples and the average distribution for each group was calculated .
then , the error calculations were performed on the new set of averaged distributions .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) to reduce the noise .
( see supplementary fig .
s1 for histograms with linear bins . )
the basin size does not include the size ( period ) of the attractor . the evolved attractor - count and attractor - period distributions
are shifted toward those of viable networks with lower @xmath23 , indicating that evolved networks display a slightly more ordered character .
the basin - size distributions for the evolved networks seem to overlap with the ones for the viable networks except for the largest basins , which become less probable after selection .
the lines connecting the symbols are guides to the eye.[fig : stats1],title="fig : " ] and @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
each curve represents an average over 20,000 realizations for the `` random '' networks , and 10,000 realizations for the `` viable '' and `` evolved '' ones . for the evolved networks , we did not use clan averages to avoid a bias
instead , we picked one sample network from each evolved clan .
error bars were calculated by grouping the data : each data set was divided into groups of 1000 samples and the average distribution for each group was calculated .
then , the error calculations were performed on the new set of averaged distributions .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) to reduce the noise .
( see supplementary fig .
s1 for histograms with linear bins . )
the basin size does not include the size ( period ) of the attractor . the evolved attractor - count and attractor - period distributions
are shifted toward those of viable networks with lower @xmath23 , indicating that evolved networks display a slightly more ordered character .
the basin - size distributions for the evolved networks seem to overlap with the ones for the viable networks except for the largest basins , which become less probable after selection .
the lines connecting the symbols are guides to the eye.[fig : stats1],title="fig : " ] and @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
each curve represents an average over 20,000 realizations for the `` random '' networks , and 10,000 realizations for the `` viable '' and `` evolved '' ones .
for the evolved networks , we did not use clan averages to avoid a bias .
instead , we picked one sample network from each evolved clan .
error bars were calculated by grouping the data : each data set was divided into groups of 1000 samples and the average distribution for each group was calculated . then , the error calculations were performed on the new set of averaged distributions .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) to reduce the noise .
( see supplementary fig .
s1 for histograms with linear bins . )
the basin size does not include the size ( period ) of the attractor . the evolved attractor - count and attractor - period distributions
are shifted toward those of viable networks with lower @xmath23 , indicating that evolved networks display a slightly more ordered character .
the basin - size distributions for the evolved networks seem to overlap with the ones for the viable networks except for the largest basins , which become less probable after selection . the lines connecting the symbols are guides to the eye.[fig : stats1],title="fig : " ] and @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
each curve represents an average over 20,000 realizations for the `` random '' networks , and 10,000 realizations for the `` viable '' and `` evolved '' ones . for the evolved networks , we did not use clan averages to avoid a bias
instead , we picked one sample network from each evolved clan .
error bars were calculated by grouping the data : each data set was divided into groups of 1000 samples and the average distribution for each group was calculated .
then , the error calculations were performed on the new set of averaged distributions .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) to reduce the noise .
( see supplementary fig .
s1 for histograms with linear bins . )
the basin size does not include the size ( period ) of the attractor . the evolved attractor - count and attractor - period distributions
are shifted toward those of viable networks with lower @xmath23 , indicating that evolved networks display a slightly more ordered character .
the basin - size distributions for the evolved networks seem to overlap with the ones for the viable networks except for the largest basins , which become less probable after selection .
the lines connecting the symbols are guides to the eye.[fig : stats1],title="fig : " ] and @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
each curve represents an average over 20,000 realizations for the `` random '' networks , and 10,000 realizations for the `` viable '' and `` evolved '' ones . for the evolved networks , we did not use clan averages to avoid a bias
instead , we picked one sample network from each evolved clan .
error bars were calculated by grouping the data : each data set was divided into groups of 1000 samples and the average distribution for each group was calculated .
then , the error calculations were performed on the new set of averaged distributions .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) to reduce the noise .
( see supplementary fig .
s1 for histograms with linear bins . )
the basin size does not include the size ( period ) of the attractor . the evolved attractor - count and attractor - period distributions
are shifted toward those of viable networks with lower @xmath23 , indicating that evolved networks display a slightly more ordered character .
the basin - size distributions for the evolved networks seem to overlap with the ones for the viable networks except for the largest basins , which become less probable after selection .
the lines connecting the symbols are guides to the eye.[fig : stats1],title="fig : " ] and @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
each curve represents an average over 20,000 realizations for the `` random '' networks , and 10,000 realizations for the `` viable '' and `` evolved '' ones . for the evolved networks
, we did not use clan averages to avoid a bias .
instead , we picked one sample network from each evolved clan .
error bars were calculated by grouping the data : each data set was divided into groups of 1000 samples and the average distribution for each group was calculated .
then , the error calculations were performed on the new set of averaged distributions .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) to reduce the noise .
( see supplementary fig .
s1 for histograms with linear bins . )
the basin size does not include the size ( period ) of the attractor . the evolved attractor - count and attractor - period distributions
are shifted toward those of viable networks with lower @xmath23 , indicating that evolved networks display a slightly more ordered character .
the basin - size distributions for the evolved networks seem to overlap with the ones for the viable networks except for the largest basins , which become less probable after selection .
the lines connecting the symbols are guides to the eye.[fig : stats1],title="fig : " ] and @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
each curve represents an average over 20,000 realizations for the `` random '' networks , and 10,000 realizations for the `` viable '' and `` evolved '' ones . for the evolved networks , we did not use clan averages to avoid a bias
instead , we picked one sample network from each evolved clan .
error bars were calculated by grouping the data : each data set was divided into groups of 1000 samples and the average distribution for each group was calculated .
then , the error calculations were performed on the new set of averaged distributions .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) to reduce the noise .
( see supplementary fig .
s1 for histograms with linear bins . )
the basin size does not include the size ( period ) of the attractor . the evolved attractor - count and attractor - period distributions
are shifted toward those of viable networks with lower @xmath23 , indicating that evolved networks display a slightly more ordered character .
the basin - size distributions for the evolved networks seem to overlap with the ones for the viable networks except for the largest basins , which become less probable after selection .
the lines connecting the symbols are guides to the eye.[fig : stats1],title="fig : " ] @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
the second row shows the distributions for the lengths of the transients starting from all possible states . the third row , however , shows the distribution for @xmath62 , the length of the transient starting from @xmath14 , the original initial state .
again , changes in the distributions indicate that evolved networks display a slightly more ordered character .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) and the other plots were histogrammed using linear ( 0,1,2 ... ) bins .
see the caption of fig .
[ fig : stats1 ] for the details of the data analysis .
the lines connecting the symbols are guides to the eye.[fig : stats2],title="fig : " ] @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
the second row shows the distributions for the lengths of the transients starting from all possible states .
the third row , however , shows the distribution for @xmath62 , the length of the transient starting from @xmath14 , the original initial state .
again , changes in the distributions indicate that evolved networks display a slightly more ordered character .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) and the other plots were histogrammed using linear ( 0,1,2 ... ) bins .
see the caption of fig .
[ fig : stats1 ] for the details of the data analysis .
the lines connecting the symbols are guides to the eye.[fig : stats2],title="fig : " ] @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) . the second row shows the distributions for the lengths of the transients starting from all possible states .
the third row , however , shows the distribution for @xmath62 , the length of the transient starting from @xmath14 , the original initial state .
again , changes in the distributions indicate that evolved networks display a slightly more ordered character .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) and the other plots were histogrammed using linear ( 0,1,2 ... ) bins .
see the caption of fig .
[ fig : stats1 ] for the details of the data analysis .
the lines connecting the symbols are guides to the eye.[fig : stats2],title="fig : " ] @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
the second row shows the distributions for the lengths of the transients starting from all possible states .
the third row , however , shows the distribution for @xmath62 , the length of the transient starting from @xmath14 , the original initial state .
again , changes in the distributions indicate that evolved networks display a slightly more ordered character .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) and the other plots were histogrammed using linear ( 0,1,2 ... ) bins .
see the caption of fig .
[ fig : stats1 ] for the details of the data analysis .
the lines connecting the symbols are guides to the eye.[fig : stats2],title="fig : " ] @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
the second row shows the distributions for the lengths of the transients starting from all possible states .
the third row , however , shows the distribution for @xmath62 , the length of the transient starting from @xmath14 , the original initial state .
again , changes in the distributions indicate that evolved networks display a slightly more ordered character .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) and the other plots were histogrammed using linear ( 0,1,2 ... ) bins .
see the caption of fig .
[ fig : stats1 ] for the details of the data analysis .
the lines connecting the symbols are guides to the eye.[fig : stats2],title="fig : " ] @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
the second row shows the distributions for the lengths of the transients starting from all possible states .
the third row , however , shows the distribution for @xmath62 , the length of the transient starting from @xmath14 , the original initial state .
again , changes in the distributions indicate that evolved networks display a slightly more ordered character .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) and the other plots were histogrammed using linear ( 0,1,2 ... ) bins .
see the caption of fig .
[ fig : stats1 ] for the details of the data analysis .
the lines connecting the symbols are guides to the eye.[fig : stats2],title="fig : " ] @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
the second row shows the distributions for the lengths of the transients starting from all possible states .
the third row , however , shows the distribution for @xmath62 , the length of the transient starting from @xmath14 , the original initial state .
again , changes in the distributions indicate that evolved networks display a slightly more ordered character .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) and the other plots were histogrammed using linear ( 0,1,2 ... ) bins .
see the caption of fig .
[ fig : stats1 ] for the details of the data analysis .
the lines connecting the symbols are guides to the eye.[fig : stats2],title="fig : " ] @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
the second row shows the distributions for the lengths of the transients starting from all possible states .
the third row , however , shows the distribution for @xmath62 , the length of the transient starting from @xmath14 , the original initial state .
again , changes in the distributions indicate that evolved networks display a slightly more ordered character .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) and the other plots were histogrammed using linear ( 0,1,2 ... ) bins .
see the caption of fig .
[ fig : stats1 ] for the details of the data analysis .
the lines connecting the symbols are guides to the eye.[fig : stats2],title="fig : " ] @xmath61 5 , and 10 ( columns 1 , 2 , and 3 , respectively ) .
the second row shows the distributions for the lengths of the transients starting from all possible states .
the third row , however , shows the distribution for @xmath62 , the length of the transient starting from @xmath14 , the original initial state .
again , changes in the distributions indicate that evolved networks display a slightly more ordered character .
the data plotted on log - log scale were histogrammed using exponential bins ( 0 , 1 , 2 - 3 , 4 - 7 , ... ) and the other plots were histogrammed using linear ( 0,1,2 ... ) bins .
see the caption of fig .
[ fig : stats1 ] for the details of the data analysis .
the lines connecting the symbols are guides to the eye.[fig : stats2],title="fig : " ] as seen in fig .
[ fig : before - and - after - plots](a ) , networks of all connectivities have similar values of mutational robustness , @xmath63 , before selection .
after selection , however , the networks with higher connectivity reach a much greater mean mutational robustness @xcite .
sparsely connected networks do not show much improvement because the low connectivity leaves little room for optimization .
another significant change is seen in the transient time , @xmath62 ( figs .
[ fig : stats2](g ) , ( h ) , and ( i ) and [ fig : before - and - after - plots](b ) ) @xcite . for viable networks ,
the transient time increases with the connectivity .
this is not surprising since state spaces of densely connected networks typically have basins with longer branches ( fig .
[ fig : state - space - layout](b ) ) .
however , selection brings the mean transient time down to around 2 , independent of the connectivity of the network .
low transient time is one of the properties of highly robust networks @xcite .
mutations are not the only kind of perturbations that genetic regulatory networks experience .
all genetic systems are also exposed to noise created by both internal an external sources .
we measure robustness to noise dynamical robustness using two parameters .
the first one is the dynamical robustness , @xmath64 , of a gene expression state , @xmath65 , to small perturbations ( random one - bit flips ) , which is simply the fraction of nearest neighbors in hamming distance of @xmath65 that lie in the same basin as @xmath65 @xcite . in other words , @xmath64 is the probability that a random flip of a gene on @xmath65 yields a state that drains into the same attractor as @xmath65 .
we also employ the mean robustness of a set of states , for instance , robustness of the gene expression trajectory , @xmath66 , which is the average of @xmath64 over all states in the trajectory ( including @xmath14 and @xmath21 ) . similarly , the robustness of the principal basin , @xmath67 , is the average of @xmath64 over all states in the basin , and the robustness of the entire state space , @xmath68 , is the average of @xmath64 over all possible states of the network .
( we usually omit the term `` dynamical '' when we talk about robustness of a state or a set of states since the mutational robustness of a _ state _ is not defined . )
the second parameter we use to estimate dynamical robustness is the normalized size of the principal basin , @xmath69 , which is the basin containing both @xmath14 and @xmath21 .
the size of a basin is the total number of states it contains .
therefore , @xmath69 is the fraction of all possible perturbations to the trajectory that leave @xmath21 unchanged , i.e. , the probability that flipping an arbitrary number of genes at random in a state on the trajectory does not change the attractor that the network settles into
. clearly , the parameters we employ to measure robustness to noise are quite different than conventional measures of stability , such as the damage - spreading rate .
this is because we are interested in the endpoints of the gene expression trajectory , but not in the exact path taken .
therefore , the relationship between the damage spreading and robustness to noise is not trivial .
in fact , it is quite counterintuitive as explained below . as shown in figs .
[ fig : stats1](g ) , ( h ) and ( i ) , densely connected networks have a greater number of larger basins as their basin - size distributions essentially follow a power law for about two decades .
the effect of the connectivity on @xmath69 is similar , as shown in fig .
[ fig : before - and - after - plots](c ) . networks with low connectivity have small principal basins both before and after selection .
more densely connected networks have larger @xmath69 before selection .
we see a small increase in networks with @xmath60 , while fully connected networks ( @xmath70 increase their @xmath71 significantly after evolution .
however , compared to the change in mutational robustness , the effect of selection on the size of the principal basin is modest for highly connected networks .
although greater connectivity increases the damage spreading ( sensitivity to small perturbations ) , the dynamical robustness of the trajectory , @xmath72 , as well , increases with the connectivity ( fig .
[ fig : before - and - after - plots](d ) ) . networks with @xmath73 have a small @xmath74 both before and after selection . the @xmath72 curve for the viables
has a maximum around @xmath75 , which also shows no improvement after selection .
this indicates that the networks with @xmath76 are intrinsically robust to noise , but not evolvable .
viable networks with higher connectivity have monotonically decreasing values of @xmath74 , but the selection improves their average robustness significantly , above or up to the same level as the networks with @xmath75 .
the robustnesses of the initial state , @xmath77 , and the final state , @xmath78 ( figs .
[ fig : before - and - after - plots](e ) and ( f ) ) follow the same trend .
the robustnesses of the principal basin , @xmath79 , and state space , @xmath80 , ( figs .
[ fig : before - and - after - plots](g ) and ( h ) ) do not show as much improvement with selection .
this implies that the effect of selection is more local ( on the specific gene - expression trajectory ) than global in terms of the change in dynamics .
our results above agree with the findings by @xcite , who used very similar parameters to measure robustness to noise .
we note that most of the robustness measures discussed above do not increase monotonically with @xmath81 for the viable networks ( i.e. , before selection ) .
for example , @xmath82 for viable networks has a maximum around @xmath83 ( fig .
[ fig : before - and - after - plots](a ) ) .
this probably indicates that these quantities depend on more than one parameter ( like the principal basin size and the magnitude of the damage spreading ) , some increasing and some decreasing with increasing @xmath81 . for the evolved networks , however , @xmath82 , @xmath71 , and @xmath74 increase monotonically with @xmath84 at least for the range tested in this paper .
this indicates that networks of higher connectivity are more evolvable when mutational robustness is considered .
however , these networks are more chaotic on average compared to the ones of low connectivity , regardless of selection .
mean mutational robustness ( @xmath85 transient time @xmath86 , normalized size of the principal basin @xmath87 , robustnesses of the trajectory @xmath88 , initial state @xmath89 , final state @xmath90 , principal basin @xmath91 , and state space @xmath92 before ( viable ) and after evolution ( evolved ) for @xmath16 and @xmath73 , 2.5 , 5 , 7.5 and 10 averaged over 10,000 networks .
the error bars are equal to one standard deviation ( not standard error ) .
@xmath69 is normalized by @xmath56 .
the lines connecting the symbols are guides to the eye.,title="fig : " ] mean mutational robustness ( @xmath85 transient time @xmath86 , normalized size of the principal basin @xmath87 , robustnesses of the trajectory @xmath88 , initial state @xmath89 , final state @xmath90 , principal basin @xmath91 , and state space @xmath92 before ( viable ) and after evolution ( evolved ) for @xmath16 and @xmath73 , 2.5 , 5 , 7.5 and 10 averaged over 10,000 networks .
the error bars are equal to one standard deviation ( not standard error ) .
@xmath69 is normalized by @xmath56 .
the lines connecting the symbols are guides to the eye.,title="fig : " ] mean mutational robustness ( @xmath85 transient time @xmath86 , normalized size of the principal basin @xmath87 , robustnesses of the trajectory @xmath88 , initial state @xmath89 , final state @xmath90 , principal basin @xmath91 , and state space @xmath92 before ( viable ) and after evolution ( evolved ) for @xmath16 and @xmath73 , 2.5 , 5 , 7.5 and 10 averaged over 10,000 networks .
the error bars are equal to one standard deviation ( not standard error ) .
@xmath69 is normalized by @xmath56 .
the lines connecting the symbols are guides to the eye.,title="fig : " ] mean mutational robustness ( @xmath85 transient time @xmath86 , normalized size of the principal basin @xmath87 , robustnesses of the trajectory @xmath88 , initial state @xmath89 , final state @xmath90 , principal basin @xmath91 , and state space @xmath92 before ( viable ) and after evolution ( evolved ) for @xmath16 and @xmath73 , 2.5 , 5 , 7.5 and 10 averaged over 10,000 networks .
the error bars are equal to one standard deviation ( not standard error ) .
@xmath69 is normalized by @xmath56 .
the lines connecting the symbols are guides to the eye.,title="fig : " ] mean mutational robustness ( @xmath85 transient time @xmath86 , normalized size of the principal basin @xmath87 , robustnesses of the trajectory @xmath88 , initial state @xmath89 , final state @xmath90 , principal basin @xmath91 , and state space @xmath92 before ( viable ) and after evolution ( evolved ) for @xmath16 and @xmath73 , 2.5 , 5 , 7.5 and 10 averaged over 10,000 networks .
the error bars are equal to one standard deviation ( not standard error ) .
@xmath69 is normalized by @xmath56 .
the lines connecting the symbols are guides to the eye.,title="fig : " ] mean mutational robustness ( @xmath85 transient time @xmath86 , normalized size of the principal basin @xmath87 , robustnesses of the trajectory @xmath88 , initial state @xmath89 , final state @xmath90 , principal basin @xmath91 , and state space @xmath92 before ( viable ) and after evolution ( evolved ) for @xmath16 and @xmath73 , 2.5 , 5 , 7.5 and 10 averaged over 10,000 networks .
the error bars are equal to one standard deviation ( not standard error ) .
@xmath69 is normalized by @xmath56 .
the lines connecting the symbols are guides to the eye.,title="fig : " ] mean mutational robustness ( @xmath85 transient time @xmath86 , normalized size of the principal basin @xmath87 , robustnesses of the trajectory @xmath88 , initial state @xmath89 , final state @xmath90 , principal basin @xmath91 , and state space @xmath92 before ( viable ) and after evolution ( evolved ) for @xmath16 and @xmath73 , 2.5 , 5 , 7.5 and 10 averaged over 10,000 networks .
the error bars are equal to one standard deviation ( not standard error ) .
@xmath69 is normalized by @xmath56 .
the lines connecting the symbols are guides to the eye.,title="fig : " ] mean mutational robustness ( @xmath85 transient time @xmath86 , normalized size of the principal basin @xmath87 , robustnesses of the trajectory @xmath88 , initial state @xmath89 , final state @xmath90 , principal basin @xmath91 , and state space @xmath92 before ( viable ) and after evolution ( evolved ) for @xmath16 and @xmath73 , 2.5 , 5 , 7.5 and 10 averaged over 10,000 networks .
the error bars are equal to one standard deviation ( not standard error ) .
@xmath69 is normalized by @xmath56 .
the lines connecting the symbols are guides to the eye.,title="fig : " ] and @xmath73 , 2.5 , 5 , 7.5 , and 10 .
the @xmath93-value is much smaller than 0.01 for all correlation coefficients with an absolute value above 0.03 .
@xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 denote changes in normalized size of the transient time , size of the principal basin , robustnesses of the trajectory , initial state , final state , principal basin , and state space , and damage spreading , respectively .
the dynamical robustness of the trajectory includes the contribution from the final state .
the lines connecting the symbols are guides to the eye .
see text for details.[fig : correlations],title="fig : " ] and @xmath73 , 2.5 , 5 , 7.5 , and 10 .
the @xmath93-value is much smaller than 0.01 for all correlation coefficients with an absolute value above 0.03 .
@xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 denote changes in normalized size of the transient time , size of the principal basin , robustnesses of the trajectory , initial state , final state , principal basin , and state space , and damage spreading , respectively .
the dynamical robustness of the trajectory includes the contribution from the final state .
the lines connecting the symbols are guides to the eye .
see text for details.[fig : correlations],title="fig : " ] and @xmath73 , 2.5 , 5 , 7.5 , and 10 .
the @xmath93-value is much smaller than 0.01 for all correlation coefficients with an absolute value above 0.03 .
@xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 denote changes in normalized size of the transient time , size of the principal basin , robustnesses of the trajectory , initial state , final state , principal basin , and state space , and damage spreading , respectively .
the dynamical robustness of the trajectory includes the contribution from the final state .
the lines connecting the symbols are guides to the eye .
see text for details.[fig : correlations],title="fig : " ] and @xmath73 , 2.5 , 5 , 7.5 , and 10 .
the @xmath93-value is much smaller than 0.01 for all correlation coefficients with an absolute value above 0.03 .
@xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 denote changes in normalized size of the transient time , size of the principal basin , robustnesses of the trajectory , initial state , final state , principal basin , and state space , and damage spreading , respectively .
the dynamical robustness of the trajectory includes the contribution from the final state .
the lines connecting the symbols are guides to the eye .
see text for details.[fig : correlations],title="fig : " ] and @xmath73 , 2.5 , 5 , 7.5 , and 10 .
the @xmath93-value is much smaller than 0.01 for all correlation coefficients with an absolute value above 0.03 .
@xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 denote changes in normalized size of the transient time , size of the principal basin , robustnesses of the trajectory , initial state , final state , principal basin , and state space , and damage spreading , respectively .
the dynamical robustness of the trajectory includes the contribution from the final state .
the lines connecting the symbols are guides to the eye .
see text for details.[fig : correlations],title="fig : " ] and @xmath73 , 2.5 , 5 , 7.5 , and 10 .
the @xmath93-value is much smaller than 0.01 for all correlation coefficients with an absolute value above 0.03 .
@xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 denote changes in normalized size of the transient time , size of the principal basin , robustnesses of the trajectory , initial state , final state , principal basin , and state space , and damage spreading , respectively .
the dynamical robustness of the trajectory includes the contribution from the final state .
the lines connecting the symbols are guides to the eye .
see text for details.[fig : correlations],title="fig : " ] and @xmath73 , 2.5 , 5 , 7.5 , and 10 .
the @xmath93-value is much smaller than 0.01 for all correlation coefficients with an absolute value above 0.03 .
@xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 denote changes in normalized size of the transient time , size of the principal basin , robustnesses of the trajectory , initial state , final state , principal basin , and state space , and damage spreading , respectively .
the dynamical robustness of the trajectory includes the contribution from the final state .
the lines connecting the symbols are guides to the eye .
see text for details.[fig : correlations],title="fig : " ] and @xmath73 , 2.5 , 5 , 7.5 , and 10 .
the @xmath93-value is much smaller than 0.01 for all correlation coefficients with an absolute value above 0.03 .
@xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 denote changes in normalized size of the transient time , size of the principal basin , robustnesses of the trajectory , initial state , final state , principal basin , and state space , and damage spreading , respectively .
the dynamical robustness of the trajectory includes the contribution from the final state .
the lines connecting the symbols are guides to the eye .
see text for details.[fig : correlations],title="fig : " ] the recent theoretical and computational work on gene regulatory networks @xcite indicates that there is a strong link between mutational robustness and robustness to noise . in order to see the relation between these two quantities , we calculated the correlation coefficients between the change in mutational robustness ( @xmath102 ) and changes in different measures of robustness to noise , as well as transient time and the damage spreading as functions of linkage density .
changes in the transient time and mutational robustness are highly correlated in agreement with the earlier studies @xcite .
the principal basin size does not seem to have much effect on @xmath103 , as the correlation between their changes is quite weak ( fig .
[ fig : correlations](b ) ) .
although a larger principal basin size means greater robustness to noise , this result suggests that it is not strongly selected for .
changes in robustnesses of the initial state , final state , principal basin , and state space are also correlated with @xmath102 ( figs .
[ fig : correlations](c ) , ( d ) , ( e ) , and ( f ) ) . the weak correlation between the changes in the damage spreading and mutational robustness ( fig . [ fig : correlations](h ) ) imply that it is not the change in dynamical behavior that brings mutational robustness .
in this paper , we have analyzed changes in state - space properties of model genetic regulatory networks under selection for an optimal phenotype . both numerical stability analysis and
the state - space statistics indicate that the difference between the global dynamical properties of mutationally robust networks that have undergone selection and their random ancestors are quite small .
furthermore , the correlation between the changes in the damage spreading and the mutational robustness is weak .
therefore , changes in the global dynamical properties do not seem to be responsible for the increase in mutational robustness after selection .
dynamics of many random threshold networks , as well as random boolean networks , depend largely on their connectivity distributions .
variants of the random threshold networks used in this paper have been shown to have a chaotic phase above @xmath104 , depending on the model details @xcite . essentially , rtns have a chaotic phase for sufficiently large @xmath23 @xcite just as rbns @xcite .
the connectivity of a network affects the evolvability of its mutational robustness , as well as its dynamical character . for viable ( essentially random ) networks ,
the mutational robustness is very similar for all connectivities @xcite . for the evolved networks ,
however , it increases monotonically with increasing connectivity , creating drastic differences for large @xmath23 .
the dynamical robustness of the gene - expression trajectory , the initial state , and the final state follow similar trends .
these results clearly indicate that the stability of the system as measured by the damage spreading does not capture its dynamical characteristics in this context .
as pointed out in a recent paper by @xcite , selection decreases the transient time by picking the `` proper '' interaction constants to construct a shorter ( or direct ) path from @xmath14 to @xmath21 .
both @xcite and @xcite stated that mutationally robust networks are the ones that have found a path for the gene expression trajectory at a safe distance from the basin boundary , so that small perturbations can not kick them into a different basin .
thus , even if the selection operates only on the stability of the stationary gene - expression pattern , robustness to mutations intrinsically requires stability of the gene - expression trajectory against small perturbations .
there is also some experimental evidence supporting the association between genetic and non - genetic change @xcite .
although it might seem like robustness to noise evolves as a by - product of robustness to mutations , the converse case is also true @xcite .
thus , robustness to noise and robustness to mutations seem to evolve mutually when certain conditions are met @xcite .
the analysis of the cell - cycle regulatory network of budding yeast provides further evidence for the chaotic behavior of gene networks @xcite .
the simplified form of this network has 11 nodes ( genes or proteins ) , one checkpoint ( an external input , in this case , the cell size ) , and 34 links including self - degrading interactions . using a dynamical model similar to the one used in this paper , @xcite _ _ _ _
showed that the stationary state of the network has a basin occupying 86% of the state space .
@xcite studied an ensemble of networks that can perform the same function , i.e. , the 12-step sequence of transitions in the expression trajectory , and found that these functional networks have larger basins for the stationary state ( consequently , broadly distributed basin sizes ) , fewer attractors , longer transient times , and a larger damage - spreading rate compared to their randomized counterparts .
they concluded that those dynamical features emerge due to the functional constraints on the network .
here , we showed that those features , which are signs of chaotic dynamics , can arise under the presence of structural perturbations ( mutations ) if the connectivity of the network is large enough , even when the constraints on the function are minimal , i.e. , when the only selection is on the phenotype .
although the length of the gene - expression trajectory of the yeast cell - cycle network needs further explanation in terms of mutational robustness , it appears like chaotic dynamics may be a design principle underlying seemingly `` boring '' and ordered behavior generally seen in models of gene regulatory networks , where a simple cascade of expression terminates at a stationary state @xcite .
the effect of network topology on evolvability of robustness is another aspect of the problem , which we do not discuss in this paper @xcite .
however , we would like to point out that recent studies indicate that certain topological features , such as connectivity , are not very crucial in determining the response of cellular networks to genetic or non - genetic change @xcite , and there may be other factors shaping their topological structure @xcite . our results also imply that the `` life at the edge of chaos '' hypothesis @xcite , which suggests adaptability ( evolvability ) is maximized in the critical regime does not seem to be necessary , at least not to explain evolvability of robustness to mutations and noise . indeed
, recent studies concerning dynamics of genetic regulatory networks do not indicate any special feature brought by criticality @xcite .
the `` edge of chaos '' concept was primarily developed to describe the phase in which cellular automata can perform universal computation @xcite , and it may not be related to dynamics and evolution of biochemical pathways as was once thought .
to summarize , our study indicates that conventional measures of stability may not be very informative about robustness to mutations or noise in gene regulatory networks when one considers the steady gene - expression pattern as the robust feature of the network .
also , the dynamics underlying the simple gene - expression trajectories can be very rich , reflecting a complex state - space structure .
the online version of this article contains supplementary material : linearly binned histograms of the data represented in figs .
[ fig : stats1 ] and [ fig : stats2 ] , probability densities for the hamming distance between the initial and final states , the magnitude of the damage spreading before and after evolution , and probability densities for the latter .
we thank t. f. hansen , s. bornholdt , g. brown , d. balcan , b. uzunolu , p. oikonomou , and a. pagnani for helpful discussions .
this research was supported by u.s .
national science foundation grant nos .
dmr-0240078 and dmr-0444051 , and by florida state university through the school of computational science , the center for materials research and technology , and the national high magnetic field laboratory .
albert , r. , othmer , h. g. , 2003 .
the topology of the regulatory interactions predicts the expression pattern of the segment polarity genes in
_ drosophila melanogaster_. j. theor .
223 ( 1 ) , 1 18 .
aldana , m. , coppersmith , s. , kadanoff , l. p. , 2003 .
boolean dynamics with random couplings . in : kaplan ,
e. , marsden , j. e. , sreenivasan , k. r. ( eds . ) , perspectives and problems in nonlinear science . a celebratory volume in honor of lawrence sirovich .
springer , berlin heidelberg new york , pp .
2389 .
azevedo , r. b. r. , lohaus , r. , srinivasan , s. , dang , k. k. , burch , c. l. , 2006 .
sexual reproduction selects for robustness and negative epistasis in artificial gene networks .
nature ( london ) 440 ( 7080 ) , 87 90 .
balcan , d. , kabakiolu , a. , mungan , m. , erzan , a. , 2007 .
the information coded in the yeast response elements accounts for most of the topological properties of its transcriptional regulation network .
plos one 2 ( 6 ) , e501 . de visser , j. a. g. m. , hermisson , j. , wagner , g. p. , meyers , l. a. , bagheri - chaichian , h. , blanchard , j. l. , chao , l. , cheverud , j. m. , elena , s. f. , fontana , w. , gibson , g. , hansen , t. f. , krakauer , d. , lewontin , r. c. , ofria , c. , rice , s. h. , von dassow , g. , wagner , a. , whitlock , m. c. , 2003 . perspective : evolution and detection of genetic robustness .
evolution 57 ( 9 ) , 1959 1972 .
ellson , j. , gansner , e. r. , koutsofios , e. , north , s. c. , woodhull , g. , 2003 .
graphviz and dynagraph static and dynamic graph drawing tools . in : junger , m. , mutzel , p. ( eds . ) , graph drawing software .
springer - verlag , berlin heidelberg new york , pp .
127148 .
von dassow , g. , odell , g. m. , 2002 .
design and constraints of the _ drosophila _ segment polarity module : robust spatial patterning emerges from intertwined cell state switches . j. exp .
294 ( 3 ) , 179 215 . | robustness to mutations and noise has been shown to evolve through stabilizing selection for optimal phenotypes in model gene regulatory networks .
the ability to evolve robust mutants is known to depend on the network architecture .
how do the dynamical properties and state - space structures of networks with high and low robustness differ ?
does selection operate on the global dynamical behavior of the networks ?
what kind of state - space structures are favored by selection ?
we provide damage propagation analysis and an extensive statistical analysis of state spaces of these model networks to show that the change in their dynamical properties due to stabilizing selection for optimal phenotypes is minor .
most notably , the networks that are most robust to both mutations and noise are highly chaotic . certain properties of chaotic networks , such as being able to produce large attractor basins , can be useful for maintaining a stable gene - expression pattern .
our findings indicate that conventional measures of stability , such as the damage - propagation rate , do not provide much information about robustness to mutations or noise in model gene regulatory networks . |
the cosmological abundance of relativistic and `` dark '' particles as active or sterile neutrinos or any other light particle ( as axions , for instance ) is usually parameterized in terms of the number of relativistic degrees of freedom @xmath0 , being @xmath1 the standard value for three ( active ) neutrino species .
several recent cosmological data analyses have constrained the value of @xmath0 with increasing accuracy ( see , for instance , refs .
@xcite ) providing a clear evidence for a dark relativistic background at high statistical significance .
the latest measurements of cosmic microwave background ( cmb ) anisotropies at arc - minute angular scales from the south pole telescope ( spt ) @xcite and the atacama cosmology telescope ( act ) @xcite , when combined with other cosmological data sets , yield the constraint @xmath2 at @xmath3 confidence level ( cl ) @xcite ( see also @xcite and @xcite ) , showing evidence for dark radiation ( i.e. @xmath4 ) at more than seven standard deviations , suggesting values higher than those expected in the standard scenario .
this hint , that clearly must be tested by future cmb measurements as those expected , for example , by the planck satellite @xcite , is interesting since an extra relativistic component in the standard three active neutrino model could be explained by a sterile neutrino .
models with one additional @xmath5 ev massive sterile neutrino , i.e. the so - called ( 3 + 1 ) models , were introduced to explain lsnd short baseline antineutrino data @xcite by means of neutrino oscillations . up to date cosmological constraints on the number of massive sterile neutrino species
have been presented in refs .
however , a larger value for @xmath0 could also arise from different physics , related to axions @xcite , decay of non - relativistic matter @xcite , gravity waves @xcite , extra dimensions @xcite , dark energy @xcite or asymmetric dark matter models @xcite .
the constraints on @xmath0 are also affected by possible interactions of the dark radiation component with the dark matter component , see refs .
@xcite .
information on the dark relativistic background , however , can be obtained not only from its effects on the expansion rate of the universe but also from its clustering properties .
for example , if dark radiation is made of massless neutrinos it should behave as relativistic particles also from the point of view of perturbation theory . following the definition introduced by hu @xcite
, this means that the dark radiation component should have an effective sound speed @xmath6 and a viscosity parameter @xmath7 such that @xmath8 .
these perturbation parameters can be constrained through measurements of the cmb anisotropies since dark radiation is coupled trough gravity with all the remaining components @xcite .
this clearly opens a new window for testing the dark radiation component , since , for example , a smaller value for @xmath7 could indicate possible non standard interactions ( see e.g. @xcite ) .
a value of @xmath7 different from zero , as expected in the standard scenario , has been detected in @xcite and confirmed in subsequent papers @xcite .
more recently , a value of the effective sound speed @xmath6 smaller than the standard value of @xmath9 has been claimed in ref .
@xcite . from this discussion is clear that current cosmological data analyses are sensitive to dark radiation properties and that the latest constraints on these parameters are showing interesting deviations from their expected standard values .
it is therefore timely to investigate the impact of non - standard dark radiation properties in the determination of cosmological parameters related to different sectors as dark energy or inflation . in this paper
we derive bounds from the most recent cosmological data on the dark radiation parameters @xmath0 , @xmath6 , and @xmath7 relaxing the usual assumption of a @xmath10cdm cosmology , analysing the correlations among the clustering parameters @xmath6 and @xmath7 and other key cosmological parameters , as the dark energy equation of state or the scalar spectral index .
we also study these correlations in light of future planck and core cmb data
. we shall generate mock cmb data for a cosmology with dark radiation perturbation parameters different from their standard values and then we fit these simulated data via the usual mcmc analysis to an extended non minimal cosmology with standard dark radiation parameters , but varying both the constant and the time varying dark energy equation of state , or the scalar spectral index and its running .
the paper is organized as follows .
section [ sec : i ] describes the details of the analysis carried out here , including the cosmological parameters and data sets used in the analyses . in section
[ sec : ii ] the different cosmological scenarios considered here are analyzed and the most important degeneracies among the parameters are carefully explored .
section [ sec : iii ] summarizes our main conclusions .
we perform our analyses considering three different scenarios : we first analyze the wilkinson microwave anisotropy probe ( wmap ) data together with the luminous red galaxy clustering results from the sloan digital sky survey ii ( sdssii ) @xcite and with a prior on the hubble constant @xmath11 from the hubble space telescope ( hst ) @xcite , referring to it as the `` baseline model '' .
then we add to these data sets spt data @xcite and we will refer to this case as the `` baselinespt model '' .
we will explore as well the impact of supernovae ia ( snia ) luminosity distance measurements from the union 2 compilation @xcite in constraining the dark radiation parameters , and we will refer to this case in the following as the `` baselinespt - snia model '' . we have modified the boltzmann ` camb ` code @xcite incorporating the two extra dark radiation perturbation parameters @xmath6 and @xmath7 and we have extracted the cosmological parameters from current data using a monte carlo markov chain ( mcmc ) analysis based on the publicly available mcmc package ` cosmomc ` @xcite .
we sample the following six - dimensional standard parameters : the baryon and cold dark matter densities ( @xmath12 and @xmath13 ) , the ratio between the sound horizon and the angular diameter distance at decoupling @xmath14 , the optical depth @xmath15 , the scalar spectral index @xmath16 , and the amplitude of the primordial spectrum @xmath17 .
we consider purely adiabatic initial conditions and we impose spatial flatness . we also include the effective number of relativistic degrees of freedom @xmath0 , the effective sound speed @xmath6 and the viscosity parameter @xmath7 .
table [ tab : priors ] shows the flat priors considered on the different cosmological parameters .
finally , we generate a mock data set for the ongoing planck cmb experiment , with @xmath7 different from its standard value and with @xmath18 and @xmath19 .
then we fit these mock data using a mcmc analysis to different extensions of the minimal cosmological model in which the dark radiation is standard .
the three possible extensions we consider are : _ ( a ) _ a @xmath10cdm model with a running spectral index @xmath20 , _ ( b ) _ the @xmath21cdm model in which we include the possibility of a dark energy equation of state parameter @xmath21 different from @xmath22 , and _
( c ) _ the @xmath23cdm model in which we assume an equation of state evolving with redshift .
the reconstructed values of the dark energy equation of state and of the running spectral index will be , in general , different from the values used in the mocks and , in the case of the dark energy equation of state @xmath21 , different from the value expected within the @xmath10cdm model .
we shall also explore the impact of future cmb data from the core mission @xcite , performing an equivalent forecast to the one we present here for planck .
.flat priors for the cosmological parameters considered here .
[ cols="^,^",options="header " , ]
we consider a cosmological model described by the following set of parameters : @xmath24,\ : { n_{\rm { eff } } } , \ : { c_{\rm vis}^2 } , \ : { c_{\rm eff}^2}\},\ ] ] notice from the results in the first column of tab .
[ tab : constraints1 ] that in the `` baseline model '' , ( i.e. the one with wmap7+sdssii+@xmath11 data ) , the preferred value for the effective number of relativistic degrees of freedom is @xmath25 , considerably higher than the standard model prediction @xmath26 .
the addition of spt data in the `` baselinespt model '' decreases the preferred value of @xmath0 , making it closer to its canonical value .
finally , the addition of snia data in the `` baselinespt - snia model '' brings the value of @xmath0 even closer to @xmath27 , however , it is still @xmath28 away from the former value .
these results seem to agree with the excess of radiation claimed in the literature in previous analyses @xcite . regarding the dark radiation perturbation parameters @xmath7 and @xmath6 ,
their preferred values are close to their standard values of @xmath29 , being @xmath6 much better constrained from current data than @xmath7 .
there exists a degeneracy between @xmath6 and @xmath16 and between @xmath7 and @xmath16 , see the top and middle panels of fig .
[ fig : ns_ceff_cvis ] .
these degeneracies are related to the fact that if @xmath16 increases ( decreases ) , the power at low multipoles decreases ( increases ) , while the power at high @xmath30 increases ( decreases ) to a lesser extent .
this effect could be compensated by an increase ( decrease ) in the viscosity parameter @xmath7 , that leads to a decrease ( increase ) of the power at all scales . concerning the degeneracy between @xmath6 and @xmath16 , it is mainly related to the degeneracy between @xmath6 and @xmath7 ( see the lower panel of fig . [
fig : ns_ceff_cvis ] ) .
there also exist degeneracies between @xmath21 and @xmath7 and between @xmath21 and @xmath0 , see fig .
[ fig : w ] . as we shall explain in the following section , a value of @xmath31 shifts the positions of the cmb acoustic peaks to lower multipoles @xmath30
; this effect could be compensated by a decrease of @xmath7 or by an increase of @xmath0 .
the degeneracy between @xmath21 and @xmath7 gets alleviated when information on high @xmath30 multipoles from spt is considered in the analysis .
a change on the scalar spectral index @xmath16 can also be compensated by a change in @xmath0 , see the upper panel of fig . [
fig : neff ] .
this degeneracy only affects the amplitude of the cmb peaks : a higher @xmath0 will reduce the amplitude of the cmb peaks at @xmath32 due to a higher silk damping , which in turn is due to the increased expansion rate @xcite .
lcccc & baseline & baselinespt & baselinespt - snia + & model & model & model + + @xmath21 & @xmath33 & @xmath34 & @xmath34 + + @xmath35 & @xmath36 & @xmath37 & @xmath38 + + @xmath7 & @xmath39 & @xmath40 & @xmath41 + + @xmath6 & @xmath42 & @xmath43 & @xmath44 + + @xmath16 & @xmath45 & @xmath46 & @xmath47 + lcccccc & @xmath10cdm + @xmath20 ( planck ) & @xmath10cdm + @xmath20 ( core ) & @xmath21cdm ( planck ) & @xmath21cdm ( core ) & @xmath23cdm ( planck ) & @xmath23cdm ( core ) + + @xmath21 & -1 & -1 & @xmath48 & @xmath49 & & + + @xmath35 & 3.04 & 3.04 & 3.04 & 3.04 & 3.04 & 3.04 + + @xmath16 & @xmath50 & @xmath51 & @xmath52 & @xmath53 & @xmath52 & @xmath54 + + @xmath20 & @xmath55 & @xmath56 & 0 & 0 & 0 & 0 + + @xmath57 & & & & & @xmath58 & @xmath59 + + @xmath60 & & & & & @xmath61 & @xmath62 + + and @xmath63 cl bounds in the @xmath64 , @xmath65 and @xmath66 planes , respectively .
the blue , red and green regions refer to the baseline , baselinespt and baselinespt - snia models , respectively.,title="fig : " ] and @xmath63 cl bounds in the @xmath64 , @xmath65 and @xmath66 planes , respectively .
the blue , red and green regions refer to the baseline , baselinespt and baselinespt - snia models , respectively.,title="fig : " ] and @xmath63 cl bounds in the @xmath64 , @xmath65 and @xmath66 planes , respectively .
the blue , red and green regions refer to the baseline , baselinespt and baselinespt - snia models , respectively.,title="fig : " ] and @xmath63 cl bounds in the @xmath67 and @xmath68 planes , respectively .
the blue , red and green regions refer to the baseline , baselinespt and baselinespt - snia models , respectively.,title="fig : " ] and @xmath63 cl bounds in the @xmath67 and @xmath68 planes , respectively .
the blue , red and green regions refer to the baseline , baselinespt and baselinespt - snia models , respectively.,title="fig : " ] and @xmath69 planes , respectively .
the blue , red and green regions refer to the baseline , baselinespt and baselinespt - snia models respectively.,title="fig : " ] and @xmath69 planes , respectively .
the blue , red and green regions refer to the baseline , baselinespt and baselinespt - snia models respectively.,title="fig : " ] in the following we shall present the reconstructed values of @xmath16 , @xmath20 , @xmath21 , @xmath57 and @xmath60 which will result from a fit of planck and core mock data ( generated with non standard values for the dark radiation perturbation parameters , @xmath70 ) to a cosmology with a standard value for the dark radiation parameter @xmath71 but with a running spectral index or a time varying dark energy component .
we do not consider here @xmath72 due to the tighter current bounds on this parameter , when compared to the current constraints on @xmath7 .
for this scenario we consider the following set of parameters : @xmath73,\ : n_{\rm run}\}~.\ ] ] in general , the spectrum of the scalar perturbations is not exactly a power law but it varies with scale
. therefore one must consider the scale dependent running of the spectral index @xmath74 .
following @xcite , the power spectrum for the scalar perturbations reads @xmath75 being @xmath76 mpc@xmath77 the pivot scale . plane from mcmc fits of planck ( blue regions ) and core ( red regions ) cmb mock data . ]
the correlation between @xmath16 and @xmath20 is shown in fig .
[ fig : ns_nrun ] .
as stated in @xcite , the parameter that is constrained by cosmological data is the effective spectral index @xmath78 .
this is the reason for the circular allowed regions in the @xmath79 plane .
the first and second columns of tab .
[ tab : contstraints ] show that , if a cosmology with @xmath80 but with non standard dark radiation perturbation parameters ( @xmath70 ) is fitted to a cosmology with standard dark radiation parameters but with @xmath81 , the reconstructed value of the running spectral index will differ from zero at a high statistical significance .
finally , for the case of the simulated cosmology here with @xmath82 , the reconstructed value of @xmath16 is consistent with a harrison - zeldovich scale invariant primordial power spectrum within one sigma .
setting the properties of dark radiation is therefore mandatory since it is highly correlated with the spectral index of the spectrum of primordial perturbations , key to distinguish among the different inflationary models .
here we consider a cosmological model including a dark energy fluid characterized by a constant equation of state @xmath21 as a free parameter .
we consider the following set of parameters : @xmath73,\ : w\}~.\ ] ] as stated in a previous work @xcite , there exists a degeneracy between the number of the extra dark radiation species and the dark energy equation of state , see fig .
[ fig : w ] ( upper panel ) .
one of the main effects of a @xmath83 comes from the change of the epoch of the radiation matter equality , and consequently , from the shift of the cmb acoustic peaks , see ref .
@xcite for a detailed study .
the position of the acoustic peaks is given by the so - called acoustic scale @xmath84 , which reads @xmath85 where @xmath86 and @xmath87 are the comoving angular diameter distance to the last scattering surface and the sound horizon at the recombination epoch @xmath88 , respectively .
although @xmath86 almost remains the same for different values of @xmath0 , @xmath87 becomes smaller when @xmath0 is increased .
thus the positions of the acoustic peaks are shifted to higher multipoles ( smaller angular scales ) by increasing the value of @xmath0 @xcite .
a dark energy component with @xmath31 will decrease the comoving angular diameter distance to the last scattering surface @xmath86 , shifting the positions of the cmb acoustic peaks to larger angular scales , i.e. to lower multipoles @xmath30 , compensating , therefore , the effect induced by an increase of @xmath0 .
the reconstructed mcmc values for @xmath21 ( see the third and fourth columns of tab . [
tab : contstraints ] ) are larger than the value used in the input cosmology @xmath18 , excluding the @xmath10cdm scenario with high significance . a dark radiation component which deviates from its standard behavior
could therefore be confused with the presence of a dark energy fluid with @xmath89 .
but in the @xmath90 plane .
] we also consider models of the dark energy in which the equation of state of the dark energy component varies with time .
we use a parameterization that has been extensively explored in the literature @xcite : @xmath91 where @xmath57 is the equation of state parameter at present , while @xmath92 @xcite .
we consider the following set of parameters : @xmath93 , w_0 , w_a\}~.\ ] ] the fifth and sixth columns of tab .
[ tab : contstraints ] show the reconstructed values of @xmath57 and @xmath60 after fitting the plank and core mock data generated with a non standard viscosity parameter @xmath70 but with @xmath18 to a cosmology with standard dark radiation but with the time varying dark energy equation of state @xmath23 used here .
the correlation between @xmath57 and @xmath60 is shown in fig .
[ fig : wa ] .
the reconstructed values that we find for planck ( core ) mock data are @xmath94 and @xmath95 ( @xmath96 and @xmath97 ) at 68 @xmath98 cl , values which are consistent with the current constraints on these two dark energy parameters , see ref .
therefore it is crucial to unravel the nature of the dark radiation component since if it turns out to be non standard , future cosmological data might be misinterpreted as a time varying dark energy fluid .
but in the @xmath99 plane . ]
dark radiation in the standard model of elementary particles is made of three light neutrinos . however
, many extensions of the standard model predict an extra dark radiation component parameterized in terms of the relativistic degrees of freedom .
such is the case of sterile neutrinos , axions or other light degrees of freedom produced along the thermal history of the universe .
this extra dark radiation component will be characterized not only by its abundance but also by its clustering properties , as its effective sound speed and its viscosity parameter .
if dark radiation is made of sterile neutrinos it should have an effective sound speed @xmath6 and a viscosity parameter @xmath7 such that @xmath8 .
however , other relativistic species might not behave as neutrinos , being @xmath7 and @xmath6 different from their canonical values .
current bounds on the number of relativistic species and on the dark radiation perturbation parameters @xmath6 and @xmath7 have been computed using up to date cosmological data .
we find a strong degeneracy between @xmath7 and the scalar spectral index of primordial perturbations , as well as between @xmath7 and the dark energy equation of state @xmath21 .
the last degeneracy is alleviated when cmb data from the spt experiment is added in the mcmc analyses .
a question which naturally arises from the presence of these degeneracies is whether or not future cmb data will be able to distinguish among different dark radiation scenarios .
we have generated mock cmb data for the ongoing planck experiment and the future core mission with non standard values for the dark energy perturbation parameters .
then , we have fitted these data to a canonical dark radiation scenario with @xmath100 but with a running spectral index or with a dark energy component with @xmath89 , finding that non standard values for the dark radiation perturbation parameters may be misinterpreted as a scale invariant power spectrum of primordial fluctuations or as cosmologies with a running spectral index or a time varying dark energy component with high significance .
is supported by the consolider ingenio project csd2007 - 00060 , by prometeo/2009/116 , by the spanish ministry science project fpa2011 - 29678 and by the itn invisibles pitn - ga-2011 - 289442 .
this work is supported by prin - inaf , `` astronomy probes fundamental physics '' .
support was given by the italian space agency through the asi contracts euclid- ic ( i/031/10/0 ) .
g. mangano , a. melchiorri , o. mena , g. miele and a. slosar , jcap * 0703 * , 006 ( 2007 ) [ arxiv : astro - ph/0612150 ] . j. hamann , s. hannestad , g. g. raffelt and y. y. y. wong , jcap * 0708 * , 021 ( 2007 ) [ arxiv:0705.0440 [ astro - ph ] ] .
b. a. reid , l. verde , r. jimenez and o. mena , jcap * 1001 * , 003 ( 2010 ) [ arxiv:0910.0008 [ astro-ph.co ] ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , astrophys . j. suppl . * 192 * , 18 ( 2011 ) [ arxiv:1001.4538 [ astro-ph.co ] ] .
j. hamann , s. hannestad , j. lesgourgues , c. rampf and y. y. y. wong , jcap * 1007 * , 022 ( 2010 ) [ arxiv:1003.3999 [ astro-ph.co ] ] .
j. hamann , jcap * 1203 * , 021 ( 2012 ) [ arxiv:1110.4271 [ astro-ph.co ] ] .
k. m. nollett and g. p. holder , arxiv:1112.2683 [ astro-ph.co ] .
e. giusarma , m. archidiacono , r. de putter , a. melchiorri and o. mena , arxiv:1112.4661 [ astro-ph.co ] .
r. keisler _ et al .
_ , arxiv:1105.3182 [ astro-ph.co ] . j. dunkley _ et al .
_ , arxiv:1009.0866 [ astro-ph.co ] .
m. archidiacono , e. calabrese and a. melchiorri , arxiv:1109.2767 [ astro-ph.co ] .
z. hou , r. keisler , l. knox , m. millea and c. reichardt , arxiv:1104.2333 [ astro-ph.co ] . t. l. smith , s. das and o. zahn , phys . rev .
d * 85 * , 023001 ( 2012 ) [ arxiv:1105.3246 [ astro-ph.co ] ] .
[ planck collaboration ] , astro - ph/0604069 .
a. aguilar _ et al . _
[ lsnd collaboration ] , phys .
d * 64 * , 112007 ( 2001 ) [ arxiv : hep - ex/0104049 ] .
a. melchiorri , o. mena , s. palomares - ruiz , s. pascoli , a. slosar and m. sorel , jcap * 0901 * , 036 ( 2009 ) [ arxiv:0810.5133 [ hep - ph ] ] . j. hamann , s. hannestad , g. g. raffelt , i. tamborra and y. y. y. wong , phys . rev .
lett . * 105 * , 181301 ( 2010 ) [ arxiv:1006.5276 [ hep - ph ] ] .
e. giusarma , m. corsi , m. archidiacono , r. de putter , a. melchiorri , o. mena , s. pandolfi , phys .
* d83 * , 115023 ( 2011 ) . [ arxiv:1102.4774 [ astro-ph.co ] ] . s. hannestad , a. mirizzi , g. g. raffelt and y. y. y. wong , jcap * 1008 * ( 2010 ) 001 [ arxiv:1004.0695 [ astro-ph.co ] ] ; a. melchiorri , o. mena and a. slosar , phys .
d * 76 * ( 2007 ) 041303 [ arxiv:0705.2695 [ astro - ph ] ] .
w. fischler , j. meyers , phys .
* d83 * ( 2011 ) 063520 .
[ arxiv:1011.3501 [ astro-ph.co ] ] .
p. binetruy , c. deffayet , u. ellwanger and d. langlois , phys .
b * 477 * ( 2000 ) 285 [ arxiv : hep - th/9910219 ] ; t. shiromizu , k. i. maeda and m. sasaki , phys .
d * 62 * ( 2000 ) 024012 [ arxiv : gr - qc/9910076 ] ; v. v. flambaum and e. v. shuryak , europhys .
* 74 * ( 2006 ) 813 [ arxiv : hep - th/0512038 ] .
e. calabrese , d. huterer , e. v. linder , a. melchiorri and l. pagano , phys .
d * 83 * , 123504 ( 2011 ) [ arxiv:1103.4132 [ astro-ph.co ] ] .
m. blennow , e. fernandez - martinez , o. mena , j. redondo and p. serra , arxiv:1203.5803 [ hep - ph ] .
g. mangano , a. melchiorri , p. serra , a. cooray and m. kamionkowski , phys .
d * 74 * ( 2006 ) 043517 [ astro - ph/0606190 ] .
p. serra , f. zalamea , a. cooray , g. mangano and a. melchiorri , phys .
d * 81 * , 043507 ( 2010 ) [ arxiv:0911.4411 [ astro-ph.co ] ] .
w. hu , astrophys . j. * 506 * ( 1998 ) 485 [ astro - ph/9801234 ] .
w. hu , d. j. eisenstein , m. tegmark and m. j. white , phys . rev .
d * 59 * , 023512 ( 1999 ) [ astro - ph/9806362 ]
. j. f. beacom , n. f. bell , s. dodelson , phys . rev .
* 93 * ( 2004 ) 121302 .
[ astro - ph/0404585 ] ; s. hannestad , jcap * 0502 * ( 2005 ) 011 .
[ astro - ph/0411475 ] ; a. basboll , o. e. bjaelde , s. hannestad and g. g. raffelt , phys .
d * 79 * ( 2009 ) 043512 [ arxiv:0806.1735 [ astro - ph ] ] ; r. trotta , a. melchiorri , phys .
* 95 * ( 2005 ) 011305 .
[ astro - ph/0412066 ] .
b. a. reid _ et al .
_ , mon . not .
* 404 * , 60 ( 2010 ) [ arxiv:0907.1659 [ astro-ph.co ] ] .
a. g. riess _ et al .
_ , astrophys .
j. * 699 * , 539 ( 2009 ) [ arxiv:0905.0695 [ astro-ph.co ] ] .
r. amanullah , c. lidman , d. rubin , g. aldering , p. astier , k. barbary , m. s. burns and a. conley _ et al .
_ , astrophys .
j. * 716 * , 712 ( 2010 ) [ arxiv:1004.1711 [ astro-ph.co ] ] .
a. lewis , a. challinor and a. lasenby , astrophys .
j. * 538 * , 473 ( 2000 ) [ arxiv : astro - ph/9911177 ] .
a. lewis and s. bridle , phys .
d * 66 * , 103511 ( 2002 ) [ arxiv : astro - ph/0205436 ] .
f. r. bouchet _ et al .
_ [ the core collaboration ] , arxiv:1102.2181 [ astro-ph.co ] .
a. kosowsky and m. s. turner , phys .
d * 52 * , 1739 ( 1995 ) [ astro - ph/9504071 ] .
s. hannestad , s. h. hansen , f. l. villante and a. j. s. hamilton , astropart .
* 17 * , 375 ( 2002 ) [ astro - ph/0103047 ] .
k. ichikawa , t. sekiguchi and t. takahashi , phys .
d * 78 * , 083526 ( 2008 ) [ arxiv:0803.0889 [ astro - ph ] ] .
m. chevallier and d. polarski , int .
j. mod .
d * 10 * , 213 ( 2001 ) [ gr - qc/0009008 ] .
e. v. linder , phys .
lett . * 90 * , 091301 ( 2003 ) [ astro - ph/0208512 ] .
a. albrecht , g. bernstein , r. cahn , w. l. freedman , j. hewitt , w. hu , j. huth and m. kamionkowski _ et al .
_ , [ astro - ph/0609591 ] .
e. v. linder , phys .
d * 73 * , 063010 ( 2006 ) [ astro - ph/0601052 ] . | recent cosmological data have provided evidence for a `` dark '' relativistic background at high statistical significance .
parameterized in terms of the number of relativistic degrees of freedom @xmath0 , however , the current data seems to indicate a higher value than the one expected in the standard scenario based on three active neutrinos .
this dark radiation component can be characterized not only by its abundance but also by its clustering properties , as its effective sound speed and its viscosity parameter .
it is therefore crucial to study the correlations among the dark radiation properties and key cosmological parameters , as the dark energy equation of state or the running of the scalar spectral index , with current and future cmb data .
we find that dark radiation with viscosity parameters different from their standard values may be misinterpreted as an evolving dark energy component or as a running spectral index in the power spectrum of primordial fluctuations . |
uncertainty in quantum theory can be attributed to two different issues : the irreducible indeterminacy of individual quantum processes , postulated by born @xcite , and the complementarity , introduced by bohr @xcite , which implies that we can not simultaneously perform precise measurements of noncommuting observables .
for that reason , estimation of the quantum state from the measurement outcomes , which is sometimes called quantum tomography , is of paramount importance @xcite .
moreover , in practice , unavoidable imperfections and finite resources come into play and the performance of tomographic schemes should be assessed and compared . at a fundamental level , mutually unbiased bases ( mubs )
provide perhaps the most accurate statement of complementarity .
this idea emerged in the seminal work of schwinger @xcite and it has gradually turned into a keystone of quantum information . apart from being instrumental in many hard problems @xcite ,
mubs have long been known to provide an optimal basis for quantum tomography @xcite .
when the hilbert space dimension @xmath0 is a prime power , @xmath1 , it is known that there exist sets of @xmath2 mubs @xcite .
these hilbert spaces are realized by systems of @xmath3 particles , which , in turn , allow for special entanglement properties .
more specifically , we consider here @xmath3 qubits , as they are the building blocks of quantum information processing .
it was known that for three qubits , four different set of mubs exist with very different entanglement properties @xcite .
the analysis was further extended to higher number of qubits @xcite and confirmed by different approaches @xcite . for the experimentalist
, this information is very important , because the complexity of an implementation of a given set of mubs will , of course , greatly depend on how many of the qubits need to be entangled .
in this work , we use fisher information to set forth efficient tools for assessing the quality of a wide class of tomographic schemes , paying special attention to @xmath3-qubit mubs . despite the widespread belief that mubs are equally sensitive to all the state features ,
we show that mubs with different entanglement properties differ in their potential to characterize particular aspects of the reconstructed state .
we illustrate this with the fidelity estimation of three - qubit states , making evident the relevance of mub - entanglement classification and illustrating the possibility to optimize mub tomography with respect to the observables of interest .
let us start with a brief outlook of the basic methods we need for the rest of the discussion .
we deal with a @xmath0-dimensional quantum system , represented by a positive semidefinite @xmath4 density matrix @xmath5 .
a very convenient parametrization of @xmath5 can be achieved in terms of a traceless hermitian operator basis @xmath6 , satisfying @xmath7 and @xmath8 : @xmath9 where the @xmath10-dimensional generalized bloch vector @xmath11 ( @xmath12 ) is uniquely determined by @xmath13 .
the set @xmath14 coincides with the orthogonal generators of su(@xmath0 ) , which is the associated symmetry algebra @xcite . the measurements performed on the system are described , in general , by positive operator - valued measures ( povms ) , which are a collection of positive operators @xmath15 , resolving the identity @xmath16 @xcite .
each povm element represents a single output channel of the measuring apparatus ; the probability of detecting the @xmath17th output is given by the born rule @xmath18 . in a sensible estimation procedure
, we have @xmath19 identical copies of the system and repeat the measurement on each of them .
the statistics of the outcomes is then multinomial , i.e. , @xmath20 where @xmath21 is the probability of detection at the @xmath17th channel and @xmath22 the actual number of detections . here , @xmath23 is the probability of registering the data @xmath24 provided the true state is @xmath25 .
the estimation requires the introduction of an estimator ; i.e. , a rule of inference that allows one to extract a value for @xmath25 from the outcomes @xmath24 .
the random variable @xmath26 is an unbiased estimator if @xmath27 . the ultimate bound on the precision with which one can estimate @xmath25 is given by the cramer - rao bound @xcite , which , in terms of the covariance matrix @xmath28 , can be stated as @xmath29 where f stands for the fisher matrix @xcite @xmath30 since @xmath31 , it might superficially appear that computing the fisher matrix ( and hence the covariances of the estimated parameters ) is straightforward
. however , in practice , this can become quite a difficult task : the cost of computing @xmath32 matrix multiplications and inversions , even with the best - known exact algorithm @xcite , scales as @xmath33 . for a system of @xmath3 qubits
, this computational cost goes as @xmath34 , which sets an upper limit on the dimension for which the evaluation of the reconstruction errors is feasible .
for instance , analyzing a system of just five qubits requires about a billion of arithmetic operations with individual elements , which makes the problem intractable along these lines .
this especially applies when a large number of repeated evaluations is required , as in monte carlo simulations .
this numerical cost can be considerably reduced by employing a special parametrization for @xmath5 . to this end , we restrict ourselves to informationally complete ( ic ) measurements , which are those for which the outcome probabilities are sufficient to determine an arbitrary quantum state @xcite . given a system of dimension @xmath0 , any ic measurement must have at least @xmath35 output channels ; when it has just @xmath36 of them , it will be called a minimal ic reconstruction scheme .
it turns out that the error analysis in this case is particularly simple and can be done analytically avoiding time - expensive computations .
we remark that we are not addressing here the resources needed for a complete tomography ( which scale exponentially ) ; rather , our aim is to ascertain what can be better estimated from ic measurements . indeed , for an ic minimal scheme @xmath37 ( @xmath38 ) , there exists a unique representation of any quantum state in terms of the basic probabilities @xmath39 : @xmath40 normalization reduces by one the number of independent parameters describing the state : @xmath41 ( @xmath42 and @xmath43 . in this way
, we get @xmath44 ( @xmath45 ) and @xmath46 , leading us to @xmath47 we thus conclude that the errors of any ic minimal scheme are given by the covariance matrix of the underlying true multinomial distribution governing the measurement outcomes . notice that this might not apply to other overdetermined setups , as , for instance , optical homodyne tomography .
once the fisher matrix is known , the errors in any observation can be estimated .
let us consider the measurement of the average value @xmath48 of some generic observable @xmath49 .
if the reconstructed state is @xmath50 , the predicted outcomes are @xmath51 and the expected errors are @xmath52 ^ 2 \right \rangle \ , , \ ] ] where the averaging here is over many repetitions of the reconstruction . by expanding the observable in the povm elements , @xmath53 , the true and predicted outcomes can be given in terms of true and predicted measurement probabilities as @xmath54 and @xmath55 , where @xmath56 and @xmath57 , respectively . denoting @xmath58 the differences between the true and inferred probabilities , the expected error of the observable @xmath49 defined by eq
becomes @xmath59 next , we rearrange eq . so that only @xmath60 linearly independent probabilities @xmath61 are involved .
notice that @xmath62 implies @xmath63 and hence @xmath64 . in this way
, we get @xmath65 employing the cramer - rao lower bound , we finally obtain @xmath66 where @xmath67 is the inverse fisher matrix in the probability representation given by eq . .
occasionally , working in the measurement representation might be preferable .
expanding the true and estimated states in the measured povm elements , @xmath68 and @xmath69 , we seek to express @xmath70 in terms of the reconstruction errors @xmath71 .
if @xmath72 is the matrix connecting the measurement and probability representations , then @xmath73 and we have @xmath74 , which is computed effectively provided @xmath75 is sparse . by inserting this into eq .
, we get the desired result .
one pertinent example for which this probability representation turns out to be very efficient is for mubs .
as heralded before , we consider a system of @xmath3 qubits ; since the dimension @xmath76 is a power of a prime sets of mubs @xmath77 exist and explicit construction procedures are at hand @xcite .
we denote the corresponding projectors by @xmath78 where the greek index @xmath79 labels one of the @xmath80 families of mubs and @xmath17 denotes one of the @xmath0 orthogonal states in this family .
unbiasedness translates into @xmath81 notice that , in agreement with the usual mub terminology , each set of eigenstates is normalized to unity @xmath82 , so that the povm becomes normalized to @xmath2 , rather than to unity .
our previous convention is readily recovered by using @xmath83 as the total number of copies . as the total number of projections [ @xmath84 minus the number of constraints [ @xmath85 matches the minimal number [ @xmath86 of independent measurements ,
this mub tomography is indeed an ic minimal scheme .
in addition , there are @xmath2 sets of vectors , each resolving the unity , so that @xmath87 holds for each observable @xmath79 .
consequently , the fisher information matrix in the probability representation takes a block diagonal form @xmath88 and in each block @xmath79 , @xmath89 and @xmath90 are given by eq . .
measurement errors can now be easily estimated . indeed , expanding a generic state in the mub basis as @xmath91 with @xmath92 , we find that @xmath93 and the total error appears as a sum of independent contributions @xmath94 of individual mub eigensets , with @xmath95 we can reinterpret these results in an alternative way . without lack of generality , we consider one diagonal block and drop the index @xmath79 .
we introduce the operator @xmath96 by @xmath97 , with @xmath98 , @xmath99 , and @xmath100 .
this can be represented by the rectangular @xmath101-dimensional matrix @xmath102 .
if @xmath103 is the standard scalar product in this @xmath0-dimensional vector space , we may simply write @xmath104 where @xmath105 .
notice that the inferred variance takes now the form of born s rule and the effective inverse fisher matrix @xmath106 becomes the relevant object governing the errors .
for example , the mean hilbert - schmidt distance from the estimate to the true state becomes @xmath107 \rangle= { \mathop{\mathrm{tr } } \nolimits}(\tilde{\mathrm{f}}^{-1 } ) \ , , \ ] ] which gives a unitary invariant error , as it might be anticipated @xcite .
another instance of interest is the error of the fidelity measurement @xmath108 . here
@xmath109 the constant term @xmath110 can be dropped because @xmath111 enters eq .
( [ eq : uf ] ) only through differences .
these formulas , together with eq . ,
provide timely tools for analyzing the performance and optimality of different mub reconstruction schemes .
for example , one could be interested in which observable can be most ( least ) accurately inferred from a mub tomography .
this is tantamount to minimizing ( maximizing ) eq .
subject to a fixed norm of @xmath112 .
the principal axes of the error ellipsoid are defined by the eigenvectors and eigenvalues of the effective inverse fisher matrix @xmath106 .
notice that there is always one zero eigenvalue per diagonal block , corresponding to a constant vector @xmath113 ( @xmath114 , which , in turn , corresponds to measuring the trace of the signal density matrix @xmath115 and @xmath116 .
this is consistent with the fact that the trace is constrained to unity and hence error free .
thus , the least ( most ) favorable measurements from the point of view of a particular detection scheme are those given by the largest ( second smallest ) eigenvectors of @xmath106 .
similarly , one may be interested in the distribution @xmath117 $ ] of errors among all possible inferred measurements . from eq .
( [ scalarproduct ] ) , such a distribution is given precisely by the restricted numerical range or `` real shadow '' @xcite of @xmath106 .
once the formalism has been set up , let us see how it works for mub reconstruction schemes .
the goal is to assess the performance of different sets of mubs for moderately - sized quantum systems . for definiteness ,
let us look at the case of three qubits .
it is known @xcite that mub sets can be divided into nonequivalent classes with respect to entanglement properties . in the eight - dimensional hilbert space of three qubits ,
any complete set is comprised of 9 mubs .
we label the different sets by @xmath118 , where @xmath119 denotes the number of separable bases ( every eigenvector of theses bases is a tensor product of singe - qubit states ) , @xmath120 the number of biseparable bases ( one qubit is factorized and the other two are in a maximally entangled state ) and @xmath121 the number of nonseparable bases . in this notation , there are four classes : @xmath122 , @xmath123 , @xmath124 , and @xmath125 ( see the supplemental material for a detailed description ) . on physical grounds , one could expect that the performances of these four classes with respect to entanglement - specific state properties will also be different . to clarify the question we apply the procedure developed so far . to tie the observable to entanglement properties
, we consider the estimation of the fidelity error taking the inferred observable to be the projection on the true state , @xmath126 .
the confidence in the inferred fidelity can be used at the same time as a simple criterion of the overall quality of the tomographic setup .
class performs best on ghz , bipartite and separable states , but worst on w states .
the differences are up to @xmath127 .
[ fig1 ] ] in our simulations , genuinely tripartite ghz and w entangled three - qubit states were randomly generated , as well as bipartite and separable ones , and subjected to mub measurements .
for each mub class , the expected error of the inferred fidelity was estimated and averaged over @xmath128 randomly distributed states .
the result is shown in fig .
[ fig1 ] . as we can appreciate , in average the @xmath122 scheme yields smaller fidelity errors for ghz - like states ( for bipartite and separable also ) and larger errors for w states compared to other mub sets .
one also notices that the difference between ghz , bipartite and separable is small , while w are still different . to understand this result better ,
we have also investigated the noise in the measured data and calculated the correlations between the entropy of measurement probabilities and fidelity errors , restricting ourselves , for simplicity , to ghz and w states .
figure [ fig2 ] confirms strong correlations between those two magnitudes and shows that for ghz states , @xmath122 tomograms are less noisy on average than @xmath123 tomograms .
again , the opposite is true for w states .
there is not a simple connection between state entanglement and mub classes , as one might naively anticipate .
this demonstrates the key role played by error estimation along the lines presented here . and
@xmath123 bases , respectively .
[ fig2 ] ] finally , we consider a set of randomly chosen measurements @xmath129 , whose expectation values @xmath130 are inferred from the estimated state @xmath50 and analyze the probability distribution @xmath117 $ ] of the expected variances @xmath131 .
this distribution , being the real shadow @xmath132 of the effective inverse fisher matrix , tells us how the errors are distributed in the set of all possible measurements and hence describes in detail the performance of a given reconstruction scheme . in fig .
[ fig3 ] , the difference @xmath133-p_{(3,0,6)}[(\delta z)^2]$ ] of the @xmath122 and @xmath123 shadows is approximated by histograms obtained from @xmath134 variances calculated for randomly generated measurements @xmath135 and randomly choosen ghz and w states for each mub set . for ghz states , the @xmath122 scheme yields very small errors and very large errors more often than the @xmath123 scheme .
the opposite is true for w states
. such complementary behavior of different mub sets makes it possible to optimize the tomography setup .
for instance , quantities with low inherent noise are more precisely determined by the @xmath122 [ @xmath123 ] set provided the true state is ghz - like ( w - like ) .
the fidelity measurement discussed above is a salient example of that optimization .
we stress that this kind of analysis , where many repeated evaluations of the fisher matrix must be performed , is chiefly suited to the proposed method , for the standard approach would become quickly unfeasible due to the scaling with the system dimension . and
@xmath123 histograms of variances for ghz and w states .
all histograms are normalized to unity .
positive @xmath137 for some @xmath70 means error of this size is more typical for @xmath122 scheme than for @xmath123 scheme and _ vice versa_. [ fig3 ] ]
to summarize , we have presented a complete fisher - information - based toolbox for the proper assessment of any ic tomographic scheme . in this context
, we have discussed the case of set of mubs ; although all of them are ic , their performance for some particular tasks turns out to be dependent on the entanglement distribution among the states of the set .
we believe that the ideas and techniques developed here will be relevant in the experimental implementation and optimization of quantum protocols in higher - dimensional hilbert spaces .
financial support from the eu fp7 ( grant q - essence ) , the spanish dgi ( grant fis2011 - 26786 ) , the ucm - bsch program ( grant gr-920992 ) , the mexican conacyt ( grant 106525 ) , the czech ministry of trade and industry ( grant fr - ti1/384 ) and the technology agency of the czech republic ( grant te01020229 ) is acknowledged .
27ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop _ _ , ed .
( , , ) and , eds . , @noop _ _ , , vol .
( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) http://link.aps.org/doi/10.1103/physreva.84.022327 [ * * , ( ) ] @noop * * , ( )
for the sake of completeness , we briefly review the structure of the sets of mubs for three qubits , following essentially the approach in ref . @xcite . because states belonging to the same basis are usually taken to be orthonormal , to study the property of mutually unbiasedness " it is possible to use either mutually unbiased bases or the operators which have the basis states as eigenvectors .
we thus need @xmath138 operators to obtain the whole set of states . in the case of power of prime dimension ,
this set can be constructed as @xmath2 classes of @xmath139 commuting operators . in this way we get tables with nine rows of three mutually commuting ( tensor products of ) operators .
we have suppressed the tensor multiplication sign in all the tables . by construction , the simultaneous eigenstates of the operators in each row give a complete basis , and each basis is mutually unbiased to each other .
the number on the left enumerates the bases , while the number on the right denotes how many subsystems the bases can be factorized into . as stated before , we label the different sets of mubs by @xmath118 , where @xmath119 denotes the number of separable bases ( every eigenvector of theses bases is a tensor product of singe - qubit states ) , @xmath120 the number of biseparable bases ( one qubit is factorized and the other two are in a maximally entangled state ) and @xmath121 the number of nonseparable bases . in this way , the allowed structures are @xmath140 and the corresponding tables are given in the following . | an efficient method for assessing the quality of quantum state tomography is developed .
special attention is paid to the tomography of multipartite systems in terms of unbiased measurements .
although the overall reconstruction errors of different sets of mutually unbiased bases are the same , differences appear when particular aspects of the measured system are contemplated .
this point is illustrated by estimating the fidelities of genuinely tripartite entangled states . |
with @xmath116 if @xmath43 is represented by a mps in mixed normalised form as discussed in the main text with @xmath80}$ ] being the blocked tensor of sites @xmath51 and @xmath58 .
legs corresponding to indices of site @xmath51 are indicated in blue , to indices of site @xmath58 in red . ] implementing the optimisation methods in grassmann manifolds as described in refs .
@xcite we parametrise the grassmann manifold @xmath117 by @xmath118 isometries @xmath119 which form the stiefel manifold @xmath120 . in order to implement a the conjugated gradient search for identifying a ( local ) minimum of @xmath121 in @xmath117 the derivatives @xmath122 are needed . as the resulting mode transformation acts on two sites only , we are interested in the case @xmath123 . given an isometry @xmath124 , i.e.@xmath125 we can construct a unitary @xmath126 with the first @xmath127 columns being equal to the columns of @xmath128 by @xmath129 with @xmath130 and @xmath131 which corresponds to a generalised householder reflection of the subspace spanned by the columns of @xmath128 . note that if @xmath132 turns out to be singular , we can always transform the columns of @xmath128 using a random @xmath133 unitary , e.g. @xmath134 with @xmath135 random , by which we chose a new representative in @xmath136 of the same element in @xmath137 given a general invertible matrix @xmath138 , the elements of which depend on a real parameter @xmath5 , we can evaluate the derivative of the inverse matrix to @xmath139 y(t)^{-1}.\ ] ] from fig .
[ fig:4norm ] we can read off @xmath140 m [ g(u)\otimes g(u ) ] n ) , \end{aligned}\ ] ] where @xmath141})$ ] corresponds to the inner ring of tensors in fig . [ fig:4norm ] which depends on the coupled tensor @xmath80}$ ] and @xmath142 orders the outer legs as shown .
we can then evaluate the derivative of this cost function with respect to the parameters @xmath143 and @xmath144 using @xmath145 if @xmath146 , @xmath147 and @xmath148 and 0 otherwise , where @xmath149 , @xmath150 $ ] and @xmath151 denotes the number of elements of @xmath128 smaller @xmath152 and @xmath153 with @xmath154 to avoid redundant calculations , dmrg implementations use complementary operators which are expanded and reloaded during the sweeps @xcite .
we denote by @xmath155 the set of all modes which are associated to sites @xmath73 with @xmath92 and by @xmath156 the set of modes that belong to sites @xmath73 with @xmath93 .
in addition we define the abbreviations @xmath157}^{\alpha_1}\ldots a_{[m-1]}^{\alpha_{m-1}})_a{|\alpha_1\rangle}\otimes\ldots\otimes{|\alpha_{m-1}\rangle}\ ] ] and @xmath158}^{\alpha_{m+1}}\ldots a_{[n]}^{\alpha_{n}})_a{|\alpha_{m+1}\rangle}\otimes\ldots\otimes{|\alpha_{n}\rangle}.\ ] ] the four different types of complementary operators , @xmath159 with @xmath160 , @xmath161 $ ] and @xmath162\backslash\{i_m\}$ ]
are defined as follows @xmath163}\bigl[({|l_a\rangle\langlel_b|}\otimes { \mathbbm{1}})\sum\limits_{k , l\in l_m } v_{i , j , k , l } c_i^\dag c_j^\dag c_l c_k\bigr],\\ q^{l_m , a , b}_{i , k } & = { \text{tr}}_{[m-1]}\bigl[({|l_a\rangle\langlel_b|}\otimes { \mathbbm{1}})\sum\limits_{j , l\in l_m } v_{i , j , k , l } c_i^\dag c_j^\dag c_l c_k\bigr],\\ r^{l_m , a , b}_{i } & = { \text{tr}}_{[m-1]}\bigl[({|l_a\rangle\langlel_b|}\otimes
{ \mathbbm{1}})\sum\limits_{j , k , l\in l_m } v_{i , j , k , l } c_i^\dag c_j^\dag c_l c_k\bigr],\\ s^{l_m , a , b}_{i } & = { \text{tr}}_{[m-1]}\bigl[({|l_a\rangle\langlel_b|}\otimes { \mathbbm{1}})\sum\limits_{j\in l_m } t_{i , j } c^\dag_ic_j\bigr],\end{aligned}\ ] ] where the partial traces are evaluated according to the jordan wigner representation , plus their corresponding counterparts on sites @xmath73 with @xmath93 which are defined analogously by replacing @xmath155 by @xmath156 , @xmath164 by @xmath165 the partial trace over sites @xmath166 $ ] by the partial trace over sites @xmath167\backslash[m+1]$ ] . during a dmrg run the complementary operators for different @xmath51 are saved and reused in later steps . in order to update the complementary operators efficiently without loading and saving many operators from and to the disk , we evaluated some of the basis changes in a lazy fashion . during a right sweep , the complementary operators @xmath168
, @xmath169 , @xmath170 and @xmath171 are loaded while a mode transformation transforming the orbitals @xmath172 was accumulated ( resulting from the previous @xmath173 steps ) .
we obtain the complementary operators within the updated basis by transforming them similar to general operators as discussed in the main text by @xmath174\backslash i_m } ( u_{i}^\dag)_{i , i^\prime } ( u_{i}^\dag)_{j , j^\prime } p^{i_m , a , b}_{i^\prime , j^\prime}({\mathbbm{1}})\nonumber\\ & \:= [ ( u_i^\dag\otimes u_i^\dag ) p^{l_m , a , b}({\mathbbm{1}})]_{i , j } , \\ q^{i_m , a , b}(u ) & = u^\dag_{i}q^{i_m , a , b}({\mathbbm{1 } } ) u_i , \\
r^{i_m , a , b}(u ) & = u_i^\dag r^{i_m , a , b}({\mathbbm{1 } } ) , \\
s^{i_m , a , b}(u ) & = u_i^\dag s^{i_m , a , b}({\mathbbm{1}}),\end{aligned}\ ] ] with @xmath175 for a left and right sweep respectively and @xmath176 and @xmath177 the corresponding accumulated transformations .
note that , for a right sweep , the operators @xmath178 , @xmath179 , @xmath180 and @xmath181 can be formed with the rotated coefficients of the hamiltonian and need no further rotation .
the run time for a two - site dmrg per dmrg - step including an update of the complementary operators when stepping from site @xmath51 to @xmath182 during a sweep as @xmath183 @xcite .
the steps of the algorithm presented in table 1 in the main text scale in addition to that as follows .
note that @xmath8 is typically 2 ( spin up and down electrons ) for a qc - dmrg - but we will keep track of the dependence of the cost on @xmath8 for completeness .
the rotation of the blocked matrix and computation of @xmath60 comes at a cost of @xmath184 .
the transformation of the conjugated operators are performed with a cost of @xmath185 for general transformations , where the cost can be lowered to @xmath186 , once a @xmath187 or more general symmetry is enforced on the transformations . due to the locality of the mode transformation
, we can transform coefficients of the hamiltonian in a time scaling as @xmath188 . the computational cost induced from the local mode transformations which optimise @xmath189 iteratively
are therefore given by @xmath190 , so for the relevant parameters the number of orbitals @xmath9 and the bond dimension @xmath191 the additional cost scale as @xmath192 which is lower in cost than the two - site dmrg step with @xmath193 . in order to visualize the different physical bases we present below plots for the two site correlations present in the ground state approximations within the corresponding basis .
the two site correlation is measured here by the mutual information @xmath194 with @xmath195 $ ] and @xmath196 denotes the von neumann entropy of the reduced state @xmath197^\backslash i}|\psi\rangle\langle \psi |$ ] .
we present the correlation patterns of the final states of the individual calculations shown in fig . 2 in the main text .
[ fig : mutican0][fig : mutiloc10 ] show the mutual information of the final states obtained at @xmath198 in the initial and optimised basis starting in the hf and localised orbitals .
[ fig : mutietr0 ] , [ fig : mutietr10 ] display the mutual information patterns obtained in the calculation with bounded truncation error @xmath199 .
the correlation patterns obtained in the localised ( fig . [ fig : mutiloc0 ] ) and optimised bases ( fig .
[ fig : mutican15 ] , [ fig : mutiloc10 ] , [ fig : mutietr10 ] ) show very similar features , highlighting the fact that the localised basis in this system will be close to optimal .
note , however , that although the correlations for the localised and optimised version of the localised basis are almost the same ( compare fig . [ fig : mutiloc0 ] and fig .
[ fig : mutiloc10 ] ) , the latter allows for a more efficient approximation of the ground state using a qc - dmrg ; displaying the fact that the mutual information does not contain the full information about the optimality of the basis . | tensor network states and specifically matrix - product states have proven to be a powerful tool for simulating ground states of strongly correlated spin models .
recently , they have also been applied to interacting fermionic problems , specifically in the context of quantum chemistry .
a new freedom arising in such non - local fermionic systems is the choice of orbitals , it being far from clear what choice of fermionic orbitals to make . in this work , we propose a way to overcome this challenge .
we suggest a method intertwining the optimisation over matrix product states with suitable fermionic gaussian mode transformations .
the described algorithm generalises basis changes in the spirit of the hartree - fock method to matrix - product states , and provides a black box tool for basis optimisation in tensor network methods .
capturing strongly correlated quantum systems is one of the major challenges of modern theoretical and computational physics .
recent years have seen a surge of interest in the development of potent numerical methods based on tensor networks to approximate ground states of interacting lattice models @xcite , building upon the success of the density - matrix renormalisation group ( dmrg ) @xcite .
it has become clear that such ideas are also applicable to fermionic systems @xcite , and even to systems of quantum chemistry @xcite , lacking the locality present in lattice models in condensed - matter systems .
such tools allow in principle to approximate the full configuration interaction solution to good accuracy with reasonable effort , going in instances beyond conventional approaches to quantum chemistry , such as coupled cluster @xcite , configuration interaction or density - functional theory @xcite , as convincingly shown by first implementations of dmrg algorithms in quantum chemistry ( qc - dmrg ) @xcite . yet
, there is a new obstacle to be overcome : tensor network methods have originally been tailored to capture local interactions , and consequently ground states exhibiting short - range correlations and entanglement area laws @xcite .
systems in quantum chemistry pose new challenges due to the inherent long - ranged interactions , which are present no matter in what basis the systems are expressed .
new questions hence arise concerning the optimal topology and _ physical ( orbital ) basis _ used to construct the tensor network state @xcite . in this work
, we propose a novel approach towards making use of tensor network methods in quantum chemistry , by suggesting an adaptive scheme of updating basis transformations `` on the fly '' in conjunction with tensor network updates . in this way
, we bring together advantages of matrix product states which can capture strongly correlated states , but are tailored to short - ranged correlations and low entanglement and fermionic gaussian mode transformations for which entanglement is no obstacle , but non - gaussian correlations are .
we hence go significantly beyond previous approaches towards optimising fermionic bases in tensor network approaches to quantum chemistry .
previous dmrg implementations in quantum chemistry allowing for an optimisation of the physical basis restrict the mode transformations to permutations and separate the optimisation over the basis and state such that multiple dmrg runs are necessary @xcite . as a first attempt basis optimisations
using a few transformations have been implemented for tree tensor networks , however , this has been found to be unstable @xcite .
mixing fermionic orbitals from an active space the space considered here with further ones from an additional external space has also been studied @xcite . in these approaches
orbital transformations are carried out again between different dmrg runs .
in contrast to this we perform the mode transformations within the active space in parallel to the state optimisation and directly optimise the entanglement structure of the tensor network .
we focus on matrix - product states , but explain in what way the idea is generally applicable .
we also discuss the role of symmetries and the geometry of the problem at hand .
the basis optimisation is incorporated into the standard two - site qc - dmrg and can be added to existing implementation without increasing the computational costs of the dmrg .
the resulting scheme can be used in parallel to a ground state search or as a pre - processing step in which the physical basis is optimised in a first phase restricting the bond dimension of the mps used to medium values and calculating the final ground state in the optimised basis with higher accuracy .
_ system .
_ in this work , we are concerned with strongly correlated interacting fermionic models with a finite number of relevant modes as they appear in the quantum chemistry context . in second
quantised form the hamitonian takes the form @xmath0 where @xmath1 is a fermionic annihilation operator associated to the mode labeled @xmath2 satisfying the canonical anti - commutation relations @xmath3 and @xmath4 and the coupling @xmath5 and @xmath6 are such that @xmath7 is hermitian .
@xmath8 denotes the number of different fermion species present for each of the @xmath9 orbitals , e.g. spin up and down electrons .
the one particle modes form the basis of single particle hilbert space @xmath10 .
any fermionic state will be an element of the fermionic fock space @xmath11 , where @xmath12 denotes the exterior product and @xmath13 , with a basis formed of all slater determinants @xmath14 , where @xmath15 , of the initial single particle modes .
we refer to this basis as the physical basis .
the jordan wigner transformation establishes an isomorphism between @xmath16 and the hilbert space of @xmath9 qudits @xmath17 with @xmath18 . by choosing any ordering of the orbitals
such systems can be viewed as one - dimensional lattices of @xmath9 sites with long - range interactions . _
mps and general idea .
_ for a one - dimensional quantum lattice with @xmath9 sites , where each site is described by a @xmath19dimensional hilbert space @xmath20 , a matrix product state ( mps ) vector takes the general form @xmath21}\ldots a^{\alpha_n}_{[n]}{|\alpha_1\rangle}\otimes\ldots\otimes { |\alpha_n\rangle},\label{eq : definitionmps}\ ] ] where @xmath22}^{\alpha_m}\in { \mathbb{c}}^{d_{m-1}\times d_m}$ ] and @xmath23 form a basis of @xmath24 and @xmath25 .
if the _ bond dimension _ @xmath26 is allowed to vary arbitrarily over different sites , every quantum state of the lattice can be written as in eq .
@xcite . restricting the maximal bond dimension along the chain to a fixed value @xmath27
creates the sub - manifold @xmath28 of the full state space .
approximations of the ground state of a given hamiltonian within this sub - manifold can be found using the density matrix renormalisation group algorithm ( dmrg ) which , as an alternating least square method , optimises the entries of the mps tensors @xmath29})_m$ ] iteratively @xcite .
the freedom one has in this construction is a redefinition of the fermionic modes by a linear transformation .
linear transformations of a set of fermionic annihilation operators @xmath30 to a new set @xmath31 satisfying the canonical anti - commutation relations are captured by @xmath32 with a unitary mode - transformation @xmath33 .
this change of the single particle modes induces a transformation of the physical basis of @xmath16 . under this change of basis
a fermionic state vector @xmath34 transforms to @xmath35 with the gaussian unitary transformation @xmath36}$ ] acting in fock space .
the transformation on @xmath37 induced from the jordan wigner transformation is given by @xmath38 where @xmath39 .
we now turn to describing ground states of fermionic hamiltonians with mps expressed in a given basis , where the approximatability of the states strongly depends on the choice of basis @xcite . specifically , denoting the hamiltonian written in terms of the transformed modes by @xmath40 , we are interested in the solutions of @xmath41 note that the hartree - fock method is readily included in eq . by the case
@xmath42 , when @xmath43 is restricted to be a slater determinant . identifying the optimal or close - to optimal basis for a general hamiltonian and @xmath44 in the sense of eq . would provide a deeper understanding of the entanglement structure of ground states appearing in quantum chemistry , but since this is a non - convex problem , approximate solutions are accessible only . here , we take an approach that iteratively finds close to optimal solutions numerically by optimising over the ansatz - class depicted in fig . [
fig : mpsgu ] . is a gaussian transformations defined by a mode transformation @xmath33 as described in the main text . ]
_ compositions of local mode transformations .
_ in order to calculate approximations to the solutions of eq . and avoiding stability and performance issues of a direct global optimisation
, we perform successive local mode transformations in parallel to a two - site qc - dmrg and use a few additional global reorderings of the orbitals as in refs .
@xcite to leave local minima during the optimisation - process .
given a state vector @xmath43 , a site - index @xmath45 $ ] and a cost function @xmath46 which will be discussed below we solve @xmath47 with @xmath48 denoting the @xmath49dimensional identity matrix and @xmath50 needs to be chosen depending on the symmetries of the system .
the global basis change is then composed of local unitaries which are solutions of for different @xmath51 and act non - trivially on overlapping areas of the lattice and intermediate global reorderings of the lattice - sites .
the cost function is chosen according to the following paradigm .
the bond dimension needed for a bipartition of the system to approximate a state up to a predefined accuracy can be upper bounded using the rnyi entropies @xmath52 of the reduced state for @xmath53 @xcite , where @xmath54 .
we therefore iteratively minimise the @xmath55 entropy over the chosen bipartition by using the cost function @xmath56 where @xmath57 denotes the schmidt spectrum of @xmath43 for a bipartiting cut between sites @xmath51 and @xmath58 . with increasing dimension of @xmath59 , which growth with the number of species per orbital @xmath8 , and bond dimension
@xmath27 the optimisation of @xmath60 becomes slow .
efficiency can be gained by minimising @xmath61 of which we can calculate the gradient @xmath62 analytically and efficiently in the bond dimension as shown in the appendix .
the optimisation of @xmath63 will not lead to certified bounds on the required bond dimension , but will favour stronger decays in the schmidt spectrum similar as the minimisation of @xmath64 .
the results presented here have been obtained by optimising the one norm of the schmidt spectrum , @xmath65 .
the optimisation of @xmath66 can be applied in if @xmath59 has a higher dimension ; which appear for @xmath67 if the system lacks symmetries .
both the choice of the cost function and symmetries influence the choice of @xmath59 as argued in the following .
_ optimisation set .
_ in the presence of symmetries , choosing a physical basis which can be labeled by good quantum numbers decouples different symmetry sectors in the coefficient tensors of the hamitonian and the mps and allows for more efficient computations .
only mode transformations which commute with the generators of the symmetry transformations will preserve the structure imposed by the symmetry . in general qc - dmrg algorithms only exploit a subgroup of the full symmetry group of a specific hamiltonian , such as conservation of the number of particles , spin reflection symmetries , abelian point group symmetries or a @xmath68 spin rotation symmetry @xcite
( see also ref .
@xcite ) .
here , we consider for the states the case of particle number conservation of each species , which is an abelian symmetry and allows for an easy implementation of symmetric mps @xcite and want the local mode transformations to respect the @xmath68 symmetry of the considered systems .
the admissible transformations in this case are of the form @xmath69 with @xmath70 acting on one species of fermions .
the cost functions @xmath46 chosen above depend only upon the schmidt - spectrum of the state for a cut between sites @xmath51 and @xmath58 and are therefore insensitive to mode transformations of the form @xmath71 with @xmath72 acting only on the modes associated to the lattice site @xmath73 . to obtain a non - redundant parametrisation of the unitaries used in the optimisation in eq .
we restrict it to the set of left cosets @xmath74 which is isomorphic to the grassmann manifold @xmath75 .
efficient implementations of optimisation algorithms such as the conjugate gradient method within grassmann manifolds using @xmath76 parameters are described in refs .
@xcite .
if we restrict ourselves to mode transformations which preserve the @xmath68-symmetry , the relevant mode transformations are parametrised by @xmath77 , leaving @xmath78 optimisation parameters in each step .
focussing on this case here with medium values for @xmath27 , we can obtain @xmath79 of @xmath60 by using gradient free schemes such as the nelder - mead method , due to the small number of parameters .
_ algorithm . _ combined with an approximation of the ground state of a given hamiltonian , local mode transformations naturally extend a two - site dmrg .
a single two - site dmrg step results in a blocked tensor @xmath80}\in { \mathbb{c}}^{d^2\times d_{m-1}\times d_{m+1}}$ ] . in the generic case , restoring the mps format in eq . by decomposing the blocked tensor into local tensors @xmath22}^\prime\in
{ \mathbb{c}}^{d\times d_{m-1}\times d^\prime_m}$ ] and @xmath81}^\prime\in { \mathbb{c}}^{d\times d^\prime_m \times d_{m+1}}$ ] will lead to @xmath82 such that the found state needs to be projected into @xmath28 by discarding the @xmath83 smallest values of the resulting schmidt spectrum @xmath84 .
the projection yields an truncation error @xmath85 , where @xmath86 are the discarded singular values .
if we allow for a local mode transformation before the blocked tensor is decomposed , the truncation error can be reduced . 1 over neighbouring sites @xmath45 $ ] : 2get blocked tensor @xmath87}$ ] ( e.g. from two - site dmrg ) 3calculate ( local ) minimum @xmath79 of @xmath88 4 @xmath89 : 5transform @xmath43 with @xmath90 by updating + @xmath80}^{\alpha,\beta } = \sum_{\alpha^\prime,\beta^\prime=1}^dg(u_\text{opt}^\text{loc})_{(\alpha,\beta),(\alpha^\prime,\beta^\prime)}a_{[m , m+1]}^{\alpha^\prime,\beta^\prime}$ ] and + transform relevant operators with @xmath91 6calculate @xmath22}$ ] , @xmath81}$ ] by decomposing @xmath80}$ ] with + truncation and update mps with new tensors + using the gauge - invariance of mps , we bring the mps to a mixed normalised form , i.e. matrices of sites @xmath92 are left - normalised whereas matrices associated to sites @xmath93 are right - normalised @xcite .
we can then calculate @xmath84 from the blocked tensor @xmath80}$ ] .
we optimise the basis by solving eq .
while keeping expectation values of the state , such as the energy , constant by using @xmath94 with @xmath95 . as the mode transformation acts non -
trivially only on sites @xmath51 , @xmath58 the transformed state vector @xmath43 can be represented by @xmath96}(u)&=&a_{[k]}({\mathbbm{1}}),\,\,k\in[n]\backslash\{m , m+1\},\\ a_{[m , m+1]}^{\alpha,\beta}(u)&=&\sum_{\alpha^\prime,\beta^\prime}g(u)_{(\alpha,\beta),(\alpha^\prime,\beta^\prime ) } a_{[m , m+1]}^{\alpha^\prime,\beta^\prime}({\mathbbm{1}}).\end{aligned}\ ] ] operators such as the hamiltonian can be transformed efficiently using their second quantised representation . for an operator @xmath97 with @xmath98 ,
the coefficients transform under a mode - transformation according to @xmath99 .
as most operators of interest , e.g. , the hamiltonian , contain terms with small @xmath100 and @xmath101 those transformations can be implemented efficiently ; with cost scaling as @xmath102 for local transformations . be atoms with interatomic distance of @xmath103 .
all calculations have been performed with a @xmath104 symmetric open boundary mps and local mode transformations which keep the @xmath68 symmetry of the hamiltonian .
both diagrams show results of the described optimisation for the physical basis . in the left panel
we show the bond dimension needed for a bounded truncation error @xmath105 and @xmath106 when starting in the hf basis .
the dark blue dashed line corresponds to a calculation in the hf basis , the blue dotted and light blue line correspond to the first and the tenth iteration of the calculation with basis optimisation .
the right panel compares the relative error in energy @xmath107 obtained by calculations with @xmath108 , where the reference value for the ground state energy @xmath109 has been obtained from a calculation with @xmath110 in the localised basis .
the dark blue dashed and red dashed - dotted line show the results for a calculation in the hf and localised basis respectively . in light blue and orange we plot the relative error of the 15th and 10th iteration of the calculation with basis optimisation starting in the hf and localised basis respectively .
, scaledwidth=50.0% ] standard qc - dmrg algorithms use complementary operators @xcite in order to reduce the computational cost of each dmrg step . in the appendix
we show that complementary operators transform as general operators under local mode transformations and argue that local mode transformation can be found and applied in a time not exceeding the computational cost of a single dmrg step .
this allows us to keep the structure and the computational complexity of the two - site dmrg algorithm and perform basis optimisations essentially for free with the algorithm in table [ tbl : algorithm ] .
_ numerical results .
_ we use a qc - dmrg algorithm which uses the dynamical block state selection approach @xcite and configuration interaction based dynamically extended active space @xcite procedure to accelerate the convergence and adapt the basis of the physical space by the algorithm described above . as a test - system , we have chosen the electron - configuration of a beryllium ring built from 6 be atoms .
this system has recently been investigated @xcite and a strong dependence of the convergence of the dmrg from the initial basis was observed .
we investigate the molecule in a stretched geometry with an interatomic distance of 3.3 .
as initial bases we use the hartree - fock ( hf ) basis of the system and a localised basis derived from the hf orbitals by a foster - boys localisation @xcite .
such localised orbitals are widely used in qc - dmrg calculations and are known to yield a better convergence for the be ring @xcite .
starting from the according initial basis we iteratively apply the following scheme : we run the standard qc - dmrg for 2 sweeps , perform 8 additional sweeps together with the local mode transformation as described in table [ tbl : algorithm ] and reorder the basis according to its mutual information patterns @xcite .
hereby we either fix the truncation error @xmath111 made in each step or set a hard - cut on @xmath27 .
the results of our calculations are show in fig .
[ fig : results ] . in the left panel of fig . [
fig : results ] we show how the bond dimension behaves for a ground state search with a bounded truncation error @xmath105 and @xmath106 for a calculation in the hf basis and optimised bases obtained by the above scheme starting from the hf orbitals .
it is key to the method proposed that the optimisation of the basis leads to a significant decrease of needed resources already in the first iteration , where after the tenth iterations of basis optimisation the needed bond dimension is more than one order of magnitude smaller than in the unoptimised orbitals . for realistic applications of the scheme intermediate high bond dimensions during the calculation
need to be avoided .
the right panel of fig .
[ fig : results ] shows the relative error in energy reached when performing a calculation with @xmath108 starting in the hf and localised basis . as noted before the localised basis allows for a more efficient approximation of the ground state than the hf basis .
the basis optimisation allows us to further significantly optimise both the hf and the localised basis .
starting from the hf orbitals we obtain a basis for which the relative error in energy drops by one order of magnitude from @xmath112 to @xmath113 . beginning at the localised basis allows us to reduce the relative error in energy from @xmath114 to @xmath115 .
note that the energy in the optimised basis starting from the hf orbitals is slightly worse than the energy obtained in the localised orbitals , reflecting the fact that finding the optimal basis is a hard global optimisation problem .
we have repeated similar calculations for different configurations of the be ring ; at the equilibrium configuration and close to the avoided crossing . in each case
we have been able to find a physical basis allowing for a more efficient approximation of the ground state .
this illustrates that the above scheme can significantly and efficiently optimise a given initial basis .
as the local mode transformation can be added with no increase of the computational cost to an existing two - site dmrg and typically yield already in the first iteration of the basis optimisation a significant improvement of the basis our scheme extends the standard qc two - site dmrg . _ conclusion and perspectives .
_ in this work , we have presented a scheme that adapts the physical basis an mps is formulated in by applying gaussian transformations . incorporating local gaussian transformations into the two - site dmrg algorithm allows us to optimise both the basis and the mps iteratively .
the resulting algorithm successfully optimises the physical basis such that distinctly better approximations of the ground state of a given system by an mps can be identified
. it should be manifest from the description of the method that the same idea is equally applicable to other tensor networks , due to the locality of the transformations .
in particular , _ tree - tensor network _ approaches @xcite can readily be combined with the methods laid out here .
similarly , they are expected to be helpful for 2-d lattice systems @xcite .
in addition the above scheme can be directly combined with recent developments for the time - evolution of mps @xcite in order to obtain a time - evolution with variational physical basis .
our general strategy of combining tensor networks with fermionic transformations complements the recent interesting approach of ref .
@xcite , which is similar in mindset , but where these two components are put together in the opposite order . there
, a matrix - product operator is applied onto a free fermionic wave function .
in contrast to that approach , we here retain efficient contractibility , however . the approach taken in this work
can also be seen as a variational principle that allows to find the optimal _ fermionic tensor network _ in ref .
@xcite , where a fixed fermionic basis change is being made use of . widening the scope , these tools seem also helpful in related approaches making use of a _ big data _ machinery to capture strongly correlated quantum systems . for example , _ compressed sensing ideas _ can help finding localised wannier functions @xcite , which in turn can be made use of in density functional theory @xcite . in conjunction with the tools developed here , a combined approach close to optimally representing fermionic correlated states seems within reach .
_ acknowledgments .
_ we would like to thank g. chan , e. fertitta , v. murg , a. nagy , v. nebendahl , j. rodriguez - laguna , r. schneider , t. szilvsi , and f. verstraete for discussions . j. e. and c. k. acknowledge support from the eu ( siqs , raquel , aqus ) , the erc ( taq ) , the dfg ( ei 519/7 - 1 , crc 183 ) , the templeton foundation , and the studienstiftung des deutschen volkes , l. v. and . l. from the czech science foundation , grant no .
16 - 12052s and the hungarian research fund ( otka ) , grant no .
nn110360 and k100908 .
50ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop *
* , ( ) @noop `` , '' ( ) , @noop * * , ( ) @noop ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop `` , '' ( ) , @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) |
radio emission in active galaxies may hold the key to understanding the formation , growth and evolution of active galactic nuclei and the engines widely believed to power them : supermassive black holes . in his early paper ,
lynden bell ( 1969 ) argued that low - power active galactic nuclei were simply `` old collapsed quasars '' the remnants of what was left after the powerful activity waned and the supermassive black hole was left in a state of lower activity .
however , the problems of early evolutionary scenarios @xcite in explaining the transformation of nuclear features , in particular the differences in emission lines in quasars and radio galaxies@xcite , have favored over time scenarios where the differences between high- and low - power agn are interpreted in terms of the relative orientation of the nuclear engine and the observer @xcite .
however , evolution in radio agn has been known for many years @xcite , and recent high redshift surveys have shown increasing evidence for an evolution in the power of agn , their comoving density , and accretion properties , with a shift in activity from high mass to low mass agn generally referred to as `` agn downsizing '' ( e.g. @xcite ) .
this downsizing in agn activity appear as part of a more general trend in downsizing of galaxy properties over cosmic times .
star - formation and mass assembly appear to have downsized as well @xcite : more massive galaxies were in palace at earlier times and also quenched their star - formation earlier than low - mass galaxies , a fact that suggests a physical co - evolution of all galaxy properties and agn activity as we discussed in paper ii .
this evolution of agn power and features seems at odd with a static scenario typical of orientation models where high- and low - power agn are seen in first approximation as two classes given by different viewing angles .
furthermore , even though orientation clearly affects the properties of agn and is able to account for many observed features of agn @xcite , the generality of such models has been challenged in a number of studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
further investigation is therefore warranted to clarify the interplay between host and agn properties and the role of orientation and evolution in shaping the properties of active nuclei . to this end , we have carried out a study of a local population of narrow - line ( seyfert ii and liners ) , and broad - line ( seyfert i and quasars ) active nuclei drawn from the sloan digital sky survey @xcite second data release ( sdss , 2dr ) . in paper i ( reviglio & helfand 2009a ) we have presented the multiwavelength database for the narrow - lined population that we have constructed for this study .
we discussed the systematics affecting the selection of active galactic nuclei in redshift surveys , quantifying the incompleteness given by the pollution of sdss spectra from light of the host and showing that after correction for such effects , @xmath050% of early - type galaxies show spectroscopic signatures of agn activity , a fact suggesting that most , if not all , massive galaxies must undergo an agn phase in their evolution .
interestingly , we found that , at all host luminosities , emission lines are less pronounced in redder hosts , suggesting that redder galaxies harbor systems with lower ( or less efficient ) accretion .
we further explored this effect in paper ii ( reviglio & helfand 2009b ) , where we showed evidence for an evolution of the spectroscopic signature of agn , with emission lines becoming less and less remarkable in systems with older stellar populations .
this evolution is accompanied by less powerful radio activity and a change in the optical morphology of galaxies which is more substantial in less luminous hosts , in agreement with downsizing scenarios . in this third paper of the series
we explore the spectroscopic and radio properties of the population of narrow - line active galactic nuclei discussed in paper i and compare their properties with a sample of broad - line agn and quasars .
we examine the properties of the radio - emitting agn along with their spectroscopic features and further explore the well - established mismatch between the radio , x - ray and spectroscopic signatures of agn @xcite .
we study the interplay between the radio emission and spectroscopic features and the physical properties of the supermassive black holes powering them .
we also review critically unified models , showing major incongruities with the data analyzed in this paper .
we discuss the implication of these results .
this paper is organized as follows : in [ sample ] we summarize the selection criteria and the main properties of the whole sample , while in [ smbh ] we derive the masses for the supermassive black holes for the agn and quasar populations .
we go on to show ( [ host ] ) that more massive black holes are more likely to power stronger and bigger jets and discuss the role of the environment in such a correlation . in [ undetected ]
we explore the properties of the undetected population of radio sources and show a correlation between the median radio power of the spectroscopically selected population and the mass of the smbh . in
[ compact ] we study the interplay between compactness of radio sources and their photometric , environmental , and spectroscopic properties , showing that the less compact is the radio source , the more unremarkable its lines , lending further support to the idea of a decreasing accretion rate in these systems . in section [ unified ] we compare the expectations of unified models of agn activity based on orientation with our sample and show several major inconsistencies .
we discuss our results in [ discussion_ch4 ] and summarize our conclusions in section [ conclusion_ch4 ] .
we will further analyze here the sample of 151,815 galaxies drawn from the sdss 2dr described in paper i. we refer to that paper for the technical details concerning the sample selection and the cross correlation with other catalogs . in this paper
we will use fully the fir , radio , x - ray , spectroscopic and morphological information we have cataloged on these objects , as well as the density of their environments calculated using the three - dimensional density estimator discussed in paper i and @xcite . of these sources ,
6% are detected in the radio band at 1.4 ghz in either the first @xcite or nvss @xcite survey , @xmath1 69% of which have been classified as agn on the basis of their spectroscopic and radio properties .
the radio luminosity function for these sources is consistent with those from other studies @xcite , as discussed in paper i. in that paper we also showed that by using the first survey s high - resolution images , the agn among these radio - emitting galaxies can be divided in two main categories based on the fwhm size of a gaussian fit to the radio source : point - like ( unresolved ) sources for fwhm@xmath2 and jet - like resolved sources for fwhm@xmath3 .
jet - like sources with typical fanaroff - riley morphologies were further classified as fr1 and fr2 sources @xcite by visually inspecting all the first fields with multiple radio sources associated with an optical galaxy .
[ image_sample_tiles ] in this paper we thus consider four main radio agn morphologies : point sources , jets , fr1s and fr2s , including respectively 1935 , 2015 , 148 , and 174 objects . in figure
[ image_sample_tiles ] we show an example of each class . a total of 1739 galaxies had unclassifiable radio emission , since they were not observed in first .
the median fwhm of a marginally resolved jet in the first survey at the median redshift of the optical survey is @xmath1 4 kpc
jets with physical sizes smaller than this would be indistinguishable from point sources .
the point source class therefore includes intrinsically small , unresolved radio jets . for the whole optical sample we have adopted the spectroscopic classification obtained by @xcite and @xcite , based on an improved version of the standard bpt emission - line ratio diagnostic @xcite .
we also adopted the values of the extinction corrected line - fluxes and the equivalent widths obtained by those authors , as well as the extinction - corrected estimates of the stellar masses of the galaxies in the sample .
line fluxes not corrected for reddening were obtained directly from the sdss database , as were measurements of line widths and information about the velocity dispersion of the bulge - components of the galaxies .
as discussed in paper i , dilution and pollution produced by light from the host galaxy s star - forming regions severely bias the selection of agn in the sdss sample , leading to the misclassification of a significant fraction of agn as passive systems with increasing distance . as shown in paper i , the effect of dilution and pollution mainly affects the number counts of early type galaxies , and preferentially affects galaxies with weak emission lines , typically in redder galaxies .
all results presented here use our sample in which we have statistically corrected for this effect .
the kauffmann et al .
2003 sample does not include broad - line systems , since it was selected to compare the properties of the agn with their host galaxies . in broad emission line systems
the light of the nucleus often outshines the light of the host and no information about the mass , velocity dispersion , and color of the host can be reliably extracted from the sdss data .
however these systems are of great interests in understanding the physical properties of different types of active galaxies .
for this reason we have selected the whole sample of 1412 broad - line agn found in the dr2 release at [email protected] . according to standard classifications of quasars
, we divided our sample into 1278 seyfert is and 134 quasars requiring quasars to have @xmath5 ( schneider et al .
2005 ) , where m@xmath6 is the absolute magnitude of the nucleus in the sdss i - band .
[ narrow_width_distr ] [ broad_width_distr ] in order to analyze the relation between line widths and the features of the population of active galactic nuclei , we have downloaded from the sdss database available line widths for the sources in both the narrow- and the broad - line samples . in figures
[ narrow_width_distr ] and [ broad_width_distr ] we show the distribution of the @xmath7 line width for the narrow - line spectroscopic sample of agn and for the whole broad - line spectroscopic sample ; it is clear that @xmath8 @xmath1 10@xmath9 km s@xmath10 in h@xmath11 neatly divides the two classes of agn , in agreement with the usual notion that a fwhm of 1000 km s@xmath10 divides the broad and the narrow population , since fwhm@xmath12 for a gaussian fit .
a few systems classified as broad - line have either h@xmath13 or h@xmath11 with fwhm@xmath4 1000 km s@xmath10 , which accounts for the few systems at low fwhm in the histograms of figure [ broad_width_distr ] .
a few narrow - line systems show fwhm@xmath0 1000 km s@xmath10 ( fig .
[ narrow_width_distr ] ) : we inspected a few of these systems and found that their spectra do now show significant line - broadening , we conclude that these systems have unreliable estimates of their line widths . for the broad emission line sample we performed the same type of cross correlation that we did for the narrow - line catalog ( cf .
paper i ) with the iras @xcite , rosat @xcite , nvss @xcite , and first @xcite surveys . we found 181 matches with first , 44 with iras and 474 with rosat .
all sources detected in first were also detected in nvss ( and vice - versa ) .
we also visually inspected the radio images and found 13 additional fr2s .
we classified as point - like 106 unresolved sources and as jets 61 resolved sources with first angular sizes @xmath02.5 .
one source has a peculiar large jet - like structure , most likely associated with a precessing jet .
virial masses of supermassive black holes can be computed from reverberation maps @xcite . according to these authors , by assuming that the motion of the gas clouds orbiting the black hole is keplerian , the width of the emission lines is an indication of the velocity dispersion of the gas @xmath14 and the size of the region producing the broad - line emission ( blr ) of the agn would be an estimate of the radius . under such condition @xmath15 .
the size of the broad - line emission region can be determined from reverberation maps and has been shown to correlate with the total nuclear luminosity at 5100 , a fact that allows the estimation of bh masses for large samples of broad - emission line objects .
the masses of supermassive black holes have also been shown to correlate with the blue luminosity of the surrounding hot stellar component @xcite and with the velocity dispersion of the bulge of the host @xcite , providing alternative methods to evaluate smbh masses in these systems .
we are interested in comparing the features of different types of agn with the masses of their supermassive black holes . for this reason we have computed smbh masses following different well - tested methods . for narrow - line objects which typically inhabit spheroids ,
we have determined the black hole mass from the velocity dispersion of the bulge component of the galaxy , @xmath16 , following the revised correlation by @xcite : @xmath17 this relation leads to estimates of @xmath18 for smbhs in the elliptical galaxies in our sample , in agreement with what is found in the studies of individual systems ( e.g. , * ? ? ?
* ) the @xmath19-@xmath8 relation in galaxies is known to be a tighter correlation than the one with the luminosity of the host
. however , not all galaxies harboring agn have measured velocity dispersions . for this reason , and to compare our results with the study of @xcite , we have also estimated the masses of the central black holes from the absolute r - band magnitude of their hosts @xcite @xmath20 in figure [ distribution_bh ] we show the distributions of smbh masses for the whole spectroscopic population of narrow - line agn , corrected for dilution and contamination as discussed in paper i , obtained with the two methods outlined above .
( left ) and @xmath21 for the spectroscopically selected agn . _ ] [ distribution_bh ] the plots show that the two estimates are largely consistent , but the absolute magnitude method produces a long tail at the low - mass end .
the skewness of the distribution suggests that this is not an effect of the large uncertainty in the mass estimate , since this would give a symmetric broadening of the distribution .
one might argue that since an absolute r magnitude can be calculated for all systems in the survey , including bulgeless late - type galaxies , these might account for the low - mass tail .
however , the systems in the low - mass tail are spectroscopically classified as agn and therefore _ must _ harbor a massive black hole .
we note that for systems with @xmath22 the median concentration parameter @xmath23 given by the ratio of @xmath24 ( the radius @xmath24 within which 50% of the light of the galaxy is emitted ) and @xmath25 ( the radius @xmath25 within which 90% of the light of the galaxy is emitted ) is 0.4310 . for comparison , late - type galaxies as defined by @xcite have c=0.447 , while early - type systems have c=0.351 .
this shows that the tail arises because of late - type galaxies with small / unremarkable bulge componenent harboring low - mass smbh .
this is supported by the difference in the @xmath26 parameter given by the arctangent of the ratio of the likelihood that a de - vaucouler s and an exponential disk profile fit the host surface brightness : the median @xmath27 for the @xmath28 is 1.457 , typical of bulgy systems , while the median p for the @xmath29 population is 0.148 , a factor 10 lower and typical of disk galaxies . for the broad - line population , we determined virial bh masses @xmath30 from the h@xmath11 line profile and the total @xmath31 luminosity of the nucleus following the revised relation by @xcite : @xmath32
the light of the agn largely dominates the emission in all the sdss bands for broad- emission - line objects .
therefore the total luminosity @xmath33 of a nucleus can be determined by scaling the sdss r - band specific luminosity @xmath34 for the object to the specific luminosity at 5100 , @xmath35 .
the scaling is done using a power - law spectrum @xmath36 typical for broad - line agn continuum with power spectrum index @xmath37 : @xmath38 .
the best choice of sdss magnitude system for objects in which the point - like nuclear emission dominates the photometric images , such as in quasars and seyfert is , is the point - spread - function magnitude ( psf - mag ) .
therefore r - band specific luminosities have been computed from the sdss r - band psf - magnitudes .
the results do not change significantly if instead of using the r - band luminosity to obtain @xmath35 we use the u - band luminosity .
we also compared our estimates for the smbh masses of seyfert i nuclei and quasars to those obtained by @xcite who used a different sample and a different method of smbh mass estimation ; we found very good agreement . for broad emission - line agn .
_ ] [ distribution_bh_broad ] in figure [ distribution_bh_broad ] we show the distribution of the supermassive black hole masses for the broad - line spectroscopic population of agn .
the distributions of smbh masses for the narrow and broad line populations span the same range of values and peak at fairly similar values , supporting the idea that it is not the mass of the smbh that drives the difference between the two classes . in figure [ narrow_jet_distr ]
we show the distribution in luminosity for 1215 radio jets more luminous than @xmath39 for the narrow- emission - line population , along with the distribution of their central black holes masses .
the distribution in luminosity of the jets falls off rapidly towards higher luminosities , while the distribution in masses of the black - holes peaks at intermediate black hole masses .
( right ) of the final sample of 1215 radio sources classified as jets in the narrow emission line systems .
@xmath40 values were derived from the galaxies r magnitudes _ ] [ narrow_jet_distr ] for comparison , in figure [ narrow_jet_distr_sigma ] we show the distribution in luminosity for a sub - sample of 749 radio jets more luminous than @xmath39 which have black hole masses measured from bulge velocity dispersions @xmath21 , as well as the distribution of their @xmath21 .
we note that the distribution in luminosity and in black hole mass are very similar to those determined using the r - magnitude method .
this demonstrates that the choice of the black hole mass estimator does not significantly affect the statistical properties of the jet sample .
it also shows that the distribution of the jets is not significantly affected by the optical selection of the sample : drawing a sub - sample from the original one with different optical properties does not change the shape of the distribution significantly .
( right ) for a subsample of 749 systems in narrow emission line sample classified as jets for which bulge velocity dispersions were available , in narrow emission line sample . _ ] [ narrow_jet_distr_sigma ] in figure [ broad_jet_distr ] we show the distribution in luminosity of radio jets more luminous than @xmath39 for the broad- emission - line population and the distribution of their central black hole masses .
the figure suggests a bimodal distribution in both radio luminosity and black hole mass , suggesting a link between the radio luminosity and the mass of the smbh .
we will investigate this further in the next section .
[ broad_jet_distr ]
there are primarily two questions we wish to answer : do more luminous galaxies host more luminous radio agn ? and , do more luminous galaxies host bigger jets ?
these questions can not be answered cleanly owing to selection effects arising from the flux - limited nature of the optical and radio surveys , which favors the selection of brighter hosts and brighter radio sources at higher redshifts .
we would be satisfied by answering two other questions : are radio agn with luminosity above a certain threshold @xmath41 more likely to be found in more luminous host galaxies ? and , are radio jets bigger than a certain size @xmath42 more likely to be found in more luminous host galaxies ? in order to answer these questions we used a chain of eight volume - limited samples with widths of 0.5 in absolute r magnitude over the interval @xmath43 .
for each of these classes we determine the fraction of the members of the class which have an agn with _ radio _
luminosity greater than @xmath41 , where @xmath41 is a radio luminosity detectable throughout the whole volume . as above ,
we use @xmath44=@xmath45 @xmath46 ; if an agn is brighter than @xmath41 , it would be detected in first at all redshifts below z=0.2 , the cut - off adopted for our optical sample .
more of these sources would be detected at higher redshifts , because of the geometry of the surveys , since at higher redshift larger volumes are sampled and therefore one has greater chance to select brighter objects .
however , this volume effect is removed by normalizing this number of brighter - than-@xmath41 agn to the number of optical galaxies in the class . for each optical absolute magnitude class
we then determine the median smbh mass and plot the fraction of sources brighter than @xmath44=10@xmath47 @xmath46as a function of this median smbh masses .
the advantage of this method is that one samples several orders of magnitude in luminosity for the host galaxies ( and therefore several orders in smbh masses , given the scaling relation between r - band luminosity of a host galaxy and its smbh mass ) .
the fact that low - luminosity agn are not seen at higher redshift is accounted for by restricting the analysis to the sub - sample of agn that would be observable at all redshifts throughout the survey .
nothing can be inferred for agn with luminosities lower than @xmath41 .
we will extend this analysis to low - luminosity objects in the next section by means of another technique . as a function of baryonic mass of the host . a steady increase up to two order of magnitudes in the fraction of bright radio agn is found .
right : when jet radio sizes are considered , we find that the fractional abundance of large jets strongly increases with increasing mass ( or equivalently luminosity ) of the hosts . _ ] [ jets_fractions_bh ] since we wish to break the degeneracy between physical properties of the host and its environment which results mainly from the density - morphology relation , we have divided our sample in two narrow classes of densities : lower densities ( 0.5@xmath4log(@xmath48)@xmath41.0 ) and higher densities ( 1.0@xmath4log(@xmath48)@xmath41.5 ) .
the results of this analysis are shown in the left column of fig.[jets_fractions_bh ] .
all errors are calculated assuming poisson statistics .
galaxy luminosities are proportional to their stellar masses , so the plots can be interpreted either in terms of host luminosity or its baryonic mass . in bins below @xmath49 ,
the fraction of agn with luminosity above @xmath41 is negligible .
with increasing luminosity this fraction increases by two order of magnitudes . for galaxies with smbh masses higher than @xmath50 more than 15%
have a bright radio agn .
similar trends are found if , instead of considering the whole population of agn , we restrict the analysis just to jet population .
more massive smbhs power stronger radio sources , in agreement with models of jet formation ( e.g. , * ? ? ?
the analysis of jet sizes was carried out in a similar way .
for each of the classes of optical luminosity , we evaluated the fraction of those galaxies with jets larger than a size @xmath42 which would be resolved throughout the whole survey .
we chose s@xmath51=7.5 kpc , for which jets would be resolved up to z=0.2 .
once again we removed the volume effect by normalizing the number of jets larger than @xmath42 in a certain optical luminosity bin by the number of galaxies in that bin . to avoid problems with low - luminosity jets that might have been missed in the first survey because of its flux - density limit
, we restricted our analysis to jets with @xmath44=30 @xmath46 .
we then plotted the fraction of galaxies with jets larger than a size @xmath42 against the median smbh mass for each class .
physical sizes for our jets have been calculated by multiplying the angular size of our source by the angular distance @xmath52 which relates to the luminosity distance @xmath53 as @xmath54 .
the results are shown in the right column of fig.[jets_fractions_bh ] . in bins below @xmath55
the fraction of agn with jet size above @xmath42 is negligible . with increasing luminosity
the fraction increases by two order of magnitudes ; more than 15% of galaxies with masses above 8.5 have a large radio jet .
we can not correct for the orientation of the jets .
however , since jet orientation is not dependent on the host luminosity or the smbh mass , the trends found can not be due to such effects .
more massive smbhs have a much greater chance of powering radio jets of larger size .
a similar increase in the fractional abundance of high - power radio agn and jets with smbh mass is seen in both high and low density environments ; the same is true for the increase in the fractional abundance of larger jets with increasing smbh mass .
this means that , to first approximation , the physical properties of the host determine the size and power of radio agn in radio galaxies , not the environment in which the jet propagates or the effects of interaction with neighboring galaxies or the intergalactic medium .
in particular , the existence of large and powerful jets depends on the mass of the smbh ( or equivalently on the mass of the host galaxy ) .
this is in agreement with simulations of jets propagating in an external medium . @xcite
have followed the propagation of jets in a stratified external medium and concluded that two regimes exist : one phase in which the jet cocoon is highly over - pressured with respect to the external medium and drives a strong shock into the ambient gas , and a second phase in which the cocoon pressure decays and reaches balance with the external value , making the shock become weak .
the separation between two regimes depends essentially on the jet kinetic power ; the transition occurs at larger distances from the galactic nucleus for high kinetic - power jets and at lower distances for low kinetic - power jets ( cf . also ferrari , 2006 ) .
therefore one would expect stronger radio jets ( which generally have high kinetic power as evinced in their high mach numbers ) to have on average larger radio sizes , since it is the first phase that sets the observed size of a radio jet .
re - acceleration of electrons or positrons in a weak shock is much less effective than in strong shocks @xcite ; in addition , the magnetic fields are much lower far from the galaxy , leading to much weaker synchrotron emission in the second phase . in order to determine if the environment is correlated with the existence of larger and more luminous jets we fixed two classes in absolute magnitude ( -22.5@xmath4r@xmath4 - 22.0 and -22.0@xmath4r@xmath4 - 21.5 ) and plotted the fraction of galaxies with powerful ( l@xmath0 10@xmath47 @xmath46 ) and large ( [email protected] kpc ) jets with different environmental density ( figure [ jets_fractions_dens ] ) .
we find a modest increase of nearly a factor two between low- and high- density environments in the likelihood of finding stronger radio sources and bigger radio jets , although this is evident only for the higher luminosity galaxy sample . as a function of density of the host environment .
a mild increase of a factor of two is found for luminous galaxies in denser environments .
right : when jet radio sizes are considered we find that the change of powering large jets is higher in massive galaxies residing in denser environments . _ ] [ jets_fractions_dens ] it appears therefore that even though the role of the environment is limited compared to the role of the mass of the smbh in triggering bigger and more powerful radio jets , its effect is not negligible . since denser galaxy environments are generally associated with denser intergalactic media the pressure of which should tend to confine the jet , not make it bigger , we suggest that the effect of the environment on powering jets may be given by the interactions of the radio galaxy with their neighbors which are more frequent in denser environments . such interactions could disturb the gas in the host , making it easier for it to fall in the inner regions of the galaxy and feed the active nucleus .
if the accretion is higher , stronger radio jets are expected in current models of jet formation ( cf .
meier , 2001 )
only 6% of the galaxies in our sample have detected radio emission : the vast majority of active galaxies are not detected at the current survey sensitivity threshold : either the flux - limited nature of the radio survey is responsible or there may exist a population of active galaxies actually lacking radio emission .
we employ radio image stacking to evaluate the median properties for the whole radio agn population , detected and undetected .
this allows us assess whether more massive galaxies power more powerful radio agn or not , a question that we have left unanswered in the previous section .
we have used the procedure described in white et al .
( 2006 ) for stacking first fields with and without radio emission and evaluate median properties of the resulting radio image . since we are evaluating _ median _ images , including the actual detections does not bias the result ; on the contrary , it allows us to explore the whole radio luminosity distribution of the sample .
it is worth stressing however that the undetected population represents @xmath194% of the sample .
we limit our analysis to the usual chain of volume - limited samples with r between -23.5 and -19.5 for optical completeness and we divide our sample according to its spectroscopic properties . for each class of luminosity
we evaluated the median smbh mass and the median radio luminosity .
median radio luminosities are calculated by fitting 2-d gaussians to the maps obtained by stacking fields in luminosity space .
[ stack_trend_lum_mass ] we find that an increase in the mass of the smbh ( or equivalently the absolute magnitude of the host ) is accompanied by a strong increase in the median luminosity of the radio agn ( cf .
[ stack_trend_lum_mass ] ) .
this trend might even be underestimated since , as shown above , the most massive galaxies tend to have large jets and with the technique used here we are only able to evaluate the central luminosity of the radio agn ; any luminosity in the lobes is missed .
this allows to conclude that more luminous central components of radio agn are associated with more massive smbh : this is in qualitative agreement with expectations of jet production theories ( e.g. , * ? ? ?
this result is in disagreement with the claim by best et al .
2005 that the distribution in radio luminosity of active nuclei is independent of the smbh mass .
galaxies are known to show a variety of radio structures . while some galaxies show point - like radio sources , other show radio jets that might stretch out well beyond the optical host . the mechanism(s ) leading to such differences is not understood .
the alignment of jets postulated by orientation models has been challenged by several studies ( e.g. * ? ? ? * ; * ? ? ?
* ) on the basis of the diversity in environment where these sources are found .
alternative explanation contemplate frustration of the jet by the interstellar medium@xcite or an evolutionary scenario ( e.g. , * ? ? ?
in [ host ] we have shown a strong correlation of jet size with the mass of the host galaxy ; however , more massive galaxies are also bigger . in this section
we compare the size of the jet with the size of the host galaxy , trying to evaluate if the compactness of the jets correlates with host and/or environmental properties .
first , we note that point - like structures in the narrow - line population can be physically small , unresolved jets or larger jets that appear smaller either because of their orientation with respect to the observer or because of their distance , since the physical size resolved throughout the whole survey up to @xmath57 is 7.5 kpc .
we note that point - like sources inhabit preferentially lower density environments than jet - like structures : their respective average environmental density is 2.87 galaxies / mpc@xmath58 and 3.50 galaxies / mpc@xmath58 ; in addition their respective median radio luminosities are different , with point - like sources average luminosity of 10@xmath59 @xmath46and jet average luminosity of 10@xmath60 @xmath46 .
since orientation does not depend on the environmental density and the luminosity of the point - like sources is expected to be higher rather than lower than the luminosity of the jet population if beaming is present , this argues against the hypothesis that orientation effects mostly account for the unresolved population .
as far as the possibility that point - sources are the distant unresolved sub - sample of the standard jet population , we note that this is not the case , since they tend to reside at somewhat lower redshifts than jets : their median redshift is @xmath1 0.11 , while the median redshift of jets is @xmath1 0.14 .
therefore point - like sources must be , for the most part , _ physically _ different from larger jets , as suggested by their lower luminosities . for the sdss sample , a measure of the optical size of the galaxy
is given by the petrosian radius ( blanton et al .
we therefore consider three main classes of objects : point - like ( unresolved ) sources , compact sources with jet size @xmath61 smaller than the r - band petrosian radius @xmath62 of their host and extended jets with @xmath63 .
we expect some degree of contamination in the second class from larger jets oriented close to the line of sight
. we will show that our results are not significantly affected by this problem .
we also expect some degree of contamination in the first class , given by large jets that might become unresolved at the highest distances .
we note however that the high - resolution of the first survey limits this problem significantly . only jet smaller than 7.5 kpc are unresolved at @xmath64 we evaluate the median absolute r magnitude for the three samples for all sources with @xmath65 : @xmath66 for the point - like class median , @xmath67 for the compact class and @xmath68 for the extended class . we find a increasing trend in median luminosity with decreasing compactness of the radio source : @xmath66=-21.98 , @xmath67=-22.20 , @xmath68=-22.41 with median absolute deviations of 0.47 , 0.41 , 0.32 , showing that the spread of the distributions in absolute magnitude for the three classes decreases with decreasing compactness , a likely effect of the contamination of the hosts of compact sources by the hosts of extended sources seen at an angle , which tend to be more luminous and therefore add to the spread in luminosity in these classes . in order to establish
if the decreasing trend in the median luminosity is statistically significant , we chose to apply a chain of wilcoxon ( mann - whitney ) tests to pairs of the median values @xmath66 , @xmath67 , @xmath68 , in order to establish if , given the variance of the distributions , their differences are significant .
first we compared @xmath66 with @xmath67 and found that the probability that the difference in the two median values is not significant is of order @xmath69 , establishing that @xmath70 .
similarly we applied the test to @xmath67 and @xmath68 and established that @xmath71 at the same confidence level .
thus , @xmath72 and the decrease in absolute magnitude ( or @xmath19 ) is significant . in order to ensure that blending of these classes does not significantly affect this conclusion we evaluated the median absolute magnitude for the three classes applying more restrictive cuts
: we required that jets in the compact class have sizes significantly smaller than the petrosian radius of the galaxy ( @xmath73 ) , therefore ensuring a lower chance of mistakenly including large jets oriented along the line of sight in this class ; we also limited in redshift the population of point - like sources by selecting only sources at redshifts @xmath74 , therefore ensuring a lower contamination in this class from larger unresolved jets that would belong to the compact class .
we find @xmath66=-21.45 , @xmath67=-22.10 , @xmath68=-22.41 : the trend is still present , and blending does not appear to introduce a significant bias .
all three classes of objects are found at all optical luminosities , but their median optical luminosities appear to decrease gradually for more and more compact sources .
this also means that the average mass of their smbh decreases with increasing compactness of the radio source .
indeed , we find that the median smbh masses for the three classes are @xmath75 , @xmath76 , and @xmath77 m@xmath78 .
there are no selection effects that would produce correlation between the compactness of an agn and the luminosity of its host .
point - like sources are seen both in high- and low - luminosity systems . for extended sources
there might be some problem with the fact the first survey s high - resolution tends to resolve out part of the flux of extended sources such that the true size of some jets might be underestimated .
this problem however does not preferentially happen in either high or low optical luminosity systems .
we performed the same type of analysis considering the u - r colors of the hosts and found respectively 2.87 , 3.03 and 3.14 mag , and a significant correlation from a chain of wilcoxon ( mann - whitney ) tests , showing that more compact sources are found in bluer systems . if we compare the prominence of the bulges along this sequence we find a small , yet significant increase in the bulge to disk ratio , with p = arctan(@xmath79 ) :
we find respectively 1.478 , 1.505 , and 1.519 .
the mad for the distribution are respectively 0.028 , 0.039 , and 0.070 , and a wilcoxon test confirms the existence of a significant trend .
using the concentration of the light profile , brings to the same conclusion .
this is also in agreement with the fact that the host colors are redder for less compact jets . for the median densities of the environment surrounding these sources ,
we find respectively 2.88 , 3.40 , 5.00 galaxies / mpc@xmath58 , and positive correlation from a chain of wilcoxon ( mann - whitney ) tests , showing that more compact sources are found in less dense environments .
interestingly , we find that our low - luminosity point - like sources are found in lower density environments , similarly to their high - luminosity counterparts studied by @xcite . therefore , when the size of the jet is compared to the size of the host , we find that , on average , less compact jets are found in redder , more luminous , more bulgy , early - type galaxies , which typically reside in denser environments , and that the transition from compact to extended jets is _ smooth _ along this portion of the hubble sequence .
we compared the spectral properties of the three classes of radio compactness .
we selected as usual a small range in absolute r magnitudes ( @xmath80 ) and built a volume - limited sample since we want to compare the spectral properties across a uniform class of hosts .
for each class of compactness previously discussed we evaluated the fraction @xmath81 where @xmath82 is the number of emission - line systems in the class and @xmath83 is the number of passive ( absorption line ) systems .
we assume poisson statistics in evaluating the errors on these fractions .
for the point - like agn we find a ratio of [email protected] , for the compact jet - like agn 1.55 @xmath84 0.09 , and for extended jets 1.25 @xmath84 0.13.the wilcoxon test evaluates this decline as significant at a level of 10@xmath85 .
the fraction of emission line systems increases with the compactness of the radio agn .
the larger the jets the more unlikely are detectable emission - lines in their spectra.we discussed the lack of emission lines in galaxies powering strong jets in paper i and in @xcite .
interestingly compact jets form a class between point - like and large jet sources , and the abundance of emission line systems is more than four times higher than the abundance of absorption - line systems in hosts containing very compact radio sources .
this can be seen more clearly in figure [ compactness_spectro ] where we show the fractional abundances for different spectroscopic types with radio agn of different compactness .
emission - line agn dominate the compact class and then progressively decrease . in the extended class
low - signal - to - noise , weak - line agn dominate together with spectroscopically passive galaxies .
we conclude that the radio and spectroscopic signatures of agn activity in radio galaxies _ anti - correlate _ , with luminous large jets showing less remarkable lines than compact radio sources at all smbh masses .
we note that in paper ii we have shown that a weakening of the lines is correlated with a decrease in star - formation activity in the host galaxy , as well as a change in morphology .
this change is more substantial for less luminous hosts in agreement with evidences for `` downsizing '' in both the mass assembly of early type galaxies @xcite and smbh growth @xcite and can be interpreted as the effect of a progressive exhaustion of the gas present in a galaxy in a secular evolution scenario .
if the gaseous reservoir of a galaxy is progressively transferred towards inner regions at higher rates when the galaxy is young and at lower rates when it becomes older , the accretion rate onto the smbh would vary over time , making the agn activity on average wane over time , in agreement with the findings of a decrease in agn strength in x - ray and radio power from other surveys .
it would also explain the findings by @xcite that point - like radio emission is associated with hosts that have detectable neutral hydrogen , while strong jets seem to lack such gas .
a progressive change in the measured accretion rates of radio galaxies has been reported by @xcite .
a variation of the accretion over time associated with a change in the radio morphology would call for an evolutionary model of agn activity where lines progressively decrease in strength and the radio emission transforms from point - like to jet - like .
anisotropy in the radiation of active nuclei has been well established in the past decades @xcite and most features of agn are currently explained in terms of orientation of the nuclear component with respect of the observer .
the results presented in this paper which link the size of a jet to the mass of the smbh powering it appear problematic for such models , since the smbh mass is a quantity independent of orientation . in this section
we examine the predictions of unified models using our large statistical sample to see how well such models perform .
first we compared the masses of the black holes for radio galaxies showing fr2 sources , selected from our narrow - line sample , and the masses of point - like quasars and found @xmath86 for the point - like radio quasars and @xmath87 @xmath88for fr ii , suggesting that these two types of objects are indeed powered by the same type of black holes , and confirming that quasars are mostly associated with ellipticals @xcite since only these systems have such massive black holes .
this is in agreement with the unified paradigm .
we compared the radio luminosities of the two classes .
if unified by orientation , the two classes must have the same radio luminosity , unless relativistic boosting is important .
the median radio luminosity of point - like quasars in our sample is @xmath89 @xmath46 , while the median radio luminosity of fr2s associated with elliptical galaxies is @xmath90 @xmath46 .
the fact that there is more than an order of magnitude difference in the two populations is inconsistent with the hypothesis that the two classes can be unified by orientation .
boosting of radio emission given by orientation would increase the radio emission of point - like quasars , not decrease it and can not be the cause of such difference .
the fact that these point sources are less luminous than the extended ones might suggest that their jets are intrinsically smaller .
the two populations also differ in the density of their environments , with a density for point - like radio quasars of 2.77 galaxies / mpc@xmath58 and density of the fr2 sources of 4.72 galaxies / mpc@xmath58 .
the environmental density of fr2s is almost twice as great as the density surrounding radio quasars .
if we restrict the analysis to point - like quasars and fr2s with l@xmath91 @xmath46to avoid troubles with the sampling of the luminosity function of the two populations , we find that the average density for the radio - loud point - like quasars is still 2.77 galaxies / mpc@xmath58 while for the fr2 is 4.35 galaxies / mpc@xmath58 , in agreement with the result by @xcite who found that the two population do not inhabit the same environments , using a completely different sample and density estimator .
this is also inconsistent with the unified model since inclination should not depend on environmental density .
we also note that the fact that we find 13 fr2 galaxies associated with broad- emission - line systems is not consistent with the geometric interpretation : broad - line systems should not display large jets , since the jet should be pointing towards the observer .
fr2s are routinely found associated with quasars , at higher redshifts , consistent with our finding , a fact that undermines the assumption that orientation effects can be used to explain the main differences between radio quasars and elliptical radio galaxies .
we also examined the idea that seyfert iis are obscured seyfert i systems from a statistical point of view , since several studies have pointed out inconsistencies with such an interpretation , finding differences in the polarization properties of the seyfert ii nuclei @xcite , differences in the host features @xcite , and differences in the distribution of the [ oiii ] emission @xcite .
the underlying assumption of the unified models is that seyfert i and seyfert ii systems host the same agn , and that the two population are different only because of the different viewing angle with respect to an absorbing dusty torus . in order to make a fair comparison with the seyfert i sample which was selected at redshifts @xmath57 , seyfert ii galaxies used in this analysis were selected in the same redshift interval , using the standard emission - line flux diagnostic @xmath92/h\beta > 3 $ ] for systems with signal - to - noise higher than 3 .
first , we compared the x - ray properties of these systems .
the fraction of x - ray emitting systems among seyfert i is 30% , while for seyfert ii is only 0.4% .
this is consistent with the idea that a very large fraction of the x - rays are obscured in seyfert ii galaxies as suggested by other studies
. this would be in qualitative agreement with the expectations of the unified models , since x - ray emission would be extincted by the obscuring torus .
secondly we compared the balmer decrements for these two types of systems .
the balmer decrement gives an estimate of the extinction for systems .
for agn in the absence of obscuration the ratio should be @xmath1 3.1 @xcite .
we find that the balmer decrement for seyfert is is 3.74 , fairly close to a non extincted system , while for seyfert iis it is 10.5 , clearly showing that seyfert ii are dust - extincted , in agreement with orientation models . in the orientation paradigm ,
a seyfert i luminosity is given by the sum of the luminosity of the nucleus and the host , while in seyfert iis most of the nuclear light is blocked by the absorbing torus .
we calculated the average reddening e(u - r ) expected for these systems from the standard relation between the extinction @xmath93 and median blamer decrement @xmath94 , assuming @xmath95 .
we adopted the standard slope @xmath96 for the extinction curve near the v band , and the conversion between the b and v bands to u and r bands from rieke & lebofsky ( 1985 ) .
we found that the expected extinction e(u - r ) is 1.2 mag .
the extinction coefficients are respectively @xmath97 and @xmath98 .
if the nucleus of a seyfert i is dimmed by obscuration to produce a seyfert ii such that contribution to the optical fluxes from the host becomes predominant , one would expect that the luminosity of seyfert is after subtraction of the part obscured , should be of the order of the luminosity of a typical seyfert ii .
we compared the average absolute magnitude in u and r of the seyfert i sample with the absolute magnitudes of the seyfert ii sample .
the median absolute u magnitude is @xmath99 . for the seyfert ii sample
the median absolute u magnitude is @xmath100 .
the extinction calculated is @xmath101 .
therefore it appears that , on average , seyfert is are not bright enough to account for the extinction expected by their balmer decrements . the situation is worse for the r - band : in this case we find @xmath102 and @xmath103 .
seyfert ii are brighter in r than seyfert i , while for unified theories the luminosity of seyfert is should always be higher than for seyfert iis , since their light is the sum of the host plus the light from the nucleus .
this result is incompatible with orientation and poses a severe problem to orientation models .
dust heated up by high energy photons re - radiates in the fir .
the assumed obscuring torus re - radiates in all directions , and therefore its orientation should not significantly affect the amount of light collected in a low resolution survey such as iras . if the same absorbing medium is present both in broad and narrow emission line systems and they differ only in orientation , approximately the same fraction of fir emitting sources should be found in both classes of objects and the luminosity in the fir should be about the same .
on the contrary , in our sample we found a higher fraction of infrared detections for the seyfert i sample .
the fraction found is [email protected] % , while for seyfert ii is [email protected] % . the difference is not very large , yet statistically significant .
moreover the average luminosity of the fir emitting galaxies among the seyfert i sample is about _ twice _ as high as that found for seyfert iis at 60@xmath104 m , a fact that contradicts the idea that the two classes of systems host the same type of active nucleus .
another quantity that can be compared is the density of the environment for broad and narrow emission - line systems : we find that the two populations inhabit similar environments , with a slight preference for seyfert ii to reside in denser environment : the median density for seyfert i is [email protected] galaxy / mpc@xmath58 , while for seyfert ii is 2.59@xmath84 0.02 galaxy / mpc@xmath58 , in agreement with early results by @xcite .
orientation does not depend on the environmental density , so this might be regarded as a potential further problem , although the difference in environmental density is not as strongly marked as for the radio quasars vs radio - galaxy sample .
we evaluated the median black hole mass for the seyfert i population and obtained @xmath105 .
for the seyfert ii population the values obtained is @xmath106 , where we used the @xmath8 estimate since its spread is significantly lower than the black hole r estimate .
the errors bars represent the maximum and minimum error associated with the method estimation . using a wilcoxon test , we find that the probability that the two median values are consistent with each other is @xmath107 . under the assumption that the two black hole estimates are consistent with each other this
would imply that seyfert i host somewhat smaller bh masses than seyfert ii .
this would be consistent with the fact they reside in slightly denser environments , since galaxies tend to have bigger bulges and therefore more massive smbh in denser environments . for radio emitting seyferts we compared the fraction of the jet to point - like sources .
if the jets of seyferts are oriented perpendicular to the obscuring torus one would expect that for the seyfert i population the fraction should be smaller than for seyfert ii , since the former should have the jet pointing towards the observer and therefore more sources among them should be unresolved . on the contrary we find a similar ratio of jet to unresolved sources in both classes : 1.21 for seyfert i and 1.26 for seyfert ii .
moreover the average jet size for seyfert i is larger than for seyfert ii : the median jet size for seyfert i is 8.66 kpc while for seyfert ii is 6.73 kpc .
this is also problematic for the orientation paradigm , if jets are mostly perpendicular to the obscuring tori .
we also confirm the difference in [ oiii ] luminosity distribution for seyfert i and seyfert ii sources discussed by @xcite .
the distributions found for different spectroscopic classes are shown in figure [ oiii_distrib ] .
the median @xmath108}$ ] for seyfert i in our sample is @xmath109 erg s@xmath10 , while for seyfert ii is more than one order of magnitude lower : @xmath110 erg s@xmath10 .
it must be noted that @xmath108}$ ] is supposed to be an isotropic indicator not affected by the torus obscuration ( unlike @xmath111 ) since its emission comes from the narrow - line region which is at _ larger _ radii than the region where the obscuring torus is supposed to lie in unified models ( if the torus should enshroud the narrow line region as well , then the opening angle for the radiation to emerge from the nucleus would be so large that the inclination of the system with respect to the observer would not make a significant difference ) .
therefore a difference in [ oiii ] luminosities between seyfert is and iis further support the idea that these systems are physically different , with seyfert i galaxies intrinsically more powerful sources than seyfert iis .
luminosities for different classes of spectroscopic objects _ ] this difference in the strength of [ oiii ] between broad and narrow - line nuclei was noted by @xcite .
@xcite , however , argued that [ oiii ] might be contaminated by nuclear star - formation and anisotropic obscuration and that [ oii ] would be a better indicator , claiming to show a lack in substantial difference in [ oii ] luminosity between the broad and the narrow line population . in figure [ oii_distrib ]
we show that there exists an even more marked difference in the luminosity of the [ oii ] between seyfert i and seyfert ii , inconsistent with the earlier claim .
interestingly , unlike all the other lines , we find that [ oii ] is stronger in seyfert iis than seyfert is .
[ oii ] has now been shown to be strongly correlated with star - formation @xcite , not agn activity .
it appears more likely that the difference in [ oii ] in seyfert ii and seyfert i is given by a different amount of nuclear star - formation , or by contamination by the the galaxy star - forming regions , which according to the study of @xcite are more frequent in seyfert iis than in seyfert is .
luminosities for different classes of spectroscopic objects _ ] another line that can be used to confirm that different classes of agn differ in the strength of their forbidden lines is [ nii ] . the distribution of the line luminosity for our sample is shown in fig .
[ nii_distrib ] .
a trend similar to the one found for [ oiii ] is found .
interestingly the trends found for the forbidden lines are similar to the ones found for the permitted lines : in figure [ ha_distrib ] we show the trend for h@xmath13 .
this suggests that strength of forbidden and permitted lines varies in a similar way across these spectroscopic classes .
luminosities for different classes of spectroscopic objects _ ] [ nii_distrib ] in unified models , seyfert is and quasars are supposed to be sources observed close to face on .
when compared to narrow - lined agn they would exhibit strong x - ray emission , point - like radio emission , bluer colors , and less extinction .
however , how do these two classes of putatively face - on agn compare to each other ?
the ratio of point - like sources to jet - like sources is 2.6 for quasars and 0.8 for seyfert i. unresolved radio sources dominate the quasars class , but not the seyfert i class .
one possible effect might be given by the fact that quasars , being hosted by more massive ellipticals are selected at higher redshift than seyfert i and their potential jet may be more likely unresolved .
if we restrict our sample to to the same redshift range 0.1@[email protected] , the results are unchanged : we still find a stronger preference for point - like radio sources in quasars than with seyfert i. the other possibility , that stronger jets are powered by smaller smbhs , is at odds with what we find in section [ host ] .
luminosities for different classes of spectroscopic objects _ ] [ ha_distrib ] if seyfert is were agn seen face on like quasars , one would expect a similar fractional amount of resolved to unresolved radio sources .
one might argue that seyfert i may be more inclined systems than quasars . in this case
one would expect their extinction to be higher . on the contrary
the balmer ratio for seyfert is is 3.8 , while for quasars it is 4.8 .
nearby quasars appear to be more extincted than seyfert is . while the data are consistent with the standard picture in which seyfert ii are dust - extincted systems while seyfert i are not , we find many inconsistencies with the assumption that orientation _ only _ can explain the differences between the two classes .
likewise , several incongruities are found in our data for the paradigm which suggests that nuclei of radio galaxies hosting strong jets are radio quasars seen at a larger angle .
in this study we have demonstrated a smooth variation in the the radio and spectroscopic properties of broad- and narrow - line agn with increasing host galaxy luminosity ( or alternatively baryonic mass ) . in particular
, we showed that , with increasing host luminosity , radio morphologies become more extended and the average radio luminosities increase ; the agn emission lines , however , become less remarkable and the extinction increases .
these trends are accompanied by a variation in the properties of the host from late- to early - type , with star - formation quenching at earlier times and colors becoming redder as we discussed in greater detail in paper ii .
the physical properties of galaxies and their nuclei thus appear to be well - correlated , suggesting a closer link between the properties of the nuclei of galaxies and their hosts , beyond the well - known bulge - smbh relation @xcite .
if late- and early - type galaxies are systems in different stages of their evolution ( or transformation ) these trends may trace the evolution of active nuclei , from point - like systems with strong line emission to jet - like systems with low or unremarkable line emission . as we move to higher redshift
, the properties of agn associated with massive smbhs do become more similar to the one observed in later types galaxies in the nearby universe : their lines become more prominent , they become less extincted , their radio emission is predominantly point - like and their host become bluer @xcite .
however , in unified models , these differences are mostly interpreted in terms of orientation , not evolution . in this paper
however , we question the generality of such models , which do not seem to meet some simple statistical tests , as already pointed out by other authors .
the increasing evidence for downsizing in many galactic features , from star - formation to smbh accretion requires models of agn activity that link the properties of the nuclei to the properties of their host galaxies .
if downsizing is simply the consequence of a different rate in the evolution of systems of different mass ( with the more massive ones evolving faster ) then agn activity would peak at different epochs for different smbh masses as observed @xcite , and the smbh assembly would follow the spheroid assembly as shown by @xcite , @xcite ( paper ii ) , and @xcite .
given the well - known density - morphology correlation , this evolution would be accompanied by a change in the typical regions where different types of agn are found at different redshifts . if the quasar phase is an early phase of nuclear activity , when a galaxy hosting a big smbh is richer in gas , then more massive galaxies , which inhabit denser environments , would undergo such a phase at earlier times and quasars clustering would appear to increase with increasing redshift , as observed @xcite . furthermore ,
as we discussed in paper ii , the star - formation rates are expected to progressively change as the host galaxy becomes depleted of cold gas , either because it is transported into the inner regions by secular instabilities or because it is heated up by the central engine @xcite . if this process happens with higher speed in bigger galaxies as suggested by studies of their stellar populations , then downsizing in star - formation @xcite
would be expected to accompany the evolution in nuclear features .
considering that high - redshift surveys are showing increasing proofs of downsizing in all galaxy features , progressive quenching of star - formation , mass assembly , agn features and that the strength of active nuclei is well - established to evolve over time , further investigation of evolutionary scenarios of agn activity @xcite is warranted .
assuming that evolution is the main factor behind the photometric , morphological and spectroscopic differences in active nuclei as discussed in the previous chapters , we here discuss a speculative model of agn activity based on spherical accretion , where the progressive depletion of cold gas in the host galaxy leads to a progressive transformation of the nuclei .
this transformation would be in agreement with the evidence for co - evolution of agn and host galaxy features presented in the previous chapters and the results from high - redshift surveys .
orientation effects can be expected in this paradigm , as we will discuss . for semplicity
, we will consider accretion from a homogeneous gaseous reservoir that has lost most of its angular momentum .
in such a case , the accretion is spherical .
recent evidence in agreement with spherical accretion in early type galaxies has been presented by @xcite .
we notice that the main feature of spherical accretion is a dependence of the accretion rates on the density of the outer regions of the gaseous reservoir from which the smbh is accreting ( bondi 1952 ) . in this way
, it links the properties of host with the accretion properties of the nuclei , the main point we illustrated in the previous sections .
furthermore in a spherically accreting system , a gradient in temperature and density of the gas is formed as the gas accumulates in the inner regions .
the structure of an active galactic nucleus would resemble that of the interior of a protostar , with shells of different density and temperature contracting under the influence of gravity until a stable configuration is reached .
the density profile of a spherically accreting system is typically a power law , @xmath112 , with @xmath113 ( larson 1969 ) . for smbh of @xmath114 @xmath88the
gas may heat up to temperatures @xmath115k in the inner shells , and lower temperatures at outer radii .
@xcite for a review of possible solutions ) .
the sustainability of such gradients depends on the accretion rates and the thermodynamic properties of the gas .
if the radiative cooling and the heating from accretion onto a smbh and/or a shock @xcite propagating in the medium are in equilibrium , a hot temperature in the inner shells can be sustained over time .
as we will show , such a hot component , if sustained , can easily account for the high luminosity of the nuclei since it is a strong emitter of bremstrahalung radiation , independent from the fact that spherical accretion efficiencies are low .
@xcite have pointed out that if gas surrounding a smbh intercepts 0.5% of the radiation , its thermodynamic equilibrium would be significantly altered : if radiative heating from the accretion is effective , a hot phase of the gas surrounding the smbh can be long - lived .
quasars are associated to deep potential wells given their high smbh masses and are very strong radiators : the expectation of a hot , turbulent , dense phase of the gas surrounding their smbh seems physically more plausible than an ensamble of small , warm ( @xmath116 k ) long - lived clouds nicely orbiting the smbh in keplerian orbits , with low filling factor as postulated in unified scenarios . radiative heating can sustain high temperatures in the surrounding regions of smbh , well beyond the assumed @xmath117 pc scale of the accretion disk .
for example , if the inner regions are fed by a cooling flow from the outer region of a galaxy @xcite , an increasing trend in temperature of the gas from outer regions ( 100 - 1000 pc ) down to radii of @xmath11 pc is found . at 1 pc ,
where their simulation stops , the temperature reaches a temperature t@xmath118 k. in this picture three main regions surround a smbh : a region filled by hot , dense gas , a region filled by gas of medium density and medium temperatures and a region filled by low density cool gas .
the relative size of these three regions depends on the hydrodynamic ( and possibly magnetohydrodynamical ) and thermodynamical evolution of the region .
as we will discuss in detail we identify these three regions as the three main components of the standard agn picture : the broad line region ( blr ) , the warm absorber and the narrow line region . the main radiative process in a hot
, dense plasma is thermal bremsstrahlung .
the maximum total emission from bremsstrahlung expected in a hot plasma per unit volume in the absence of absorption is @xmath119 erg s@xmath10 @xmath120 , where t is the temperature of the plasma , @xmath121 is the number density of the ions , @xmath122 is the number density of the electrons , and g is the gaunt factor which is approximately equal to 1 . for a neutral plasma @xmath123 . for scale values typical of a blr ( @xmath124 ions / cm@xmath58 , r@xmath1 0.1 pc ) and an average temperature typical of the hot phase ( @xmath126 k ) ,
one obtains that the bolometric emission from a region of volume @xmath127 may account for luminosities as high as @xmath128 erg s@xmath10 .
this shows that this picture has the energetics to account for the high bolometric luminosities of quasars ( @xmath129 erg s@xmath10 ) , even in the presence of substantial absorption by the surrounding medium .
furthermore the flatness of the bremsstrahlung spectrum would be in agreement with the typical flatness of agn spectrum across several bands .
the physics of dense plasmas at the high temperatures predicted by spherical accretion models is poorly understood .
however , we note that for the temperatures expected in the inner regions , one expects thermal broadening of the recombination lines to be substantial . neglecting relativistic corrections , in first approximation the dispersion of the lines and the temperature
are related by @xmath130 $ ] .
fwhm can be obtained by multiplying by @xmath131 . for temperatures of @xmath132
k one would obtain @xmath133 typical of permitted lines of quasars . in models of spherical accretion , smaller smbh black holes would have lower accretion rates and therefore are expected to be able to sustain lower temperatures in their inner regions . in agreement with this one finds that for sigma of @xmath134 km s@xmath10 typical of seyferts , the temperature of the gas must reach @xmath135 k , at least an order of magnitude lower than quasars .
we note that a model where the inner regions host a hot , dense plasma may help account for the variability observed in such nuclei .
variability in quasars is on timescales of light - years , while only light week or days in seyferts ( barvainis 1993 ) .
this would strengthen the point that the emitting region of quasars is on scales @xmath136 while for seyfert is at least an order of magnitude smaller , in agreement with the expectation that these nuclei can only sustain high temperatures in smaller volumes given their lower smbh masses .
the cooling timescale for bremsstrahlung can be estimated as @xmath137 years for n@xmath138 @xmath120 and temperatures of @xmath139k one obtains timescales of order of few hours to days .
this can simply account for the observed variability of the emission across the whole spectrum from radio to gamma ray for quasars and blazars on timescales of hours to days : if the temperature of the gas is not smoothly sustained over timescales of hours by the central engine , the observed power of the source changes significantly . if , on the contrary , the temperatures fall below @xmath140 or less , the broad component of emission lines would disappear , since the thermal broadening of the lines would now be on the order of narrow - line broadening ; however , the bremsstrahlung continuum would still be strong since it scales with @xmath141 .
the bolometric luminosity of a hot gas sphere at t@xmath142k with n@xmath143 and r@xmath144 would be just one order of magnitude lower than the luminosity of a t@xmath145 .
therefore one expects a hot region with such characteristics to produce a lower luminosity quasar with narrow lines .
given the lower temperatures in the gas surrounding the smbh , dust survival is favored as we will discuss , and the quasar may appear more extincted .
halpern ( 1981 ) has suggested the existence of a warm absorber ( t@xmath146 ) in the region in between the broad and the narrow line region of agn which would account for the absorption features observed in the x - ray part of the spectrum . in this model
the existence of a warm absorber may be simply explained , since the structure of the gas is stratified in density and temperatures . at larger radii ,
the temperature and the density of the gas drops significantly and in models of spherical accretion , a region of lower density with temperatures of @xmath147 k surrounds the hot region .
regarding this region , we note that barvanis ( 1993 ) has summarized the evidence in favor of a warm phase of the plasma , the bremsstrahlung emission of which would account for the uv bump observed in quasars spectra .
reynolds et al .
( 1997)have discussed the presence of dust in the warm absorber , under the assumption that dust grains decouple from the plasma temperature and are kept at lower temperatures .
dust would be in any case present in the outer regions where the temperature falls below the sublimation temperature of dust , which is a few thousands kelvins .
such dust can absorb the uv radiation produced in this shell and re - emit it in the infrared , creating the infrared bump observed in spectra .
alternatively , we have suggested that comptonization by the warm medium of low energy photons coming from the inner region may be responsible for the production of the uv bump .
interestingly @xcite have discussed the fact that the uv bump is _ not _ present in all quasars spectra , which suggests that it is _ not _ a features given by accretion itself , since in such event should be present in all spectra . in the scenario that we are considering ,
the presence or absence of the warm phase would be responsible for the presence or absence of the uv bump .
this region may be absent , and the uv bump not observed : krolik et al .
1981 have shown that a t@xmath148 k phase of the gas surrounding a quasars is unstable , and therefore it is likely that is short lived .
incidently this has the implication that thermal widths of order few hundreds , which would be associated to the warm phase of the gas are not easily observed since they are short lived . in such case one would expect a lack of systems with fwhm of few hundreds , as observed in the bi - modal distribution of the permitted line widths among active nuclei , ( cf .
figure [ narrow_width_distr ] and [ broad_width_distr ] ) . in this picture at radii larger than 1pc the gas may be too cold for the lines to be thermally broadened . at @xmath116 the @xmath8 of the line
is of order 10 km / s , while the observed size is @xmath149 km / s .
the broadening must come from the gas random motion and possibly infall .
this region would have densities below @xmath150 and therefore permitted and forbidden lines would originate in this region .
we associate this region to the narrow line region of current models .
the low densities allow for forbidden lines to form , while in the inner regions , the high density would not allow the associated atomic transitions .
if the size of the hot gas becomes smaller and the output of high energy photons drops , then one would expect lower [ oiii ] in low luminosity nuclei such as seyfer ii and radio galaxies than in their high luminosity counterparts , as found in the sample analyzed in this paper and others @xcite . since the temperatures for dust sublimation is of order @xmath151
k , dust can survive in this region , possibly even in the warmer zone , if it decouples from the hydrogen plasma and its temperature falls below @xmath151 k. it is on the contrary unlikely that dust would survive in the hot phase of the plasma surrounding the central engine .
therefore systems with broad emission - lines would show lower levels of extinction , as observed , since dust molecules surrounding the host phase are spread over a large volume at large distances from the nucleus . at later times , as the region shrinks and the material in the outer regions falls in and accumulates keeping lower temperatures , the dust density increases and the nucleus becomes dust enshrouded .
its lines become more extincted . in association to this the broad line component disappears , because of the lack of a substantial fraction of hot dense medium , and possibly because of the enhanced extinction .
if the gas in this region keeps some angular momentum , the geometry of this region might be ellipsoidal and the dust distribution would be toroidal . in this way the anisotropic features observed in the radiation of agn may be recovered and a contribution by orientation to their features would be expected . if a fraction of the light from the inner hot shells is scattered into the line of sight , the results of spectrophotopolarimetry @xcite may be recovered . the hidden `` type 1 '' nucleus seen in polarized light , would be given by the radiation by the innermost shells scattered into the line of sight .
this way the contribution of orientation to the features of the agn typical of unified model can be incorporated in the model .
the effect of orientation would be more significant in systems where the gas has higher angular momentum , i.e. later types galaxies .
seyfert galaxies , which typically are hosted by later type galaxies , would show more significant orientation - driven features , as observations suggest . in summary
the basic idea is that it is not the efficient conversion of matter into light what produces the agn luminosity , but the presence of relatively large volumes of hot gas in the inner 0.1 pc .
if the galaxy becomes gas poor , however , the accretion rates fall and the hot phase might not be sustained : this would bring the hot shell to be progressively confined to smaller radii and eventually disappear , making the agn fade and its broad emission line disappear . at later times , since cool gas is more likely to infall than warmer gas , the low - power agn phase may be followed by a new cycle of activity if the accretion is high enough to bring the surrounding gas to high temperatures again .
this may eventually trigger another cycle of agn activity , until the density of the gas in the outskirt is so low that the infall of material is unable to sustain a hot dense medium in the surroundings of the agn . in such case
the nucleus would still be accreting , but it would not shine .
this may help explain the existence of nuclei in elliptical galaxies which show a large amount of gas in their nuclear regions , and yet they do not power an agn ( cfr . * ? ? ?
a fraction of about 10% quasars and agn show radio emission .
we note that a stratified medium in density and temperature may explain many salient features of the radio emission in the nuclei of galaxies .
the need for a stratified medium in the inner regions of galaxies has been suggested by bicknell et al .
1997 to explain the features of compact steep spectrum ( css ) radio sources and gigahertz - peaked spectrum ( gps ) radio sources .
the main difference in these two classes is their linear size @xmath152 and the peak of their radio emission @xmath153 .
these two have been shown to correlate with law @xmath154 ( cfr .
odea 1998 ) , with more extended sources peaking at lower frequencies .
bicknell et al . ( 1997 ) showed that the properties of gps and css sources can be described by a model involving interaction of a jet with a dense interstellar medium with density described by a power law @xmath155 with @xmath156 and density of 10100 atoms @xmath120 at distances of 1 kpc .
this distribution in density would give densities above @xmath157 atoms @xmath158 in the inner 0.1 pc , in agreement with the model we discussed .
the shell structure of the ism they postulated is also in agreement with the one we would expect in our scenario . in the model discussed by bicknell et al .
( 1997 ) , the interaction of the jet and the dense interstellar medium produces a bow shock , which defines the size of the sources . as the shock front proceeds into the interstellar medium ,
the free - free self - absorption of the radio emission varies , being more effective when the jet traverses the denser shells , since the optical depth of the bremsstrahlung self - absorption scales as @xmath159 . since quasars are known to power mostly point - like sources , and yet harbor massive smbh , at the high end of the distribution in mass of smbh in galaxies , this leads to the question of why these systems most of the time do not show remarkable jets , but mostly unresolved or parsec - scale jets .
it also leads to the question of why these systems sometimes do power double lobed radio jets ( de vries et al .
2007 ) .
we note that a stratified medium as the one suggested by bicknell et al .
( 1997 ) may help explain this , if we consider that in the inner regions , high temperatures are sustained .
a shell of hot , dense gas is much more effective in frustrating a jet than cold dense gas , since pressure is proportional to the temperature . if we consider the model adopted by bicknell et al .
( 1997 ) , a jet is treated as an ellipsoidal cocoon with pressure @xmath160 expanding in a surrounding medium described by density profile @xmath161 .
the jet has a energy flux @xmath162 .
the work that a cocoon can do on the external medium is @xmath163 . for @xmath164
this becomes @xmath165 and the work is proportional to the energy flux of the jet .
the typical jet energy fluxes estimated by bicknell et al .
( 1997 ) are @xmath166 erg s@xmath10 and are in agreement with evalutation in other studies ( e.g. , celotti et al . 1993 ) . in the picture we are discussing , the early stages the cocoon expands in a hot and dense medium with mean temperature @xmath167 ( or higher ) and density @xmath168 @xmath120 . assuming that the jet is relativistic , the expansion timescale for it to emerge from the hot zone is of order @xmath169 .
the work that it needs to do to win the pressure of the external medium is of order @xmath170 , where @xmath171 is volume of the cocoon .
the external pressure from the hot plasma is @xmath172 dyne @xmath173 .
the work needed to win the external pressure would be @xmath174 . for @xmath175 ,
i.e. a cocoon with major axis twice as long as the minor axis , and a jet that expends up to the boundaries of the hot zone , the required work is of order @xmath176 erg s@xmath10 this means that a typical jet with flux energy of order @xmath177 erg s@xmath10 , would not have enough energy to break through the inner hot zone and will be confined and disrupted inside it .
the radio source would appear point - like or barely resolved .
if a parsec - scale jet is powered , it may be substantially decelerated by the motion through the hot gas zone : this may provide a mechanism to explain the observed jet deceleration in blac blazar @xcite , for example .
furthermore , only highly relativistic jets would be able to emerge from the inner regions , suggesting that , if observed , jets associated to quasars and blazars must be highly relativistic , since jets with lower kinetic energies would not be able to win the pressure of the hot region shell and would not be launched .
this selection may explain the high lorentz factors of jets associated to quasars and blazars .
a more complete discussion of this model can be found in reviglio ( 2008 ) .
a detailed modeling of activity in such scenario is beyond the scope of this paper .
we note however that in the scenario described , the results differences in [ oiii ] , [ oii ] , [ nii ] shown in section [ unified ] , the anticorrelation between jet displacement and strength of the lines and the correlation between the properties of the host and the properties of the nuclei shown in this paper may find a simpler explanation .
we have presented a comparative study of a local population of more than 150000 narrow- and broad - lined active galactic nuclei drawn from the sloan digital sky survey ( 2dr ) and their radio counterparts drawn from the first and nvss surveys .
after classification of radio morphology for more than 4000 sources , we have shown that more massive central black holes are more likely associated with more powerful and larger jets and that this correlation between radio power and supermassive black hole mass is found also in the fainter population of radio agn which is not detected at the current sensitivity limit of all - sky radio surveys .
we have compared this effect in different region of the large scale structure and found that this correlation is primarily driven by the host mass ( and associated smbh mass ) , not by the large scale environment in which the host galaxy reside .
this supports the idea that the average size of a jet is determined by the strength of its kinetic power and not by the environment in which it propagates , in agreement with results from magneto - hydrodynamic simulations of agn .
a small , yet significant increase in the fractional amount of jets in denser environment is nonetheless found , which suggests that close interactions in denser environments may be capable of powering stronger radio activity .
we have compared the strength of the radio emission in our agn with the strength of their lines and found that the more prominent the jet , the weaker the lines .
we interpret this as an effect of gas depletion . in support of this idea , we find that more compact jets are associated to galaxies with more recent episodes of star formation and therefore richer in cold gas , while large extended jets tend to be associated to systems with little ongoing star - formation and most likely gas depleted .
this is in agreement with the study of emonts et al .
2007 , who find that point - like unresolved radio sources are associated to galaxies richer in hi , while fr i sources do not show detection in hi . assuming a simple spherical accretion scenario with gas brought into the inner regions by the secular evolution of the galaxy
, the anti - correlation between the spectral signature strength and the radio signature strength may be explained if we assume that more massive smbh are progressively brought to starvation by depletion of cold gas in the host : as the density of cold gas in the host drops , the smbh would accrete at lower rates . from this , it follows that the accretion rates would be lower , the emission of the nucleus less powerful , the ionized regions surrounding the smbh thinner , and consequently the emission lines weaker . in the absence of a significant amount of dense gas
, radio jets would be less likely contained or decelerated by the interstellar medium in their inner regions and therefore may reach larger extents .
these findings point towards a physical link between the properties of the spectroscopic lines and the radio morphology of radio agn , and their hosts , a fact that might question the assumption of the unified models based on orientation . moreover , the trends in spectroscopic and radio features of nearby active nuclei appear to form a continuum with increasing galaxy mass .
this seems hard to reconcile with scenarios where most of the features of agn are interpreted in terms of various degrees of extinction given by the orientation of the nucleus with respect to the observer .
we have tested the expectation of unified models based on orientation with the statistical properties of our large sample of agn .
we showed several incongruities with the expectations of such models , which suggest that the differences of broad- and narrow - line nuclei can not simply be an effect of orientation , in agreement with studies of other authors .
in particular we find that narrow and broad - line nuclei selected in the same volume of the universe show distribution of forbidden line luminosities ( [ oiii ] , [ oii],[nii ] ) markedly different.this is at odds with the expectations of a simple orientation models , since the forbidden line emission comes from regions at much larger radii than the obscuring torus assumed in such models and therefore should be similar in broad- and narrow - line agn .
a revision of such models may be necessary .
we have presented an alternative model of agn activity which may account for the evolution of active nuclei over cosmic times and link the properties of a host galaxy with the properties of its nucleus . in the model discussed , based on a spherical accretion scenario ,
the features of high- and low - power agn are explained in terms of the emission and absorption properties of shells of gas surrounding a supermassive black hole . unlike current models of agn
, the strong emission of the agn is not given by the accretion itself , which in spherical accretion is known to be inefficient , but by the hot ( @xmath178 ) shell of gas in the inner region of the gaseous reservoir surrounding a smbh , at distances @xmath179 0.1 pc , coincident with the standard broad - line region of quasars and seyfert i nuclei .
if radiative heating is efficient @xcite , and the accretion rates substantial , high temperatures can be sustained in the inner shells over timescales longer than @xmath180 yr .
we have shown that this hot component is an efficient emitter of bremsstrahlung radiation which , after reprocessing from the medium in which propagates , can account for many observed features of active nuclei spectra , including the broad line emission , the variability on small time scales , the uv and infrared bump . outer shells would account for the other components of the standard agn picture : the warm absorber and the narrow - line region . in these regions ,
the temperature of the gas is much lower than in the innermost shells and dust can survive , and eventually be transfered into the inner regions .
if the hot region is confined to smaller radii over time , dust progressively enshroud the nucleus , making low - power agn more extincted , as observed .
if this dust component retains part of its angular momentum , its geometry may be ellipsoidal / toroidal , and geometric effects may be incorporated in the model .
unlike unified models , where the stability of such component has been long questioned , in this model the dust component is just part of a continuum distribution of gas around the smbh , and can be stable .
as the galaxy becomes depleted of cold gas , the accretion rates drop and the nuclei are brought to starvation .
high temperatures may not be supported anymore and the agn would eventually fade over time .
this transformation would be accompanied by the disappearence of the broad component of permitted lines , and a progressive increase in the dust extinction .
we noticed that a hot , dense core may help explain the radio morphologies of active nuclei , with point - like sources associated to gas rich systems and jets to gas depleted ones , since a hot core may efficiently contain and disrupt jets . in this scenario
, active nuclei would evolve from high to low power , transforming their spectral features and their radio morphologies over cosmic times , in agreement with an evolutionary path from quasars to radio galaxies envisioned by lynden - bell ( 1969 )
we wish to thank jacqueline van gorkom , bob becker , laura kay , ben johnson for helpful discussions .
funding for sloan digital sky survey project has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss website is http://www.sdss.org .
the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions ( the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , the korean scientist group , los alamos national laboratory , the max - planck - institute for astronomy , the max - planck - institute for astrophysics , new mexico state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington ) .
allen , s. w. , dunn , r. j. h. , fabian , a. c. , taylor , g. b. , & reynolds , c. s. 2006 , , 372 , 21 antonucci , r. 1993 , , 31 , 473 baldwin , j. a. , phillips , m. m. , & terlevich , r. 1981 , , 93 , 5 bardeen , j. m. 1970 , , 226 , 64 baum , s. a. , zirbel , e. l. , & odea , c. p. 1995
, , 451 , 88 becker , r. h. , white , r. l. , & helfand , d. j. 1995 , , 450 , 559 blandford , r. d. , & payne , d. g. 1982 , , 199 , 883 blandford , r. d. , & rees , m. j. 1974 , , 169 , 395 blandford , r. d. , & znajek , r. l. 1977 , , 179 , 433 brinchmann , j. , charlot , s. , white , s. d. m. , tremonti , c. , kauffmann , g. , heckman , t. , & brinkmann , j. 2004 , , 351 , 115 bruzual a. , g. 1983 , , 273 , 105 1 carter , b.j . , fabricant , d.g . , geller , m.j . , & kurtz , m.j.,2001,,559,606 celotti , a. , & fabian , a. c. 1993 , , 264 , 228 celotti , a. 2005 , , 300 , 23 chiaberge , m. , capetti , a. , & celotti , a. 2002 , , 394 , 791 cimatti , a. , daddi , e. , & renzini , a. 2006 , , 453 , l29 ciotti , l. , & ostriker , j. p. 2007
, , 665 , 1038 condon , j. j. , cotton , w. d. , greisen , e. w. , yin , q. f. , perley , r. a. , taylor , g. b. , & broderick , j. j. 1998 , , 115 , 1693 cowie , l. l. , songaila , a. , hu , e. m. , & cohen , j. g. 1996 , , 112 , 839 cowie , l. l. , barger , a. j. , bautz , m. w. , brandt , w. n. , & garmire , g. p. 2003
, , 584 , l57 dunlop , j. s. , & peacock , j. a. 1990 , , 247 , 19 dunlop , j. s. , mclure , r. j. , kukula , m. j. , baum , s. a. , odea , c. p. , & hughes , d. h. 2003 , , 340 , 1095 emonts , b. h. c. , morganti , r. , oosterloo , t. a. , van der hulst , j. m. , van moorsel , g. , & tadhunter , c. n. 2007 , , 464 , l1 fanaroff , b. l. , & riley , j. m. 1974 , , 167 , 31p fermi , e. 1949 , physical review , 75 , 1169 ferrarese , l. , & ford , h. 2005 , space science reviews , 116 , 523 ferrarese , l. , & merritt , d. 2000 , , 539 , l9 ferrari , a. 1998 , , 36 , 539 ferrari , a. 2006 , recent advances in astronomy and astrophysics , 848 , 453 greene , j. e. , & ho , l. c. 2005 , , 630 , 122 harvanek , m. , ellingson , e. , stocke , j. t. , & rhee , g. 2001 , , 122 , 2874 hasinger , g. , miyaji , t. , & schmidt , m. 2005 , , 441 , 417 he s , r. , barthel , p. d. , & fosbury , r. a. e. 1993 , , 362 , 326 hill , g. j. , & lilly , s. j. 1991 , , 367 , 1 ho , l. c. w. 2004 , coevolution of black holes and galaxies , 292 jackson , n. , & browne , i. w. a. 1990 , , 343 , 43 kallman , t. , & bautista , m. 2001 , , 133 , 221 kaspi , s. , smith , p. s. , netzer , h. , maoz , d. , jannuzi , b. t. , & giveon , u. 2000 , , 533 , 631 kauffmann , g. , et al .
2003 , , 346 , 1055 kewley , l. j. , groves , b. , kauffmann , g. , & heckman , t. 2006 , , 372 , 961 kormendy , j. , & richstone , d. 1992 , , 393 kriek , m. , et al . 2007 , , 669 , 776 559 krolik , j. h. 1999 , active galactic nuclei : from the central black hole to the galactic environment /julian
h. krolik .
princeton , n. j. : princeton university press , c1999 .
krolik , j. h. , mckee , c. f. , & tarter , c. b. 1981 , , 249 , 422 la franca , f. , andreani , p. , & cristiani , s. 1998 , , 497 , 529 livio , m. , ogilvie , g. i. , & pringle , j. e. 1999 , , 512 , 100 longair , m. s. 1966 , , 133 , 421 lynden - bell , d. 1969 , , 223 , 690 malkan , m. a. , gorjian , v. , & tam , r. 1998 , , 117 , 25 marchesini , d. , celotti , a. , & ferrarese , l. 2004 , , 351 , 733 martini , p. , kelson , d. d. , mulchaey , j. s. , & trager , s. c.,2002 , , 576 , 109l merloni , a. 2004 , , 353 , 1035 merloni , a. , rudnick , g. , & di matteo , t. 2004 , , 354 , l37 meszaros , p. , & ostriker , j. p. 1983
, , 273 , l59 mcdowell , j. c. , elvis , m. , wilkes , b. j. , willner , s. p. , oey , m. s. , polomski , e. , bechtold , j. , & green , r. f. 1989 , , 345 , l13 mclure , r. j. , & jarvis , m. j. 2002 , , 337 , 109 meier , d. l. 1999 , , 522 , 753 meier , d. l. 2001 , 20th texas symposium on relativistic astrophysics , 586 , 420 moderski , r. , & sikora , m. 1996 , , 283 , 854 moran , e. c. , halpern , j. p. , bothun , g. d. , & becker , r. h. 1992 , , 104 , 990 moshir , m. , et al . 1990 , , 22 , 1325 nemmen , r. s. , bower , r. g. , babul , a. , & storchi - bergmann , t. 2007 , , 377 , 1652 odea , c. p. 1998
, , 110 , 493 osterbrock , d. e. , & ferland , g. j. 2006 , astrophysics of gaseous nebulae and active galactic nuclei , 2nd .
ed . by d.e .
osterbrock and g.j .
ferland .
sausalito , ca : university science books , 2006 ostriker , j. p. , & ciotti , l. 2005 , royal society of london philosophical transactions series a , 363 , 667 park , m .- g . , & ostriker , j. p. 1998
, advances in space research , 22 , 951 penrose , r. 1969 , nuovo cimento , 1 , 252 petrosian , a. r. 1982 , astrofizika , 18 , 548 postman , m. , & geller , m. j. 1984 , , 281 , 95 prestage , r. m. , & peacock , j. a. 1983 , , 204 , 355 proga , d. 2005 , , 629 , 397 punsly , b. , & coroniti , f. v. 1990 , , 354 , 583 quataert , e. , & narayan , r. 2000 , , 528 , 236 rees , m. j. , begelman , m. c. , blandford , r. d. , & phinney , e. s. 1982 , , 295 , 17 rees , m. j. 1984 , , 22 , 471 reviglio , p. , & helfand , d. j. 2006 , , 650 , 717 reviglio , p. , `` active galaxies and their evolution as observed in the first and sloan digital sky surveys '' , ph.d . thesis , columbia university , new york , u.s.a . , 2008 reviglio , p. , helfand , d. j. 2009 , paper i , in publication reviglio , p. , helfand , d. j. 2009 , paper ii , in publication rybicki , g. b. , & lightman , a. p. 1986
, radiative processes in astrophysics , by george b. rybicki , alan p. lightman , pp
. 400 .
isbn 0 - 471 - 82759 - 2 .
wiley - vch , june 1986 . , ryle , m. , highlights of astronomy ( edit . by perek , l. )
( d. reidel , 1967 ) .
sadler , e. m. , et al .
2002 , , 329 , 227 salpeter , e. e. 1964 , , 140 , 796 sanders , d. b. , phinney , e. s. , neugebauer , g. , soifer , b. t. , & matthews , k. 1989 , , 347 , 29 schneider , d. p. , et al .
2005 , , 130 , 367 shakura , n. i. , & syunyaev , r. a. 1973 , , 24 , 337 shen , y. , et al .
2007 , , 133 , 2222 sikora , m. , stawarz , .
, & lasota , j .-
2007 , , 658 , 815 silk , j. , & rees , m. j. 1998 , , 331 , l1 strateva , i. , et al .
2001 , , 122 , 1861 tran , h. d. 2003 , , 583 , 632 voges , w. , et al . 1999 , , 349 , 389 tremaine , s. , et al .
2002 , , 574 , 740 urry , c. m. , & padovani , p. 1995
, , 107 , 803 urry , m. 2003 , active galactic nuclei : from central engine to host galaxy , 290 , 3 van breugel , w. , miley , g. , & heckman , t. 1984 , , 89 , 5 voges , w. , et al .
2000 , vizier online data catalog , 9029 , 0 wang , j .-
m . , chen , y .- m .
, ho , l. c. , & mclure , r. j. 2006 , , 642 , l111 wang , j. , li , h. , & xue , l. 2004 , , 617 , 113 white , r. l. , helfand , d. j. , becker , r. h. , glikman , e. , & de vries , w. 2007 , , 654 , 99 wilhite , b. c. , vanden berk , d. e. , kron , r. g. , schneider , d. p. , pereyra , n. , brunner , r. j. , richards , g. t. , & brinkmann , j. v. 2005 , , 633 , 638 york , d. g. , et al . 2000 , , 120 , 1579 zanni ,
c. , bodo , g. , rossi , p. , massaglia , s. , durbala , a. , & ferrari , a. 2003 , , 402 , 949 zeldovich , y. b. 1964 , dokl .
nauk sssr , 155 , 67 | in this third of a series of papers concerning active galaxies in the first and sloan digital sky surveys , we analyze the spectroscopic and radio properties of a sample of narrow - line active galactic nuclei ( agn ) , broad - line seyfert i galaxies , and quasars in the local universe in order to investigate the dependence of their activity on the mass , spin and accretion rates of the supermassive black holes ( smbh ) residing at the centers of their host galaxies . we show that galaxies hosting more massive smbh are more likely to power stronger and larger radio jets , and we show a strong anti - correlation between the strength of the lines of radio emitting galaxies and their radio power .
furthermore we show that the compactness of a jet is correlated with the epoch of the last episode of star - formation , suggesting a link between the presence of cold gas in a galaxy , the size of its smbh and the radio and spectroscopic features of its agn .
we use our large statistical sample to test the expectations of unified models of agn based on orientation .
while confirming that seyfert ii galaxies and radio galaxies are significantly more extincted then seyfert is and nearby quasars , we find several major inconsistencies with such a paradigm .
in particular we show a strong difference in the [ oiii],[oii ] and [ nii ] luminosities for different spectroscopic classes , a result which argues in favor of an evolution of the broad and narrow line regions of active nuclei over time .
we suggests that evolution , rather than orientation , may be the key element in shaping the properties of active nuclei , as also suggested by the results of high - redshift x - ray and radio surveys and we speculate on a model that may predict this kind of evolution . |
in the history of using computer simulation as a research tool to study the physics of turbulence , the dominant approach has been to use spectral methods . direct numerical simulation ( dns )
was introduced as a means to check the validity of turbulence theories _ directly _ from the equations of fluid dynamics @xcite .
the idea that important features of turbulence are _ universal _ encouraged researchers to study the simplest of geometries , a periodic cubic volume of homogeneous , isotropic turbulent fluid . in this case , the simplicity and efficiency of a fourier - spectral method can not be matched .
the largest direct numerical simulation of isotropic turbulence was conducted by ishihara et al .
@xcite using @xmath0 grid points at a maximum taylor - microscale reynolds number @xmath1 .
this record - breaking computation was done on _ earth simulator_a large vector machine with crossbar switch interconnect that can efficiently perform large - scale fft .
the successive generations of supercomputers have not been so fft - friendly , and this record has not been surpassed even though the peak performance of supercomputers has increased nearly 50-fold since then .
the record was matched for the first time in the u.s . by donzis et al .
@xcite , running on 16 thousand cpu cores of the ranger supercomputer in texas , a linux - cluster supercomputer with 16-core nodes .
future high - performance computing systems will have ever more nodes , and ever more cores per node , but will probably not be equipped with the bandwidth required by many popular algorithms to transfer the necessary data at the optimal rate .
this situation is detrimental to parallel scalability .
therefore , it is becoming increasingly important to consider alternative algorithms that may achieve better sustained performance on these extremely parallel machines of the future . in most standard methods of incompressible cfd
, the greatest fraction of the calculation runtime is spent solving a poisson equation .
equations of this type can be efficiently solved by means of an fft - based algorithm , a sparse linear solver , or a fast multipole method ( fmm ) @xcite . for the sake of our argument
, we will not differentiate between fft - based poisson solvers and pseudo - spectral methods because they both rely on fft .
the fast multipole method has not gained popularity due to the fact that it is substantially slower depending on implementations , at least an order of magnitude slower than fftand multigrid solvers when compared using a small cpu cluster .
we aim to show with our ongoing research that the relative performance of fmmimproves as one scales to large numbers of processes using gpu acceleration .
the highly scalable nature of the fmmalgorithm , among other features , makes it a top contender in the algorithmic toolbox for exascale systems .
one point of evidence for this argument is given by the gordon bell prize winner of 2010 , which achieved 0.7 petaflop / s with an fmmalgorithm on 200k cores of the jaguar supercomputer at oak ridge national laboratory @xcite .
in the previous year , fmmalso figured prominently at the supercomputing conference , with a paper among the finalists for the best paper award @xcite and the gordon bell prize in the price / performance category going to work with hierarchical @xmath2-body methods on gpu architecture @xcite .
the fmmalgorithm is well adapted to the architectural features of gpus , which is an important consideration given that gpus are likely to be a dominant player as we move towards exascale .
the work presented in 2009winner in the price / performance category in great measure thanks to the ingenious and resourceful system design using gaming hardware reported ( at the time of the conference ) 80 teraflop / s @xcite . that work progressed to an honorable mention in the 2010 list of awardees , with a performance of 104 teraflop / s @xcite . at the level of the present work , where we present a 1.08 petaflop / s ( single precision ) calculation of homogeneous isotropic turbulence , the fmmmoves firmly into the arena of _ petascale _ gpu computing .
the significance of this advance is that we are now in a range where the fmmalgorithm shows its true capability .
the excellent scalability of fmmusing over 4000 gpus is an advantage over the dominant fft - based algorithms .
showcasing the fmmin the simulation of homogeneous isotropic turbulence is especially fitting , given that a years - old record there remains unchallenged .
we not only match the grid size of the world record , but also demonstrate that using the fmmas the underlying algorithm enables isotropic turbulence simulations that scale to thousands of gpus . given the severe bottleneck imposed by the all - to - all communication pattern of the fftalgorithm ,
this is not possible with pseudo - spectral methods in current hardware .
the fmmis used here as the numerical engine in a vortex particle method , which is an approach to solve the navier - stokes equations using a the vorticity formulation of momentum conservation and a particle discretization of vorticity .
this is not a standard approach for the simulation of turbulence and the vortex method is not yet trusted for this application .
for this reason , we have made efforts to document the validation of our vortex - method code and , in a separate publication @xcite , compared with a trusted spectral - method code .
we looked at various turbulence statistics , including higher - order velocity derivatives , and performed a parameter study for the relevant numerical parameters of the vortex method .
that work provides evidence that the vortex method is an adequate tool for direct numerical simulation of fluid turbulence , while in the present work we focus on the performance aspects . for completeness , this sections gives a brief overview of the numerical methods .
the vortex method @xcite is a particle - based approach for fluid dynamics simulations .
the particle discretization results in the continuum physics being solved as an @xmath2-body problem .
therefore , the hierarchical @xmath2-body methods that extract the full potential of gpus can be used for the simulation of turbulence . unlike other particle - based solvers for fluid dynamics , e.g. ,
smoothed particle hydrodynamics @xcite , the vortex method is especially well suited for computing turbulent flows , because the vortex interactions seen in turbulent flows are precisely what it calculates with the vortex particles . in the vortex method ,
the navier - stokes equation is solved in the velocity - vorticity formulation , using a particle discretization of the vorticity field .
the velocity is calculated using the following equation , representing the biot - savart law of fluid dynamics : @xmath3 here , @xmath4 is the strength of vortex particles , @xmath5 is the green s function for the laplace equation and @xmath6 is the cutoff function , with @xmath7 the distance between the interacting particles , and @xmath8 the standard deviation of the gaussian function .
the navier - stokes system is solved by a simultaneous update of the particle positions to account for convection , of the particle strengths to account for vortex stretching , and of the particle width to account for diffusion .
the equation used to calculate the stretching term , @xmath9 , is : @xmath10 which was obtained by substituting the biot - savart equation ( [ eq : biotsavart ] ) for @xmath11 , and using the discrete form of the vorticity .
finally , the diffusion update is calculated according to @xmath12 we perform a radial basis function interpolation for reinitialized gaussian distributions to ensure the convergence of the diffusion calculation @xcite
. equations ( [ eq : biotsavart ] ) and ( [ eq : stretching ] ) are @xmath2-body interactions , and are evaluated using the fmm .
we use a highly parallel fmmlibrary for gpus developed in our group , called ` exafmm ` , which is available under an open - source license and is described further below . for the purpose of comparing with a spectral method
, we used a code for homogeneous isotropic turbulence developed and used at the center for turbulence research of stanford university @xcite .
the code is called ` hit3d ` and is available freely for download .
it uses a spectral galerkin method in primitive - variable formulation with pseudo - spectral methods to compute the convolution sums . for the fft , it relies on the ` fftw ` library and it provides parallel capability for cpu clusters using mpi .
parallelization of the fftis accomplished by a domain - decomposition approach , illustrated by the schematic of figure [ fig : spectral_fft ] .
domain decomposition is applied in two spatial directions , resulting in first `` slabs '' then `` pencil '' sub - domains that are assigned to each mpiprocess .
the initial conditions for our runs were generated by ` hit3d ` .
the vortex method used the same initial condition by first calculating the vorticity field in physical space , and then using radial basis function interpolation to obtain the vortex strengths .
this is different from our other publication , where we studied the accuracy of turbulence simulations using the fmm - based vortex method on gpus , looking at high - order turbulence statistics @xcite . for that case ,
the initial condition provided by ` hit3d ` , which has a fully developed energy spectrum , was not suitable for our validation exercise looking at the time evolution of the velocity derivative skewness and flatness .
for this reason , we constructed initial conditions in fourier space as a solenoidal isotropic velocity field with random phases and a prescribed energy spectrum .
this initial velocity field had a gaussian distribution and satisfied the incompressibility condition .
it is common that algorithms with low complexity ( sparse linear algebra , fft ) have low arithmetic intensity , while algorithms with high arithmetic intensity ( dense linear algebra ) tend to have high complexity .
the fmmpossesses a rare combination of @xmath13 complexity and an arithmetic intensity that is even higher than dgemm @xcite .
although this may seem like a great combination , it also implies that there is a large constant in front of the @xmath13 scaling , which results in a larger time - to - solution compared to other @xmath13 or @xmath14 methods like multigrid methods and fft .
however , as arithmetic operations become cheaper compared to data movement in terms of both cost and energy , the large asymptotic constant of the _ arithmetic _ complexity becomes less of a concern .
the calculation of velocity by means of the biot - savart equation ( [ eq : biotsavart ] ) and the computation of the stretching term ( [ eq : stretching ] ) both result in an @xmath2-body problem , which has a complexity of @xmath15 for @xmath2 particles , if solved directly .
the fast multipole method @xcite reduces the complexity to @xmath13 by clustering source and target particles , and using series expansions that are valid far ( multipole expansion ) or near ( local expansion ) a point .
the far / near relationships between points in the domain and the interactions between clusters are determined by means of a tree structure .
the domain is hierarchically divided , and different sub - domains are associated with branches in the tree , a graphical representation of which is shown in figure [ fig : kernels ] .
the algorithm proceeds as follows .
first , the strengths of the particles ( e.g. , charges or masses ) are transformed to multipole expansions at the leaf cells of the tree ( known as the particle - to - multipole , or pmkernel ) .
then , the multipole expansions of the smaller cells are translated to the center of larger cells in the tree hierarchy recursively and added ( multipole - to - multipole , or mmkernel ) .
subsequently , the multipole expansions are transformed into local expansions for all pairs of well - separated cells ( multipole - to - local , or mlkernel ) , and then to local expansions at the center of smaller cells recursively ( local - to - local , or llkernel ) .
finally , the local expansions at leaf cells are used to compute the effect of the far field on each target particle .
since mloperations can only be performed for well - separated cells , the direct neighbors at the finest level of the tree interact directly via the original equation ( particle - to - particle , or ppkernel ) .
the current implementation of the ` exafmm ` code uses expansions in spherical harmonics of the laplace green s function @xcite .
details of the extension of laplace kernels to biot - savart and stretching kernels and of the implementation on gpus can be found in previous publications @xcite . when parallelizing hierarchical @xmath2-body algorithms , the fact that the particle distribution is dynamic makes it impossible to precompute the decomposition and to balance the work - load or communication _ a priori_. warren and salmon @xcite developed a parallel algorithm for decomposing the domain into recursive subdomains using the method of orthogonal recursive bisection ( orb ) .
the resulting process can be thought of as a binary tree , which splits the domain into subdomains with equal number of particles at every bisection .
another popular technique for partitioning tree structures is to use morton ordering @xcite , where bits representing the particle coordinates are interleaved to form a unique key that maps to each cell in the tree . following the morton index monotonically
will take the form of a space - filling curve in the shape of a z " . partitioning the list of morton indices equally assures that each partition will contain an equal number of cells , regardless of the particle distribution .
for an adaptive tree , a nave implementation of the morton ordering could result in large communication costs if the z " is split in the wrong place , as illustrated in figure [ fig : partitioning ] .
sundar et al .
@xcite proposed a bottom - up coarsening strategy that ensures a clean partition at the coarse level while using morton ordering . on the other hand , an orbalways partitions the domain into a well balanced binary tree , since the number of particles is always equal on each side of the bisection
therefore , orbis advantageous from the point of view of load - balancing .
the main difference between using morton ordering and using an orbis the shape of the resulting subdomains and the corresponding tree structure that connects them , as shown in figure [ fig : partitioning ] .
morton ordering is done on cubic octrees and the corresponding tree structure can become highly non - uniform depending on the particle distribution .
conversely , an orbalways creates a balanced tree structure but the sub - domain shapes are rectangular and non - uniform . a difficulty when applying orbwithin the framework of conventional fmmis that the construction of cell - cell interaction lists depends on the cells being cubic ( and not rectangular ) .
the dual tree traversal described in the following subsection allows the use of rectangular cells , while enabling cell - cell interactions in the fmmwith minimum complications . by using orbalong with the dual tree traversal
, we obtain a partitioning+traversal mechanism that allows perfect load balancing for highly non - uniform particle distributions , while retaining the @xmath13 complexity of the fmm .
another advantage of this technique is that global indexing of cells is no longer necessary .
having a global index becomes an issue when solving problems of large size .
the maximum depth of octrees that a 64-bit integer can handle is 21 levels ( @xmath16 ) . for highly adaptive trees with small number of particles per cell
, this limit will be reached quite easily , resulting in integer - overflow with a global index . circumventing this problem by using multiple integers for index storage will require much more work when sorting ,
so this is not a desirable solution .
the dual tree traversal allows the entire fmmto be performed without the use of global indices , and is an effective solution to this problem .
our current partitioning scheme is an extension of the orb , which allows multi - sections instead of bisections . bisecting
the domain involves the calculation of the median of the particle distribution for a given direction , and doing this recursively in orthogonal directions @xmath17 is what constitutes the orthogonal recursive bisection " . therefore the extension from bisection to multi - section
can be achieved by simply providing a mechanism to search for something other than the median .
we developed a parallel version of the @xmath18-element " algorithm .
finding the @xmath18-element " is much faster than any sorting algorithm , so our technique is much faster than any method that requires sorting . always searching for the @xmath19-th element will reduce to the original algorithm based on bisections , but searching for
the @xmath20-th element will enable the domain to be split between 3 and 4 processes , for example .
therefore , our recursive multisection allows efficient partitioning when the number of processes is not a power of two .
the dual tree traversal enables cell - cell interactions in the @xmath14 barnes - hut treecode @xcite framework , thus turning it into a @xmath13 algorithm @xcite .
we give a detailed explanation of the dual tree traversal in a previous publication that focused on hybrid treecode & fmmtechniques @xcite .
it has several advantages compared to explicit construction of cell - cell interaction lists , the more common approach in the fmmframework .
firstly , it is the simplest and most general way to perform the task of finding all combinations of cell - cell interactions .
it is simple in the sense that it does not contain multiple loops for finding the parent s neighbor s children that are non - neighbors of the current cell , " as is the case with the explicit construction .
it is general in the sense that the property `` well - separated '' can be defined flexibly , instead of the rigid definition of non - neighboring cells " used traditionally in the fmm .
it is common in treecodes to define the well - separated property more precisely and adaptively by introducing the concept of multipole acceptance criterion ( mac ) @xcite .
we give a graphical description of the different types of macused in our method in figure [ fig : mac ] .
the simplest macis the one defined for barnes - hut treecodes , where @xmath21 is the size of the source cell and @xmath22 is the distance from the target particle to the center of mass of the source cell ; @xmath23 , called the opening angle , is the parameter that is used to flexibly control the definition of the well - separated property . in the fmmframework ,
a macis defined between two cells ( rather than a cell and a particle ) , where @xmath24 is the size of the target cell and @xmath22 is the distance between the center of mass of the target cell and the center of mass of the source cell .
a flexible definition of the well - separated property can not be implemented easily in the traditional fmmframework , where the cell - cell interaction lists are constructed explicitly .
this is because one must first provide a list of candidates to apply the macto , and increasing the neighbor search domain to @xmath25 neighbors instead of @xmath26 neighbors and applying the macon them is not an efficient solution .
the dual tree traversal is a natural solution to this problem , since the list of candidates for the cell - cell interaction is inherited directly from the parent .
this inheritance of cell - cell interaction candidates " is lacking from conventional fmm , and can be provided by the dual tree traversal while simplifying the implementation at the same time .
furthermore , since the dual tree traversal naturally allows interaction between cells at different levels of the tree , it automatically finds the pair of cells that appear in @xmath27-lists for adaptive trees @xcite , but with much greater flexibility ( coming from the mac ) and no overhead of generating the lists ( since it does nt generate any ) .
another advantage of the dual tree traversal is that it enables the use of non - cubic cells . as we have explained in section [ sse : orb ] , this allows using orbpartitioning in the fmmframework , which has superior load balancing properties compared to morton ordering and removes the dependence on global indexing . as mentioned above ,
global morton indices are a problem when using millions of cores and the tree depth exceeds 21 levels causing the morton index to overflow from the 64-bit integer . the combination of the dual tree traversal and orbis an elegant solution to this problem as well .
we describe the dual tree traversal in algorithm
[ al : evaluate ] , which calls an internal routine for the interaction of a pair of cells , given in algorithm [ al : interact ] .
first , a pair of cells is pushed to a stack .
it is most convenient to start from the root cell although there is no possibility that two root cells will interact . for every step in the while - loop
, a pair of cells is popped from the stack and the larger cell is subdivided . then
, algorithm [ al : interact ] is called to perform either particle - particle ( pp ) or multipole - local ( ml ) interactions .
if the cells in the pair are too close and either of the cells has children , the pair is pushed to the stack and will be handled later .
a more detailed explanation is given in a previous publication focusing on hybrid treecode & fmmtechniques @xcite .
a = b = rootcell push pair(a , b ) into a stack pop stack to get a pair(a , b ) interact(a , b ) interact(a , b ) evaluate multipole - local ( m2l ) evaluate particle - particle ( p2p ) evaluate multipole - local ( m2l ) push pair(a , b ) into a stack the partitioning techniques discussed in section [ sse : orb ] will assign a local portion of the global tree to each process .
however , the fmmevaluation requires information from all parts of the global tree , and the multipole expansions on remote processes must be communicated . unlike standard domain - decomposition methods , where only a thin halo needs to be communicated , the fmmrequires a global halo .
fortunately , this global halo becomes exponentially coarser as the distance from the target cell increases , as shown in figure [ fig : periodic ] .
therefore , the data required to be communicated from far partitions is very small ( but never zero ) .
this results in a non - homogeneous ` alltoallv`-type communication of subsets of the global tree .
once all the subsets of the global tree are communicated between the processes , one can locally reconstruct a tree structure that contains all the information required for the evaluation in the local process .
this reconstructed tree is called the _ local essential tree _ ( let ) , and it is a ( significantly smaller ) subset of the global tree that contains all the information that is necessary to perform the evaluation .
salmon and warren @xcite introduced two key techniques in this area : one for finding which cell data to send , and the other for communicating the data efficiently .
the determination of which data to send is tricky , since each process can not see what the adaptive tree structure looks like on remote processes .
the solution proposed by salmon and warren @xcite is to use a conservative estimate and communicate a larger portion of the tree than is exactly required .
this conservative estimate is obtained by means of a special macdescribed in figure [ fig : mac ] as the letmac .
there , the distance @xmath22 is defined as the distance between the center of mass of the source cell and the edge of the partition on the remote process .
a formula to calculate @xmath22 can be given as the following expression , where we assume element - wise boolean operations over array elements ( giving zero for false , and 1 for true ) and element - wise multiplication and summation : @xmath28 here , @xmath29 and @xmath30 are the minimum and maximum coordinate values for all particles in the target partition , while @xmath31 is the center of mass of the source cell .
this definition of the letmaccorresponds to assuming the case where the target cell is located at the edge of the remote partition .
we also assume that the target cell is of the same size as the source cell ( @xmath32 ) , so @xmath33 becomes @xmath34 .
these assumptions generally hold quite well and the required part of the tree is sent to the remote process most of the time . however , we must ensure that the fmmcode still works for extreme cases where necessary information fails to be sent . with this end in view
, we have added a conditional statement in the interaction calculation ( algorithm
[ al : interact ] ) as a further precaution for anomalous cases . this way , the traversal will perform the mltranslation with the smallest cell that is available
. this cell may be too large to satisfy the fmmmacso there will be a small penalty on the accuracy , but since the occurrence of such a case is so rare , it does not affect the overall result .
a schematic showing how the partitioning and communication techniques are used in the vortex method is given in figure [ fig : flow_chart ] .
first , the domain is partitioned using the recursive multisection described in section [ sse : orb ] .
then the fmmkernels for the local tree are evaluated while the letdata is being communicated in a separate openmp section .
after the letdata is communicated , the fmmkernels are evaluated again for the remaining parts of the let .
subsequently , the position , vortex strength @xmath35 , and core radius @xmath8 of the vortex particles are updated locally .
this information is communicated in the next stage when the letis exchanged .
in addition , the lagrangian vortex method needs to reinitialize the particle positions to maintain sufficient overlap of the smooth particles . for this procedure
, we can reuse the same tree structure since the particles are reinitialized to the same position every time .
therefore , the partitioning is performed only once at the beginning of this simulation .
the fmmwas originally devised to solve potential problems with free - field boundary conditions .
the method can be extended to handle periodic boundary conditions by placing periodic images around the original domain and using multipole expansions to approximate their influence @xcite , as illustrated in figure [ fig : periodic ] .
when a sufficient number of periodic images are placed , the error caused by using a finite number of periodic images becomes smaller than the approximation error of the fmmitself @xcite .
this approach to extend the fmmto periodic domains adds negligible computational overhead to the original fmm , for two reasons .
first , distant domains are clustered into larger and larger cells , so the extra cost of considering a layer of periodic images is constant , while the number of images accounted for grows exponentially .
the second reason is that only the sources need to be duplicated and the target points exist only in the original domain .
since the work load for considering the periodicity is independent of the number of particles , it becomes negligible as the problem size increases .
an earlier study showed that periodic boundary conditions add approximately 3% to the calculation time for a million particles @xcite .
the fmmconsists of six different computational kernels , as illustrated on figure [ fig : kernels ] . in the ` exafmm`code , all of these kernels are evaluated on gpu devices using cuda . out of the six kernels , a great majority of the runtime is spent executing ppand ml .
we use a batch evaluation for these two kernels , since there is no data dependency between them : they are evaluated in one batch after the tree traversal is completed .
this batch evaluation can be broken into multiple calls to the gpu device , depending on its storage capacity and the data size . with this approach , we are able to handle problem sizes of up to 100 million particles on a single gpu , if the memory on the host machine is large enough @xcite . as an example of the usage of thread blocks in the gpu execution model , we show in figure [ fig : m2l_gpu ] an illustration of the mlkernel on gpus .
each coefficient of the multipole / local expansion is mapped to a thread on the gpu and each target cell is mapped to a thread block , while each source cell is loaded to shared memory and evaluated sequentially .
all other kernels are mapped to the threads and thread blocks in a similar manner .
more details regarding the gpu implementation of fmmkernels can be found in chapter 9 of the _ gpu gems emerald edition _
book @xcite , with accompanying open - source codes .
in this section , we present results from large - scale simulations of homogeneous isotropic turbulence using the fmm - based vortex particle method , up to a problem size of @xmath0 computational points .
this is still the largest mesh - size for which dnsof turbulence has been published , even though this scale of simulation was achieved 10 years ago . as we discussed in the introduction ,
one of the reasons for this is the difficulty in scaling the fftalgorithm beyond a few thousand processes .
the current simulations with a vortex method were checked for correctness by comparing the kinetic energy spectrum with that obtained using a trusted spectral - method code ( described in [ s : spectral ] ) .
the focus here is not on the physics , however , but on demonstrating large - scale simulation of turbulence using the fmmas a numerical engine and reporting on performance aspects using gpu hardware .
the performance is described via the results of weak scaling tests using between 1 and 4096 processes with @xmath36 ( 16.8 million ) particles per process .
each process offloads to one gpu to speed - up the fmm , making it even more challenging to obtain good parallel efficiency ( it is obviously harder to scale faster code ) . despite this
, a parallel efficiency of 74% was obtained for the fmm - based vortex method on weak scaling between 1 and 4096 processes , with the full application code .
the largest calculation used @xmath37 billion particles , which as far as we know is the largest vortex - method calculation to date .
previous noteworthy results by other authors reported calculations with up to 6 billion particles @xcite , which we surpass here by an order of magnitude .
the calculations reported here were run on the tsubame-2.0system during spring and fall of 2011 , thanks to guest access provided by the grand challenge program of tsubame-2.0 .
this system has 1408 nodes , each equipped with two six - core intel xeon x5670 ( formerly westmere - ep ) 2.93ghz processors , three nvidia m2050 gpus , 54 gb of ram ( 96 gb on 41 nodes ) , and 120 gb of local ssd storage ( 240 gb on 41 nodes ) .
computing nodes are interconnected with the infiniband device grid director 4700 developed by voltaire inc .
, with non - blocking and full bisectional bandwidth .
each node has @xmath38 gbps bandwidth , and the bisection bandwidth of the system is over 200 tbps .
the total number of m2050 gpus in the system is 4224 , and the peak performance of the entire system is 2.4 petaflop / s .
we set up simulations of decaying , homogeneous isotropic turbulence using an initial taylor - scale reynolds number of @xmath39 .
given that we had guest access to the full tsubame-2.0system for a very brief period of time ( only a few hours ) to produce the scalability results , we opted for a lower reynolds number than previous turbulence simulations of this size using spectral methods .
there is limited experience using vortex methods for dnsof turbulence , but previous work has suggested that higher resolutions are needed than when using the spectral method at the same reynolds number .
we thus decided to be conservative and ensure that we obtained usable results from these one - off runs .
the calculation domain is a box of size @xmath40 ^ 3 $ ] with periodic boundary conditions in all directions , using @xmath41 periodic images in each dimension in the periodic fmm(see figure [ fig : periodic ] ) . to achieve the best accuracy from the fmmfor the present application , the order of multipole expansions
was set to @xmath42 ; this may be a conservative value , but given our limited allocation in the tsubame-2.0system , we did not have the luxury of tuning the simulations for this parameter .
the fmmkernels are run in single precision on the gpu which may raise some concerns .
we are able to achieve double - precision accuracy using single - precision computations in the fmmkernels by means of two techniques .
the multipole - expansion coefficients have exponentially smaller magnitude with increasing order of expansion ; therefore , by adding them from higher- to lower - order terms , we can prevent small numbers being added to the large numbers .
this preserves the significant digits in the final sum .
the second technique consists of normalizing the expansion coefficients to reduce the dynamic range of the variables , allowing the use of single - precision variables to get a double - precision result .
thus , we are able to achieve sufficient accuracy to reproduce the turbulence statistics ( as shown below ) and obtain the same results that would be obtained using double precision ( given that the fmmerror is larger than 6 significant digits anyway ) . the isosurface of the second invariant of the velocity gradient tensor is shown in figure [ fig : isosurface ] .
this is a snapshot at the early stages of the simulation and we do not observe any large coherent structures . in order to take a closer look at the quantitative aspects of the vortex simulation , in figure [ fig : spectrum ] we compare the kinetic energy spectrum with that of the spectral method , where @xmath43 is the eddy turnover time .
the energy spectrum of the vortex method is obtained by calculating the velocity field on a uniform lattice , which is induced by the lagrangian vortex particles .
the capability of ` exafmm ` to calculate for different sources and targets enabled such operations .
we have excellent quantitative agreement between the vortex method and spectral method and we conclude that our fmm - based particle method is capable of simulating turbulence of this scale correctly .
we turn our attention to the performance of these calculation in the next section .
, obtained with the vortex method and spectral method .
notice that the vertical axis goes down to @xmath44 , which is many times smaller than in similar plots presented by other authors . ]
we had two very short windows of time in which we were able to run weak scaling tests , one with half the system and the other with almost the full tsubame-2.0system .
the larger scaling tests used @xmath36 particles per process , on 1 to 4096 processes and executing three mpiprocesses per node , each process assigned to a gpu card within the node .
the results of the weak scaling test are shown in figure [ fig : weak_scaling ] in the form of total runtime of the fmm , and timing breakdown for different phases in the computation .
the label ` near - field evaluation ' corresponds to the ppkernel evaluation illustrated in figure [ fig : kernels ] , and the label ` far - field evaluation ' corresponds to the sum of all the other kernel evaluations , i.e. , the far field .
the ` mpicommunication ' is overlapped with the fmmevaluation ( see figure [ fig : flow_chart ] ) , so the plot shows only the amount of time that communication exceeds the local portion of the ` near - field ' and ` far - field ' evaluation . in this way , the total height of each bar correctly represents the total wall - clock time of each calculation .
note that particle updates in the vortex - method calculation take less than 0.01% in all runs and thus were invisible in the bar plots , so we ve left this computing stage out of the labels .
the ` gpu buffering ' label corresponds to the time it takes to form a queue of tasks and corresponding data buffer to be transferred to the gpu , which is a significant amount of time .
we have found this buffering to be necessary in order to achieve high efficiency in the ppand mlevaluation and fmmevaluation on gpus .
moreover , this part of the computation scales perfectly , and does not affect the scalability of the fmm .
the parts that do affect the scalability are the tree construction and mpicommunication . actually , the tree construction also involves mpicommunications for the partitioning , so the parallel efficiency in weak scaling is fully determined by mpicommunications .
figure [ fig : weak_scaling ] shows that the current fmmis able to almost completely hide the communication time up to 512 gpus .
it may be worth noting that the @xmath2-d - hypercube - type communication of the let@xcite turned out to be slower on tsubame-2.0than a simple call to ` mpi_alltoallv ` for sending the entire letat once .
this is a consequence of the network topology of tsubame-2.0(with a dual - qdr infiniband link to each node and non - blocking full - bisection fat - tree interconnect ) and also of the relatively small number of mpiprocesses .
the results shown in figure [ fig : weak_scaling ] are those obtained with ` mpi_alltoallv ` and not the hypercube - type communication .
we performed a corresponding weak scalability test for the spectral method , increasing the problem size from @xmath36 on one process to @xmath0 on 4096 processes .
we used three mpiprocesses per node to match the condition of the vortex method runs , but there is no gpu acceleration in this case . matching
the number of mpiprocesses per node should give both methods an equal advantage / handicap for the bandwidth per process .
note that using gpus for the fftwithin the spectral - method code is unlikely to provide any benefits , because performance improvements of ` cufft ` over ` fftw ` would be canceled out by data transfer between host and device ] and inter - node communications in parallel runs .
figure [ fig : compare_scaling ] shows the parallel efficiency obtained with the two methods , under these conditions .
the parallel efficiency of the fmm - based vortex method is 74% when going from one to 4096 gpus , while the parallel efficiency of the spectral method is 14% when going from one to 4096 cpus .
the bottleneck of the spectral method is the all - to - all communication needed for transposing the slabs into pencils as shown in figure [ fig : spectral_fft ] .
even though this may not be the best implementation of a parallel fft , the difference in the scalability between the spectral method and fmm - based vortex method is considerable .
the actual calculation time is in the same order of magnitude for both methods at 4096 processes : it was 108 seconds per time step for the vortex method and 154 seconds per time step for the spectral method .
therefore , the superior scalability of the fmmhas merely closed the gap with fft , being barely @xmath45 faster at this scale .
however , we anticipate that this trend will affect the choice of algorithm in the future , as we move to architectures with higher degree of parallelism .
the scaling test with half the tsubame-2.0system was done several months before and with a different revision of the code , with many changes having been incorporated since then .
we include the results here for completeness ; see figure [ fig : compare_scaling_old ] . in this case
, the number of particles per process is much smaller , at 4 million ( compared to 16.8 million particles per process in the larger test ) and we scale from 4 to 2048 processes .
the parallel efficiency of the fmm - based vortex method is 72% when going from 4 to 2048 gpus , while the parallel efficiency of the spectral method was 14% .
when we compare the parallel efficiency with the 1 to 4096 gpucase , we see that the 4 to 2048 case is scaling relatively poorly .
this is due to the number of particles per process being roughly 1/4 in the 4 to 2048 case , and also the improvement in the ` exafmm ` code during the half year gap between the two runs .
the run time per time step in this case was 27 seconds for the vortex method and 20 seconds for the spectral method to compute a domain with @xmath46 points .
note that if we read from the plot in figure 2 of donzis et al .
@xcite , their spectral - method calculation on a @xmath46 grid using 2048 cores of the ranger supercomputer takes them about 20 seconds per time step .
since this is the same timing we obtained with the ` hit3d ` code , we are satisfied that this code provides a credible reference point for the scalability of spectral dns codes .
the calculation in the fmmis mostly dominated by the floating point operations in the particle - particle interactions , while all other parts are a minor contribution in terms of flop / s ( although not negligible in terms of runtime ) .
we will thus consider in the estimate of sustained performance only the operations executed by the ppkernels .
two separate equations are being calculated for the particle - particle interactions : the biot - savart equation ( [ eq : biotsavart ] ) and the stretching equation ( [ eq : stretching ] ) . the number of floating point operations required by these two kernels is summarized in table [ tab : flops ] .
.floating point operations per ppinteraction . [ cols="^,^,^",options="header " , ] the approximate number of flop / s for one step of the vortex method calculation of isotropic turbulence is obtained by the following equation .
@xmath47 thus , the current fmm - based vortex method achieved a sustained performance of 1.08 petaflop / s ( single precision ) on 4096 gpus of tsubame-2.0 .
the authors of the ` exafmm`code have a consistent policy of making science codes available openly , in the interest of reproducibility .
the entire code that was used to obtain the present results is available from https://bitbucket.org/exafmm/exafmm .
the revision number used for the results presented in this paper is 191 for the large - scale tests up to 4096 gpus .
documentation and links to other publications are found in the project homepage at http://exafmm.org/. figure [ fig : compare_scaling ] , its plotting script and datasets are available online and usage is licensed under cc - by-3.0@xcite .
we acknowledge the use of the ` hit3d ` pseudo - spectral dns code for isotropic turbulence , and appreciate greatly their authors for their open - source policy ; the code is available via google code at http://code.google.com / p / hit3d/.
this work represents several milestones .
although the fmmalgorithm has been taken to petascale before ( notably , with the 2010 gordon bell prize winner ) , the present work represents the first time that this is done on gpu architecture .
also , to our knowledge , the present work is the largest direct numerical simulation with vortex methods to date , with 69 billion particles used in the cubic volume ; this is an order of magnitude larger than the previously reported record .
yet another significant event is reaching a range where the highly scalable fmmstarts showing advantage over fft - based algorithms . with a 1.08 petaflop / s ( single precision )
calculation of isotropic turbulence in a @xmath0 box , using 4096 gpus , we are within reach of a turning point .
the combination of application , algorithm , and hardware used are also notable
. the real challenge in exascale computing will be the optimization of data movement .
when we compare the data movement of fmmagainst other fast algorithms like multigrid and fft , we see that the fmmhas a potential advantage .
compared to fft , both multigrid and fmmhave an advantage in the asymptotic complexity of the global communication .
the hierarchical nature of multigrid and fmmresults in @xmath48 communication complexity where @xmath49 is the number of processes . on the other hand , fftrequires two global - transpose communications between @xmath50 processes , and has communication complexity of @xmath51 .
when @xmath49 is in the order of millions , it seems unrealistic to expect that an affordable network can compensate for this large gap in the communication complexity . although it is not the focus of the present article , we would like to briefly note that an advantage of fmmover multigrid is obtained from differences in the synchronization patterns .
for example , increasing the desired accuracy in iterative solvers using multigrid will result in more iterations , hence more global synchronizations .
conversely , increasing the accuracy in fmminvolves increasing the number of multipole expansions , which results in even higher arithmetic intensity in the inner kernels while the number of global synchronizations remains the same . as the amount of concurrency increases , bulk - synchronous execution
/ communication models are reaching their limit .
thus , fmmprovides a new possibility to reduce the amount of communication and synchronization in these inherently global " problems .
finally , we would like to point out that the fmmcan be used to solve the poisson equation directly , or as a preconditioner for an iterative solver .
therefore , we are not concerned at this point about the fact that vortex methods may still be comparatively inefficient for the simulation of fluid turbulence , where spectral methods will continue to dominate in the foreseeable future .
this does not detract from the conclusions about the efficiency of the fmmitself , which is the object of our study . in future work
, we would like to demonstrate the efficiency of fmmby using it as a poisson solver or preconditioner in the framework of more standard finite difference / volume / element methods .
there , the comparison against fftand multigrid methods should be of interest to a broader spectrum of the cfdcommunity .
computing time in the tsubame-2.0system was made possible by the grand challenge program of tsubame-2.0 .
the current work was partially supported by the core research for the evolution science and technology ( crest ) of the japan science and technology corporation ( jst ) .
lab acknowledges funding from nsf grant oci-0946441 , onr grant # n00014 - 11 - 1 - 0356 and nsf career award oci-1149784 .
lab is also grateful for the support from nvidiacorp . via an academic partnership award ( aug .
2011 ) .
s. g. chumakov , j. larsson , c. schmitt , and h. pitsch . lag - modeling approach for dissipation terms in large eddy simulation .
annual research briefs , center for turbulence research , 2009 .
http://www.stanford.edu/group/ctr/resbriefs/arb09.html .
t. hamada , t. narumi , r. yokota , k. yasuoka , k. nitadori , and m. taiji .
42 tflops hierarchical n - body simulations on gpus with applications in both astrophysics and turbulence . in _
sc 09 : proceedings of the conference on high performance computing networking , storage and analysis _ , pages 112 .
acm , 2009 .
t. hamada and k. nitadori .
190 tflops astrophysical @xmath2-body simulation on a cluster of gpus . in
_ high performance computing , networking , storage and analysis ( sc ) , 2010 international conference for _ , pages 19 , nov .
t. ishihara , y. kaneda , m. yokokawa , k. itakura , and a. uno .
small - scale statistics in high - resolution direct numerical simulation of turbulence : reynolds number dependence of one - point velocity gradient statistics . , 592:335366 , 2007 .
christophe g. lambert , thomas a. darden , and john a. board jr . a multipole - based algorithm for efficient calculation of forces and potentials in macroscopic periodic assemblies of particles .
, 126(2):274285 , 1996 .
i. lashuk , a. chandramowlishwaran , h. langston , t. nguyen , r. sampath , a. shringarpure , r. vuduc , l. ying , d. zorin , and g. biros . a massively parallel adaptive fast - multipole method on heterogeneous architectures . in _ proceedings of the conference on high performance computing networking , storage and analysis , sc 09 _ ,
pages 112 , portland , oregon , november 2009 .
a. rahimian , i. lashuk , s. veerapaneni , a. chandramowlishwaran , d. malhotra , l. moon , r. sampath , a. shringarpure , j. vetter , r. vuduc , d. zorin , and g. biros .
petascale direct numerical simulation of blood flow on 200k cores and heterogeneous architectures . in _ proceedings of the 2010 acm / ieee international conference for high performance computing , networking , storage and analysis _ , sc 10 , pages 111 .
ieee computer society , 2010 .
m. s. warren and j. k. salmon .
astrophysical @xmath2-body simulations using hierarchical tree data structures . in _ proceedings of the 1992 acm / ieee conference on supercomputing _ , pages 570576 , los alamitos , ca , usa , 1992 .
ieee computer society press .
m. yokokawa , k. itakura , a. uno , t. ishihara , and y. kaneda .
16.4-tflops direct numerical simulation of turbulence by a fourier spectral method on the earth simulator . in _ supercomputing ,
acm / ieee 2002 conference _ , pages 5050 .
ieee , 2002 .
rio yokota and l. a. barba .
treecode and fast multipole method for @xmath2-body simulation with cuda . in wen - mei hwu ,
editor , _ gpu computing gems emerald edition _ , chapter 9 , pages 113132 .
elsevier/ morgan kaufman , 2011 .
preprint on http://arxiv.org/abs/1010.1482[arxiv:1010.1482 ] .
rio yokota and l. a. barba .
-based vortex method for simulation of isotropic turbulence on gpus , compared with a spectral method . to appear in _ comput . & fluids _ , preprint on http://arxiv.org/abs/1110.2921[arxiv:1110.2921 ] , 2012 .
rio yokota , jaydeep p. bardhan , matthew g. knepley , l. a. barba , and tsuyoshi hamada .
biomolecular electrostatics using a fast multipole bem on up to 512 gpus and a billion unknowns . , 182(6):12711283 , 2011 . | this paper reports large - scale direct numerical simulations of homogeneous - isotropic fluid turbulence , achieving sustained performance of 1.08 petaflop / s on gpu hardware using single precision .
the simulations use a vortex particle method to solve the navier - stokes equations , with a highly parallel fast multipole method ( fmm ) as numerical engine , and match the current record in mesh size for this application , a cube of @xmath0 computational points solved with a spectral method . the standard numerical approach used in this field
is the pseudo - spectral method , relying on the fftalgorithm as numerical engine .
the particle - based simulations presented in this paper quantitatively match the kinetic energy spectrum obtained with a pseudo - spectral method , using a trusted code . in terms of parallel performance ,
weak scaling results show the fmm - based vortex method achieving 74% parallel efficiency on 4096 processes ( one gpu per mpiprocess , 3 gpus per node of the tsubame-2.0system ) .
the fft - based spectral method is able to achieve just 14% parallel efficiency on the same number of mpiprocesses ( using only cpu cores ) , due to the all - to - all communication pattern of the fftalgorithm .
the calculation time for one time step was 108 seconds for the vortex method and 154 seconds for the spectral method , under these conditions .
computing with 69 billion particles , this work exceeds by an order of magnitude the largest vortex - method calculations to date .
isotropic turbulence , fast multipole method , integral equations , gpu |
color confinement and chiral symmetry breaking are the striking non - perturbative phenomena in low - energy qcd , which are of particular importance in particle and nuclear physics @xcite .
several scenarios of the confinement mechanism have been proposed @xcite , where the ghost and gluon propagators in deep infrared need to be quantified , thus this requires a non - perturbative quantization of qcd .
whereas this issue has been investigated extensively , it remains challenging to comprehend the non - perturbative aspects from the first - principle calculations . in a pure su(3 ) gauge theory , the polyakov loop is an exact order parameter of the @xmath0 center symmetry and for deconfinement , which dictates a first order phase transition @xcite . in the presence of light dynamical quarks ,
the polyakov loop loses its interpretation as an order parameter and is smoothly changing with temperature .
contrary to the broad polyakov loop , a particular ratio of the polyakov loop susceptibilities retain a clear remnant of the underlying @xmath0 center symmetry fairly well even in full qcd with the physical pion mass @xcite .
thus , the ratio of the polyakov loop fluctuations can serve as observables to identify the onset of deconfinement in qcd . in the presence of light dynamical quarks , the transition from hadronic phase to quark - gluon plasma becomes crossover and accompanies partial restoration of chiral symmetry at finite temperature @xcite .
spontaneous chiral symmetry breaking is characterized by a non - vanishing condensation of quark - bilinear operators .
the low - lying dirac modes , which are the eigenmodes of the dirac operator with small eigenvalues , are known to be responsible for saturating the chiral condensate of light quarks @xmath1 , through the banks - casher relation @xcite
. in fact , at vanishing and small baryon chemical potential , the lattice qcd results suggest that there is an interplay between quark deconfinement and chiral crossovers as they take place in the same narrow temperature range @xcite .
also , in the maximally abelian gauge , confinement and chiral symmetry breaking , as well as the instantons , simultaneously disappear when the qcd monopoles are removed @xcite . on the other hand , there exist several observations that unbroken chiral symmetry does not dictate deconfinement : given a tower of hadron spectra with eliminating the low - lying dirac modes @xcite , the hadrons keep their identity even in chirally restored phase where parity doublers are all degenerate .
in addition , it has been shown in the su(3 ) lattice simulations that the low - lying modes have little contribution to the polyakov loop and to the confining force , indicating that the two features are rather independent @xcite . in the context of the interplay between confinement and chiral symmetry breaking @xcite
, it is important to make a reliable separation of one from another , whereas those phenomena are supposed to be correlated .
the apparent coincidence in change of properties of the polyakov loop fluctuation ratios and the chiral condensate and its susceptibility near the chiral crossover might indicate that there is a tied relation between confinement and chiral symmetry breaking in qcd .
however , such a possible relation has not been quantified yet . utilizing the dirac - mode expansion method formulated on the lattice @xcite ,
the low - lying modes can be systematically removed in calculating expectation values of different operators . in this paper , we apply the above expansion method to investigate the relation between confinement and chiral symmetry breaking in terms of the polyakov loop fluctuations and their ratios .
we put our particular attention to the contribution of the low - lying dirac modes to the polyakov loop fluctuations .
we derive the analytic relations between the real , imaginary and modulus of the polyakov loop and their fluctuations with the dirac modes on the temporally odd - number size lattice with periodic boundary conditions .
these analytical relations are applicable to both full and quenched qcd .
we show , through numerical simulations on the lattice in quenched qcd , that the low - lying dirac modes yield negligible contribution to the polyakov loop fluctuations . with these results ,
also not observed is a direct relation between confinement and chiral symmetry breaking in qcd through the polyakov loop fluctuation ratios .
the paper is organized as follows : in the next section we derive a set of analytic relations linking the polyakov loop fluctuations to the dirac eigenmodes . in sec .
iii , we examine the role of the low - lying dirac modes in the polyakov loop fluctuations and present our numerical results within quenched lattice qcd .
section iv is devoted to summary and conclusions .
we utilize the su(@xmath2 ) lattice qcd formalism and consider a square lattice with spacing @xmath3 .
each site is indicated by @xmath4 with @xmath5 .
a gauge field , @xmath6 with the gauge coupling @xmath7 , is introduced as a link - variable , @xmath8 .
we use spatially symmetric lattice , i.e. , @xmath9 and @xmath10 , with @xmath11 .
-0.9 cm 0.9 cm -1.9 cm for each gauge configuration , the polyakov loop @xmath12 is defined as @xmath13 where @xmath14 is the unit vector in direction of @xmath15 in the lattice unit , and @xmath16 is the 4-dimensional lattice volume , @xmath17 . under the @xmath0 rotation , the polyakov loop is transformed into @xmath18 with @xmath19 @xcite . in confined phase
, @xmath20 is taken . in deconfined phase ,
@xmath21 is chosen such that the transformed polyakov loop @xmath22 lies in its real sector .
we introduce the polyakov loop susceptibilities , @xmath23 , \label{chia}\\ t^3\chi_l=\frac{n_\sigma^3}{n_\tau^3}[\langle ( l_l)^2 \rangle-\langle l_l \rangle^2 ] , \label{chil}\\ t^3\chi_t=\frac{n_\sigma^3}{n_\tau^3}[\langle ( l_t)^2 \rangle-\langle l_t \rangle^2 ] , \label{chit}\end{aligned}\ ] ] where @xmath24 and @xmath25 , and consider their ratios , @xmath26 the polyakov loop susceptibility ratios ( [ ra ] ) and ( [ rt ] ) were shown to be excellent probes of the deconfinement phase transition in a pure gauge theory @xcite .
they are almost temperature independent above and below the transition and exhibit a discontinuity at the transition temperature .
this characteristic behavior is understood in terms of the global @xmath0 symmetry of the yang - mills lagrangian and the general properties of the polyakov loop probability distribution @xcite . in the presence of dynamical quarks ,
the polyakov loop is no longer an order parameter and stays finite even in the low temperature phase .
consequently , the ratios of the polyakov loop susceptibilities are modified due to explicit breaking of the @xmath0 center symmetry .
indeed , both @xmath27 and @xmath28 vary continuously with temperature across the chiral crossover , however @xmath27 interpolates between the two limiting values set by the pure gauge theory .
this property of @xmath27 is illustrated in fig .
[ fig1 ] . also seen in this figure is that , in spite of smoothening effects observed in the presence of quarks , there is an abrupt rate change with @xmath29 in @xmath27 near the chiral crossover @xmath30 mev @xcite .
the ratio @xmath27 in fig .
[ fig1 ] has its inflection point at @xmath31 mev , which is fairly in good agreement with the chiral crossover range calculated in lattice qcd .
such behavior of the polyakov loop fluctuation ratio can be regarded as an effective observable indicating deconfinement properties in qcd @xcite .
the apparent modification of @xmath27 and @xmath28 near chiral crossover may suggest that there are certain correlations between confinement and chiral symmetry breaking .
such correlations can be best verified when expanding the polyakov loop and its fluctuations in terms of the dirac eigenmodes . in the following ,
we formulate the relevant quantities , based on this expansion method , and study the influence of the low - lying dirac modes on the properties of the polyakov loop fluctuation ratios .
to derive the analytic relation between the polyakov loop and the dirac modes , we consider the temporally odd - number lattice with the normal non - twisted periodic boundary condition for link - variables , in both temporal and spatial directions @xcite .
we introduce the link - variable operator @xmath32 , with the matrix element @xmath33 the covariant derivative operator on the lattice is introduced as @xmath34 and the dirac operator @xmath35 with its matrix element @xmath36 , \label{diracopexp}\end{aligned}\ ] ] where @xmath37 and @xmath38 .
since the dirac operator is anti - hermitian , the dirac eigenvalue equation reads @xmath39 where @xmath40 . using the dirac eigenfunction @xmath41 ,
one arrives at the eigenvalue equation @xmath42 \ \ \ \ \ & \nonumber \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ = i\lambda_n
\psi_n(s ) . \label{diraceigenexp}\end{aligned}\ ] ] at finite temperature , imposing the temporal anti - periodicity for @xmath43 acting on quarks , it is convenient to add a minus sign to the matrix element of the temporal link - variable operator @xmath44 at the temporal boundary of @xmath45 @xcite : @xmath46 then , the polyakov loop in eq .
( [ polyakov ] ) is expressed as @xmath47 where tr@xmath48 denotes the functional trace , @xmath49 , and @xmath50 is taken over color index .
the minus sign stems from the additional minus on @xmath51 in eq.([eq : lvthermal ] ) .
note that the functional trace of a product of link - variable operators corresponding to non - closed path is exactly zero .
indeed , followed by eq.([linkop ] ) , one obtains @xmath52 with @xmath53 for any non - closed path of length @xmath54 .
this is understood from elitzur s theorem @xcite , that the vacuum expectation values of gauge - variant operators are zero . in the following ,
we show that the polyakov loop can be explicitly expanded in terms of eigenmodes of the dirac operator . and with periodic boundary condition .
the left configuration is gauge - variant , whereas the middle is gauge - invariant .
the right configuration represents a closed path with @xmath55-(link - variables ) because of the periodicity in a temporal direction , thus this path corresponds to a gauge - invariant polyakov loop . ]
we introduce the following key quantity , @xmath56 where @xmath57 , and @xmath58 is taken over spinor indexes . from the definition of @xmath59 in eq.([diracop ] ) , it is clear that @xmath60 operator is expressed by a sum of products of @xmath55- ( link - variables ) .
note that , on a square lattice , it is not possible to construct any closed loop using the products of odd number of link - variables ( see fig .
[ oddlattice ] for illustration ) . by construction
, one considers a square lattice with odd @xmath55 , thus @xmath60 in eq .
( [ i ] ) does not contain any contributions from the products of @xmath55-(link - variables ) along the closed paths .
however , due to the periodic boundary condition in time direction , the only exception is a gauge - invariant term proportional to @xmath61 .
this term is thus related with the polyakov loop operator .
based on the above discussion , and applying eqs .
( [ polyakovop ] ) , ( [ nonclosed ] ) and ( [ covariantop ] ) to eq . ( [ i ] ) , one finds that @xmath62 thus @xmath63 is directly proportional to the polyakov loop . on the other hand , since @xmath63 in eq .
( [ i ] ) is defined through the functional trace , it can be expressed in the basis of dirac eigenmodes as @xmath64 consequently , from eqs.([i1 ] ) and ( [ i2 ] ) one finds that , on the temporally odd - number lattice , there is a direct relation between the polyakov loop and the dirac modes @xcite , @xmath65 which is a dirac spectral representation of the polyakov loop .
the relation ( [ relorig ] ) is a mathematical identity , and is exactly satisfied for arbitrary gauge configurations .
consequently , this relation is valid , regardless of whether the link - variables are generated in full qcd or quenched qcd @xcite .
the relation ( [ relorig ] ) enables to investigate the contribution of different dirac - modes to the polyakov loop .
of particular interest is the role of the low - lying eigenvalues which are essential to identify chiral symmetry restoration in qcd at finite temperature . in principle , it is possible to calculate numerically the contribution of the dirac modes to the polyakov loop through eq .
( [ relorig ] ) by using monte carlo simulations .
however , a very large @xmath66dimension of a dirac operator implies also a large cost of numerical calculations .
this can be , however , remarkably reduced by use of the modified kogut - susskind(ks ) formalism @xcite .
we rewrite eq .
( [ relorig ] ) into the equivalent form @xmath67 where the ks dirac eigenstate @xmath68 is obtained by solving the eigenvalue equation , @xmath69 here , the ks dirac operator @xmath70 is defined as @xmath71 \label{ksdiracop}\end{aligned}\ ] ] with the staggered phase @xmath72 , @xmath73 consequently , each dirac - mode contribution to the polyakov loop is obtained by solving the eigenvalue equation of the ks dirac operator whose dimension is now @xmath74 , instead of @xmath66 , as in the original dirac operator . in terms of the ks dirac eigenfunction @xmath75 , the ks dirac matrix element @xmath76 is explicitly expressed as @xmath77 from the gauge transformation property of link - variables and the ks dirac eigenfunctions , the matrix element @xmath76 is gauge - invariant @xcite .
note that the modified ks formalism applied here is not an approximation , but is a method for spin - diagonalization of the dirac operator . in this study
, we do not use specific fermions , such as the ks fermion , but apply the modified ks formalism as a prescription to reduce the numerical cost .
the relations of the polyakov loop and dirac eigenmodes in eqs .
( [ relorig ] ) and ( [ relks ] ) are exact .
they are valid at finite temperature and density , and are independent of the particular implementation of fermions on the lattice @xcite , thus can be used to identify the interplay between deconfinement and chiral symmetry restoration in qcd .
the expansion of the polyakov loop in terms of the dirac eigenmodes , formulated in the previous section , can be also applied to the fluctuations of the real , imaginary and modulus of the polyakov loop .
multiplying eq .
( [ relorig ] ) by the @xmath0 factor @xmath78 , one obtains the relation between the @xmath0 transformed polyakov loop @xmath22 and the dirac modes , @xmath79 where @xmath80 is chosen such that , for each gauge configuration , the @xmath22 lies in a real sector . taking the real and the imaginary parts of eq .
( [ ltildedirac ] ) , the dirac spectral representation of the longitudinal and transverse polyakov loops reads @xmath81 respectively , whereas , taking the absolute value of eq .
( [ relorig ] ) , the following relation is also obtained ; @xmath82 since eqs .
( [ ltildedirac ] ) , ( [ lldirac ] ) , ( [ ltdirac ] ) and ( [ labsdirac ] ) are valid for each gauge configuration , the dirac spectral representation for different fluctuations of the polyakov loop and their ratios are directly obtained by substituting eqs .
( [ lldirac])-([labsdirac ] ) to eqs .
( [ chia])-([chit ] ) . as an example we quote an explicit expression for the dirac spectral representation of the @xmath83 ratio as @xmath84 where @xmath85 denotes an average over all gauge configurations .
the explicit analytic relations of the dirac spectral decomposition of the real , imaginary and modulus polyakov loop and their fluctuations are the key results of our studies . note here that , like eq.([relorig ] ) , these relations ( [ ltildedirac])-([labsdirac ] ) are applicable to both full and quenched qcd , since we just use elitzur s theorem @xcite : only gauge - invariant quantities survive . all these relations are derived on the temporally odd - number lattice for practical reasons . however , a particular choice of the parity for the lattice size in time direction does not alter the physics , since in the continuum limit , @xmath86 and @xmath87 , any number of large @xmath55 should give the same result @xcite .
in fact , similar relations are derived also on the even lattice , whereas a more compact form can be obtained on the temporally odd - number lattice @xcite .
it is however difficult to take the continuum extrapolation .
for instance , the continuum limit of the polyakov loop itself is still unsettled because of uncertainty of its renormalization .
however , at least , the ambiguity of the multiplicative renormalization of the polyakov loop can be avoided by considering the ratio of the polyakov - loop susceptibilities @xcite .
to study numerically the influence of different dirac modes on the polyakov loop fluctuations and their ratios , we further apply the modified ks formalism .
this amounts in replacing the diagonal dirac matrix element @xmath88 in eqs .
( [ lldirac])-([labsdirac ] ) by the corresponding ks dirac matrix element @xmath89 @xcite , as @xmath90 we analyze the contributions from the low - lying dirac modes to the polyakov loop fluctuations in the su(3 ) lattice qcd through monte carlo simulations . in the mathematical sense ,
all the obtained relations ( [ ltildedirac])-([labsdirac ] ) hold for both full and quenched qcd . in this paper , we perform su(3 ) lattice qcd calculations with the standard plaquette action at the quenched level on @xmath91 size lattice .
numerical studies are carried out both in confined and deconfined phases for different couplings @xmath92 , and the corresponding temperatures , @xmath93 .
we use the linear algebra package ( lapack ) @xcite in diagonalizing the ks dirac operator to obtain the eigenvalues @xmath94 and the eigenfunctions @xmath95 .
the lattice spacing @xmath3 is determined by the zero - temperature string tension of @xmath96 gev / fm on a large lattice at each @xmath97 .
in fact , we here calculate the static quark - antiquark potential @xmath98 on @xmath99 lattice at each @xmath100 , and fit it by the cornell potential , i.e. , the coulomb plus linear form @xcite , to extract the string tension @xmath101 . in confined phase ,
we fix @xmath102 on @xmath91 lattice , which corresponds to @xmath103 fm and @xmath104 mev .
we also calculate the creutz ratio @xmath105 , for an estimate of the string tension or the lattice spacing on the precise lattice @xcite , in spite of an additional contamination from the coulomb potential .
nevertheless , this value is consistent with the string tension @xmath101 obtained from the potential @xmath98 , and leads to @xmath106 fm . here
, the average plaquette value is obtained as @xmath107=0.53(2 ) , which is consistent with the previous su(3 ) lattice studies @xcite . in deconfined phase
, the simulations are performed at @xmath108 on @xmath109 lattice , i.e. , for @xmath110 fm and @xmath111 mev . on this lattice ,
the average plaquette value is @xmath107=0.60(2 ) , which is also consistent with the previous works @xcite . for each value of @xmath100
, we use 20 gauge configurations , which are taken every 500 sweeps after the thermalization of 5000 sweeps . since the polyakov loop and its different fluctuations are expressed as the sums over all dirac - modes , we divide the entire tower of the dirac eigenvalues into the low- and higher - lying modes with the insertion of the infrared cutoff @xmath112 .
based on eq.([relks ] ) , we introduce the @xmath112dependent polyakov loops , @xmath113 for the modulus , and @xmath114 for the real and the imaginary part , respectively , with @xmath115 . applying the cutoff dependent polyakov loops from eqs .
( [ labsdiraccutks ] ) , ( [ lldiraccutks ] ) and ( [ ltdiraccutks ] ) to eqs . ( [ chia])-([chit ] )
, we also introduce the @xmath112dependent susceptibilities @xmath116 , \label{chiacut}\end{aligned}\ ] ] where @xmath117 stands for @xmath118 , @xmath119 or @xmath120 , and their ratios @xmath121 to differentiate the importance of the low - lying dirac modes on the properties of the polyakov loop fluctuations and the chiral condensate , we introduce the cutoff - dependent chiral condensate @xmath122 . in terms of the dirac modes ,
the @xmath122 is expressed as @xcite @xmath123 where @xmath124 is the current quark mass .
the chiral condensate is strongly affected by the low - lying dirac modes . taking a typical value for the infrared cutoff @xmath125 and the quark mass @xmath126 , leads to a drastic reduction of the chiral condensate @xmath127 in a confined phase at @xmath128 @xcite .
figure [ diraceigendist ] shows the dirac eigenvalue distribution @xmath129 in confined ( @xmath130 ) and deconfined ( @xmath131 ) phases .
note here that the near - zero dirac - mode density @xmath132 is apparently finite in confined phase , whereas it is highly suppressed in deconfined phase .
we show in fig .
[ mass_vs_qbarq ] the bare chiral condensate @xmath133 per a flavor in the confined phase as a function of the quark mass @xmath124 in the lattice unit . the chiral condensate remains finite in the small-@xmath124 region .
thus , from figs .
[ diraceigendist ] and [ mass_vs_qbarq ] it is clear that the chiral symmetry is definitely broken in confined phase , whereas it is restored in deconfined phase . in the lattice unit : ( a ) the confinement phase at @xmath134 ( i.e. , @xmath103 fm ) on @xmath91 ; ( b ) the deconfinement phase at @xmath108 ( i.e. , @xmath135 fm ) on @xmath91 @xcite .
, title="fig : " ] in the lattice unit : ( a ) the confinement phase at @xmath134 ( i.e. , @xmath103 fm ) on @xmath91 ; ( b ) the deconfinement phase at @xmath108 ( i.e. , @xmath135 fm ) on @xmath91 @xcite . , title="fig : " ] per a flavor in the confinement phase plotted against the quark mass @xmath124 in the lattice unit .
the lattice qcd calculation is done at @xmath130 ( i.e. , @xmath103 fm ) on @xmath136.,width=302 ] in the same spirit , we introduce the ratio , @xmath137 to quantify the sensitivity of the polyakov loop fluctuations to the particular dirac modes .
when , with some @xmath112 , the ratio stays @xmath138 , then the low - lying dirac modes below the cutoff @xmath112 have a negligible contribution to the polyakov loop fluctuations . in fig .
[ ratio_conf ] , we show the monte carlo results for @xmath139 in a confined phase at @xmath130 , for various values of the infrared cutoff @xmath112 . for the sake of comparison , we also show in fig .
[ ratio_conf ] the @xmath140 ratio , calculated at the same temperature and with the light quark mass , @xmath141 . the ratios , @xmath139 and @xmath140 , indicate the influence of removing the low - lying dirac modes with the infrared cutoff @xmath112 on confinement and chiral symmetry breaking , respectively . and @xmath139 ratio from eqs .
( [ r_chiral ] ) and ( [ r_conf ] ) , respectively , as a function of an infrared cutoff @xmath112 introduced on dirac eigenvalues , expressed in lattice units .
the monte carlo calculations have been performed on @xmath136 lattice at @xmath130 , and for the quark mass of @xmath141 .
, width=302 ] from fig .
[ ratio_conf ] , it is clear that the @xmath140 ratio is strongly reduced by removing the low - lying dirac modes .
thus , the low - lying dirac modes , which are important modes for chiral symmetry breaking , are also dominant to quantify the chiral condensate .
in contrast to @xmath140 , the @xmath139 ratio is almost unchanged when removing the low - lying dirac modes even with relatively large cutoff @xmath142 .
thus , the essential modes for chiral symmetry breaking in qcd are not important to quantify the polyakov loop fluctuation ratios , which are sensitive observables to confinement properties in qcd .
the same result is also found in deconfined phase , as seen in table [ confircut ] , which summarizes our numerical results on different fluctuations of the polyakov loop and their ratios , obtained on the lattice at @xmath130 and @xmath131 , with 20 gauge configurations .
note here that the analytical relation ( [ relorig ] ) and the subsequent formulae of eqs.([lldirac])-([labsdirac ] ) hold for each gauge configuration , and the contribution from the low - lying dirac modes to the polyakov loop @xmath12 is found to be negligible @xcite .
this fact inevitably leads to almost equivalence between the polyakov - loop fluctuations and those without the low - lying dirac modes , although the statistical error is significant because of the small statistics .
cccc @xmath100 & & original & ir - removed + @xmath100=5.6 & @xmath143 & @xmath144 & @xmath145 + & @xmath146 & @xmath147 & @xmath148 + & @xmath149 & @xmath150 & @xmath151 + & @xmath27 & 0.6548 & 0.6549 + & @xmath28 & 1.131 & 1.131 + + @xmath100=6.0 & @xmath143 & @xmath152 & @xmath152 + & @xmath146 & @xmath153 & @xmath153 + & @xmath149 & @xmath154 & @xmath154 + & @xmath27 & 0.9834 & 0.9834 + & @xmath28 & 0.2603 & 0.2603 + [ confircut ] the differences in the influence of the low - lying dirac modes on the chiral condensate and the polyakov loop fluctuations can be understood semi - analytically . from eqs .
( [ lldirac])-([labsdirac ] ) , it is clear , that the contribution of the low - lying dirac - modes with @xmath155 is suppressed , relative to the higher - lying dirac modes , due to the damping factor @xmath156 .
in fact , a dirac matrix element @xmath157 does not yield a stronger singularity than @xmath158 , therefore the contribution from the low - lying dirac modes to the polyakov loop @xcite , as well as , to its fluctuations is negligible .
hence , the essential modes for chiral symmetry breaking do not contribute to a sensitive probe for deconfinement in qcd .
thus , this finding suggests no direct one - to - one correspondence between confinement and chiral symmetry breaking in qcd .
the main objective of these studies was to establish the relation between the polyakov loop and its fluctuations with the eigenmodes of a dirac operator . based on the lattice qcd formalism , we have derived a dirac spectral representation of the real , imaginary and the modulus of the polyakov loop and their fluctuations . although the formulation was done on a temporally odd - number lattice , this choice of the parity for the lattice size does not alter the physics in the continuum limit with any large number of @xmath55 .
the analytical decomposition of the polyakov loop and its fluctuations is fully general .
it is independent from the gauge group , the implementation of fermions on the lattice , and is also valid at finite baryon density . to quantify the influence of dirac modes over the polyakov loop fluctuations , we have performed monte carlo simulations in the su(3 ) lattice qcd .
our calculations were carried out with the standard plaquette action at the quenched level on @xmath159size lattice at two different temperatures , corresponding to confined and deconfined phases .
we have shown that the low - lying dirac modes have negligible contribution to the polyakov loop fluctuations .
this result is intact both in confined and deconfined phases . on the other hand ,
the low - lying dirac modes are essential , in both phases , to quantify the chiral condensate . these findings ,
both in analytical formulas and in numerical calculations , suggest no direct , one - to - one correspondence between confinement and chiral symmetry breaking in qcd .
however , this does not exclude a coincidence of these two properties in qcd since the abrupt change of the ground state from chiral symmetry broken to restored phase may drive the onset of deconfinement .
the above conclusion is based on the numerical simulations on a rather small - size lattice , being far from a continuum limit .
thus , our result on the polyakov loop fluctuations suffers from finite size effects .
such effects can certainly modify the values of fluctuations at a given temperature , however will not change our conclusion on the influence of the low - lying dirac modes on their properties .
the low - lying dirac modes have negligible contribution to the polyakov - loop fluctuations because of the damping factor @xmath156 which appears in eq.([relorig ] ) .
although our numerical calculation was performed at the quenched level , the derived analytic formulae are fulfilled even in the presence of dynamical quarks .
it is one of the future prospects to perform full - qcd simulations in the present formalism to further justify our conclusion .
in addition , the derived analytic relations connecting the polyakov loop and dirac modes are mathematically exact for arbitrary odd temporal size @xmath55 .
thus we expect that our conclusion is rather robust in the continuum limit @xcite .
moreover , the ambiguity of the multiplicative renormalization of the polyakov loop has been avoided in the ratio of the polyakov loop susceptibilities .
yet , it is left as an important but difficult task to extrapolate these analytic relations to the continuum .
in addition to the polyakov loop fluctuations , there are further observables which are linked to deconfinement properties in qcd and show abrupt , but smooth change across the chiral crossover .
one of such observables is the kurtosis of the net - quark number fluctuations @xcite .
besides , the qcd monopole , in the maximally abelian gauge , is a relevant degree of freedom in the low - energy qcd @xcite , and plays a fundamental role for non - perturbative phenomena such as confinement and chiral symmetry breaking .
thus , from the future perspectives , it would be of particular interest to investigate such quantities in terms of the dirac - mode expansion and to explore the influence of the low lying eigenmodes on their properties near the chiral crossover .
k. r. acknowledges partial support of the u.s .
department of energy under grant no .
de - fg02 - 05er41367 , and fruitful discussions with bengt friman and pok man lo .
d. is supported by grant - in - aid for jsps fellows ( grant no .
15j02108 ) , and h. s. is supported by the grants - in - aid for scientific research ( grant no .
15k05076 ) from japan society for the promotion of science . the work of k. r. and c. s. has been partly supported by the polish science foundation ( ncn ) under maestro grant no .
dec-2013/10/a / st2/00106 and by the hessian loewe initiative through the helmholtz international center for fair .
the lattice qcd calcula- tions were performed on nec - sx8r and nec - sx9 at osaka university .
h. suganuma , t. m. doi , t. iritani , arxiv:1404.6494 [ hep - lat ] ; proc .
( lattice 2013 ) , 374 ( 2013 ) ; proc .
( qcd - tnt - iii ) , 042 ( 2014 ) ; epj web of conf . * 71 * , 00129 ( 2014 ) ; proc .
( hadron 2013 ) , 121 ( 2014 ) . | we investigate correlations of the polyakov loop fluctuations with eigenmodes of the lattice dirac operator .
their analytic relations are derived on the temporally odd - number size lattice with the normal non - twisted periodic boundary condition for the link - variables .
we find that the low - lying dirac modes yield negligible contributions to the polyakov loop fluctuations .
this property is confirmed to be valid in confined and deconfined phase by numerical simulations in su(3 ) quenched qcd .
these results indicate that there is no direct , one - to - one correspondence between confinement and chiral symmetry breaking in qcd in the context of different properties of the polyakov loop fluctuation ratios . /#1#1
# 1#1 # 1#1 |
determining the hubble constant ( @xmath0 ) accurately is one of the most important challenges in modern cosmology , since it sets the scale for all cosmological times and distances , and it reveals for human beings the present cosmic expansion rate , the size and age of the universe and the cosmic components .
based on the early determination by hubble @xcite , the value of @xmath0 was believed to lie in the large range [ 50 , 100 ] km s@xmath2 mpc@xmath2 @xcite . utilizing improved control of systematics and different calibration techniques ,
the first precise value , @xmath3 km s@xmath2 mpc@xmath2 @xcite , was given by the local measurements from the hubble space telescope ( hst ) in 2001 .
ten years later , riess et al .
( hereafter r11 ) calibrated the type ia supernovae ( sne ia ) and obtained @xmath4 km s@xmath2 mpc@xmath2 @xcite by using three indicators , i.e. , the distance to ngc 4258 from a mega - maser measurement , the trigonometric parallaxes measurements to the milk way ( mw ) cepheids , cepheid observations and a modified distance to the large magellanic cloud ( lmc ) . in 2012 , there are three groups to measure the hubble constant : riess et al
. got @xmath5 km s@xmath2 mpc@xmath2 by utilizing the cepheids in m31 @xcite ; freedman et al .
obtained @xmath6 km s@xmath2 mpc@xmath2 by using a mid - infrared calibration for the cepheids @xcite .
subsequently , in 2013 , @xmath7 km s@xmath2 mpc@xmath2 derived by planck @xcite from the anisotropies of the cosmic microwave background ( cmb ) gave a strong tension with the local measurement from r11 s result at 2.4@xmath8 confidence level . in order to resolve or alleviate the tension , several groups carried out the measurements utilizing different techniques and methods : bennett et al .
@xcite and hinshaw et al .
@xcite gave a @xmath9 determination , i.e. , @xmath10 km s@xmath2 mpc@xmath2 by using the nine - year wilkinson microwave anisotropy probe ( wmap-9 ) measurements ; spergel et al .
@xcite found @xmath11 km s@xmath2 mpc@xmath2 by removing the @xmath12 ghz detector set spectrum used in the planck analysis ; fiorentino et al .
@xcite obtained @xmath13 km s@xmath2 mpc@xmath2 by utilizing 8 new classical cepheids observed in galaxies hosting sne ia ; different from the calibration method exhibited by r11 , tammann et al .
@xcite got a lower value @xmath14 km s@xmath2 mpc@xmath2 by calibrating the sn ia with the tip of red - giant branch ( trgb ) ; efstathiou @xcite obtained @xmath15 km s@xmath2 mpc@xmath2 by revising the geometric maser distance to ngc 4258 from humphreys et al . @xcite and using this indicator to calibrate the r11 s measurements ; rigault et al .
@xcite gave @xmath16 km s@xmath2 mpc@xmath2 by considering predominately star - forming environments .
the medium redshift data can also act as an complementary and effective tool to determine @xmath0 . combining them with the high - redshift cmb data ,
the uncertainties of @xmath0 can be reduced obviously . by making the best of baryon acoustic oscillations ( bao ) data , cheng et al .
@xcite found @xmath11 km s@xmath2 mpc@xmath2 for the standard cosmological model . utilizing the cmb and bao data and assuming the six - parameter standard cosmology , bennett et al .
@xcite got a substantially accurate result @xmath17 km s@xmath2 mpc@xmath2 .
subsequently , using other mid - redshift data including the bao peak at @xmath18 @xcite , 18 @xmath19 data points @xcite , 11 ages of old high - redshift galaxies @xcite and the angular diameter distance data from the bonamente et al .
galaxy cluster sample @xcite , lima et al .
obtained @xmath20 km s@xmath2 mpc@xmath2 @xcite in a @xmath21cdm model .
furthermore , replacing the bonamente et al .
galaxy cluster sample with the filippis et al .
one @xcite , holanda et al .
@xcite gave @xmath22 km s@xmath2 mpc@xmath2 .
this indicates that different mid - redshift data can provide different values of @xmath0 .
moreover , based on the fact that different observations should provide the same luminosity distance at some certain redshift , wu et al .
@xcite proposed a model - independent method to determine @xmath0 .
they found @xmath20 km s@xmath2 mpc@xmath2 by combining the union 2.1 sne ia data with galaxy cluster data @xcite .
recently , the improved local measurement @xmath23 km s@xmath2 mpc@xmath2 from riess et al .
2016 @xcite gives a stronger tension with the planck 2016 release @xmath24 km s@xmath2 mpc@xmath2 @xcite ( hereafter p16 ) at @xmath25 confidence level .
the improvements different from r11 s result are summarized in the following manner : @xmath26 adopting new , near - infrared observations of cepheid variables in 11 sne ia hosts ; @xmath26 increasing the sample size of ideal sne ia calibrators from 8 to 19 ; @xmath26 giving the calibration for a magnitude - redshift relation based on 300 sne ia at @xmath27 ; @xmath26 a @xmath28 reduction of the systematic uncertainty in the maser distance to the ngc 4258 ; @xmath26 increasing the sample size of cepheids in the lmc ; @xmath26 hst observations of cepheids in m31 ; @xmath26 using new hst - based trigonometric parallaxes for the mw cepheids ; @xmath26 a more robust distance to the large magellanic cloud ( lmc ) based on the late - type detached eclipsing binaries ( debs ) . in light of the higher tension than before between the local and global measurements of @xmath0
, we use the latest hii galaxy data to determine the hubble constant here .
we find that the @xmath0 values are more consistent with the r16 s result at @xmath1 confidence level by utilizing a combination of model - dependent and model - independent methods , and that the data prefers a higher underlying value of @xmath0 than r16 s result .
the rest of this paper is outlined as follows . in the next section
, we describe the data we use in this analysis . in sec . 3 ,
we constrain five cosmological models . in sec . 4
, we use the model - independent gp reconstructions to check the correctness of the @xmath0 values from the model - dependent method .
the discussions and conclusions are presented in the final section ( we use units @xmath29 ) .
our sample consists of 24 giant extragalactic hii regions ( gehr ) at redshifts @xmath30 analyzed in @xcite , and 107 low-@xmath31 hii galaxies vividly described in @xcite .
this compilation also includes 6 high-@xmath31 star - forming galaxies in the redshift range @xmath32 $ ] obtained via the x - shooter spectrography at the cassegrain focus of the european southern observatory very large telescope ( eso - vlt ) @xcite .
we also add 19 high-@xmath31 objects1 from @xcite , 6 from @xcite and 12 from @xcite into our sample , therefore , we have 25 high-@xmath31 hii galaxy measurements @xcite . in total , our curent sample includes 156 objects which are shown in table .
[ t1 ] cccc + & & & + continued from previous page + & & & + + & gehr + @xmath33 & @xmath34 & @xmath35 & @xcite + @xmath33 & @xmath36 & @xmath37 & @xcite + @xmath38 & @xmath39 & @xmath40 & @xcite + @xmath41 & @xmath42 & @xmath43 & @xcite + @xmath44 & @xmath45 & @xmath46 & @xcite + @xmath44 & @xmath47 & @xmath48 & @xcite + @xmath49 & @xmath50 & @xmath51 & @xcite + @xmath52 & @xmath53 & @xmath54 & @xcite + @xmath55 & @xmath56 & @xmath57 & @xcite + @xmath41 & @xmath58 & @xmath59 & @xcite + @xmath41 & @xmath60 & @xmath61 & @xcite + @xmath52 & @xmath60 & @xmath62 & @xcite + @xmath63 & @xmath64 & @xmath65 & @xcite + @xmath49 & @xmath66 & @xmath67 & @xcite + @xmath55 & @xmath64 & @xmath68 & @xcite + @xmath41 & @xmath69 & @xmath70 & @xcite + @xmath63 & @xmath71 & @xmath72 & @xcite + @xmath55 & @xmath73 & @xmath74 & @xcite + @xmath38 & @xmath75 & @xmath76 & @xcite + @xmath55 & @xmath77 & @xmath78 & @xcite + @xmath63 & @xmath79 & @xmath80 & @xcite + @xmath63 & @xmath81 & @xmath82 & @xcite + @xmath63 & @xmath83 & @xmath84 & @xcite + @xmath63 & @xmath85 & @xmath86 & @xcite + & low-@xmath31 hii galaxies + @xmath87 & @xmath88 & @xmath89 & @xcite + @xmath90 & @xmath91 & @xmath92 & @xcite + @xmath93 & @xmath94 & @xmath95 & @xcite + @xmath90 & @xmath91 & @xmath92 & @xcite + @xmath96 & @xmath97 & @xmath98 & @xcite + @xmath99 & @xmath100 & @xmath101 & @xcite + @xmath102 & @xmath103 & @xmath104 & @xcite + @xmath105 & @xmath106 & @xmath107 & @xcite + @xmath108 & @xmath109 & @xmath110 & @xcite + @xmath111 & @xmath112 & @xmath113 & @xcite + @xmath114 & @xmath115 & @xmath116 & @xcite + @xmath117 & @xmath118 & @xmath119 & @xcite + @xmath120 & @xmath121 & @xmath122 & @xcite + @xmath123 & @xmath124 & @xmath125 & @xcite + @xmath126 & @xmath127 & @xmath128 & @xcite + @xmath129 & @xmath130 & @xmath131 & @xcite + @xmath132 & @xmath133 & @xmath134 & @xcite + @xmath135 & @xmath136 & @xmath137 & @xcite + @xmath138 & @xmath139 & @xmath140 & @xcite + @xmath141 & @xmath142 & @xmath143 & @xcite + @xmath144 & @xmath145 & @xmath146 & @xcite + @xmath147 & @xmath148 & @xmath149 & @xcite + @xmath150 & @xmath151 & @xmath152 & @xcite + @xmath153 & @xmath154 & @xmath155 & @xcite + @xmath156 & @xmath157 & @xmath158 & @xcite + @xmath159 & @xmath160 & @xmath161 & @xcite + @xmath162 & @xmath163 & @xmath164 & @xcite + @xmath165 & @xmath166 & @xmath167 & @xcite + @xmath168 & @xmath169 & @xmath170 & @xcite + @xmath171 & @xmath172 & @xmath173 & @xcite + @xmath174 & @xmath175 & @xmath176 & @xcite + @xmath177 & @xmath178 & @xmath179 & @xcite + @xmath180 & @xmath181 & @xmath182 & @xcite + @xmath183 & @xmath184 & @xmath185 & @xcite + @xmath186 & @xmath187 & @xmath188 & @xcite + @xmath189 & @xmath190 & @xmath191 & @xcite + @xmath192 & @xmath193 & @xmath194 & @xcite + @xmath195 & @xmath196 & @xmath197 & @xcite + @xmath198 & @xmath199 & @xmath200 & @xcite + @xmath123 & @xmath201 & @xmath202 & @xcite + @xmath203 & @xmath204 & @xmath205 & @xcite + @xmath206 & @xmath207 & @xmath208 & @xcite + @xmath209 & @xmath210 & @xmath211 & @xcite + @xmath212 & @xmath213 & @xmath214 & @xcite + @xmath215 & @xmath216 & @xmath217 & @xcite + @xmath218 & @xmath219 & @xmath220 & @xcite + @xmath221 & @xmath222 & @xmath223 & @xcite + @xmath224 & @xmath225 & @xmath226 & @xcite + @xmath227 & @xmath228 & @xmath229 & @xcite + @xmath230 & @xmath231 & @xmath232 & @xcite + @xmath209 & @xmath210 & @xmath211 & @xcite + @xmath233 & @xmath234 & @xmath235 & @xcite + @xmath236 & @xmath237 & @xmath238 & @xcite + @xmath239 & @xmath240 & @xmath241 & @xcite + @xmath242 & @xmath243 & @xmath244 & @xcite + @xmath245 & @xmath246 & @xmath247 & @xcite + @xmath248 & @xmath249 & @xmath250 & @xcite + @xmath251 & @xmath252 & @xmath253 & @xcite + @xmath254 & @xmath255 & @xmath256 & @xcite + @xmath257 & @xmath258 & @xmath259 & @xcite + @xmath260 & @xmath261 & @xmath262 & @xcite + @xmath263 & @xmath264 & @xmath265 & @xcite + @xmath266 & @xmath267 & @xmath268 & @xcite + @xmath269 & @xmath270 & @xmath271 & @xcite + @xmath272 & @xmath273 & @xmath274 & @xcite + @xmath275 & @xmath276 & @xmath277 & @xcite + @xmath278 & @xmath279 & @xmath280 & @xcite + @xmath281 & @xmath282 & @xmath283 & @xcite + @xmath284 & @xmath285 & @xmath286 & @xcite + @xmath287 & @xmath288 & @xmath289 & @xcite + @xmath290 & @xmath291 & @xmath292 & @xcite + @xmath293 & @xmath294 & @xmath295 & @xcite + @xmath296 & @xmath297 & @xmath298 & @xcite + @xmath299 & @xmath300 & @xmath301 & @xcite + @xmath302 & @xmath303 & @xmath304 & @xcite + @xmath305 & @xmath306 & @xmath307 & @xcite + @xmath308 & @xmath309 & @xmath310 & @xcite + @xmath311 & @xmath312 & @xmath313 & @xcite + @xmath314 & @xmath315 & @xmath316 & @xcite + @xmath317 & @xmath318 & @xmath319 & @xcite + @xmath320 & @xmath321 & @xmath322 & @xcite + @xmath323 & @xmath324 & @xmath325 & @xcite + @xmath326 & @xmath327 & @xmath328 & @xcite + @xmath329 & @xmath330 & @xmath268 & @xcite + @xmath331 & @xmath332 & @xmath333 & @xcite + @xmath334 & @xmath335 & @xmath336 & @xcite + @xmath337 & @xmath338 & @xmath339 & @xcite + @xmath340 & @xmath341 & @xmath342 & @xcite + @xmath343 & @xmath344 & @xmath345 & @xcite + @xmath346 & @xmath347 & @xmath348 & @xcite + @xmath349 & @xmath350 & @xmath351 & @xcite + @xmath352 & @xmath353 & @xmath354 & @xcite + @xmath355 & @xmath356 & @xmath357 & @xcite + @xmath358 & @xmath359 & @xmath360 & @xcite + @xmath361 & @xmath362 & @xmath363 & @xcite + @xmath364 & @xmath365 & @xmath366 & @xcite + @xmath367 & @xmath368 & @xmath369 & @xcite + @xmath370 & @xmath228 & @xmath371 & @xcite + @xmath372 & @xmath373 & @xmath374 & @xcite + @xmath375 & @xmath376 & @xmath377 & @xcite + @xmath378 & @xmath379 & @xmath380 & @xcite + @xmath381 & @xmath382 & @xmath383 & @xcite + @xmath384 & @xmath385 & @xmath386 & @xcite + @xmath387 & @xmath388 & @xmath389 & @xcite + @xmath390 & @xmath391 & @xmath392 & @xcite + @xmath393 & @xmath394 & @xmath395 & @xcite + @xmath396 & @xmath397 & @xmath398 & @xcite + @xmath399 & @xmath400 & @xmath401 & @xcite + & high-@xmath31 hii galaxies + @xmath402 & @xmath403 & @xmath404 & @xcite + @xmath405 & @xmath406 & @xmath407 & @xcite + @xmath408 & @xmath409 & @xmath410 & @xcite + @xmath411 & @xmath412 & @xmath413 & @xcite + @xmath414 & @xmath415 & @xmath416 & @xcite + @xmath417 & @xmath418 & @xmath419 & @xcite + @xmath420 & @xmath421 & @xmath422 & @xcite + @xmath423 & @xmath424 & @xmath425 & @xcite + @xmath426 & @xmath427 & @xmath428 & @xcite + @xmath429 & @xmath430 & @xmath431 & @xcite + @xmath432 & @xmath433 & @xmath434 & @xcite + @xmath435 & @xmath436 & @xmath437 & @xcite + @xmath438 & @xmath439 & @xmath440 & @xcite + @xmath441 & @xmath442 & @xmath443 & @xcite + @xmath444 & @xmath445 & @xmath446 & @xcite + @xmath447 & @xmath448 & @xmath449 & @xcite + @xmath450 & @xmath451 & @xmath452 & @xcite + @xmath453 & @xmath454 & @xmath455 & @xcite + @xmath456 & @xmath457 & @xmath458 & @xcite + @xmath459 & @xmath460 & @xmath461 & @xcite + @xmath462 & @xmath463 & @xmath464 & @xcite + @xmath465 & @xmath466 & @xmath467 & @xcite + @xmath468 & @xmath469 & @xmath470 & @xcite + @xmath471 & @xmath472 & @xmath473 & @xcite + @xmath474 & @xmath475 & @xmath476 & @xcite + .1@xmath8 confidence range of free parameters of five different cosmological models . [
cols="^,^,^,^,^,^,^",options="header " , ] [ t2 ] in [ ] , the authors has pointed out that , for gehr and hii galaxies , the @xmath477 relation can be applied into measuring the distance , and it can be expressed as @xmath478 where @xmath479 and @xmath480 denote the balmer emission line luminosity for these objects and the velocity dispersion of the young star - forming cluster from the measurements of the line width , respectively .
then the corresponding observational distance modulus is shown as @xmath481 where @xmath482 is the measured flux in the @xmath483 line .
the theoretical distance modulus @xmath484 for a gehr or hii galaxy can be written as @xmath485 the hubble luminosity distance in a friedmann - robertson - walker ( frw ) universe is @xmath486 where @xmath487 denotes the model parameters , @xmath0 is the hubble constant , the dimensionless hubble parameter @xmath488 , the present - day cosmic curvature @xmath489 , and for @xmath490 , @xmath491 , which corresponds to a closed , flat and open universe , respectively . in order to constrain a specific cosmological model , we adopt the maximal likelihood method and the corresponding @xmath492 function to be minimized for the gehr and hii galaxy data is @xmath493 ^ 2 , \label{5}\ ] ] where @xmath494 and @xmath495 are the @xmath1 statistical error and the observed value of distance modulus at a given redshift @xmath496 for every object , respectively . and
2@xmath8 confidence ranges for parameter pairs ( @xmath0 , @xmath497 ) of the @xmath21cdm model , ( @xmath0 , @xmath497 ) of the non - flat @xmath21cdm model , ( @xmath0 , @xmath498 ) of the @xmath498cdm model , ( @xmath0 , @xmath498 ) of the @xmath498cdm model , ( @xmath0 , @xmath497 ) of the dv model and ( @xmath0 , @xmath499 ) of the hde model , respectively .
the blue symbols `` @xmath500 '' represent the best - fit points.,title="fig : " ] and 2@xmath8 confidence ranges for parameter pairs ( @xmath0 , @xmath497 ) of the @xmath21cdm model , ( @xmath0 , @xmath497 ) of the non - flat @xmath21cdm model , ( @xmath0 , @xmath498 ) of the @xmath498cdm model , ( @xmath0 , @xmath498 ) of the @xmath498cdm model , ( @xmath0 , @xmath497 ) of the dv model and ( @xmath0 , @xmath499 ) of the hde model , respectively .
the blue symbols `` @xmath500 '' represent the best - fit points.,title="fig : " ] and 2@xmath8 confidence ranges for parameter pairs ( @xmath0 , @xmath497 ) of the @xmath21cdm model , ( @xmath0 , @xmath497 ) of the non - flat @xmath21cdm model , ( @xmath0 , @xmath498 ) of the @xmath498cdm model , ( @xmath0 , @xmath498 ) of the @xmath498cdm model , ( @xmath0 , @xmath497 ) of the dv model and ( @xmath0 , @xmath499 ) of the hde model , respectively .
the blue symbols `` @xmath500 '' represent the best - fit points.,title="fig : " ] and 2@xmath8 confidence ranges for parameter pairs ( @xmath0 , @xmath497 ) of the @xmath21cdm model , ( @xmath0 , @xmath497 ) of the non - flat @xmath21cdm model , ( @xmath0 , @xmath498 ) of the @xmath498cdm model , ( @xmath0 , @xmath498 ) of the @xmath498cdm model , ( @xmath0 , @xmath497 ) of the dv model and ( @xmath0 , @xmath499 ) of the hde model , respectively .
the blue symbols `` @xmath500 '' represent the best - fit points.,title="fig : " ] and 2@xmath8 confidence ranges for parameter pairs ( @xmath0 , @xmath497 ) of the @xmath21cdm model , ( @xmath0 , @xmath497 ) of the non - flat @xmath21cdm model , ( @xmath0 , @xmath498 ) of the @xmath498cdm model , ( @xmath0 , @xmath498 ) of the @xmath498cdm model , ( @xmath0 , @xmath497 ) of the dv model and ( @xmath0 , @xmath499 ) of the hde model , respectively .
the blue symbols `` @xmath500 '' represent the best - fit points.,title="fig : " ] cdm model , non - flat @xmath21cdm model , @xmath498cdm model , dv model and hde model , respectively . ] and @xmath501 correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
lower panels ( from left to right ) : the reconstructions of the dark energy eos correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
the points with red error bars are 156 hii galaxy measurements .
the shaded regions are reconstructions with @xmath504 and @xmath505 confidence level .
the blue lines and magenta lines represent the underlying true model ( the mean value of reconstructions ) and the @xmath21cdm model , respectively .
, title="fig : " ] and @xmath501 correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
lower panels ( from left to right ) : the reconstructions of the dark energy eos correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
the points with red error bars are 156 hii galaxy measurements .
the shaded regions are reconstructions with @xmath504 and @xmath505 confidence level .
the blue lines and magenta lines represent the underlying true model ( the mean value of reconstructions ) and the @xmath21cdm model , respectively . ,
title="fig : " ] and @xmath501 correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
lower panels ( from left to right ) : the reconstructions of the dark energy eos correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
the points with red error bars are 156 hii galaxy measurements .
the shaded regions are reconstructions with @xmath504 and @xmath505 confidence level .
the blue lines and magenta lines represent the underlying true model ( the mean value of reconstructions ) and the @xmath21cdm model , respectively .
, title="fig : " ] and @xmath501 correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
lower panels ( from left to right ) : the reconstructions of the dark energy eos correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
the points with red error bars are 156 hii galaxy measurements .
the shaded regions are reconstructions with @xmath504 and @xmath505 confidence level .
the blue lines and magenta lines represent the underlying true model ( the mean value of reconstructions ) and the @xmath21cdm model , respectively .
, title="fig : " ] and @xmath501 correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
lower panels ( from left to right ) : the reconstructions of the dark energy eos correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
the points with red error bars are 156 hii galaxy measurements .
the shaded regions are reconstructions with @xmath504 and @xmath505 confidence level .
the blue lines and magenta lines represent the underlying true model ( the mean value of reconstructions ) and the @xmath21cdm model , respectively .
, title="fig : " ] and @xmath501 correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
lower panels ( from left to right ) : the reconstructions of the dark energy eos correspond to the cases of @xmath24 , @xmath502 and @xmath503 km s@xmath2 mpc@xmath2 , respectively .
the points with red error bars are 156 hii galaxy measurements .
the shaded regions are reconstructions with @xmath504 and @xmath505 confidence level .
the blue lines and magenta lines represent the underlying true model ( the mean value of reconstructions ) and the @xmath21cdm model , respectively .
, title="fig : " ]
in this section , for the purpose to explore the values of @xmath0 , we place constraints on five cosmological models by using the latest hii galaxy measurements .
the hubble parameter for the spatially flat @xmath21-cold - dark - matter ( @xmath21cdm ) model is @xmath506 while for the non - flat @xmath21cdm model it can be expressed as @xmath507 where @xmath497 and @xmath508 denote dimensionless matter density ratio parameter and the dimensionless curvature density ratio parameter today , respectively .
we consider the simplest parameterization of dark energy equation of state ( eos ) @xmath509 , namely the @xmath498cdm model , and the corresponding hubble parameter for the spatially flat @xmath498cdm model can be written as @xmath510 another interesting model is the so - called decaying vacuum ( dv ) model , which is based on the a simple assumption about the form of the modified matter expansion rate @xcite .
the hubble parameter for the dv model is @xmath511 where @xmath512 is a small positive constant describing the deviation from the standard matter expansion rate .
we also take into account the holographic dark energy ( hde ) model inspired by the extraordinary thermodynamics of black holes @xcite , and the corresponding hubble parameter can be shown as @xmath513 where the parameter @xmath499 plays a main role in the hde model .
the minimal values of the derived @xmath492 and the best - fit values of the parameters for these five cosmological models are listed in table .
we also exhibit the @xmath1 and @xmath514 contour plots in fig .
one can easily find that , by using the latest hii galaxy data , the @xmath0 values for these five cosmological models are all consistent with the r16 s local measurement at @xmath1 confidence level .
in addition , we also conclude that these five models prefer a higher best - fit @xmath0 value than r16 s result .
furthermore , we find that these five models fit the current data well and can not be distinguished from each other in the hii galaxies hubble diagram ( see fig .
this means that there exists a very high degeneracy among these five models using only the latest hii galaxy data .
we have adopted a model - dependent method to determine the @xmath0 values for 5 different cosmological models by utilizing the latest hii galaxy data . in this section
, we would like to use the model - independent gp method to check the correctness of the @xmath0 values from the model - dependent method .
we use the publicly available package gapp ( gaussian processes in python ) to implement our reconstruction @xcite .
the gp is a generalization of a gaussian distribution , which is a distribution of a random variable , and exhibits a distribution over functions .
the gp can reconstruct directly a function from the observed data without assuming a specific parameterization for the underlying function . at each reconstruction point
@xmath31 , the reconstructed function @xmath515 is a gaussian distribution with a mean value and gaussian error .
the key of the gp is a covariance function @xmath516 depending on two hyper - parameters @xmath517 and @xmath518 , which characterize the coherent scale of the correlation in @xmath519-direction and typical change in @xmath520-direction , respectively . in @xcite ,
the authors has verified the matrn ( @xmath521 ) covariance function is a better choice to carry out the reconstruction than the usual squared exponential covariance function .
therefore , we adopt the matrn ( @xmath521 ) covariance function to exhibit the first gp reconstruction using the hii galaxy data in the literature : @xmath522 . \label{11}\ ] ] as the previous works @xcite , using eq .
( [ 3 ] ) and the expression of the normalized comoving distance @xmath523 , we transform the theoretical distance modulus @xmath484 to @xmath524 in the following manner @xmath525 and the dark energy eos can be expressed as @xmath526d'}{3d'\{(1+z)^2[\omega_{k0}+(1+z)\omega_{m0}]d'^2-(1+\omega_{k0}d^2)\ } } , \label{2}\ ] ] where the prime represents the derivative with respect to the redshift @xmath31 . as noted in @xcite , we set the initial conditions @xmath527 and @xmath528 throughout the reconstruction processes .
note that the @xmath0 value affects obviously the reconstruction results by affecting the transformed @xmath524 .
additionally , we have assumed @xmath529 and @xmath530 @xcite in our gp reconstructions . in fig .
[ f3 ] , we consider the effects of 3 different @xmath0 values on the reconstructions of @xmath531 , @xmath501 and dark energy eos , i.e. , r16 s result @xmath24 km s@xmath2 mpc@xmath2 , p16 s result @xmath23 km s@xmath2 mpc@xmath2 and the prediction of the @xmath21cdm model @xmath532 km s@xmath2 mpc@xmath2 using the hii galaxy data . from upper panels of fig . [ f3 ] , one can easily find that , when @xmath532 km s@xmath2 mpc@xmath2 , the reconstructions of @xmath531 and @xmath501 are better than the other two cases . in the lower left and medium panels of fig . [ f3 ] , when @xmath24 km s@xmath2 mpc@xmath2 and @xmath502 km s@xmath2 km s@xmath2 mpc@xmath2 , we find that the underlying true model is consistent with the @xmath21cdm model in the low - redshift range at @xmath514 confidence level .
however , when @xmath532 km s@xmath2 mpc@xmath2 , this occurs in the low - redshift range at @xmath1 confidence level .
hence , we conclude that the gp reconstructions prefer a higher @xmath0 value than the measurements in previous works ( e.g. , r16 and p16 ) , and that the model - independent method have verified the correctness of @xmath0 values obtained by model - dependent method .
it is worth noting that , because the current hii galaxy data points are mainly located at low redshifts ( low-@xmath31 and gehr samples ) and there is a lack of medium and high redshifts data , we can not provide a accurate constraint on @xmath531 , @xmath501 and the dark energy eos .
precise measurements of @xmath0 is one of the most important and intriguing tasks . the recent local measurement implemented by riess et al . exhibits a high tension with the planck s result from cmb anisotropy data at @xmath25 confidence level .
our motivation is to use the latest hii galaxy measurements to determine the value of @xmath0 .
we explore the value of @xmath0 by using a combination of model - dependent and model - independent method .
first of all , we constrain five cosmological models by using the newest compilation of hii galaxy measurements and obtain the corresponding values of @xmath0 .
we find that the @xmath0 values for these five cosmological models are all consistent with the r16 s local measurement at @xmath1 confidence level , and that these five models prefer a higher best - fit @xmath0 value than r16 s result ( see table [ t2 ] and fig .
[ f1 ] ) . in light of this , to check the correctness of @xmath0 values obtained by model - dependent method , we firstly implement the gp reconstructions using the hii galaxy data in the literature .
we find that , when @xmath532 km s@xmath2 mpc@xmath2 , the reconstructions of @xmath531 and @xmath501 are better than the other two cases ( e.g. r16 and p16 ) , and the reconstructed dark energy eos is more consistent with the @xmath21cdm model in the low - redshift range at @xmath1 confidence level than the other two cases ( see fig .
hence , one can conclude that the gp reconstructions prefer a higher @xmath0 value than r16 s and p16 s results , and that the model - independent method have verified the correctness of @xmath0 values obtained by model - dependent method .
in @xcite , using 69 nearby hii galaxies and 23 gehr in 9 galaxies , the authors have obtained a value @xmath533 , which is also consistent with our results at @xmath1 confidence level . however , unlike their results , we have obtained higher best - fit @xmath0 values by utilizing a combination of model - dependent and model - independent method .
the tendency of high @xmath0 values may be attributed to different data systematics , a lack of medium and high redshift data or underlying new physics .
in addition , from the point of view of the hii galaxies hubble diagram , we find that there exists a very high degeneracy among these five models using only the latest hii galaxy measurements ( see fig .
[ f2 ] ) . in the future
, we expect more and more high - precision data can provide more useful information for us and help us distinguish different cosmological models better .
we are thankful to professors s. d. odintsov and bharat ratra for beneficial discussions on cosmology .
the author deng wang warmly thanks prof .
jing - ling chen for talks on bell inequality .
this work is supported in part by the national science foundation of china . | we use the latest hii galaxy measurements to determine the value of @xmath0 adopting a combination of model - dependent and model - independent method . by constraining five cosmological models , we find that the obtained values of @xmath0 are more consistent with the recent local measurement by riess et al . 2016 ( hereafter r16 ) at @xmath1 confidence level , and that these five models prefer a higher best - fit value of @xmath0 than r16 s result .
to check the correctness of @xmath0 values obtained by model - dependent method , for the first time , we implement the model - independent gaussian processes ( gp ) using the hii galaxy measurements .
we find that the gp reconstructions also prefer a higher value of @xmath0 than r16 s result .
therefore , we conclude that the current hii galaxy measurements support a higher cosmic expansion rate . |
since the discovery of superconductivity in the mgb@xmath1 compound , several authors have investigated the properties of such superconductor in the presence of magnetic fields . it has been shown that the electric , magnetic and transport properties of mgb@xmath1 are strongly affected by the magnetic field @xcite .
such enhanced field dependence of the properties of mgb@xmath1 has been ascribed to the peculiar two - gap structure , with a larger gap , associated with the two - dimensional @xmath3 band , and a smaller gap , associated with the three - dimensional @xmath4 band , which is rapidly suppressed by the applied magnetic field @xcite .
measurements of the microwave ( mw ) surface resistance , @xmath5 , allow investigating the processes responsible for mw energy losses in superconductors as well as determining specific properties of the samples . it is well known that the mw surface resistance of superconductors in the mixed state depends on the magnetic field through several mechanisms @xcite .
the different vortex states , in the different regions of the h - t plane , determine the temperature and field dependencies of @xmath5 .
therefore , measurements of @xmath6 may provide important information on the fluxon dynamics in the different regimes of fluxon motion .
it has been shown that mw losses in mgb@xmath1 superconductor are strongly affected by the magnetic field in the whole range of temperatures below @xmath2 , even for relatively low field values @xcite . in particular , an unusually enhanced field dependence of @xmath5 has been observed at @xmath7 , where pinning effects should hinder the energy losses . in this paper , we report experimental results of the field - induced variations of the mw surface resistance in high - density ceramic mgb@xmath1 . below @xmath8 , we have observed a magnetic hysteresis in the @xmath0 curves , whose peculiarities can not straightforwardly be justified by models reported in the literature . on the contrary ,
the experimental results obtained near @xmath2 are well accounted for in the framework of the coffey and clem model @xcite , with fluxons moving in the flux - flow regime , provided that the anisotropy of the upper critical field is taken into due account .
the field - induced variations of @xmath5 have been studied in a bulk ceramic mgb@xmath1 sample of approximate dimensions 2@[email protected] @xmath10 , with @xmath11 39 k. the sample has been extracted from a high - density pellet ( 2.4 g/@xmath12 ) obtained by reactive infiltration of liquid mg on a powdered b preform @xcite . after the reaction in a sealed stainless steel container , lined with a nb foil , a thermal treatment has been performed for two hours at @xmath13 900 @xmath14c .
the mw surface resistance has been measured using the cavity perturbation technique .
the cavity , of cylindrical shape with golden - plated walls , is tuned in the te@xmath15 mode resonant at 9.6 ghz ( @xmath16-factor @xmath17 40,000 at @xmath18 k ) .
the sample is put in the center of the cavity by a sapphire rod , in the region of maximum mw magnetic field .
the cavity is placed between the poles of an electromagnet , which generates dc magnetic fields up to @xmath17 10 koe .
two additional coils allow reducing to zero the residual field and working at low magnetic fields .
the quality factor of the cavity has been measured by an hp@xmath19d network analyzer .
@xmath5 has been investigated as a function of the dc magnetic field , @xmath20 , in the range 0 @xmath21 10 koe , at fixed values of temperature . before each measurement
was performed , the sample was zero - field cooled ( zfc ) to the desired value of temperature ; the external field was then increased up to 10 koe and successively decreased to zero , at constant temperature . the sample and field geometry is shown in fig.1 a ) ; in this geometry the mw current induces a tilt motion of all the fluxons present in the sample , as shown in fig.1 b ) .
fig.2 shows the field dependence of @xmath5 , normalized to the normal - state surface resistance , @xmath22 , measured at @xmath23 = 40 k , for different values of the temperature .
open and solid symbols refer to results obtained on increasing and decreasing @xmath20 , respectively . as one can see , up to @xmath24 4 k below @xmath2
the @xmath0 curves show a magnetic hysteresis .
the inset shows the temperature dependence of the height of the hysteresis loop at @xmath20 = 0 , @xmath25 , i.e. the variation of @xmath26 after a complete cycle of the dc magnetic field from 0 to 10 koe and back . on increasing the temperature up to @xmath13 34 k
, @xmath25 takes a constant value ; on further increasing the temperature it quickly decreases ; the hysteretic behavior vanishes at temperatures very close to @xmath2 .
both the increasing and decreasing - field branches of the @xmath0 curves show a plateau in the field ranges 0 @xmath21 @xmath27 and 0 @xmath21 @xmath28 , respectively . since the sample was zfc , @xmath27 marks the first - penetration field .
fig.3 a ) shows @xmath27 as function of temperature ; the line has been obtained by fitting the experimental data with the law @xmath29 $ ] ; we have obtained , as best - fit parameters , @xmath30 = 390 @xmath31 20 oe and @xmath32 = 2.4 @xmath31 0.2 .
the @xmath32 value gives a temperature dependence of @xmath27 consistent with that of @xmath33 .
@xmath30 is slightly larger than the values of @xmath34 reported in the literature for mgb@xmath1 , suggesting that in our sample surface - barrier effects are weak . at different values of temperature .
open and solid symbols refer to measurements performed at increasing and decreasing field , respectively .
the inset shows the height of the hysteresis loop at @xmath20 = 0 as a function of temperature.,width=264,height=510 ] fig.3 b ) shows the temperature dependence of @xmath28 , though the meaning of this characteristic parameter is not clear , one can note that its temperature dependence is very similar to that of @xmath27 . and
@xmath28 , defined in fig.2.,width=245,height=245 ]
microwave losses induced by dc magnetic fields in superconductors in the mixed state are proportional to the complex penetration depth of the mw field , @xmath35 ; they are influenced by the fluxon motion and the very presence of vortices which bring along normal fluid in their core @xcite .
@xmath35 can be expressed in terms of the normal - fluid density , flux - flow resistivity and @xmath36 ratio , where @xmath37 is the working frequency and @xmath38 the depinning frequency , with @xmath39 the labush parameter and @xmath40 the viscous - drag coefficient . for @xmath41
the induced mw current makes fluxons moving in the flux - flow regime .
otherwise , the fluxon motion is ruled by the strength of the restoring - pinning forces .
our results show that for @xmath42 35 k the @xmath0 curves exhibit an hysteretic behavior , suggesting that pinning effects are important in this range of temperatures . on the contrary , for @xmath43 35
k the hysteresis vanishes and fluxons should move in the flux - flow regime . actually , the results near @xmath2 can be accounted for by using the coffey and clem theory @xcite , with fluxons moving in the flux - flow regime , provided that the anisotropy of the upper critical field is taken into due account . following the same procedure of ref.@xcite , based on the assumption that the anisotropic ginzburg - landau theory is valid for mgb@xmath1 in this range of temperatures ,
we have fitted the experimental data by taking the anisotropy factor , @xmath44 , as fitting parameter .
the best - fit curves , along with the experimental data , are shown in fig.4 .
for all the curves , the best - fit value of the anisotropy factor is @xmath44 = 3 , showing that in our sample the anisotropy of @xmath45 is constant in a range of temperatures of about 3 k below @xmath2 .
the hysteresis of the @xmath0 curves , observed for @xmath42 35 k , should be related to the different value of the induction field at increasing and decreasing fields . both the normal fluid density and the flux - flow resistivity depend on the number of fluxons moving under the action of the mw current . in the field geometry we used the mw current
induces a tilt motion of all the fluxons present in the sample ( see fig.1 ) .
so , the hysteresis of @xmath0 should be directly related to the hysteretic behavior of the magnetization curve due to pinning effects .
however , several anomalies can be highlighted in the @xmath0 curves at low temperatures .
firstly , from fig.2 one can see that an unusually enhanced field variation of @xmath5 is observed , even at the lowest temperatures ; a magnetic field of 10 koe induces an @xmath5 variation of about 1/3 of the normal - state value .
we suggest that such enhanced field - induced energy losses are ascribable to the effects of the two - gap structure of mgb@xmath1 .
indeed , it has been shown that the structure of vortices in mgb@xmath1 is characterized by two coherence lengths , associated with each band @xcite .
on increasing the applied field , the giant cores start to overlap at field values much smaller than @xmath45
. therefore , modest fields can suppress superconductivity between the vortices , resulting in an enhancement of the normal - fluid density @xcite .
furthermore , this particular vortex structure can modify the interaction among vortices and pinning centers , giving rise to an unconventional shape of the pinning potential well .
another anomaly concerns the shape of the decreasing - field branch of the @xmath0 hysteresis loop . by supposing fluxons in the critical state ,
after the external field starts decreasing the variation of @xmath46 should be initially slow and then faster , no matter the field dependence of the critical current .
so , a downward curvature of the decreasing - field branch of the @xmath0 curve is expected , at variance with our results .
the origin of this anomalous behavior , as well as the presence of the wide plateau in the decreasing - field @xmath0 curves , is not yet understood and needs further investigation . in conclusion , we have reported a detailed study of the field - induced variations of the mw surface resistance in high - density ceramic mgb@xmath1 .
an unusual behavior of the @xmath0 curves has been highlighted at low temperatures , which can not be explained in the framework of models reported in the literature .
we suggest that it is ascribable to the unconventional vortex structure related to the presence of the two gaps . on the contrary
, the results obtained at temperatures close to @xmath2 have been accounted for in the framework of the coffey and clem model , with fluxons moving in the flux - flow regime , by taking into account the anisotropy of upper critical field . | field - induced variations of the microwave surface resistance , @xmath0 , have been investigated in high - density ceramic mgb@xmath1 . at low temperatures , several peculiarities of the @xmath0 curves
can not be justified in the framework of models reported in the literature .
we suggest that they are ascribable to the unconventional vortex structure in mgb@xmath1 , related to the presence of two gaps . on the contrary
, the results near @xmath2 can be accounted for by the coffey and clem model , with fluxons moving in the flux - flow regime , provided that the anisotropy of the upper critical field is taken into due account .
, , , , mgb@xmath1 , microwave surface impedance , fluxon dynamics 74.25.ha , 74.25.nf , 74.60.ge , 74.70.ad |
spiral galaxies require a continuous supply of gas to maintain star formation over long timescales and to explain their metallicity distribution and evolution ( see * ? ? ? * and references therein ) .
what form this gas supply takes has not been observationally established , but it may be that the dominant form of supply is cold gas accretion ( e.g. , * ? ? ? * ) . in the milky way ( mw ) and the nearest galaxies high velocity clouds ( hvcs ) are observed @xcite , some with low metallicity @xcite . the hvcs may contribute a small fraction of the fuel needed for star formation , but a more complete census is required to be able to draw general conclusions . while distinct hi clouds of this nature are not commonly detected in the outskirts of galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , another morphological feature thick gas disks is becoming increasingly recognized as a typical structural feature of spiral galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
these gas disks can contribute up to @xmath1 of the total hi mass of their host galaxies , and have distinct kinematics ( e.g. , * ? ? ?
these slowly rotating thick disks may be connected to the accretion history @xcite and provide an observational tracer of coronal material feeding the disk .
searching for clouds and for thick hi disks requires specialized deep observations .
the halogas ( hydrogen accretion in local galaxies ; * ? ? ?
* ) survey was undertaken to provide the required capability for a large enough sample to begin to draw statistical conclusions .
the core of the survey are westerbork synthesis radio telescope ( wsrt ) observations of 22 edge - on and moderately - inclined nearby galaxies , each observed for 120 hours in the 21-cm hi line .
the typical @xmath2 column density sensitivity is @xmath3 for typical linewidth of @xmath4 , making halogas the deepest interferometric galaxy hi survey available to date .
halogas has an optimal angular resolution of @xmath5 ( 1.5 kpc at a typical distance of 10 mpc ) , unsuitable for detailed high - resolution studies of the gas - star formation connection like things ( the hi nearby galaxy survey ; * ? ? ?
* ) , but ideal for faint diffuse emission .
the typical halogas mass sensitivity for unresolved clouds with the same linewidth allows the detection of hvc analogues in the outskirts of the survey galaxies , @xmath6 the primary halogas observations were completed as of early 2013 .
a number of ancillary data products have been collected to supplement our deep hi observations .
deep multi - band optical imagery have been obtained at the _ isaac newton telescope _ ( int ) and kitt peak national observatory ( kpno ) ; see e.g. @xcite . sensitive _ galex _
uv data have also been collected to supplement deep survey data already available for some halogas targets . together with the hi line observations , broadband full - polarization continuum data are also available for many of the survey galaxies , allowing the detailed investigation of magnetic fields and cosmic rays in the same galaxies .
the interferometric halogas data are being supplemented with single dish data from effelsberg and the the green bank telescope ( gbt ) useful not only to provide a missing short - spacing correction , but to enhance the search for clouds and streams in the outer parts of halogas galaxies .
sophisticated data analysis techniques are required in order to achieve reliable structural parameters from the sensitive hi data cubes provided by halogas .
isolating thick hi disks in particular requires detailed modeling of the 3d gas distribution and kinematics .
this modeling work is mostly performed using the tirific ( tilted ring fitting code ; * ? ? ?
* ) software . two examples that illustrate the need for
detailed tilted ring modeling are shown in figure [ fig : fig1 ] . in the case of ugc 7774 , the strong warping already apparent in the plane of the sky ( see also * ? ? ?
* ) also causes the vertical thickening . in the other galaxy ,
ngc 1003 , the disk structure recovered from a detailed 3d analysis shows a substantial warp ( @xmath7 ) along the line of sight , causing the disk to appear thicker in projection than the actual vertical distribution of the hi layer .
another interesting example , ngc 5023 ( not shown ) , illustrates an extreme case : recovery of spiral structure in an edge - on disk ( see * ? ? ?
the modeling works best when these spiral structures are present and are required to fully understand the kinematics , emphasizing that this level of detail is essential to properly describe the 3d structure of the hi in galaxies .
halogas galaxies are both found to host thick hi disks @xcite and to have little extraplanar hi @xcite .
this dichotomy provides powerful leverage on the origin of hi thick disks in galaxies .
-band image of ugc 7774 displayed on a logarithmic stretch and overlaid with halogas hi total intensity contours , starting at @xmath8 and increasing by multiples of 2 .
the angular resolution of the hi data is @xmath9 . _
right _ : dss2 @xmath10-band image of ngc 1003 displayed on a logarithmic stretch and overlaid with halogas hi total intensity contours .
the hi from the entire galaxy is shown in white contours , starting at @xmath11 and increasing by multiples of 4 .
the black contours show the hvc analogues discussed in the text ( [ subsection : accretion ] ) , starting at about the same column density and increasing by multiples of 2 .
the resolution of the hi data is @xmath12.,title="fig : " ] -band image of ugc 7774 displayed on a logarithmic stretch and overlaid with halogas hi total intensity contours , starting at @xmath8 and increasing by multiples of 2 .
the angular resolution of the hi data is @xmath9 . _
dss2 @xmath10-band image of ngc 1003 displayed on a logarithmic stretch and overlaid with halogas hi total intensity contours .
the hi from the entire galaxy is shown in white contours , starting at @xmath11 and increasing by multiples of 4 .
the black contours show the hvc analogues discussed in the text ( [ subsection : accretion ] ) , starting at about the same column density and increasing by multiples of 2 .
the resolution of the hi data is @xmath12.,title="fig : " ] the entire halogas sample has been inspected in order to characterize the presence and characteristics of extraplanar hi .
we use both the 3d modeling mentioned in
[ section : overview ] , as well as an approximate disk - halo separation of the type presented by @xcite . on this basis
we have searched for evidence of a common origin of the extraplanar hi by comparing with the general properties of the sample galaxies . from the halogas sample
we find a tentative connection between the star formation rate density and the presence of thick hi disks , such that thick hi disks appear to be present in galaxies above a threshold sf energy injection .
we note that a similar relationship has been identified for radio continuum halos @xcite . in the case of the hi ,
there appears to be a strong dependence on the star formation rate density , but a weaker dependence on the stellar mass density that traces the strength of the gravitational potential .
another key result of the halogas survey is the identification of hi clouds and streams in the vicinity of the sample galaxies that could originate through the cold gas accretion process , and to determine the impact on the star formation history of the sample .
as previously noted , the occurrence of gas clouds in the outskirts of galaxies is low .
the halogas survey , despite its high sensitivity , does not substantially change this picture .
the halogas catalog of isolated gas features is still in preparation ( as we work toward a statistically sound census across the full sample ) , but we can already state qualitatively that there is a low incidence of clouds and streams .
the contribution of visible hi accretion to the star formation fueling process in nearby galaxies appears to be minimal . as a representative example
, we highlight ngc 1003 as shown in figure [ fig : fig1 ] .
a small number of morphologically and kinematically distinct gas features have been identified from an inspection of the 3d hi dataset ; these are shown with black contours in fig .
[ fig : fig1 ] .
these features do not appear to have stellar counterparts , even in our deep optical images .
the hi masses and distances from ngc 1003 of these clouds are similar to the properties of the hvcs around the mw .
the mass of these hvc analogues adds up to only @xmath13 , which over a dynamical time ( @xmath14 ) contributes only 2% of the star formation rate of ngc 1003 ( @xmath15 ; * ? ? ? * ) . alternatively ,
if the gas contained in these clouds were used to completely fuel the current sfr , they would need to be replenished after only @xmath16 .
these rates neglect a possible contribution from an ionized gas component to which we do not have observational constraints .
we note however that the contribution of ionized gas to the total mass of mw hvcs can be substantial ( e.g. , * ? ? ?
* ; * ? ? ?
* ) ; qso absorption line studies of low column density hi probed by halogas where possible would be very useful to constrain this aspect of the mass budget .
the scientific themes addressed by the halogas project will help to focus the groundbreaking science questions to be addressed with the square kilometre array ( ska ) and its pathfinder and precursor projects . for descriptions of surveys targeting deeper observations of larger nearby galaxy samples , as well as investigations of low column density igm material ,
see the discussions by @xcite and @xcite , respectively .
the westerbork synthesis radio telescope is operated by astron ( netherlands institute for radio astronomy ) with support from the netherlands foundation for scientific research ( nwo ) . | we present an overview of the halogas ( hydrogen accretion in local galaxies ) survey , which is the deepest systematic investigation of cold gas accretion in nearby spiral galaxies to date . using the deep hi data that form the core of the survey , we are able to detect neutral hydrogen down to a typical column density limit of about @xmath0 and thereby characterize the low surface brightness extra - planar and anomalous - velocity neutral gas in nearby galaxies with excellent spatial and velocity resolution . through comparison with sophisticated kinematic modeling
, our 3d halogas data also allow us to investigate the disk structure and dynamics in unprecedented detail for a sample of this size .
key scientific results from halogas include new insight into the connection between the star formation properties of galaxies and their extended gaseous media , while the developing halogas catalogue of cold gas clouds and streams provides important insight into the accretion history of nearby spirals .
we conclude by motivating some of the unresolved questions to be addressed using forthcoming 3d surveys with the modern generation of radio telescopes . |
it now seems well established that the appearance of active galactic nuclei ( agn ) depends strongly on orientation .
classes of apparently different agn might actually be intrinsically similar , only seen at different angles with respect to the line of sight .
the basic idea , based on a variety of observations and summarized in figure 1 of urry & padovani ( 1995 ) , is that emission in the inner parts of agn is highly anisotropic .
the current paradigm for agn includes a central engine , surrounded by an accretion disk and by fast - moving clouds , probably under the influence of the strong gravitational field , emitting doppler - broadened lines .
more distant clouds emit narrower lines . absorbing material in some flattened configuration ( usually idealized as a torus )
obscures the central parts , so that for transverse lines of sight only the narrow - line emitting clouds are seen ( narrow - lined or type 2 agn ) , whereas the near - ir to soft - x - ray nuclear continuum and broad - lines are visible only when viewed face - on ( broad - lined or type 1 agn ) . in radio - loud objects we have the additional presence of a relativistic jet , roughly perpendicular to the disk , which produces strong anisotropy and amplification of the continuum emission ( `` relativistic beaming '' ) . in general , different components
are dominant at different wavelengths .
namely , the jet dominates at radio and @xmath0-ray frequencies ( although it does contribute to the emission in other bands as well ) , the accretion disk is thought to be a strong optical / uv / soft x - ray emitter , while the absorbing material will emit predominantly in the ir .
it then follows that a proper understanding of agn will only come through multifrequency studies ( e.g. , padovani 1997 ) .
this axisymmetric model of agn implies widely different observational properties ( and therefore classifications ) at different aspect angles .
hence the need for `` unified schemes '' which look at intrinsic , isotropic properties , to unify fundamentally identical ( but apparently different ) classes of agn .
seyfert 2 galaxies have been `` unified '' with seyfert 1 galaxies ( see antonucci 1993 , and references therein , and granato , these proceedings , for more recent results ) . as regards the radio - loud population , fanaroff - riley type i ( fr i : i.e. , low - power ) radio galaxies have been unified with bl lacertae objects , a class of agn characterized by very weak emission lines , while fanaroff - riley type ii ( fr ii : i.e. , high - power ) radio galaxies have been unified with radio quasars . in the latter case , flat - spectrum radio quasars ( fsrq )
are thought to be oriented at relatively small angles w.r.t . to the line of sight ( @xmath1 ) , while steep - spectrum radio quasars ( ssrq ) should be at angles intermediate between those of fsrq and fr ii radio galaxies .
this paper is _ not _ meant to be a review of unified schemes : the interested reader might consult , for example , the reviews by antonucci ( 1993 ) and urry & padovani ( 1995 ) .
my aim is to discuss some of the very recent results in the field of unified schemes for radio - loud agn which have appeared in the literature in the last year or so ( and therefore were not included in urry & padovani 1995 ) .
vermeulen ( 1995 ) have discovered a broad h@xmath2 line in bl lacertae , the class prototype itself , comparable in luminosity and velocity width to that of the seyfert 1 galaxy ngc 4151 .
although weak , broad emission lines have been previously detected in other bl lacs ( stickel 1993 ) , the presence of such a line in bl lacertae itself has attracted the attention of many bl lac pundits .
the main distinguishing feature of bl lacs is in fact their weak , or even absent , emission lines , generally characterized by equivalent widths for the stronger lines @xmath3 .
the question then is : is bl lac still a bl lac ?
the reassuring answer is : yes .
the newly detected broad line , in fact , has @xmath4 , consistent with the 5 limit .
this detection , however , has some quite interesting implications for our understanding of bl lacs ( and of agn in general : see salvati , these proceedings ) .
this relatively strong line should have been detected in previous spectra , taken when the source was of roughly similar brightness .
corbett ( 1996 ) have estimated that the h@xmath2 luminosity ( and equivalent width ) must have increased by a factor of 4 from earlier observations .
if the source of ionization were related to the non - thermal continuum , the equivalent width of all emission lines should be independent of the ionizing power , as both line and continuum luminosity would be proportional to it .
corbett ( 1996 ) then suggest the presence of an additional , variable continuum source : the observed @xmath5 could in fact be explained by an accretion disk whose flux is only 4% of the non - thermal , beamed continuum at the wavelength of h@xmath2 .
such a small contribution to the total flux in the optical band affects only slightly the shape of the optical continuum by flattening the slope in the 3,500 7,000 range , but should produce a marked depolarization of the synchrotron component ( assuming thermal emission is unpolarized ) .
this is a strong prediction of the model although , as corbett point out , this dilution might be difficult to detect due to the strong variability of the wavelength dependence of the synchrotron polarization observed in bl lacs .
the possibility that bl lacs possess an accretion disk , even though its contribution to the total flux might be quite small , is suggestive .
furthermore , the presence of broad lines in bl lacs , and the lack of them ( to the best of my knowledge ) in fr is , might provide the only evidence , so far , of the presence of an obscuring torus in low - luminosity radio sources .
finally , the detection of an emission line at the same redshift as the host galaxy absorption lines shows that at least bl lac ( and all the other objects sharing this property : see e.g. , stickel 1993 ) is not gravitationally micro - lensed ( urry & padovani 1995 ) .
owen , ledlow & keel ( 1996 ) have looked for bl lacs in 183 low - luminosity radio galaxies in relatively rich ( @xmath6 of richness class 1 and 2 ) clusters at @xmath7 .
they found no object which met the `` standard definition '' of bl lacs , based on a measure of a non - thermal component ( 4000 break @xmath8 ) and weakness of the lines ( @xmath3 ) , although they found 4 objects with weak evidence for non - thermal activity ( 4000 break @xmath9 : typical ellipticals have break values @xmath10 ) , two of them with relatively strong lines ( unlike bl lacs ) .
owen expected 8 bl lacs on the basis of a relativistic beaming model ( urry , padovani & stickel 1991 ) or 3/18 at @xmath11 erg s@xmath12 hz@xmath12 simply on the basis of the observed luminosity functions of fr is and bl lacs .
their null results are inconsistent with these expectations at the @xmath13 and @xmath14 level respectively , apparently putting in serious trouble the bl lac
fr i unification .
this inconsistency might be at least partly related to the definition of a bl lac , which is likely to be not that clear - cut and should perhaps be slightly relaxed , for example by increasing the value of the 4000 break ( as pointed out by march 1996 ) .
more importantly , it seems now quite likely that bl lacs avoid the richest clusters ( at least at low redshifts : e.g. , pesce , falomo & treves 1995 ; smith , odea & baum 1995 ; wurtz 1996b ) and reside mostly in relatively poor ( richness class @xmath15 ) clusters , which however seem to include a large fraction of fr i galaxies ( hill & lilly 1991 ) .
this behaviour , although still unexplained , is nevertheless consistent with the findings of wurtz , stocke & yee ( 1996a ) that most bl lacs have host galaxies too faint to be bright cluster galaxies ( bcgs ) in rich clusters : their host galaxy magnitudes are more typical of bcgs in poorer clusters , which are less luminous .
in summary , the sample of owen ( 1996 ) probably included a large number of unlikely bl lac hosts : the significance of their null results should then be re - evaluated .
it has long been recognized that bl lacs appear to come in two versions , one being more extreme , having larger optical polarization , core - dominance , and variability , and usually associated with objects selected in radio surveys , the so - called radio - selected bl lacs ( rbl ) . the less extreme version , mostly selected in x - ray surveys ,
was then obviously named accordingly ( x - ray - selected bl lacs : xbl ) .
this division was clearly not satisfactory , as it was not based on intrinsic physical properties but solely on the selection band , and did not uniquely characterize a source .
recent all - sky surveys , in fact , include many bl lacs previously selected at radio frequencies and which could then now be classified both as rbl _ and _ xbl .
padovani & giommi ( 1995 ; see also ledden & odell 1985 for the first subdivision of bl lac classes based on @xmath17 ) have then proposed to classify bl lacs in hbl ( or high - energy peak bl lacs ) and lbl ( or low - energy peak bl lacs ) , based on a simple division in terms of their x - ray - to - radio flux ratio .
this reflects their different multifrequency spectra , with lbl ( most rbl ) having a peak ( i.e. , emitting the bulk of their energy ) at infrared - optical frequencies , and hbl ( most xbl ) having a peak at ultraviolet - x - ray frequencies .
note that such a distinction does not mean that there are two separate bl lac classes : it is more likely that there is a continuous distribution of , for example , peak frequency , which turns out to be bimodal because of selection effects .
what is the relation between the two classes ? this is being ( and has been ) discussed by various authors ( see padovani & giommi 1995 and urry & padovani 1995 , and references therein ) .
briefly , the two opposite views are that : 1 .
lbl are _ intrinsically _ more numerous ( `` different energy distribution '' scenario ) ; 2 .
hbl are _ intrinsically _ more numerous ( `` different angle '' scenario ) . in the former view ,
radio selection gives an `` unbiased '' view of the bl lac population ( in terms of population ratios ) and hbl and lbl are oriented w.r.t .
the line of sight within approximately the same angles . in the latter ,
x - ray selection is unbiased , and hbl are seen at larger angles .
currently available samples are not deep enough to overthrow the balance in favour of one or the other of the two explanations , which diverge in their predictions only at relatively faint radio and x - ray fluxes . some of the recent papers on the subject include : padovani & giommi ( 1996 ) and lamer , brunner & staubert ( 1996 ) , on the different _ rosat _ spectra of hbl and lbl , which are consistent with their different multifrequency spectra ; sambruna , maraschi & urry ( 1996 ) on , amongst other things , the difficulty in explaining the observed differences in the multifrequency spectra of hbl and lbl only by changing viewing angle ; kollgaard ( 1996 ) on radio constraints to the two pictures , which would seem to favour the `` different angle '' hypothesis . finally , wurtz ( 1996a )
find no differences in the host galaxy properties of hbl and lbl , while fossati et al .
( these proceedings ) , present yet a new explanation for the relation between hbl and lbl .
the question of the comparison between linear sizes of radio galaxies and quasars lies at the core of unified schemes : if radio quasars are oriented at small angles to the line of sight , then they should have systematically smaller large - scale radio structures than radio galaxies , thought to be in the plane of the sky . a simple relation exists between the ratio of median sizes of radio galaxies and quasars , @xmath18 , and the half - opening angle of the obscuring torus ( or equivalently the critical angle w.r.t . the line of sight separating the two classes ) , @xmath19 .
( in fact , the smaller @xmath19 , the higher will be @xmath18 . )
the critical angle is also obviously related to the quasar fraction @xmath20 ( @xmath21 ) , from which it follows that @xmath18 and @xmath20 have to be inversely related ( that is the smaller @xmath20 , the smaller @xmath19 , the higher @xmath18 ) .
the important point of barthel s ( 1989 ) paper was to show that both @xmath18 and @xmath20 were consistent with a critical angle @xmath22 for the 3c sample at @xmath23 .
barthel s findings were criticized by some authors , particularly by singal ( 1993 ) , who argued that not only the quasar fraction seemed to be redshift dependent but the apparent agreement between the angles inferred from the number ratios and those derived from the linear sizes disappeared extending the comparison to other redshift ranges .
it must be stressed that the whole subject of linear sizes of radio sources is quite complex , having been addressed by many authors quite often with contrasting results ( see neeser et al
. 1995 for a very good description of some of the selection effects which can affect this sort of analysis ) . as summarized in urry & padovani ( 1995 ) and discussed in singal ( 1996 ) , various solutions to the problems raised by singal ( 1993 ) have been proposed .
singal ( 1996 ) , however , examining the quasar fractions observed in low - frequency radio samples , noted that not only the quasar fraction decreases at lower fluxes ( which would show that a single @xmath19 is untenable ) but the ratio of median sizes of radio galaxies and quasars actually seems to _ decrease _ as well , contrary to the expectations of unified schemes . in other words , while @xmath18 should be inversely dependent on @xmath20 , @xmath18 actually _ increases _ with @xmath20 ( see figure 2 of singal 1996 ) .
note that no change of @xmath19 with redshift or power will bring the data in agreement with the predictions of unified schemes , as the observed dependence is opposite to the predicted one .
does this mean that unified schemes are totally wrong ?
apparently not , if one takes into account the evolution of the single radio sources .
recent results ( e.g. , fanti 1995 ; readhead 1996 ) suggest that radio sources grow in size during their active phase , with an accompanying decrease in power due to expansion losses .
gopal - krishna , kulkarni & wiita ( 1996 ) have parameterized this behaviour , with the further assumption that the half - opening angle of the torus depends on luminosity ( which could be explained by the fact that more powerful engines `` erode '' away the inner parts of the torus : see urry & padovani 1995 and references therein ) .
the result is that gopal - krishna can explain the apparently anomalous relation between @xmath18 and @xmath24 pointed out by singal ( 1996 ) with reasonable input parameters . within this model
, the median radio sizes of quasars can actually approach , or even exceed , that of radio galaxies , as reported by singal ( 1996 ) for some samples .
note that the inclusion of source evolution in unified schemes , which seems unavoidable , might complicate the use of the core - to - extended flux ratio , @xmath25 , as an orientation indicator , as in this view the extended flux would also be time - dependent , that is @xmath25 could change with time for the same angle to the line of sight .
there are some ways to overcome the presence of obscuring material and be able to see the central engine even in type 2 sources .
one of these is spectropolarimetry , where one looks for broad lines in that small fraction of the nuclear flux which is in some cases `` reflected '' towards our line of sight by dust and/or electrons .
recent results in this field are discussed by cimatti , these proceedings .
i concentrate here on another way to see through the dust by moving to longer wavelengths , where the torus becomes more transparent : infrared spectroscopy .
infrared broad lines have been detected before in type 2 radio sources ( see urry & padovani 1995 for references ) but it is clear that one would like to be able to answer such questions as : how many type 2 sources actually show broad lines when observed in the infrared ? what is the distribution of blr extinction in these sources ?
the best way to address these points is through the use of an unbiased , flux - limited sample , such as that studied by hill , goodrich & depoy ( 1996 ) .
these authors present infrared spectrophotometry of pa@xmath2 , plus optical spectroscopy of h@xmath2 and h@xmath26 , for a flux - limited sample of 11 3cr fr ii radio sources with @xmath27 , which includes 8 narrow - line radio galaxies , 2 broad - line radio galaxies , and one quasar .
infrared broad lines are detected in 3 narrow - line radio galaxies ( or @xmath28 of the type 2 sources ) , and extinction values are derived for narrow and broad lines .
the distribution of broad - line extinction , @xmath29 , is in reasonable agreement with a simple unification model .
extension of this work to higher redshifts ( and therefore luminosities ) should permit a direct test of the idea that the half - opening angle of the torus increases with power ( see sect .
if that is the case , at higher powers there should be fewer sources with relatively large extinctions ( @xmath30 ) , as compared to their lower power counterparts .
the relation between bl lacs and fsrq is not clear .
the two classes share similar properties ( apart from the presence of strong emission lines in the latter class ) .
sambruna ( 1996 ) have suggested a continuity in the multifrequency spectra of hbl , lbl , and fsrq , with the frequency of the peak of the emission shifting to higher values for decreasing luminosities .
this continuity seems to have been severely weakened by the discovery of `` hbl - like '' quasars ( padovani , giommi & fiore , these proceedings ) , that is of radio quasars ( including some fsrq ) with multifrequency spectra similar to those of hbl .
padovani , giommi & fiore ( 1997 ) have also found that , contrary to previous results , even the soft x - ray spectra of fsrq and bl lacs of the lbl type are similar .
finally , vagnetti ( these proceedings ) , has presented his latest results on the possible evolutionary relation between the two classes .
i have listed here some of the open questions / hot topics on which work is in progress .
these will be hopefully discussed at the next italian meeting on agn , to be held in 1998 ! *
obscuration in fr is .
the presence of broad lines in at least some bl lacs ( sect .
2.1 ) indicates that some obscuring material must be present in fr is ( as well as in fr iis ) .
a sax core program will address this issue . also , iso observations of fr is should provide further constraints . *
bl lac ( and fr i ) environment .
as stressed in section 2.2 , it might well be that bl lacs avoid the richest clusters .
more work in this field is needed , as the samples used to study the environment of bl lacs are still relatively small .
various ongoing projects will address this point in more detail .
comparable studies on the environment of fr i radio galaxies using sizeable and well defined samples should also be performed .
* hbl / lbl dichotomy .
as mentioned in section 2.3 , we need deeper and larger bl lac samples to address this point . amongst the various groups working on this , both perlman , padovani , giommi , ( 1997 ) and caccianiga , maccacaro , wolter , ( these proceedings ) are cross - correlating x - ray and radio catalogues to select new bl lac ( and fsrq ) samples . *
( spectro)polarimetry of complete samples of radio - galaxies .
as mentioned in section 3.2 , spectropolarimetry allows one to see emission from the nuclear region , particularly broad lines , even in type 2 sources .
so far , however , spectropolarimetric studies have concentrated on individual , sometimes peculiar , sources .
we would need a spectropolarimetric study of a complete , well - defined sample , to quantify how common reflected broad lines are , for example .
as far as i know , nobody is doing this , but i hope that i am misinformed on this ! tadhunter and di serego alighieri et al .
, however , are carrying out a polarimetric study of a complete sample of radio sources , which should still give interesting , quantitative information .
* vlbi studies of complete sample of radio sources .
these are important to study , among other properties , the distribution of superluminal speeds to estimate beaming parameters and constrain unification .
various surveys are in progress , for example the caltech - jodrell bank survey of strong ( @xmath31 jy ) radio sources ( e.g. , polatidis 1995 ) or the survey of low - power radio galaxies by the bologna group ( e.g. , venturi 1995 ) . in conclusion , the unification of radio - loud sources is a very active field , with interesting results appearing even on a relatively short time scale . in 1995
alone , about 65 papers on beaming and unification of radio sources have been published ( 51 in refereed journals ) , an all - time record judging from the data relative to the last ten years ( source : astrophysics data system ) .
various modifications to our relatively simple and idealized unification picture seem to be in order : not all fr i radio galaxies might host bl lacs , after all , but probably only those in relatively poor clusters ; obscuring material might be present in fr i radio galaxies as well , and not only in their higher - luminosity counterparts ; not unexpectedly , radio sources do evolve in time , growing in size and dimming in power , which might help solving some inconsistencies in their linear sizes relative to those of radio galaxies ; infrared broad lines are indeed detected in about half of the low - redshift narrow line radio galaxies ; but we still do not know if bl lacs of the lbl type are intrinsically more numerous than bl lacs of the hbl type ! finally ,
although so far unified schemes for radio - loud and radio - quiet agn have been considered separately , there might be some connection between the two .
falcke , sherwood & patnaik ( 1996 ) have recently suggested that relativistic boosting , normally associated with radio - loud agn , might be present in the radio cores of some radio - quiet sources as well .
this idea , with its far - reaching implications , might open up new research paths which could help us to solve the long - standing problem of the radio - loud / radio - quiet dichotomy of the agn population .
antonucci , r. : 1993 , 31 473 barthel , p. d. : 1989 , 336 606 corbett , e. a. , et al . : 1996 , 281 737 falcke , h. , sherwood , w. , patnaik , a. r. : 1996 , 471 106 fanti , c. , fanti , r. , dallacasa , d. , schilizzi , r. t. , spencer , r. e. , stanghellini , c. : 1995 , 302 317 gopal - krishna , kulkarni v. k. , wiita , p. j. : 1996 , 463 l1 hill , g. j. , goodrich , r. w. , depoy , d. l. : 1996 , 462 163 hill , g. j. , lilly , s. j : 1991 , 367 1 kollgaard , r. i. , palma , c. , laurent - muehleisen , s. a. , feigelson , e. d. : 1996 , 465 115 lamer , g , brunner , h. , staubert , r. : 1996 , 311 384 ledden , j. e. , odell , s. l. : 1985 , 298 630 march , m. j. m. , browne , i. w. a. , impey , c. d. , smith , p. s. : 1996 , 281 425 neeser , m. j. , eales , s. a. , law - green , j. d. , leahy , j. p. , rawlings , s. : 1995 , 451 76 owen , f. n. , ledlow , m. j. , keel , w. c. : 1996 , 111 53 padovani , p. : 1997
, in _ new horizons from multi - wavelength sky surveys _ , b. mclean et al
, in press padovani , p. , giommi , p. : 1995 ,
444 567 padovani , p. , giommi , p. : 1996 , 279 526 padovani , p. , giommi , p. , fiore , f. : 1997 , _ mon . not .
_ , in press perlman , e. s. , et al .
: 1997 , in _ new horizons from multi - wavelength sky surveys _ , b. mclean et al .
, in press pesce , j. e. , falomo , r. , treves , a. : 1995 , 110 1554 polatidis , a. g. , et al .
: 1995 , 98 1 readhead , a. c. s. , taylor , g. b. , pearson , t. j. , wilkinson , p. n. : 1996 , 460 634 sambruna , r. m. , maraschi , l. , urry , c. m. : 1996 , 463 444 singal , a. k. : 1993 , 262 l27 singal , a. k. : 1996 , 278 1069 smith , e. p. , odea , c. d. , baum , s. : 1995 , 441 113 stickel , m. , fried , j. w. , khr , h. : 1993 , 98 393 urry , c. m. , padovani , p. : 1995 , 107 803 urry , c. m. , padovani , p. , stickel , m. : 1991
, 382 501 venturi , t. , et al .
: 1995 , 454 735 vermeulen , r. c. , et al . : 1995 , 452 l5 wurtz , r. , stocke , j. t. , yee , h. k. c. : 1996a , 103 109 wurtz , r. , stocke , j. t. , ellingson , e. e. , yee , h. k. c. : 1996b , _ astrophys . | after briefly summarizing the main tenets of unified schemes of active galactic nuclei , i review some recent results in the field of unification of radio - loud sources , both for the low - luminosity ( bl lacs and fanaroff - riley type i radio galaxies ) and high - luminosity ( radio quasars and fanaroff - riley type ii radio galaxies ) populations . # 1 # 2 _ mem . soc . astron . it .
_ * # 1 * , # 2 # 1 # 2 _ the messenger _ * # 1 * , # 2 # 1 # 2 _ astron .
nach . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
lett . _ * # 1 * , l#2 # 1 # 2 _ astron .
astrophys .
rev . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
suppl .
ser . _
* # 1 * , # 2 # 1 # 2 _ astron .
j. _ * # 1 * , # 2 # 1 # 2 _ ann .
rev .
astron .
astrophys . _ * # 1 * , # 2 # 1 # 2 _ astrophys .
j. _ * # 1 * , # 2 # 1 # 2 _ astrophys .
j. lett .
_ * # 1 * , l#2 # 1 # 2 _ astrophys .
j. suppl .
_ * # 1 * , # 2 # 1 # 2 _ astrophys .
space sci .
_ * # 1 * , # 2 # 1 # 2 _ adv .
space res .
_ * # 1 * , # 2 # 1 # 2 _ bull .
astron .
inst .
czechosl . _ * # 1 * , # 2 # 1 # 2 _ j. quant .
spectrosc .
. transfer _ * # 1 * , # 2 # 1 # 2 _ mon . not .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ mem .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ phys .
lett .
rev . _ * # 1 * , # 2 # 1 # 2 _ publ .
astron .
. japan _ * # 1 * , # 2 # 1 # 2 _ publ .
astr .
soc .
pacific _ * # 1 * , # 2 # 1 # 2 _ nature _ * # 1 * , # 2 |
for a function @xmath17 , its restriction to a subset @xmath18 of its domain is denoted @xmath19 .
we say that @xmath19 is * increasing * if @xmath20 for all @xmath21 with @xmath22 .
define @xmath23 the main goal of this paper is to investigate @xmath24 when @xmath25 is a random walk .
the simple random walk is the most natural case , but our results apply to walks with steps of mean zero and finite ( positive ) variance , that is , @xmath26 such that @xmath27 is an i.i.d . sequence with @xmath28 and @xmath29 . by normalising @xmath27
we may clearly assume that @xmath30 .
we say that @xmath31 is the simple random walk if @xmath32 .
the famous erds - szekeres theorem @xcite implies that @xmath33 must contain either an increasing or a decreasing subsequence of length at least @xmath34 .
this is sharp for general sequences , and it is easy to see that there are even @xmath0 step simple walks on @xmath10 for which this is achieved . by symmetry ,
increasing and decreasing subsequences have the same length distribution , but this does not immediately imply that a similar bound holds in high probability . in random settings
, there have been extensive studies of the longest increasing subsequence in a uniformly random permutation @xmath35 .
this is easily equivalent also to the case of a sequence @xmath36 of i.i.d.(non - atomic ) random variables .
a rich theory rose from the study of this question , which is closely related to last passage percolation and other models .
it was proved by vershik and kerov @xcite and by logan and shepp @xcite that @xmath37 and @xmath38 in probability as @xmath39 . in this case ,
much more is known .
baik , deift and johansson @xcite proved that the fluctuations of @xmath40 scaled by @xmath41 converge to the tracy - widom @xmath42 distribution , first arising in the study of the gaussian unitary ensemble .
we refer the reader to romik s book @xcite for an excellent survey of this problem . on the other hand
, it appears that this problem has not been studied so far even for a simple random walk @xmath31 .
the expected length of the longest strictly increasing subsequence of @xmath33 is at most the expected size of the range @xmath43 , hence is @xmath44 .
thus we consider ( weakly ) increasing subsequences .
taking the set of record times , or alternatively the zero set of @xmath31 both yield increasing subsequences of expected length @xmath45 .
it is not immediate how to do any better .
the largest level set of @xmath31 still has size @xmath45 .
see [ f : lis ] for the longest increasing subsequence in one random walk instance .
note that the set of record times yields a similar lower bound for a general random walk with mean zero and finite variance . on some reflection ,
one finds a number of arguments that yield the weaker bound @xmath46 for a simple random walk @xmath31 . for example , first one can show that with high probability , in any interval @xmath47 , any value @xmath48 is visited at most @xmath49 times .
assume @xmath50 is such that @xmath51 is increasing .
for each @xmath52 define the interval @xmath53 $ ] , where @xmath54 is the first ( and @xmath55 is the last ) time @xmath56 such that @xmath57 . by monotonicity
the intervals @xmath58 are disjoint .
the length of the subsequence is then bounded by @xmath59 , where the number of intervals is at most @xmath60 . as @xmath61 with high probability , the cauchy - schwarz inequality gives the upper bound @xmath62 .
however , going beyond the exponent @xmath63 requires more delicate arguments even in the case of a simple random walk .
a related question is to bound the dimension of sets along which brownian motion is increasing , balka and peres @xcite showed that the hausdorff dimension of such sets is at most @xmath64 .
however , the methods used there are not powerful enough to prove the leading upper bound even in the case of a simple random walk .
other results are known in the continuous , deterministic case , and for topologically generic functions ; see elekes @xcite , kahane and katznelson @xcite , and mth @xcite .
the main goal of this paper is to prove the following theorem .
it is more convenient to state it using an exponential scale , but it is easy to switch back to the standard scale . [
t : ub ] let @xmath26 be a random walk with @xmath65 and @xmath30 . for all @xmath0 large enough for all @xmath66 we have @xmath67{n\log_2 n } } }
\right)\leq \left(\frac{2}{n^2}\right)^\ell.\ ] ] moreover , if @xmath68 for some @xmath69 then we have @xmath67{n\log_2 n } } } \right)\leq 2^{-\ell n}.\ ] ] in the following corollaries let @xmath31 be a random walk as in theorem [ t : ub ] .
[ c:1 ] for all @xmath70 with probability @xmath71 we have @xmath72 [ c : e ] for all @xmath73 and @xmath0 large enough @xmath74 in the other direction , we show that in the case of a simple random walk , with high probability there are increasing subsequences somewhat longer than the trivially found ones .
[ t : lb ] let @xmath75 be a simple random walk .
for any @xmath70 for all @xmath0 large enough @xmath76 consequently , for all large enough @xmath0 we have @xmath77 in [ s : high ] we consider higher dimensional random walks .
let @xmath78 and let @xmath79 .
we say that @xmath31 is _ increasing _ on a set @xmath80 if all the coordinate functions of @xmath51 are non - decreasing , i.e. @xmath31 is increasing with respect to the coordinate - wise partial order on @xmath81 .
generalizing @xmath82 , we define @xmath23 since the restriction of a random walk to a single coordinate is again a random walk , if @xmath31 is a @xmath83-dimensional random walk with mean @xmath84 and bounded second moment then [ c:1 ] implies that @xmath85 with probability @xmath71 . for a large class of two - dimensional random walks we are able to prove a lower bound as well .
however , the problem of determining the correct exponent remains open .
[ t:2d_lb ] let @xmath86 be a two - dimensional random walk with steps @xmath87 for which * the mean @xmath88 is the zero vector , * the covariance matrix @xmath89 is the identity matrix , * the coordinates of @xmath90 have finite @xmath91 moments for some @xmath92 .
then there is a constant @xmath93 such that for any @xmath73 and @xmath94 @xmath95 consequently , for all @xmath94 we have @xmath96 finally , in [ s : open ] we state some open questions .
oa thanks the organizers of the _ probability , combinatorics and geometry _ meeting at the bellairs institute , as well as several participants , in particular simon griffiths who proposed this problem , and louigi addario - berry , guillaume chapuy , luc devroye , gbor lugosi and neil olver for useful discussions .
the present collaboration took place mainly during visits of oa and rb to microsoft research .
we are indebted to andrs mth and boris solomyak for useful suggestions .
the goal of this section is to prove [ t : ub ] .
the key is a multi - scale argument , the time up to @xmath97 is split into @xmath98 intervals .
we consider the number of these intervals that intersect our set @xmath18 , as well as the sizes of intersections . repeating this allows us to get ( inductively ) better and better bounds .
the dependence on the randomness of the walk is done through some estimates on the local time , which we derive in the following subsection . throughout this section ,
fix a random walk @xmath99 with @xmath28 and @xmath30 .
various constants below depend only on the law of @xmath27 .
we will use the following theorems in this section .
[ t : p ] there is a constant @xmath100 such that for all @xmath101 and @xmath102 we have @xmath103 for the following theorems see ( * ? ? ?
a.2.5 ) and its corollaries .
[ t : lambda1 ] for all @xmath101 and @xmath104 we have @xmath105 [ t : lambda2 ] assume that @xmath106 for some @xmath69 .
then there is a constant @xmath107 such that for all @xmath101 and @xmath108 we have @xmath109 let @xmath110 and @xmath111 . a * time interval of order @xmath112 *
is of the form @xmath113 a * value interval of order @xmath112 * is of the form @xmath114 note that a time interval is a subset of @xmath115 , while a value interval is a real interval . for all @xmath116 let @xmath117 be the set of time intervals of order @xmath118 contained in @xmath119 .
clearly @xmath120 . the * scaled local time *
@xmath121 is the number of order @xmath118 intervals in @xmath117 in which @xmath31 takes at least one value in @xmath122 : @xmath123 our intermediate goal is to prove the following uniform estimate on scaled local times .
[ p : local_time_bound ] there is a @xmath124 such that for all @xmath0 large enough @xmath125 we begin with an estimate on the expectation of a single scaled local time .
let @xmath126 and @xmath127 denote the expectation and probability for a random walk started at @xmath128 .
[ l : es ] for some absolute constant @xmath100 and any @xmath129 we have @xmath130 the proof is straightforward , and simplified by the idea that conditioned on the event that some time interval contributes to @xmath121 , with probability bounded from @xmath84 , the random walk is still nearby at the end of the time interval . by the the strong markov property and translation invariance it suffices to consider the case @xmath131 .
let @xmath132 be the event that @xmath133 and let @xmath134 be the event that @xmath135 contributes to @xmath136 . by theorem [ t : p ]
there is a constant @xmath137 such that for all @xmath138 and @xmath139 we have @xmath140 since the central limit theorem implies that @xmath141 , there are @xmath142 and @xmath143 such that for all @xmath144 we have @xmath145 it follows that that @xmath146 , so @xmath147 where @xmath148 .
the proof is complete .
next we estimate the tail of a single scaled local time .
[ l : s_tail ] there is an absolute constant @xmath149 such that for all @xmath129 and @xmath150 we have @xmath151 let @xmath152 , where @xmath100 is the constant of [ l : es ] and @xmath153 denotes rounding up . by markov
s inequality we have @xmath154 , establishing the claim for @xmath155 .
we proceed inductively : assume that the claim holds for some @xmath156 .
observe the walk starting at time @xmath157 either until we reach time @xmath158 or until @xmath159 sub - intervals of order @xmath118 contribute to @xmath121 .
the latter happens with probability at most @xmath160 . by the strong markov property
the conditional probability that there are @xmath161 additional sub - intervals contributing to @xmath121 is at most @xmath64 , proving the claim for @xmath162 .
let @xmath163 , where @xmath149 is the constant of [ l : s_tail ]
. we apply [ l : s_tail ] with @xmath164 to each of the relevant @xmath165 .
since @xmath166 , there are @xmath167 choices for each of @xmath112 and @xmath168 . as @xmath169 , there are at most @xmath97 options for @xmath170 . if @xmath171 then @xmath172 with @xmath173 have scaled local time @xmath84 .
this is likely , as [ t : lambda1 ] yields that @xmath174 .
these imply that for all @xmath0 large enough we have @xmath175 clearly we may also require @xmath124 .
next , we use [ p : local_time_bound ] to rule out the existence of very long increasing subsequences in the random walk .
we need the following definition .
let @xmath31 be a function and let @xmath176 be a finite set such that @xmath177 .
the * variation * of @xmath31 restricted to @xmath18 is defined as @xmath178 note that if @xmath19 is increasing then @xmath179 equals the diameter of @xmath180 .
the upper bound of [ t : ub ] follows from the following proposition .
[ p : n_product ] fix @xmath181 .
assume that a walk @xmath17 is such that 1 .
[ as:1 ] the event of [ p : local_time_bound ] occurs , 2 .
[ as:2 ] @xmath182
. then we have @xmath183 let @xmath184 be a set such that @xmath19 is increasing . for @xmath185 let @xmath186 be the set of intervals of order @xmath187 that intersect @xmath18 , and its size with a convenient normalization .
clearly @xmath188 and @xmath189 .
in order to prove the claim we prove inductively bounds on @xmath190 .
let @xmath156 and index the elements of @xmath191 , and suppose that interval @xmath192 contains @xmath193 intervals in @xmath194 , so that @xmath195 .
by assumption for any @xmath172 we have that @xmath196 is visited in at most @xmath197 sub - intervals of @xmath192 .
it follows that if @xmath198 then @xmath199 must visit multiple value intervals of order @xmath187 .
this leads to a variation bound @xmath200 assumption yields that @xmath201 .
thus @xmath202 inequalities and imply that @xmath203 using @xmath204 and dividing by @xmath205 yields @xmath206 where we have used that @xmath207 .
as @xmath208 , the above inequality implies @xmath209 for every @xmath210 .
in particular we get for @xmath211 @xmath212 finally , we use [ p : n_product ] to derive an estimate on the likelihood of long increasing subsequences in a random walk .
first we prove the theorem for @xmath155 .
let @xmath213 where @xmath153 denotes rounding up .
note that @xmath214 .
we consider @xmath31 up to time @xmath215 .
for @xmath0 large enough with probability @xmath216 the event of [ p : local_time_bound ] occurs to @xmath217 .
moreover , [ t : lambda1 ] implies that @xmath218 thus with probability at most @xmath219 the conditions and conclusion of [ p : n_product ] hold for @xmath217 .
suppose additionally that @xmath68 for some @xmath69 .
then [ t : lambda2 ] yields that for some constant @xmath107 and for all @xmath0 large enough @xmath220 thus with probability at least @xmath221 the conditions and conclusion of [ p : n_product ] hold for @xmath217 .
let @xmath0 be such that [ p : n_product ] holds for @xmath217 .
since @xmath222 is increasing in @xmath0 , we obtain that for @xmath0 large enough @xmath223{n\log_2 n } } + o ( \sqrt{\smash[b]{n/\log_2 n } } ) } \\ &
< 2^{n+4\sqrt{\smash[b]{n\log_2 n } } } - 1 . \end{aligned}\ ] ] this proves [ t : ub ] if @xmath155 . for the general case
fix @xmath224 , it is enough to prove that for all @xmath66 we have @xmath225 then setting @xmath226{n\log_2 n } } } \rfloor$ ] concludes the proof , where @xmath227 denotes rounding down .
let @xmath228 .
if @xmath229 is already defined then let @xmath230 be the minimal integer @xmath231 so that @xmath232 . since @xmath233 increases by at most @xmath234 when incrementing @xmath231 , we actually have @xmath235 . by the strong markov property at @xmath229
, we see that @xmath236 are i.i.d .
copies of @xmath237 .
however , @xmath238 requires @xmath239 for all @xmath240 , with probability at most @xmath241 .
this implies , and the proof is complete .
the goal of this section is to prove [ t : lb ] . for simplicity , we present our argument only for the simple random walk on @xmath10
. however , it seems that the argument should extend with minor changes to any walk with bounded integer steps of 0 mean , and finite variance .
the construction relies on values appearing multiple times in the walk , and fails more fundamentally if the walk is not supported on multiples of some @xmath242 .
let @xmath243 denote the * hitting time * of @xmath0 by the simple random walk . let @xmath244 be the * 2-order * of @xmath245 , that is , the number of times it is divisible by @xmath246 .
we use the convention that @xmath247 .
[ l : excursions ] consider a simple random walk from @xmath248 conditioned to hit @xmath249 before returning to @xmath248 , and stopped when it reaches @xmath249 .
let @xmath250 be the times of the first and last visits to @xmath128 .
then : 1 .
[ i:1 ] the number of visits to @xmath128 is geometric with mean @xmath251 .
[ i:2 ] the walk on @xmath252 $ ] is a walk conditioned to hit @xmath128 before returning to @xmath248 , and stopped when it reaches @xmath128 .
[ i:3 ] the walk on @xmath253 $ ] is a walk from @xmath128 conditioned to hit @xmath249 without returning to @xmath128 , and stopped when it reaches @xmath249 .
[ i:4 ] the two sub - walks and the geometric variable are independent . in order to prove the first statement we first consider a simple random walk from @xmath128 up to the time @xmath254 when it reaches either @xmath248 or @xmath249 .
this walk has probability @xmath255 of returning to @xmath128 without hitting @xmath256 , at which time another excursion from @xmath128 begins .
therefore the number of visits to @xmath128 on @xmath257 $ ] is geometric with mean @xmath251 .
moreover , the number of visits to @xmath128 is independent of whether the walk hits @xmath249 or @xmath248 , so when conditioning on hitting @xmath249 the distribution is still geometric with mean @xmath251 , which proves the first statement .
now we return to our original walk from @xmath248 .
excursions from @xmath128 either return to @xmath128 , or hit @xmath249 , or hit @xmath248
. the partition into excursions around @xmath128 does not give any information on the trajectory within each excursion , except for its type , and the other claims follow .
[ l : lb_stopped ] let @xmath17 be a simple random walk . for all @xmath0 we have @xmath258 and for any @xmath70 , @xmath259 we construct an increasing subsequence of @xmath260 as follows .
informally , we take some times @xmath261 to be in our index set , greedily in decreasing order of the 2-order of @xmath262 . for each integer
@xmath263 we construct an interval @xmath264 \subset [ 0,\tau_{2^n}]$ ] .
the intervals are such that if @xmath265 then @xmath266 . given such intervals , we have that @xmath31 is increasing along @xmath267 , where @xmath268 we start by setting @xmath269 $ ] and @xmath270 = \{\tau_{2^n}\}$ ] , where @xmath271 is the last visit to @xmath84 before @xmath272 .
let @xmath273 and @xmath274 be such that @xmath275 and assume by induction that @xmath276 $ ] are already defined for all @xmath277 for which @xmath278 .
now we define @xmath279 .
let @xmath280 and @xmath281 , then clearly @xmath282 .
thus @xmath283 $ ] and @xmath284 $ ] are already defined by the inductive hypothesis .
let @xmath285 $ ] , where @xmath286 is the first hitting time of @xmath128 after @xmath287 and @xmath288 is the time of the last visit to @xmath128 before @xmath289 .
see [ f : dyadic ] for an example . in a simple random walk stopped at @xmath290 .
all visits to @xmath291 are used , then all compatible visits to @xmath292,@xmath293 , followed by @xmath294 and a single visit to each odd value .
there exist longer subsequences of length @xmath295 in this case.,scaledwidth=60.0% ] assuming @xmath275 , we show that the law of @xmath31 restricted to @xmath296 $ ] is that of a simple random walk started at @xmath297 conditioned to hit @xmath298 before returning to @xmath297 , and stopped when hitting @xmath298 .
this is seen inductively using [ l : excursions ] , and since the walk after the last visit to @xmath84 before @xmath272 can not return to @xmath84 . from the above
, we deduce that the number of visits to @xmath128 in @xmath279 is geometric with mean @xmath299 , and these are all independent .
since there are @xmath300 values @xmath301 with @xmath275 we get that @xmath302 as any geometric @xmath90 satisfies @xmath303 and our geometric random variables are all independent , we obtain that @xmath304 the second claim now follows by chebyshev s inequality .
fix @xmath70 . for large enough @xmath305
let @xmath306 be an integer such that @xmath307 then we have that @xmath308 applying [ l : lb_stopped ] for this @xmath112 we get @xmath309 moreover , ( * ? ? ?
* thm . 2.17 ) and
imply that @xmath310 since @xmath311 and @xmath312 as @xmath313 , plugging the previous bounds in gives for @xmath0 large enough @xmath314 finally , applying the above inequality for @xmath315 implies that @xmath316
the main goal of this section is to prove [ t:2d_lb ] .
as noted , the upper bound in the one - dimensional case holds trivially in any dimension . for sequences @xmath317 we use the notation
@xmath318 if @xmath319 as @xmath39 .
the lower bound is based on the following estimate by denisov and wachtel , see ( * ? ? ?
* example 2 ) and see there the history of similar estimates for brownian motion and random walks . [ t:2dgreedy ]
let @xmath320 be a two - dimensional random walk satisfying the conditions of [ t:2d_lb ] .
let @xmath254 be the hitting time of the positive quadrant : @xmath321 .
then there is some @xmath93 so that @xmath322 more generally , for a higher dimensional random walk @xmath323 , define the hitting time @xmath324 denisov and wachtel proved in @xcite that @xmath325 for some @xmath93 and @xmath326 , where @xmath242 is the exponent corresponding to brownian motion staying outside a quadrant up to time @xmath231 ( assuming again that the walk is normalized so that @xmath327 and @xmath328 , and that @xmath329 for some @xmath70 )
. consequently , the following lemma completes the proof of [ t:2d_lb ] ( with @xmath330 ) , and gives a similar lower bound for random walks in higher dimensions .
let @xmath323 be a random walk in @xmath81 , and let @xmath331 be such that @xmath332 then there is a constant @xmath93 such that for any @xmath70 and @xmath94 @xmath333 consequently , for all @xmath94 we have @xmath334 fix @xmath70 .
define the greedy increasing subsequence with time indices given by the recursion @xmath335 setting @xmath336 , we see that if @xmath337 then @xmath338 .
this gives a set @xmath339 with i.i.d .
increments @xmath340 with the law of @xmath254 .
choose @xmath341 such that for all @xmath101 @xmath342 and define the truncated variables @xmath343 .
then @xmath344 the @xmath345 also form an i.i.d . sequence and satisfy @xmath346 where @xmath347 depends only on @xmath242 and @xmath348 . by markov s inequality ,
@xmath349 combining and we obtain @xmath350 as noted , this is a bound on @xmath351 . hence holds with @xmath352 . applying with @xmath353 yields the second claim
there are many potential extensions of our results .
two central open problems are to reduce the gap between the lower and upper bounds in dimension one , and to determine the right order of magnitude in higher dimensions .
let @xmath354 be a random walk with zero mean and finite ( positive ) variance .
is there a constant @xmath286 such that , with probability @xmath71 , @xmath355 does this upper bound hold at least when @xmath31 is a simple random walk ?
let @xmath78 and let @xmath356 be a @xmath83-dimensional simple random walk .
what is the order of magnitude of @xmath24 with probability @xmath71 ? can we find a non - trivial lower bound in dimensions @xmath357 ?
v. v. petrov , on an estimate of the concentration function of a sum of independent random variables , _ teor .
_ , * 15 * ( 1970 ) , no . 4 , 718721 .
english translation in _ theor .
appl . _ * 15 * ( 1970 ) , no . 4 , 701703 .
a. m. vershik , s. v. kerov , asymptotic behavior of the plancherel measure of the symmetric group and the limit form of young tableaux ( russian ) , _ dokl .
nauk sssr _ * 223 * ( 1977 ) , 10241027 .
english translation in _
soviet math .
_ * 233 * ( 1977 ) , 527531 . | given a sequence of @xmath0 real numbers @xmath1 , we consider the longest weakly increasing subsequence , namely @xmath2 with @xmath3 and @xmath4 maximal
. when the elements @xmath5 are i.i.d .
uniform random variables , vershik and kerov , and logan and shepp proved that @xmath6 .
we consider the case when @xmath1 is a random walk on @xmath7 with increments of mean zero and finite ( positive ) variance . in this case , it is well known ( e.g. , using record times ) that the length of the longest increasing subsequence satisfies @xmath8 .
our main result is an upper bound @xmath9 , establishing the leading asymptotic behavior .
if @xmath1 is a simple random walk on @xmath10 , we improve the lower bound by showing that @xmath11 .
we also show that if @xmath12 is a simple random walk in @xmath13 , then there is a subsequence of @xmath14 of expected length at least @xmath15 that is increasing in each coordinate .
the above one - dimensional result yields an upper bound of @xmath16 .
the problem of determining the correct exponent remains open . |
the purpose of the engineering reservoir program @xcite is to protect a specific state against the decoherence stemming from the natural coupling between a quantum system and the reservoir . to engineer a reservoir , a given system ,
whose state is to be protected , is compelled to engage in additional interactions besides that with the natural reservoir .
the engineering reservoir technique is then applied to make these additional interactions prevail , modifying the dissipative liouvillian in a specific way that drives the system to the desired equilibrium with the engineered reservoir .
rightly connected to the engineering hamiltonian program @xcite , the reservoir engineering has been developed for trapped ions @xcite and atomic two - level ( @xmath0 ) systems @xcite . recently , under the assumption of a squeezed engineered reservoir , a way to observe the adiabatic geometric phase acquired by a protected state has been proposed @xcite under the assumption of an adiabatic evolution of the system - reservoir parameters .
it is worth mentioning recent results on the possibility to drive an open many - body quantum system into a given pure state by an appropriate design of the system - reservoir coupling @xcite . in this paper
we detail the general recipe presented in ref.@xcite showing how to build nonadiabatic coherent evolutions of a two - level ion trapped into a leaky cavity , in contrast with the applications of the engineering reservoir technique which require adiabatic conditions @xcite .
furthermore , we show how to implement the proposal of ref .
@xcite for controlling the slow changes of a protected @xmath0 system through the parameters of the engineered reservoir .
we also investigate the influence of thermal reservoirs at finite temperature on the fidelity of the protected superposition state .
this paper is organized as follows . in section
ii we briefly present the engineering reservoir program , including the advances presented in ref . @xcite . in section
iii we show in details a general recipe to engineer reservoirs for a fermionic system interacting with a bosonic mode . as an application we derive an effective reservoir for a time - dependent superposition state of the internal degrees of freedom of an ion trapped into a leaky cavity . in section
iv we present numerical results supporting our approximations and an analysis of the temperature effects on the protected time - dependent superposition . finally , in section v we present our conclusions
before we present the technique to protect superpositions of quantum states evolving nonadiabatically , we review basics results concerning the engineering reservoir program . to this end , we focus our attention on a system interacting with a reservoir , both system and reservoir being modeled by harmonic oscillators . the coupling , as usual ,
is assumed to be linear in position - position . from this model
one can deduce a master equation , obtained by tracing over the reservoir variables , to study the evolution of the single harmonic oscillator . by driving the system with external fields and/or allowing the system to interact with additional systems , each one possessing its own reservoir , it is possible to obtain , through some approximations , an effective master equation which is the starting point to protect a given state by means of engineered reservoirs , as we detail in the following .
the master equation describing the evolution of the density operator @xmath1 of a given system coupled with its natural reservoir in the interaction picture and at zero temperature is@xmath2 where @xmath3 ( @xmath4 ) is the creation ( annihilation ) operator in the fock states of the system .
as it is well known , the only steady state resulting from eq.([eqmestra0 ] ) is the vacuum @xmath5 , which is an eigenstate of the annihilation operator with zero eigenvalue : @xmath6 , @xmath7 .
the main goal of the standard engineering reservoir @xcite is to obtain , in the interaction picture and at zero temperature , an effective master equation in the form @xmath8 where @xmath9 is the effective decay rate of the engineered reservoir which is coupled to the quantum system in a specific way characterized by the time - independent system operator @xmath10 .
proceeding in analogy with eq.([eqmestra0 ] ) , the only pure steady state of this system is the eigenstate @xmath11 of the operator @xmath10 with null eigenvalue , ensuring that there is no further eigenstate @xmath12 of @xmath10 such that @xmath13 \left\vert \phi\right\rangle = 0 $ ] . as a consequence
, @xmath11 is the asymptotic state of the system @xcite .
@xcite adds an improvement on the engineering reservoir program by allowing to remove the adiabatic constraint in the decoherence - free evolution mentioned above . to understand this simple , yet effective improvement
, here we review the theory developed in ref .
consider the engineered time - dependent master equation in the interaction picture @xmath14 + \frac{\gamma}{2}\left [ 2\mathcal{o}\left ( t\right ) \rho\mathcal{o}^{\dagger}\left ( t\right ) -\mathcal{o}^{\dagger}\left ( t\right ) \mathcal{o}\left ( t\right ) \rho\right .
\nonumber\\ & \left .
-\rho\mathcal{o}^{\dagger}\left ( t\right ) \mathcal{o}\left ( t\right ) \right ] , \label{eqmestra2}\ ] ] where the hermitian hamiltonian @xmath15 must be engineered aiming to synthesize without adiabatic impositions the desired time - dependence of the operator @xmath16 , with @xmath17 , @xmath18 , and @xmath19 being the time - ordering operator .
such a relation between @xmath20 and @xmath21 justifies the above mentioned intimate connection between both programs of engineering hamiltonians and reservoirs ; in fact , the reservoir engineering technique relies on engineered hamiltonians .
* * we stress that we have achieved the time evolution of eq .
( [ eqmestra2 ] ) in a particular way that the engineered hamiltonian @xmath15 prompts the specific operator @xmath22 and its protected evolving eigenstate @xmath23 ( with null eigenvalue ) .
the key feature to be noted here is that , through the unitary transformation @xmath24 , we recover the time - independent form of the master equation:@xmath25 therefore , the protected stationary state , @xmath26 ( @xmath27 ) , turns out to be a nonstationary state @xmath28 in the original interaction picture of ( [ eqmestra2 ] ) , where @xmath29 .
getting rid of the adiabatic constraints , we are thus allowed to manipulate the evolution of the protected state @xmath30 through appropriate engineered hamiltonian and reservoir .
it is worth mention that the time dependence of the protected state is closely related to the properties of @xmath31 and @xmath22 . if @xmath32 = \left [ i\overset{\cdot}{r}(t)r^{\dag}(t),\mathcal{o}\left ( t\right ) \right ] = i$ ] @xmath33 , then it is straightforward to see that @xmath34 , i.e. , @xmath35is stationary apart from a global phase factor .
next we develop a general theory of reservoir engineering for a system interacting with a bosonic mode , i. e. , we show how to use the bosonic decay to build up the general master equation ( [ eqmestra1])*. * then we apply this theory to protect quantum states of a @xmath0 system , showing how to obtain an effective interaction between this system and a cavity mode which leads to the desired state protection mechanism .
the starting point to develop a general theory of reservoir engineering for a fermionic system ( from now on `` the system '' ) interacting with a dissipative bosonic field is the engineered effective hamiltonian@xmath36 where @xmath3 ( @xmath4 ) is the creation ( annihilation ) operator of the bosonic field , while @xmath37 and @xmath38 are operators associated with the system whose state we wish to protect , and @xmath39is the effective coupling between the bosonic mode and the system . as an example ,
some of us showed how to build up bimodal interactions in cavity quantum electrodynamics using three- or two - level atoms @xcite .
we note that the interaction ( [ hef1 ] ) is a bilinear form similar to the interaction of a bosonic field with a natural reservoir which leads to the well - know liouvillian describing the amplitude damping mechanism .
thus , the engineered interaction ( [ hef1 ] ) must generate an effective dissipative liouvillian which competes with the natural one .
when the engineered decay rate is significantly larger than the natural one , the effective dissipative liouvillian will govern the dynamics of the system .
once achieved the important step of building the interaction between the system of interest and the quantized bosonic mode , we now add the dissipative mechanism of the bosonic mode through the master equation @xmath40 \nonumber\\ & + \frac{\kappa}{2}\left ( 2a\rho^{(sa)}a^{\dagger}-a^{\dagger}a\rho ^{(sa)}-\rho^{(sa)}a^{\dagger}a\right ) , \label{master - geral}\ ] ] written in the same representation we obtained @xmath41 ; the factor @xmath42 stands for the natural decay rate of the bosonic mode and the supra index in @xmath43 indicates its dependence from both the field ( @xmath4 ) and system ( @xmath44 ) operators . from the above equation we straightforwardly derive the evolution equation for the matrix elements @xmath45 in the fock basis : @xmath46 \label{master1}\\ & + \kappa\sqrt{\left ( n+1\right ) \left ( m+1\right ) } \rho_{n+1,m+1}^{(s)}-\frac{\kappa}{2}\left ( n+m\right )
\rho_{n , m}^{(s)}.\nonumber\end{aligned}\ ] ] to engineer the reservoir , we assume that the decay constant @xmath42 of the bosonic mode is significantly larger than the effective coupling * * @xmath47 in eq .
( [ master1 ] ) .
this condition is easily achieved for cavities with low quality factor @xmath48 , where @xmath49 is the bosonic mode frequency .
together with the good approximation of a reservoir at zero temperature , this regime enables to consider only the matrix elements @xmath50 inside the subspace @xmath51 of fock states .
actually , eq .
( [ master1 ] ) can very well be described by the set of equations @xmath52 which are similar to those written in the atomic basis in the reservoir engineering program for trapped ions @xcite . the strong decay rate @xmath42 enables the adiabatic elimination of the elements @xmath53 and @xmath54 from the equations above .
the formal adiabatic elimination is equivalent to assume @xmath55 , allowing us to write @xmath56 .
we can thus eliminate both @xmath57 and @xmath58 from the dynamics of the diagonal elements of the density matrix by substituting them in the above equations . since @xmath59 , the reduced density operator for the system of interest thus results@xmath60 where @xmath61 represents the effective decay rate corresponding to the engineered reservoir . with this result
we can easily see that the reservoir engineering technique depends basically on the manipulation of the interaction between the bosonic mode and the system whose state we wish to protect .
the two kinds of master equations presented above , ( [ eqmestra1 ] ) and ( [ eqmestra2 ] ) , may be recovered through eq .
( [ eff - master ] ) by identifying the operator @xmath37 with @xmath10 and @xmath62 , respectively .
such correspondence is established when the operator @xmath10 is written in the usual interaction picture , while @xmath62 is written in an arbitrary representation where it is stationary and eq .
( [ eqmestra2 ] ) remains valid .
below we show how to engineer a reservoir which allows to protect an arbitrary evolving superposition state of the internal degrees of freedom of a trapped ion / atom without the adiabatic constraints .
to implement the ideas discussed in the previous section , we use a two - level ( @xmath0 ) system ( s ) characterized by the transition frequency @xmath63 between the ground @xmath64 and excited @xmath65 states . this system can be either a @xmath0 neutral atom in a dipole trap @xcite or a @xmath0 ion in a harmonic trap with frequency @xmath66 @xcite . the transition @xmath64 @xmath67 @xmath68 is driven by a classical field of frequency @xmath69 , with coupling strength @xmath70 . the @xmath0 system is made to interact with a cavity mode field * * ( * * @xmath4 * )
* of frequency @xmath71 under the jaynes - cummings hamiltonian with rabi frequency @xmath72 .
assuming the case of a two - level trapped ion , the hamiltonian modelling this system is given , within the rotating - wave approximation ( @xmath73 ) , by @xmath74 where @xmath75(@xmath76 ) is the creation ( annihilation ) operator of the vibrational mode whose position operator is @xmath77 , @xmath78 being the ionic mass and @xmath79 the unit vector along the vibrational direction .
the wave vectors @xmath80 and @xmath81 stand for the cavity mode and the classical amplification field ( with relative phase @xmath82 ) , respectively , while @xmath83 ( @xmath84 and @xmath44 labeling the states @xmath72 or @xmath85 ) .
the vibrational mode is decoupled from the remaining degrees of freedom of our model by assuming the wave vectors @xmath80 and @xmath81 to be perpendicular to @xmath86 ; otherwise , a sufficiently small lamb - dicke parameter keeps the motional state almost unchanged . under this assumption
we arrive at the hamiltonian@xmath87 which will be the starting point for our purposes . in the following ,
we show how to engineer reservoirs suitable for obtaining nonadiabatic evolutions of the internal ionic states . to engineer the appropriate interaction between the @xmath0 ion and the cavity mode we apply the unitary transformation given by @xmath88 $ ] , leading to the hamiltonian @xmath89 , \nonumber\ ] ] with @xmath90 ( @xmath91 ) being the detuning between the atomic transition ( cavity mode ) and the laser field .
moving to another frame of reference defined by the unitary transformation @xmath92
t\right\ } , $ ] the foregoing calculations are significantly simplified .
this procedure leads to the hamiltonian@xmath93 \operatorname{e}^{-i\phi_{c}}+\mathrm{h{.c.}}\text { , } \label{h2}\ ] ] where @xmath94 , @xmath95 , and @xmath96 .
the atomic operators are defined by @xmath97 , with @xmath98 and @xmath99 . by assuming a large detuning between the cavity mode and the laser field , such that @xmath100 , together with the additional choice @xmath101 , we obtain under the rwa the effective hamiltonian@xmath102 with @xmath103 .
the effective coupling thus depends on the parameter @xmath104 , whose value follows from the laser detuning @xmath105 which must be significantly smaller than the rabi frequency @xmath70 .
next , observing that @xmath106 $ ] , we focus on the values @xmath107 which allow for an effective coupling of the same order of the atom - cavity field coupling , i.e. , @xmath108 . once achieved the building of the interaction between the two - level system and the cavity mode , we now take into account their interaction with a thermal reservoir at temperature @xmath109 through the master equation@xmath110 \nonumber\\ & + \frac{\kappa}{2}\left ( 2a\widetilde{\rho}^{(sa)}a^{\dagger}-a^{\dagger } a\widetilde{\rho}^{(sa)}-\widetilde{\rho}^{(sa)}a^{\dagger}a\right ) \label{master11}\\ & + \frac{\gamma}{2}\left ( 2\widetilde{\sigma}_{ge}\widetilde{\rho}^{(sa)}\widetilde{\sigma}_{eg}-\widetilde{\sigma}_{ee}\widetilde{\rho}^{(sa)}-\widetilde{\rho}^{(sa)}\widetilde{\sigma}_{ee}\right ) , \nonumber\end{aligned}\ ] ] where again the supra index in @xmath111 denotes the density operator for both system and field and the tilde is used to describe the operators in the same representation of @xmath41 * * , * * i.e. , @xmath112**. * * here the constants @xmath42 and @xmath113 are the decay rates of the cavity mode and the tl system , respectively . now , except by the superoperator describing the decay of the two - level system , we can see that eq .
( [ master11 ] ) is equivalent to eq .
( [ master - geral ] ) , with @xmath114 and @xmath115 . as described above , assuming a strong decay rate @xmath42 we can adiabatically eliminate the cavity mode variables .
this procedure is needed to make clear what is the asymptotically protected atomic superposition .
then , the dynamics of the reduced density operator for the @xmath0 system results to be ( @xmath116)@xmath117 so that @xmath118 represents the effective damping of the engineered reservoir and @xmath119 * * is the last term in the r.h.s .
( [ master11 ] ) after tracing over the cavity field variables .
note that the operator @xmath120 and consequently the protected state @xmath121 ( @xmath122 ) , exhibit no temporal dependences in the convenient representation where we have described the evolution ( [ mestra_1laser ] ) .
however , as we shall see in the following , a coherent nonadiabatic evolution is recovered in the interaction picture , where @xmath123 and @xmath124 $ ] , even under spontaneous decay , provided that @xmath125 . here
we note that the combined unitary operations @xmath126 act on the hilbert space of the atom and the mode .
however , once we are interested in the two - level system only and since the field and atomic operators commute with each other , we are omitting , for convenience , the corresponding state of the mode . _
ideal case : _ * * in the ideal case where @xmath127 , the solution of eq .
( [ mestra_1laser ] ) leads to the @xmath0 steady state @xmath128 @xmath129 . by returning to the interaction picture
we thus obtain @xmath130 , where * * @xmath131 with @xmath132 .
this state describes a nonadiabatic evolution that depends on the detuning @xmath105 between the atomic transition and the classical field .
note that under the restriction @xmath107 imposed above , the states around the north pole of the bloch sphere are not allowed steady states .
_ nonideal case : _ to appreciate the robustness of the present reservoir engineering technique , it is necessary to take into account the damping stemming from the natural reservoir when @xmath133 . here
we will consider the reservoir at zero temperature .
the effect of finite temperature in the fidelity of the protected state will be analyzed in the next section .
for convenience , we will analyze such effects in the frame on which we have defined @xmath134 . to this end , we use @xmath135 in eq .
( [ mestra_1laser ] ) , which can thus be written as@xmath136 \nonumber\\ & + \frac{\gamma_{eng}}{2}\left ( 2\sigma_{+-}\rho_{at}\sigma_{-+}-\sigma _ { --}\rho_{at}-\rho_{at}\sigma_{--}\right ) \label{mestra_1laser_u}\\ & + \frac{\gamma}{2}\left ( 2\sigma_{ge}\rho_{at}\sigma_{eg}-\sigma_{ee}\rho_{at}-\rho_{at}\sigma_{ee}\right ) \text{.}\nonumber\end{aligned}\ ] ] by projecting eq .
( [ mestra_1laser_u ] ) on the atomic basis we find the following set of equations corresponding to the matrix elements @xmath137 , @xmath138 , and @xmath139 : @xmath140 \rho_{at}^{++}\right . \\ & \left .
-\widetilde{\gamma}\chi\sqrt{4-\chi^{2}}\left ( \rho_{at}^{+-}+\rho_{at}^{-+}\right ) \right .
\text{,}\\ \overset{\cdot}{\rho}_{at}^{+- } & = \left .
\widetilde{\gamma}\left ( \chi ^{2}-2\right ) -2\widetilde{\gamma}\chi\sqrt{4-\chi^{2}}\rho_{at}^{++}\right .
\\ & \left .
-\left [ \frac{\gamma_{eng}}{2}+2\widetilde{\gamma}\left ( \chi ^{2}-12\right ) -i2\xi\right ] \rho_{at}^{+-}\right . \\ & \left
+ \widetilde{\gamma}\sqrt{4-\chi^{2}}\rho_{at}^{-+}\right .
\text{,}\ ] ] with @xmath141 . imposing the condition @xmath142
, we can also determine the asymptotic solutions for @xmath143 and @xmath144 , greater simplified considering a large cooperative parameter @xmath145 , given by@xmath146 where @xmath147 ^{2}$ ] and @xmath148 .
under the condition @xmath149 , we see that both @xmath150 and @xmath151 are much smaller than unity .
actually , taking into account the noise effects introduced by the reservoir , the steady state is approximately described by @xmath152 .
the net effect of the noise introduced by the atomic decaying mechanism is computed through the fidelity@xmath153 = 1-\epsilon_{++}\text{.}\ ] ] the approximation leading to eq .
( [ mestra_1laser_u ] ) is better as higher the decay rate @xmath42 ; however , for the dynamics of the @xmath0 system to be driven by the engineered reservoir , the magnitude of @xmath42 must be chosen within the restriction @xmath154 . within the optical regime @xcite , for example ,
where @xmath155 s@xmath156 and @xmath157 s@xmath156 , we obtain for a cavity decay constant @xmath158 , the strength @xmath159 and a fidelity around @xmath160 .
therefore , under the excellent approximation @xmath161 and reversing the unitary transformation @xmath21 , the state @xmath162 , written in the interaction picture eq .
( [ estado_prot ] ) allows for a nonadiabatic coherent evolution of a @xmath0 system under spontaneous decay , which can be manipulated through the parameter @xmath105 .
such an evolution corresponds to flips in the atomic states , representing trajectories on different planes parallel to the equator on the bloch sphere , governed by the hamiltonian @xmath163 .
to validate the approximations carried out in the previous section , we proceed to numerically evaluate the full hamiltonian eq .
( [ hamilto_partida ] ) , in the interaction picture , taking into account the thermal reservoir for both the atom and the field , such that@xmath164 \nonumber\\ & + \frac{\kappa\left ( \overline{n}_{a}+1\right ) } { 2}\left ( 2a\rho ^{(sa)}a^{\dagger}-a^{\dagger}a\rho^{(sa)}-\rho^{(sa)}a^{\dagger}a\right ) \nonumber\\ & + \frac{\kappa\overline{n}_{a}}{2}\left ( 2a^{\dagger}\rho^{(sa)}a - aa^{\dagger}\rho^{(sa)}-\rho^{(sa)}aa^{\dagger}\right ) \nonumber\\ & + \frac{\gamma\left ( \overline{n}_{s}+1\right ) } { 2}\left ( 2\sigma _ { ge}\rho^{(sa)}\sigma_{eg}-\sigma_{ee}\rho^{(sa)}-\rho^{(sa)}\sigma _ { ee}\right ) \nonumber\\ & + \frac{\gamma\overline{n}_{s}}{2}\left ( 2\sigma_{eg}\rho^{(sa)}\sigma _ { ge}-\sigma_{gg}\rho^{(sa)}-\rho^{(sa)}\sigma_{gg}\right ) \text { , } \label{mastert}\ ] ] where@xmath165 and @xmath166 and @xmath167 are the average number of thermal photons for the bosonic field and atomic reservoirs , respectively .
( dashed line ) and @xmath168 ( dotted line ) . the constant value @xmath169 indicated by the dashed - dotted line refers to the case where the coupling @xmath72 between the atomic system and the cavity mode is null . ]
we emphasize that , differently from the master equation ( [ master11 ] ) which has been derived under the rwa approximation leading to the effective hamiltonian ( [ hef2 ] ) , the above master equation ( [ mastert ] ) describes exactly the dynamics of the whole system thus allowing us to quantify the errors introduced by our approximations
rwa in eq .
( [ hef2 ] ) and adiabatic elimination of the field variables in eq .
( [ mestra_1laser ] ) .
since we are assuming the atomic frequency near the resonance with the cavity field frequency , both their reservoirs will have the same average thermal photons , such that from now on we take @xmath166 @xmath170 @xmath171 . by projecting the above equation in the fock @xmath172 and electronic @xmath173 bases , we are lead to an infinity set of coupled equations for the matrix elements . to solve numerically
this system of infinity coupled differential equations we must truncate the fock basis somewhere .
the strong decay rate @xmath42 allows us to safely do it since the matrix elements corresponding to highly excited fock states are virtually zero .
we then numerically solve the master equation ( [ mastert ] ) following the method presented in @xcite .
since we are interested in the evolution of the atomic two - level system only , we trace numerically over the bosonic field variables .
therefore , after solving the full master equation we end up with a density matrix for the two - level system with elements @xmath174 where @xmath175 . here
we analyze the robustness of the protected atomic superposition given by eq .
( [ estado_prot ] ) .
our strategy consists in computing the robustness of the protected state under thermal effects considering the following different regimes : i ) @xmath176 _ constant _ , which corresponds to the static case , ii ) by allowing @xmath177 to slowly vary with time , corresponding to @xmath178 , and iii ) allowing @xmath179 to rapidly vary in time , corresponding to @xmath180 .
the robustness of the protected state is computed through the fidelity @xmath181 $ ] where @xmath182 follows from the numerical solution of the full master equation ( [ mastert ] ) . in units of @xmath113
, we assumed the reasonable decay rates @xmath183 , @xmath184 and @xmath185 .
starting with the static case , we consider @xmath186 with constant @xmath82 , such that @xmath187 . assuming @xmath188 and the @xmath0 atom prepared in the ground state , in fig .
1 we present the numerical results of the fidelity against the parameter @xmath189 for both cases of absolute zero ( solid line ) and finite temperature , with the average number of thermal photons @xmath190 ( dashed line ) and @xmath168 ( dotted line ) . as expected , we verify that the temperature effects reduce substantially the fidelity of the atomic superposition . to stress the effectiveness of our protocol , we call the attention to the case where the coupling @xmath72 between the atomic system and the cavity mode is null . in this case we observe that the fidelity remains @xmath169 ( dashed - dotted line ) showing that the cavity mode is a crucial ingredient , together with the classical pumping , to protect the desired superposition state . undergoing adiabatic and nonadiabatic evolutions at zero temperature .
we consider the values @xmath191 ( solid line ) , @xmath192 ( dashed line ) , and finally , @xmath193 ( dotted line ) . ] in fig .
2 we plot the fidelity of the protected state @xmath194 undergoing slow and fast evolutions .
we have considered three distinct values for the detuning between the atom and the classical field : @xmath195 , leading to a slow evolution with @xmath191 ( solid line ) , @xmath196 , departing from the adiabatic regime with @xmath192 ( dashed line ) , and finally @xmath197 , with @xmath198 ( dotted line ) .
all these curves in fig .
2 were plotted considering @xmath199 . as we are concerned with slow and fast evolutions dictated by the time varying parameter @xmath200 , we focused our attention in @xmath179 taking @xmath188 .
we stress that when the condition @xmath178 is weakened , meaning that some parameters are rapidly varying in time , we found that although the equilibrium is reached more slowly , the fidelity does not drop off quickly , attaining a value about @xmath201 even for @xmath193 , corroborating again the effectiveness of our protocol . undergoing nonadiabatic evolutions , with @xmath197 ( @xmath198 ) , for absolute zero ( solid line ) and finite temperature , with the average number of thermal photons @xmath190 ( dashed line ) and @xmath168 ( dotted line ) . ] in order to see the temperature effects in the nonadiabatic evolutions , in fig .
3 we draw the curves of the fidelity for @xmath197 ( @xmath202 ) for different values of mean number of thermal photons , i.e. , for absolute zero ( solid line ) , @xmath190 ( dashed line ) and @xmath168 ( dotted line ) . again , as in the static case ( fig .
1 ) , we observe that the fidelity decreases with the increase of the mean number of thermal photons , showing that the temperature is the most important source of decoherence in our present protocol .
this conclusion is also drawn from fig.4 , where a functional dependence of the fidelity of the protected state @xmath203 undergoing nonadiabatic evolutions against the reservoirs mean photon number is displayed . from this figure , where we have used @xmath204 ( solid line ) , @xmath205(dashed line ) , @xmath206 ( dotted ) and @xmath207 ( dashed - dotted line ) , we note that the fidelity decays faster with the increasing number of thermal photons , as we have mentioned above . undergoing nonadiabatic evolutions , versus the average thermal photons @xmath208 characterizing the finite temperatures with @xmath204 ( solid line ) , @xmath205(dashed line ) , @xmath206 ( dotted ) and @xmath207 ( dashed - dotted line ) . ]
in this paper we deepened the analysis of the engineering reservoir technique we have proposed in ref .
this proposal has been accomplished by deriving a time - dependent master equation which leads to a decoherence - free evolving superposition state which can be nonadiabatically controlled by the system - reservoir parameters . in the present contribution
we have provided a general recipe to engineer arbitrary effective reservoirs for a fermionic system by manipulating its interaction with a bosonic mode .
more specifically , we showed how to protect a superposition state of a two - level ion trapped into a leaky cavity .
the robustness of our scheme was analyzed by means of the fidelity considering the case where the protected state does not depend on time , as well as the case of slowly and rapidly time varying evolutions .
to support the approximations used to derive our analytical results , we have numerically solved the full master equation , obtaining an excellent agreement with the approximations we have carried out . in the present contribution we also analyzed the temperature effects of the reservoirs on the fidelity of the static and time varying protected states .
we concluded that the temperature is the most important source of decoherence in our present protocol .
we hope that this contribution can be useful for information processing with trapped ions inside optical cavities , for example for implementing deutsch algorithm @xcite or universal dissipative quantum computing @xcite .
f. o. prado , n. g. de almeida , m. h. y. moussa , and c. j. villas - boas , phys .
a * 73 * , 043803 ( 2006 ) ; r. m. serra , c. j. villas - boas , n. g. de almeida , and m. h. y. moussa , phys . rev . a * 71 * , 045802 ( 2005 ) ; c. j. villas - bas and m. h. y. moussa , euro .
journal d * 32 * , 147 ( 2005 ) . | in this paper we detail some results advanced in a recent letter [ phys . rev
. lett . * 102 * , 073008 ( 2009 ) ] showing how to engineer reservoirs for two - level systems at absolute zero by means of a time - dependent master equation leading to a nonstationary superposition equilibrium state .
we also present a general recipe showing how to build nonadiabatic coherent evolutions of a fermionic system interacting with a bosonic mode and investigate the influence of thermal reservoirs at finite temperature on the fidelity of the protected superposition state .
our analytical results are supported by numerical analysis of the full hamiltonian model . |
carbon can bond in different hybridization forms , including diamond - like ( @xmath0 ) and graphite - like ( @xmath1 ) , resulting in c based materials with extremely different physical and chemical properties . after equilibrium conditions
are achieved , disrupted @xmath0 bonds may reconstruct as the more stable @xmath1 bonds .
the passage of swift ions through matter causes severe bond breakage , which is generally referred to as a `` thermal - spike '' that accompanies the stopping of the ion in matter . for the case of diamond
, the local bond disruption of the @xmath0 bonds , and subsequent reconstruction upon equilibration , may result in the formation of an @xmath1 rich region along the ion - track .
indeed , the passage of mev or gev heavy ions through diamond - like ( @xmath0 rich ) layers has been shown by high resolution afm and stm scans to result in local graphitization @xcite .
this is evidenced by the formation of protrusions at the ion - impact points ( swelling ) and by the formation of electrically conductive channels .
both these factors indicate that the preferred @xmath1 bonding has occurred along the pathways of the ions , in the otherwise @xmath0 rich material @xcite .
the width of the transformed material , due to the impact of gev au ions , has been deduced from the data an was found to be of the order of 85 nm .
[ rafi3 ] illustrates the material transformation ( density and electrical conductivity ) measured by atomic force and scanning tunneling microscopy .
surface morphology ( left ) and local electrical conductivity ( right ) of the transformed diamond - like amorphous carbon due to ion impact ( taken from ref .
@xcite ) . ]
the damage inflicted on diamond by sub - mev ion - implantation is known to result in electrical conductive regions along the ion track and around the stopping points of the ions @xcite .
the nature of these conductive regions , which have been shown to give rise to hopping conductivity , is controversial .
prins postulates that they mainly contain a mixture of vacancies and interstitial c atoms which eventually collapse to from vacancy clusters ( vacloids ) @xcite . on the other hand , prawer and kalish
have concluded , based on measurements of the dependence of the conductivity of damaged diamond on ion - damage - implantation dose that the ion - affected diamond is highly conductive due to the formation of graphite - like regions along the ion - tracks @xcite .
computer simulations of the damage in diamond caused by the energetic displacement of a few c atoms resulting in damage to a small volume in diamond @xcite ( of 1.4 nm@xmath2 ) show that the size and nature of these regions depend on the initial recoil energy imparted to the c atoms and on the sample temperature . in some cases
the formation of clusters of @xmath1 bonded c atoms forming a six - fold rings has been observed in these simulations . in the present study
we simulate the passage of a swift ion through a diamond or a diamond - like tetrahedrally bonded amorphous c layer ( @xmath3 ) by locally `` heating '' a section of the sample up to very high temperatures ( comparable to the temperatures expected to prevail in the thermal spike ) .
we compute the @xmath1 and @xmath0 fractions in the affected volume and follow their evolution as a function of cooling rate and time .
we find that , upon cooling , a region which contains both @xmath1 and @xmath0 bonded c forms along the track .
the nature of the material in this regrown region depends on the initial heating of the track and on its cooling rate . under some conditions , complete regrowth back to perfect diamond is found , whereas in other cases , a @xmath1 rich region remains along the ion track .
similar results are found when a thermal spike is simulated in amorphous carbon which is then allowed to cool down under a range of different conditions .
the relevance of these computational results , obviously limited to small samples and to relatively short cooling times , to the actual ion passage through c based materials is discussed .
we start below by briefly reviewing the physical processes involved in the formation of the thermal spike and the typical parameters of such a spike initiated in diamond by the passage of a swift heavy ion ; we then describe the computations performed , discuss their relevance to ion - implantation experiments ; present and discuss our results .
the `` thermal spike '' which accompanies the passage of an ion through matter can be viewed as a short - time local melting of the ion damaged region followed by a rapid cooling .
quenching from the liquid phase has been widely used to create amorphous materials ( including amorphous carbon ) in computer simulations , using empirical potentials @xcite , _ ab initio _ @xcite and tight - binding molecular dynamics @xcite .
it has been shown that such a procedure produces an amorphous carbon similar to material produced experimentally by the deposition of a carbon film either by evaporation in an electron beam , by carbon arc @xcite or by ion beam deposition with ion energies between several ten and several hundred ev @xcite .
in contrast to this large body of computer simulations of amorphous carbon , there have been only a few simulations of the material resulting from damaging diamond by ion - implantation .
specific simulations of damage in material induced by ion implantation can be divided into two categories : ( i ) the commonly used trim code @xcite and ( ii ) the molecular dynamics ( md ) technique @xcite .
the trim ( transport of ions in matter ) code is a monte carlo program that calculates the trajectories of the primary ion and of the recoiling host atoms involved in the irradiation event .
the required input parameters are the ion type and energy and the properties of the host material i.e. its composition , density , and the displacement energy of the constituent atoms .
trim provides good estimates of the damage and of the implant distributions and it is widely used in planning the dopant profile in ion - implantation experiments .
however , trim can not account for the structure of the material ( amorphous vs. crystalline ) nor model the dynamic annealing of the implantation affected material .
in particular , it can not predict the nature of the defects in diamond because of the variety of possible bonding configurations of carbon that may induce the diamond to graphite transition .
a modification of trim has recently been developed by shaanan @xcite which enables following the evolution , in space and time , of the damage cascade caused by the passage of a single ion through matter .
the results of the application of this code to simulate the passage of an 1mev xe ion through diamond show that extremely high temperatures ( of the order of many tens of thousand of degrees ) are induced during very short times ( less than 1 ps ) along the ion track . as we will discuss below ,
these resemble the time / temperature regime modeled in our computations of the damage inflicted on diamond and @xmath3 by heavy ion irradiation .
an md study of radiation damage caused by the passage of a single carbon atom through diamond was performed by wu and fahy @xcite , using the tersoff potential . in that study the computations
were restricted to the investigation of the damage threshold energy necessary to displace a single c atom from its lattice site .
smith @xcite simulated the results of the bombardment of diamond using a similar computational method but limiting the analysis to the ejection mechanisms of c atoms from a diamond surface ( carbon self - sputtering ) .
@xcite have investigated the growth of @xmath4 layers . by computing the mean square displacement of the atoms during the cooling of the melt and the time required for it to relax ,
the life - time of the thermal spike in diamond was found to be less than 1 picosecond , for ion impacts below 400 ev . in these latter study
_ ab initio _
md techniques were used .
saada _ et al _ @xcite carried out md simulations , using the tersoff potential , of a damage region embedded inside a diamond matrix created by the energetic displacement of carbon atoms with a kinetic energy of up to 416 ev ( 8 times the displacement energy of carbon atoms in diamond ) in the bulk of a diamond crystal .
such regions are likely to appear in diamond at the final stages of the implantation - related damage cascade .
the conclusion of that study was that partial amorphization of a volume with a radius of 1 - 2 nm occurs around the stopping point of the recoiling ion . with an increasing number of successively displaced atoms ,
the damage volume and the number of threefold coordinated atoms was found to increase .
the structure of the damaged regions ( i.e. the ratio of @xmath1 and @xmath0 bonded c atoms ) was found in ref .
@xcite to depend on the sample temperature .
these computer studies strengthen our understanding of the processes occurring inside or at the end of an ion track in diamond .
however , the transitions from @xmath0 to @xmath1 bonds within thermal spike regions have not yet been investigated in detail .
the purpose of the present study is to investigate transformations of diamond and diamond - like amorphous carbon under the extreme conditions which prevail during and after the passage of a heavy fast ion through the materials .
we present a simulations of the thermal spike which occurs during the passage of a heavy fast ion through either diamond or amorphous carbon by locally heating a slab of the material to very high temperatures and allowing it to rapidly cool down . a large increase in volume
accompanies the @xmath0- to @xmath1 bond conversion @xcite , but the volume of the damaged region in the center of the diamond is restricted by the surrounding diamond ( or amorphous carbon ) lattice , therefore simulations are carried out under constant volume conditions . in order to set the temperature , a randomly directed velocity is assigned to each atom . the value of the velocity is randomly selected in accordance with a maxwell distribution for the desired temperature .
the velocities can be rescaled to be related to the ambient temperatures .
periodic boundary conditions are applied to the samples in all three directions . in order to describe the interactions between carbon atoms we use a tight- binding model @xcite , which has been shown to be transferable to successfully simulate different carbon polytypes : diamond , graphite , fullerenes @xcite as well as disordered carbon structures such as liquid and amorphous carbon phases @xcite .
the md step was 0.5 @xmath5 10@xmath6 s. the method used here to distinguish between @xmath0 and @xmath1 bonded atoms is based on determination of the coordination number of each atom .
the radial distribution function ( rdf ) of all carbon samples exhibits a clear gap , centered at about 0.19 nm , separating the first and the second peak corresponding to the first and second neighbor shells .
all atoms within the sphere of radius 0.19 nm are thus assumed to comprise the nearest neighborhood of a given atom .
therefore , the number of neighbors of each atom within a distance of 0.19 nm determines the coordination number .
atoms that have four nearest neighbors within this sphere are assumed to be @xmath0 bonded , while atoms which have only three are assumed to be @xmath1 bonded .
the atomic visualization package ( * aviz * ) @xcite was used extensively in all stages of this project .
a visualization of our carbon samples with color coding for different atomic bonding helped indentify clusters of either @xmath1 or @xmath0 coordinated atoms , rings , graphite - like planes .
color pictures of the results of the present study can be found on the web site @xcite .
six different samples with a density of 3.5 g / cc , initially arranged as a perfect diamond crystal , were prepared .
their sizes were 2 @xmath5 2@xmath5 6 ( 192 atoms ) , 2 @xmath5 2@xmath5 7 ( 224 atoms ) , 2 @xmath5 2@xmath5 8 ( 256 atoms ) , 2 @xmath5 3@xmath5 6 ( 288 atoms ) and 3 @xmath5 3@xmath5 6 ( 432 atoms ) diamond unit cells .
the `` thermal spike '' was created by heating the central layers up to `` temperatures '' of 1.2 - 2.6 ev ( 14000 - 30000 k ) .
the 32 upper and 32 lower atoms of each sample ( the upper and lower layers of a height of 3.55 ) were frozen , i.e. the motion of these atoms was forbidden .
the initial geometry of the samples is shown in fig .
the outermost frozen diamond layers had the effect on their neighboring hot atoms of forcing them to return to their initial positions i.e. to regrow epitaxially back to diamond . in this way a temperature gradient from the edges to the center of the sample was created .
once the liquid phase reached equilibrium ( this process was controlled by monitoring the total energy of the system ) , the central layers were cooled to room temperature at a cooling rate of 10 k / fs .
the height of the hot layers varied between 4 and 6 diamond unit cells and the width of the hot layers varied from 2 to 3 diamond unit cells .
the full time of cooling was 1.5 ps ( 3000 md steps ) .
however , the central layers ceased to move after the 750 first md steps , so the lifetime of our `` thermal spike '' was 0.3 ps .
this value is similar to the estimates of marks @xcite and saada @xcite , 0.2ps and 0.21ps , respectively .
initial configuration of the samples ; grey ( purple ) balls represent the frozen atoms , white ( yellow ) balls represent atoms that are free to move .
( purple and yellow are shown online ) ] four initial samples of diamond - like amorphous carbon at density 3.5 g / cc were prepared by heating diamond up to 5000 k followed by cooling of the liquid phase to 300 k during 5 ps with an exponential cooling profile .
each @xmath3 amorphous sample contained 192 atoms with a geometry 7.1 @xmath5 7.1 @xmath5 21.3 .
the samples contained 66@xmath74 % of @xmath0 coordinated atoms . in order to achieve a larger percentage of @xmath0 coordinated atoms ( above 70 % ) ,
two additional samples were prepared using the the following procedure : the amorphous structure was prepared with a density of 4.0 g / cc by quenching a liquid phase down from 10000 k to 700 k with a cooling rate of about 500 k / ps , then the density was reduced from 4 to 3.5 g / cc by uniform expansion , after which the temperature of the system was raised to 2500 k , and thereafter the structure was recooled to room temperature @xcite .
the two structures obtained at a final density of 3.5 g / cc consisted of 84 % and 77.5 % of @xmath0 bonded atoms respectively . to simulate a `` thermal spike '' in
the diamond - like amorphous carbon samples , the central region of each sample was heated to `` temperatures '' of 1.2 - 2.60 ev per atom ( 14000 - 30000 k ) .
the upper and lower layers of a height of 3.55 @xmath8 each , were kept frozen .
for the six amorphous carbon samples the height of the hot layers was 14.2 @xmath8 ( 128 atoms in average ) .
a representation of a typical initial sample geometry is given by figure [ init ] .
dark atom at top and bottom indicate the frozen layers , the light central ones are free to move and when heated , there is a temperature gradient from the edges to the center of the samples . following the heating stage the structures
were cooled to room temperature at a cooling rate of 10 k / fs .
the time of calculations and the lifetime of the `` thermal spike '' were the same those described above for the case of the damaged region between two layers of perfect diamond .
we checked carefully , by monitoring of the total energy of the system , that the heating time was sufficiently long to enable the liquid structure to reach equilibrium ( early attempts to prepare samples of amorphous carbon by quenching from a liquid carbon that was not in equilibrium led to inconclusive results ) .
our aim is to determine the nature of transitions between different carbon bonds as the result of the passage of fast ions in samples of different initial structures . to do this ,
we need to examine the region of passage and its intermediate neighbourhood on an atomistic scale .
however , this microscopic region is contained in a larger carbon matrix , and while the microscopic region corresponds directly to that in the experiment , the surrounding matrix may differ in scale .
we address the issues of system size and temperature in relation to the parameters of experimental samples . in the simulations , on one hand , a maximum of 432 atoms can be contained in a single sample due to computational limitations . on the other hand , the periodic boundary conditions mean that the sample is in a sense infinite , and , in particular , has no surfaces .
thus the simulation samples are both smaller and larger than the laboratory experiment .
we are investigating the local region around the ion path ( which has a linear extent of 7.2 @xmath8 along the track directon prior to repetition ) .
the geometry was chosen so that this region is well surrounded , in the direction perpendicular to the track , by diamond or amorphous carbon matrix , which is frozen in place .
in particular , there are 7.4 @xmath8 between replicas of the tracks , a region that is sufficient to avoid interaction since the range of the potential is only 3 . in order to ensure that the samples are of sufficient size to enable reliable interpretations , a series of different sizes were used for each situation and the effect of the scaling of the system size was carefully investigated . although this is not exactly equivalent to the finite size scaling @xcite that is commonly applied to phase transitions and critical phenomena , and is closer to the scaling used by rosenblum et al @xcite in a study of thermal stress at diamond interfaces .
the excellent agreement of both of these scaling approaches with results from laboratory experiments gives us confidence in the concept .
the collapse , as a function of system size , and the smoothness of the scaling that will be demonstrated below confirms that the results can be reliably extrapolated even further if needed .
temperature is introduced into the simulations as kinetic energy . in the laboratory
, the region around the ion track is extremely hot for a very short time .
the high heating `` temperatures '' of the simulation are thus large amounts of kinetic energy imported the heated atoms . here , too , we have studied the scaling of `` temperature '' with system size .
not surprisingly , the actual `` temperature '' needed to attain specific structures is a function of system size .
again , collapse of graphs to a single functional form will be demonstrated and provides validation of our approach .
when the diamond lattice in the central hot layers was melted , @xmath1 and @xmath9 bonded atoms appeared in the liquid layers . upon cooling some ( or all ) of these @xmath1 and @xmath9 bonded atoms transformed back to @xmath0 coordinated c atoms .
the final fraction of threefold and twofold coordinated atoms depends on the kinetic energy per atom ( `` temperature '' ) in the heating process .
the higher the heating `` temperature '' of the atoms the larger was the ratio of @xmath10 , that prevailed after cooling .
samples initially heated to a relatively low `` temperature '' returned to diamond , whereas for hotter samples some non diamond bonds remained after cooling .
for example , let us consider the 2 @xmath5 2@xmath5 6 sample containing 192 atoms , 128 of which were free to move ( i.e. were heated and subsequently cooled ) .
the percentages of four- , three- , and twofold coordinated atoms in the hot layers after cooling from different `` temperatures '' are presented in table [ tab141 ] . whenever the heating `` temperature '' was lower than 1.72 ev per atom all @xmath1 and @xmath9 bonds disappeared in the process of cooling , and the final , cooled , structure returned to a perfect diamond lattice , as shown in fig .
[ perc141 ] where the percentages of different hybridizations are graphed as a function of cooling `` temperature '' . at higher initial heating `` temperatures '' the heated layers remained , after cooling , partially or entirely amorphous .
the fraction of three- and two - fold coordinated atoms in the structure are shown in fig .
[ perc141_1 ] which shows the percentages of @xmath1 bonded atoms during the cooling process from different initial heating `` temperatures '' .
.fraction of the four- , three- , and twofold coordinated atoms in the heated layers after cooling in the sample with 192 atoms . [
cols="^,^,^,^",options="header " , ] now let us consider structure of a central amorphous carbon layer sandwiched between two layers of frozen amorphous carbon .
after cooling , the central damaged region remained amorphous , but the @xmath10 ratio changed .
the dependence of this generated structure on the heating `` temperature '' was similar to that of the previous stage of the simulation ( amorphous carbon between two layers of diamond ) .
namely , if the central atoms were heated to higher `` temperatures '' , the fraction of the @xmath1 bonded atoms increased ( correspondingly , the fraction of the @xmath0 bonded atoms decreased ) . some of these structures were highly inhomogeneous , in particular when the initial sample contained a large fraction of fourfold bonded atoms and was heated to a relatively high `` temperature '' .
in fig . [ amam ] the visualization of some of these amorphous carbon samples is presented .
we can see that the frozen layers of the sample with 77.5 % of @xmath0 bonded atoms initially , contained after heating and subsequent cooling 75 % of @xmath0 bonded atoms , while the central layers after heating to 2.33 ev and subsequent quenching contained only 39 % of the fourfold atoms .
the radial ( rdf ) and the angle ( adf ) distribution functions of the central layers of this amorphous carbon sample after cooling from the `` temperatures '' of 1.55 ev and 2.33 ev are plotted in figs .
[ rdfamor ] and [ adfamor ] respectively .
the peaks corresponding to the higher temperatures of heating were lower , broader , and slightly shifted toward the shorter , graphitelike , bond length and larger , graphitelike , bond angle relative to those at the lower temperatures .
this indicates a higher fraction of the @xmath1 bonded atoms .
it is obvious that the percentage of differently coordinated atoms in the frozen layers has influence on the results of the central region .
the samples initially containing a larger fraction of the @xmath1 bonded atoms , contained a larger fraction of the @xmath1 bonded atoms after cooling ( at the same temperature of heating ) .
[ graphamor ] shows the final percentage of @xmath1 coordinated atoms as a function of heating `` temperature '' plotted for samples with different percentage of @xmath0 and @xmath1 bonded atoms before heating of the central layers .
the data related to the smaller initial fraction of @xmath1 atoms have a lower final fraction of the @xmath1 .
samples of amorphous carbon located between two layers of @xmath3 amorphous carbon : the initial sample ( 77.5 % of @xmath0 ) ( left ) , and the samples after cooling from 2.07 ev ( in the center ) , and from 2.33 ev ( right ) .
grey ( purple ) balls represent fourfold coordinated atoms , white ( yellow ) balls represent threefold coordinated atoms , black balls ( of which there are only four in the right most sample ) represent twofold coordinated atoms .
boundaries between frozen layers and layers of atoms which are free to move are denoted by arrows . ]
radial distribution function @xmath11 of central layers of the sample containing 77.5 % of @xmath0 coordinated atoms initially ( 79 % of @xmath0 in the central layers ) cooled from two different `` temperatures '' : after cooling from 1.55 ev the central layers contained 78 % of @xmath0 coordinated atoms , while after cooling from 2.33 ev the central layers contained 39 % of @xmath0 coordinated atoms . ] angular distribution function of central layers of the sample containing 77.5 % of @xmath0 coordinated atoms initially ( 79 % of @xmath0 in the central layers ) cooled from two different `` temperatures '' . ] the final percentage of @xmath1 coordinated atoms as a function of heating `` temperature '' plotted for samples with different percentage of @xmath0 and @xmath1 bonded atoms before heating of the central layers .
circles in solid line-38 % of @xmath1 in initial amorphous sample ( 36% in the central layers ) , diamonds in dashed line-36% of @xmath1 in initial amorphous sample ( 31% in the central layers ) , squares in dash - dot line - 30% of @xmath1 in initial amorphous sample ( 27.5% in the central layers ) , triangles in dashed line - 22.5% of @xmath1 in initial amorphous sample ( 21% in the central layers ) stars in solid line - 16% of @xmath1 in initial amorphous sample ( 14% in the central layers ) . ] during the calculations many samples of amorphous carbon with different structures were generated .
the width of the band gap was `` measured '' for most of them by computing of density of states ( dos ) as function of energy .
for example , the graph of density of states for two samples of amorphous carbon , containing 86 % of @xmath0 bonded atoms and 53 % of @xmath0 bonded atoms is plotted in fig .
the bandgap of the sample with 86 % of @xmath0 bonded atoms was estimated visually from the graph to be 2.3 ev . the sample with 53 % of @xmath0 bonded atoms
does not indicate the presence of any band gap , this sample is expected to be electrically conducting .
the width of the band gap depends not only on the fraction of the differently coordinated atoms , but also on their individual configuration , i.e on the local clustering of @xmath1 bonded atoms .
[ finalbandgap ] shows a plot of the width of the bandgap as a function of the fraction of @xmath1 coordinated atoms in various @xmath3 samples .
there is a noticeable scattering of the points between 60 and 75 % of @xmath0 bonded atoms .
we suspect that in this interval the fraction of @xmath1 bonded atoms increases sufficiently to begin to create clusters . if the fraction of the @xmath1 atoms is smaller than 25 % , than only separated @xmath1 bonded atoms are embedded in the @xmath0 amorphous network .
if the fraction of the @xmath1 atoms is larger than 40 % , the @xmath1 clusters begin to connect one with another .
the band gap disappears when the fraction of the @xmath1 bonded atoms reaches 45 % . the density of states of the samples of amorphous carbon , containing 86 % of @xmath0 bonded atoms and 53 % of @xmath1 bonded atoms .
( @xmath12 is chosen arbitrary . ) ] the width of the band gap as a function of the percentage of @xmath0 bonded atoms .
the error bars represent an unaccuracy in visual determination of the band gap from dos . ]
the experimental results @xcite described above show that the the high local energy deposition of the ions into diamond or amorphous carbon along their paths causes a rearragement of the carbon atoms and can lead to a transformation of the insulating diamond or amorphous diamond - like @xmath0 form of carbon into a conducting layer rich in @xmath1 bonded carbon atoms . in the simulations we modeled the transformation of
highly ordered diamond structures into amorphous structures with the presence of graphitic like @xmath1-bonds . in the samples which were initially amorphous ,
rearragement of the amorphous structure and an increase of the fraction of @xmath1 bonded atoms has occurred .
the experiments @xcite show that the resistivity , both in diamond and in amorphous carbon networks , decreases with ion dose . in diamond there
exists a critical damage density @xmath13 below which the diamond restores this pristine structure after annealing @xcite . at @xmath14 a sharp drop in resistivity is observed and at higher doses graphitic conduction is measured in the irradiated volume .
kalish et @xmath15 @xcite investigated the possibility that the sharp drop in resistivity could be explained in terms of a percolation transition between conducting islands produced by the impinging ion , but found that the @xmath16 vs @xmath17 dependence was not sharp enough to be consistent with such a simple percolation theory .
by contrast , hoffman et @xmath15 @xcite in their experiment on 1kev ar irradiated diamond proved that the amorphization is a very sudden process and at @xmath18 the diamond collapses to an amorphous but still highly insulating form similar to amorphous diamond ( @xmath3 ) . only at doses exceeding those required for amorphization ,
does the amorphous @xmath0 structure transform to an amorphous @xmath1 structure which is electrically conducting . in our simulation at low `` heating '' temperatures that corresponds to a slower or lighter ion impact , the diamond also reconstructs this structure after cooling . at higher heating `` temperatures ''
the structure remains amorphous after cooling and the fraction of @xmath1 bonded atoms in the damaged region increases with the heating `` temperatures '' .
the faster the ion the larger the damaged region . in our simulation
increasing of the heated volume leaded to increasing of @xmath1 fraction in this ion track and to increasing of conductivity .
experimental results @xcite show that in contrast to diamond , for @xmath3 where no crystalline template exists , there is nothing to inhibit the @xmath0 to @xmath1 bond conversion , hence there is no reason to expect a critical dose at which a sharp increase in the electrical conductivity should start .
moreover , the decrease of the electrical resistivity in @xmath3 occurs at much lower implantation dose than in diamond @xcite .
the results of electron energy - loss spectroscopy allows to estimate the fraction of @xmath1 bonds , @xmath19 , in ion implanted @xmath3 . at
low ion doses the @xmath19 increases linearly with ion dose , at a dose 10@xmath20 ions / cm@xmath2 , @xmath19 becomes less sensitive and increases only slightly with dose . in our calculation
@xmath3 becomes conductor at much lower heating `` temperatures '' than diamond , this means that in contrast to @xmath3 , diamond needs an additional energy to break its crystal lattice and lead to amorphization upon cooling .
the conductivity in our amorphous carbon samples deduced from the shrinkage of the band gap increases with percentage of @xmath1 bonded atoms in the damaged region .
there are some slight discrepancies with experimental studies of electronic structure of damaged region .
for example , mcculloch _ et al _ @xcite found a band gap of 2.5 ev at 85 % of @xmath0 bonded atoms , whereas we find that a value of 3.3 ev corresponds the fraction of 84 % of the @xmath0 atoms .
we have simulated a damaged region occuring in diamond and in diamond - like amorphous carbon by local heating of a slab in the sample of thickness comparable to that created by the `` thermal spike '' due to the passage of an energetic ion followed by quenching back to room temperature .
the `` temperature '' of the heating varied from 1.2 to 2.59 ev / atom ( 14000 - 30000 k ) . for the case of diamond we found that at a heating `` temperature '' higher than a characteristic temperature @xmath22 a small region in the center of the sample remains amorphous . the size and the structure of the amorphous region depends on the heating `` temperature '' ; the size and the fraction of three- and two - coordinated atoms in the region increase with the heating `` temperature '' . when we increased the number of the free - to - move layers , the fraction of three- and two - coordinated atoms in the central layers increased .
the curves of the dependence of the fraction of the @xmath1 bonded atoms in the central layers on the temperature of heating have similar shape for different sizes of the hot layers .
the same procedure was applied to simulate a damaged region in amorphous carbon samples .
the central layers of each sample were melted at a very high temperatures ( 1.29 - 2.33 ev / atom ) and then cooled by keeping the outer layers frozen .
the percentage of three- and twofold coordinated atoms in the obtained damaged region was found to increase with heating `` temperature '' and with the initial fraction of the threefold coordinated atoms in the sample .
most of the structures of amorphous carbon generated in different stages of our simulation were highly inhomogeneous and clustering of @xmath1 bonded c atoms has been observed .
the band gap measured for all the amorphous carbon samples became narrower when the population of @xmath1 bonded atoms increased .
it vanished when the fraction of @xmath1 bonded atoms reached a value of 45 % . beyond this fraction
the damage affected volume becomes conductive .
the results obtained in this study support the experimental conclusions that in the damaged region formed after the passage of a heavy fast ion either through diamond or amorphous carbon , amorphization and sometimes conversion of @xmath0 bonds to @xmath1 bonds occurs .
another conclusion which follows from the simulations is that after passing of a faster ( or heavier ) ion ( that corresponds to higher temperature of the `` thermal spike '' ) , the damaged region is larger and more @xmath0 bonds are transformed to @xmath1 .
although the present simulations do not exactly reproduce the conditions of the experimental `` thermal spike '' , qualitatively our results are in good agreement with experiments .
we are grateful to dr .
g. wagner and dr .
d. segev ( formely known as saada ) for their contribution to this project .
we thank prof .
a. horsfield and prof .
m. finnis for providing us with the oxon package .
this work was supported in part by the israel science foundation and the german israel foundation ( gif ) . | we present tight - binding molecular dynamics simulations of the structural modifications that result from the `` thermal spike '' that occurs during the passage of a heavy fast ion through a thin diamond or amorphous carbon layer , and the subsequent regrowth upon cooling .
the thermal spike and cooling down are simulated by locally heating and then quenching a small region of carbon , surrounded either by diamond or by a mostly @xmath0 bonded amorphous carbon network . for the case of the thermal spike in diamond
we find that if the `` temperature '' ( kinetic energy of the atoms ) at the center of the thermal spike is high enough , an amorphous carbon region containing a large fraction of threefold coordinated c atoms ( @xmath1 bonded ) remains within the diamond network after cooling .
the structure of this amorphous layer depends very strongly on the `` temperature '' of heating and on the dimensions of the thermal spike .
scaling is found between curves of the dependence of the percentage of @xmath1 bonded atoms in the region of the thermal spike on the heating `` temperature '' for different volumes .
when the thermal spike occurs in an initially amorphous sample the structure of the damaged region after cooling exhibits the above dependencies as well as being a function of the structure of the original amorphous carbon layers . |
in the solar system , co - orbital asteroids share the orbit of a planet , going around the sun in almost exactly one orbital period of their host planet , i.e. , move inside the 1:1 mean motion resonance with a planet .
these minor bodies are not truly captured by the gravity of the host planet as natural satellites are , but from the planet s point of view they loop around , sometimes in remarkably stable orbits .
co - orbital motion is characterised by a suitable angular variable , the resonant or critical angle , that in this case is the difference between the mean longitudes of the asteroid and its host planet or relative mean longitude .
the mean longitude of an object planet or minor body is given by @xmath1 , where @xmath2 is the mean anomaly , @xmath3 is the longitude of the ascending node , and @xmath4 is the argument of perihelion ( see , e.g. , murray & dermott 1999 ) . for these objects ,
the value of the critical angle librates or oscillates ; for a regular passing body , the relative mean longitude circulates or takes any value in the range ( 0 , @xmath5 ) .
even if some of them can experience relatively close flybys , co - orbital asteroids tend to be small , transient visitors and they do not pose a particularly high hazard for their host planets .
dynamically speaking , the co - orbital state is surprisingly rich and it includes three elementary configurations quasi - satellite or retrograde satellite , trojan or tadpole , and horseshoe and a multiplicity of hybrid arrangements or compound orbits ( for instance , combination of quasi - satellite and tadpole orbits , morais & morbidelli 2002 ) .
recurrent transitions between the main resonant states are possible ( namouni et al .
1999 ; namouni & murray 2000 ) .
the quasi - satellite dynamical state is observed when the value of the relative mean longitude , @xmath6 , oscillates around 0(see , e.g. , mikkola et al .
if the libration is around @xmath7 we call the object a trojan and it follows a tadpole orbit ( see , e.g. , murray & dermott 1999 ) ; such an object is classified as an @xmath8 trojan when the value of @xmath6 librates around + 60 , or as an @xmath9 trojan if the value oscillates around @xmath10 ( or 300 ) .
the trojan configuration is the most prevalent co - orbital state among long - term , stable co - orbitals .
however , the most usual co - orbital dynamical state among transient co - orbitals is the one characterised by a libration amplitude larger than 180around a value of @xmath11 and often enclosing @xmath7 ; such objects follow horseshoe orbits ( see , e.g. , murray & dermott 1999 ) . at regular intervals and before moving away , a horseshoe librator follows a corkscrew - like trajectory in the vicinity of its host planet for a certain period of time , a decade or more in the case of the earth ; these episodes are recurrent .
hollabaugh and everhart ( 1973 ) predicted the existence of asteroids moving in long - lasting horseshoe orbits associated with the earth .
none was found until 3753 cruithne ( 1986 to ) was identified by wiegert et al .
( 1997 ) as an asteroidal companion to our planet following a horseshoe orbit .
nearly two decades later , the number of known asteroids which are earth co - orbitals remains relatively small .
our planet hosts one trojan
2010 tk@xmath12 ( connors et al .
2011) , four quasi - satellites 164207 ( 2004 gu@xmath13 ) ( connors et al .
2004 ; mikkola et al .
2006 ; wajer 2010 ) , 277810 ( 2006 fv@xmath14 ) ( wiegert et al .
2008 ; wajer 2010 ) , 2013 lx@xmath15 ( connors 2014 ) , and 2014 ol@xmath16 ( de la fuente marcos & de la fuente marcos 2014) , and eight horseshoe librators 3753 ( wiegert et al . 1997 , 1998 ) , 85770 ( 1998 up@xmath17 ) , 54509 yorp ( 2000 ph@xmath18 ) ( wiegert et al . 2002 ; margot & nicholson 2003 ) , 2001 go@xmath0 ( wiegert et al . 2002 ; margot & nicholson 2003 ; brasser et al .
2004 ) , 2002 aa@xmath19 ( connors et al . 2002 ;
brasser et al . 2004 ) , 2003 yn@xmath20 ( brasser et al . 2004 ; connors et al .
2004 ) , 2010 so@xmath21 ( christou & asher 2011 ) , and 2013 bs@xmath22 ( de la fuente marcos & de la fuente marcos 2013 ) .
most of these objects are not large enough to cause widespread damage if they strike our planet , but there are some outliers . while objects such as 2003 yn@xmath20 , 2002 aa@xmath19 or 2001 go@xmath0 are probably not large enough to reach the ground in the rare event of a collision with a significant fraction of their original kinetic energy remaining , others , e.g. , 3753 ( @xmath23 mag ) or 85770 ( @xmath24 mag ) are significantly larger than either the tunguska ( see , e.g. , yeomans 2006 ) or chelyabinsk ( brown et al . 2013 ) impactors and may cause considerable destruction over populated and/or coastal areas . based on their current orbital solutions ,
the 13 objects cited are present - day but transient earth co - orbital asteroids ; the values of their respective @xmath6 librate or oscillate as described above although some of them are in the process of transitioning between co - orbital states or follow compound orbits .
these co - orbitals are often considered rare curiosities , but their dynamical behaviour and affordable accessibility from the earth ( e.g. stacey & connors 2009 ) make them very good candidates for future interplanetary space activities such as in situ study , sample return missions , or even commercial mining ( e.g. lewis 1996 ; elvis 2012 , 2014 ; garca yrnoz et al .
2013 ; harris & drube 2014 ) . here , we show that the recently discovered asteroid 2015 so@xmath0 is performing the corkscrew motion characteristic of earth co - orbitals of the horseshoe type .
this object is the 14th known earth co - orbital and the 9th horseshoe librator .
this paper is organised as follows . in sect .
2 , we present the available data on 2015 so@xmath0 and the methodology followed in this study . the dynamical evolution of this minor body
is studied in sect .
the details of the mechanism triggering the transitions between the horseshoe and the quasi - satellite dynamical states are explored in sect .
4 . in sect . 5
, we consider the impact of errors on our results .
a discussion is presented in sect .
6 and in sect .
7 we summarise our conclusions .
asteroid 2015 so@xmath0 was discovered on 2015 september 21 by b. miku and s. maticic observing with the 0.6-m f/3.3 cichocki telescope at the rni vrh observatory in slovenia ( miku et al .
it had a @xmath25 magnitude of 19.4 when first observed .
it is a small object with @xmath26 = 23.9 mag , which translates into a diameter in the range 50111 m for an assumed albedo of 0.200.04 .
its orbit is moderately well determined with 84 observations acquired during 9 d ( see table [ elements ] ) and it is typical of a minor body that moves co - orbitally with the earth .
the source of the heliocentric keplerian osculating orbital elements and uncertainties in table [ elements ] is the jpl small - body database and they are referred to the epoch jd2457200.5 , i.e. a time prior to its discovery and subsequent close encounter with our planet . at that time , the object was an apollo asteroid moving in an orbit with a value of the semi - major axis @xmath27 = 1.00079 au , very close to that of our planet ( 0.99957 au ) , relatively low eccentricity , @xmath28 = 0.11 , and moderate inclination , @xmath29 = 92 . since the close encounter with the earth ,
2015 so@xmath0 has become an aten asteroid .
0.10truecm .heliocentric ecliptic keplerian orbital elements of asteroid 2015 so@xmath0 .
values include the 1@xmath30 uncertainty ( epoch = jd2457200.5 , 2015-june-27.0 ; j2000.0 ecliptic and equinox .
source : jpl small - body database . ) [ cols="<,<,<",options="header " , ] [ elements ] in the following , we study the short - term dynamical evolution of this recently discovered near - earth asteroid ( nea ) . our analysis is based on results of direct @xmath31-body calculations that use the most updated ephemerides and include perturbations from the eight major planets , the moon , the barycentre of the pluto - charon system , and the three largest asteroids .
the extensive numerical integrations presented here have been performed using the hermite scheme described by makino ( 1991 ) and implemented by aarseth ( 2003 ) .
the standard version of this direct @xmath31-body code is publicly available from the ioa web site .
non - gravitational forces , relativistic and oblateness terms are not included in the calculations ; for further details , see de la fuente marcos & de la fuente marcos ( 2012 ) .
initial conditions ( positions and velocities in the barycentre of the solar system ) have been obtained from the jet propulsion laboratory ( jpl ) horizons system ( giorgini et al .
1996 ; standish 1998 ) and they are referred to as the jd 2457200.5 epoch ( 2015-june-27.0 ) , which is the @xmath32 = 0 instant in our figures .
although the current orbit of 2015 so@xmath0 is relatively poor , if enough test orbits are studied , the best estimates of the past and future dynamical evolution of this object can be determined .
this assumption is based on the widely accepted idea that statistical results of an ensemble of collisional @xmath31-body simulations are accurate , even though individual simulations are not ( see , e.g. , boekholt & portegies zwart 2015 ) . for values of the standard deviation of the orbital elements as moderate as those in table [ elements ] , if consistent behaviour is systematically found within reasonable limits , the dynamical nature of the object can be firmly established .
in addition to the calculations completed using the nominal orbital elements in table [ elements ] , we have performed 50 control simulations with sets of orbital elements obtained from the nominal ones within the accepted error limits ( up to 9@xmath30 ) that reflect the observational uncertainty in astrometry ( see sect . 5 for details ) .
in the figures when an orbit is labelled ` @xmath33 ' , where @xmath34 is an integer , it has been obtained by adding ( + ) or subtracting ( @xmath35 ) @xmath34-times the uncertainty from the orbital parameters ( the six elements ) in table [ elements ] . all the control orbits exhibit consistent behaviour within a few hundred years of @xmath36 , which is the lyapunov time time - scale for exponential divergence of initially close orbits for this object .
in order to study the dynamics of 2015 so@xmath0 we have performed @xmath31-body calculations in both directions of time for 50 kyr using the physical model described above .
figure [ orbit ] , top panel , shows the evolution of the relative mean longitude , @xmath6 , during one period of the horseshoe orbit .
the value of @xmath6 behaves as expected of a classical horseshoe librator ( see , e.g. , murray & dermott 1999 ) .
the motion of 2015 so@xmath0 over the time range ( -170 , 57 ) yr as seen in a coordinate system rotating with the earth in space ( middle panel ) and projected onto the ecliptic plane ( bottom panel ) is plotted in fig .
[ orbit ] ( nominal orbit in table [ elements ] ) .
asteroid 2015 so@xmath0 is an earth co - orbital ; it is performing the corkscrew motion characteristic of earth co - orbitals of the horseshoe type ( see fig . [ orbit ] , middle panel ) .
all the investigated control orbits ( within @xmath37 ) exhibit the same behaviour within the timeframe mentioned above . based solely on the number of simulations performed
, we estimate the probability of 2015 so@xmath0 being a present - day horseshoe librator to the earth at @xmath38% . the motion in fig .
[ orbit ] is the result of the superposition of a @xmath39 yr epicyclic motion describing a kidney - like path and a slow oscillation in mean longitude with a half period close to 113 yr ( the time to reach the other end of the horseshoe ) .
the shape of the epicycle - like path is affected by the value of the eccentricity of the orbit .
minor bodies subjected to this type of dynamics may have had their origin in the earth - moon system ( see , e.g. , margot & nicholson 2003 ) although other sources in the main asteroid belt are possible . as pointed out by connors et al .
( 2004 ) the effects derived from orbital chaos severely limit our ability to determine the source of these neas . in the time interval ( -170 , 57 ) yr as described by the nominal orbit in table [ elements ] .
evolution of the relative mean longitude , @xmath6 , during one period of the horseshoe orbit ( top panel ) . the motion of 2015 so@xmath0 in a coordinate system rotating with the earth , in space ( middle panel ) and displayed projected onto the ecliptic plane ( bottom panel ) . the orbit and position of our planet are also indicated .
all the investigated control orbits ( @xmath37 ) exhibit the same behaviour within this timeframe . ]
the short - term dynamical evolution of the nominal orbit in table [ elements ] is presented in fig .
[ short ] .
figure [ short ] , panel d , shows that 2015 so@xmath0 , like other horseshoe librators , alternates between the aten and apollo dynamical classes .
prior to its close encounter with the earth on 2015 september 30 , earth s gravity was decreasing the asteroids s orbital energy in such a way that its semi - major axis was changing from @xmath40 au ( apollo - type orbit ) to a value of @xmath41 au ( aten - type orbit ) . as an aten
, it will reach the other end of the horseshoe path in about 113 years from now , going from the leading side of the earth to the trailing side . at that time
, earth s gravity will increase the asteroid s orbital energy so its current value of the semi - major axis of @xmath41 au will become @xmath40 au .
figure [ short ] , panel a , shows that 2015 so@xmath0 has only experienced relatively distant close encounters with our planet during the time interval displayed ( 4000 yr ) , well beyond the value of the radius of the hill sphere of the earth , 0.0098 au .
the figure also shows that although 2015 so@xmath0 currently follows a horseshoe orbit with respect to the earth , it will become a quasi - satellite of our planet ( the value of @xmath6 librating around 0 ) in about 350 yr ( see panel c ) .
this dynamical state will persist for about 150 yr .
figure [ short ] , panel g , shows that the argument of perihelion of 2015 so@xmath0 oscillates around @xmath42 or 270 .
therefore , this object reaches perihelion when it is farthest from the ecliptic plane , i.e. , close encounters with inner planets are not possible at perihelion or aphelion .
when an object exhibits this behaviour , it is said to be affected by the kozai mechanism or trapped in a kozai resonance ( kozai 1962 ) . because of the kozai resonance ,
both eccentricity and inclination oscillate with the same frequency but out of phase ( see fig . [
short ] , panels e and f ) ; when the value of the eccentricity reaches its maximum the value of the inclination is the lowest and vice versa ( @xmath43 constant , see fig .
[ short ] , panel b ) .
although certainly present as verified by the behaviour of the invariant , the kozai resonance is not very strong .
( panel b ) .
the resonant angle , @xmath6 ( panel c ) .
the orbital elements @xmath27 ( panel d ) , @xmath28 ( panel e ) , @xmath29 ( panel f ) , and @xmath4 ( panel g ) .
the distances to the descending ( thick line ) and ascending nodes ( dotted line ) are plotted in panel h ; earth s aphelion and perihelion distances are also shown . ] figure [ first ] , central panels , shows the results of the entire integration .
we confirm that 2015 so@xmath0 does not experience particularly close encounters with the earth ; a large number of transitions from horseshoe to quasi - satellite and back again are observed .
the object exhibits kozai - like behaviour throughout the entire time interval with its nodes confined between earth s aphelion and perihelion most of the time .
the distance between the sun and the nodes for a prograde orbit is given by @xmath44 where the `` + '' sign is for the ascending node ( where the orbit crosses the ecliptic from south to north ) and the `` @xmath35 '' sign is for the descending node .
longer integrations not shown here indicate that it may have reached earth s co - orbital region 500200 kyr ago and it could leave in another 100200 kyr to follow orbits interior to that of the earth . based on the integrations performed , the earth s co - orbital region extends from @xmath41 au to @xmath40 au .
figure [ short ] , panel c , shows that in general 2015 so@xmath0 follows a horseshoe orbit with respect to our planet but has periods of quasi - satellite behaviour ; in 4000 years of simulated time we observe three instances of the quasi - satellite phase with an average duration of 150 yr per episode .
asteroid 2015 so@xmath0 is not the first transient earth co - orbital of the horseshoe type that suffers repeated transitions to the quasi - satellite dynamical state , brasser et al .
( 2004 ) have documented this behaviour for 2001 go@xmath0 , 2002 aa@xmath19 and 2003 yn@xmath20 . these transitions had been predicted and explained by namouni ( 1999 ) and christou ( 2000 ) , and depend on the influence of other planets , in particular the giant planets and venus , owing to the overlapping of multiple secular resonances .
this is not surprising as ito & tanikawa ( 1999 ) pointed out that the terrestrial planets share the effect of the secular perturbation from jupiter . in a follow - up work ,
tanikawa & ito ( 2007 ) further extended this analysis concluding that , regarding the secular perturbation from jupiter , the inner planets are a planetary group or collection of loosely connected mutually dynamically dependent planets .
the effects of the overlapping secular resonances persist even outside the region where @xmath6 librates and the dynamics of some passing bodies which are kozai librators because their argument of perihelion oscillates is still controlled by them ( de la fuente marcos & de la fuente marcos 2015a ) .
brasser et al . (
2004 ) noticed that , in the cases studied by them , the transition to the quasi - satellite phase was possible both at the leading and at the trailing sides of the earth , although most transitions took place at the trailing side of our planet .
figure [ short ] , panel c , shows that quasi - satellite episodes start when the nodes are farthest from each other ( see panel h ) and also farthest from the path of the earth . in the solar system and for a minor body moving in an inclined orbit , like 2015 so@xmath0 , close encounters with major planets
are only possible in the vicinity of the nodes .
therefore , the start of the quasi - satellite phase of the co - orbital motion coincides with the time when the average gravitational effects of the earth on the object are the weakest possible .
the transition back to the horseshoe phase is triggered when both nodes are close to earth s perihelion ; then , encounters with the earth - moon system are possible at both nodes and the sustained action of these encounters increases the asteroid s orbital energy eventually triggering the transition and returning the minor body to a horseshoe trajectory . in our simulations ,
the transition from horseshoe librator to quasi - satellite always takes place when the orbital energy of the asteroid increases after having followed a trajectory of the aten type for nearly 100 yr . in other words , all the observed transitions from horseshoe to quasi - satellite took place at the trailing side of the earth , never at the leading side . during the quasi - satellite phase ,
the orbital eccentricity is the highest and the inclination , the lowest ; the value of the argument of perihelion decreases as predicted by namouni ( 1999 ) .
also consistent with namouni ( 1999 ) , during the transitions from horseshoe to quasi - satellite , the argument of perihelion is stationary and a maximum ; the value of the eccentricity must be higher than a critical one to start the transition .
transitions are only observed when the descending node is beyond earth s aphelion and the ascending node is interior to earth s perihelion which is likely the result of overlapping secular resonances .
it may be argued that the orbit of 2015 so@xmath0 currently available is based on a short arc and , therefore , this makes our conclusions rather weak or even entirely questionable
. however , not all orbital solutions with short data - arc spans are equally poor .
the object discussed here follows a relatively stable orbit that is only directly perturbed by the earth moon system .
in addition to the integrations performed making use of the nominal orbital parameters in table [ elements ] , we have computed 50 control simulations with sets of orbital elements obtained from the nominal ones within the quoted uncertainties and assuming gaussian distributions for them up to @xmath37 .
representative results of these calculations are displayed in figs .
[ first ] , [ second ] and [ third ] .
the orbital evolution of 2015 so@xmath0 within a few thousand years of @xmath45 is not too different from that in fig .
[ short ] even if deviations in the values of the orbital elements as large as @xmath46 or even @xmath47 are considered . for the extreme case of @xmath37 ,
the overall dynamical behaviour is still consistent : the object remains within earth s co - orbital zone for the entire integration .
these results clearly indicate that our conclusions are robust and reliable .
a more detailed analysis of the effects of errors on our short - term results is performed here by studying how the changing initial parameters of the test orbits influence the variation of the osculating orbital elements over time .
two additional sets of 100 shorter control simulations are discussed here . as in the previous calculations ,
the initial orbital elements of each control orbit have been computed varying them randomly , within the ranges defined by their mean values and standard deviations .
for example , a new value of the orbital eccentricity has been found using the expression @xmath48 , where @xmath49 is the eccentricity of the test orbit , @xmath50 is the mean value of the eccentricity from the available orbit ( table [ elements ] ) , @xmath34 is a suitable integer ( for instance , 3 , 6 or 9 ) , @xmath51 is the standard deviation of @xmath28 ( table [ elements ] ) , and @xmath52 is a ( pseudo ) random number with normal distribution in the range @xmath351 to 1 .
figure [ errors ] shows the short - term evolution of the orbital elements @xmath27 , @xmath28 , @xmath29 , @xmath3 and @xmath4 of the object studied here .
the thick black curves show the average results of the evolution of 100 control orbits computed as described above .
the thin red curves show the ranges ( minimum and maximum ) in the values of the parameters at a given time .
figure [ errors ] , left - hand and central panels , shows the results for @xmath53 and @xmath54 spreads , respectively , in the initial values of the orbital elements and confirms once more that even if our study is based on a relatively short - arc orbit , its results are robust and reliable .
although our conclusions rest on a 9-day observational arc for 2015 so@xmath0 , we can reliably state that this object follows a horseshoe trajectory with respect to the earth .
in general , fig . [ errors ]
shows that the value of the lyapunov time of 2015 so@xmath0 is very asymmetrical ( about 200 yr versus several hundred years ) ; the orbital evolution is significantly more chaotic into the future than it was in the past .
sitarski ( 1998 , 1999 , 2006 ) has pointed out that the procedure used above is equivalent to considering a number of different virtual minor planets moving in similar orbits , but not a sample of test orbits incarnated from a set of observations obtained for a single minor planet .
the correct statistical alternative is to consider how the elements affect each other , applying the monte carlo using the covariance matrix ( mccm ) approach ( bordovitsyna et al .
2001 ; avdyushev & banschikova 2007 ) , or to follow the procedure described in sitarski ( 1998 , 1999 , 2006 ) . as a consistency test ,
we have used an implementation of the mccm approach to recompute the orbital evolution of this object generating control orbits with initial parameters from the nominal orbit adding random noise on each initial orbital element making use of the covariance matrix ( for details , see de la fuente marcos & de la fuente marcos 2015b ) .
figure [ errors ] , right - hand panels , shows the results of these simulations and , in general , the difference is not very significant . the lyapunov time asymmetry is also confirmed .
if we compare the orbital evolution of 2015 so@xmath0 as depicted in figs . [ first ] , [ second ] and [ third ] with that of other horseshoe librators of the earth , for instance 2001 go@xmath0 , 2002 aa@xmath19 , and 2003 yn@xmath20 ( brasser et al . 2004 ) or 2013 bs@xmath22 ( de la fuente marcos & de la fuente marcos 2013 ) one rapidly realises that 2015 so@xmath0 is far more stable .
given the fact that the values of the semi - major axis of these objects are very similar , the source of the enhanced stability must be in the values of other orbital elements .
asteroids 2001 go@xmath0 , 2003 yn@xmath20 and 2013 bs@xmath22 all move in orbits with @xmath55 ; 2002 aa@xmath19 has @xmath56 , but its eccentricity is rather low , 0.013 .
in contrast , 2015 so@xmath0 has @xmath28 = 0.11 and @xmath29 = 92 .
this combination of values of @xmath28 and @xmath29 appears to be particularly favourable regarding orbital stability .
however , it places the object far from the earth even at close approaches ; in nearly two centuries around the current epoch , the distance of closest approach is 0.037 au .
if objects like 2015 so@xmath0 are numerous , then they will be intrinsically more difficult to discover than those similar to 2013 bs@xmath22 that reach perigee at 0.013 au .
asteroid 2013 bs@xmath22 is far less stable than 2015 so@xmath0 , compare fig . [ first ] with fig . 2 in de la fuente marcos & de la fuente marcos ( 2013 ) .
almost certainly , the most stable known earth s co - orbital is 2010 so@xmath21 that stays as horseshoe librator for at least 120 kyr and possibly up to 1 myr ( christou & asher 2011 ) , remaining in the same co - orbital configuration during this time span .
asteroid 2015 so@xmath0 may be almost as stable as 2010 so@xmath21 but it often switches co - orbital configuration .
in addition , the value of its distance from the earth at perigee remains under 0.5 au for about 22 yr when it reaches the ends of its horseshoe orbit , every 113 yr . during its current stint as neighbour of our planet , its final perigee at @xmath57 au will be on 2026
september 16 .
in other words , objects like 2015 so@xmath0 are characterised by relatively long favourable accessibility windows making the implementation of a hypothetical space mission easier . on the other hand ,
our analysis leaves the question of the possible origin of 2015 so@xmath0 open .
it is clearly a transient co - orbital to our planet , but its dynamical behaviour is quite unusual .
an origin in the earth - moon system can not be entirely ruled out based on the available information alone ; its orbital evolution was significantly more stable in the past .
finally , it can be argued that presenting this object here is rather premature because its orbit is still not well known ( see table [ elements ] ) .
however , this object is only directly perturbed by the earth moon system and the time interval between closest approaches to the earth
moon system is currently 113 yr .
it means that the characteristic time - scale between favourable visibility windows is over a century and that its orbit ( see above ) is quite stable .
if one misses the opportunity of studying the object during the next few years , waiting for over a hundred years will be required to recover it . in addition
, the analysis of the influence of errors on our conclusions clearly indicates that our study is justified and its conclusions robust .
one of the objectives of this research is to bring this interesting minor body to the attention of the astronomical community , encouraging follow - up observations .
spectroscopic studies during its next perigee should be able to confirm if an origin in the earth - moon system is plausible .
in this research , we have used @xmath31-body simulations and statistical analyses to study the orbital evolution of 2015 so@xmath0 .
the main conclusions of our study can be summarised as follows : * asteroid 2015 so@xmath0 currently follows a horseshoe trajectory with respect to the earth ( probability @xmath38% ) , the ninth asteroid known to do so . * asteroid 2015 so@xmath0 is a transient co - orbital that experiences a rather peculiar orbital evolution characterised by alternating horseshoe and quasi - satellite episodes . *
asteroid 2015 so@xmath0 is subjected to a kozai resonance ; its argument of perihelion oscillates around @xmath42 or 270 . *
asteroid 2015 so@xmath0 may have had its origin in the earth - moon system and it is one of the most stable earth horseshoe librators discovered to date .
we thank two anonymous referees for their constructive and helpful reports , and s. j. aarseth for providing the main code used in this research . in preparation of this paper , we made use of the nasa astrophysics data system , the astro - ph e - print server and the mpc data server .
aarseth , s.j . : gravitational n - body simulations , p. 27 .
cambridge university press , cambridge ( 2003 ) avdyushev , v.a . ,
banschikova , m.a .
41 * , 413 ( 2007 ) boekholt , t. , portegies zwart , s. : comput .
* 2 * , 2 ( 2015 ) bordovitsyna , t. , avdyushev , v. , chernitsov , a. : celest .
astron . * 80 * , 227 ( 2001 ) brasser , r. , innanen , k.a . ,
connors , m. , veillet , c. , wiegert , p. , mikkola , s. , chodas , p.w . : * 171 * , 102 ( 2004 ) brown , p.g .
, et al . :
* 503 * , 238 ( 2013 ) christou , a. : * 144 * , 1 ( 2000 ) christou , a.a . , asher , d.j . : * 414 * , 2965 ( 2011 ) connors , m. : * 437 * , l85 ( 2014 ) connors , m. , chodas , p. , mikkola , s. , wiegert , p. , veillet , c. , innanen , k. : meteoritics planet . sci .
* 37 * , 1435 ( 2002 ) connors , m. , veillet , c. , brasser , r. , wiegert , p. , chodas , p. , mikkola , s. , innanen , k. : meteoritics planet .
sci . * 39 * , 1251 ( 2004 ) connors , m. , wiegert , p. , veillet , c. : * 475 * , 481 ( 2011 ) de la fuente marcos , c. , de la fuente marcos , r. : * 427 * , 728 ( 2012 ) de la fuente marcos , c. , de la fuente marcos , r. : * 434 * , l1 ( 2013 ) de la fuente marcos , c. , de la fuente marcos , r. : * 445 * , 2985 ( 2014 ) de la fuente marcos , c. , de la fuente marcos , r. : * 580 * , a109 ( 2015a ) de la fuente marcos , c. , de la fuente marcos , r. : * 453 * , 1288 ( 2015b ) elvis , m. : * 485 * , 549 ( 2012 ) elvis , m. : * 91 * , 20 ( 2014 ) garca yrnoz , d. , sanchez , j.p . , mcinnes , c.r . :
astron . * 116 * , 367 ( 2013 ) giorgini , j.d .
, et al . : * 28 * , 1158 ( 1996 ) harris , a.w . ,
drube , l. : * 785 * , l4 ( 2014 ) hollabaugh , m. , everhart , e. : astrophys . lett . *
15 * , 1 ( 1973 ) ito , t. , tanikawa , k. : * 139 * , 336 ( 1999 ) kozai , y. : * 67 * , 591 ( 1962 ) lewis , j.s . : mining the sky : untold riches from the asteroids , comets , and planets , p. 82 .
addison - wesley , reading ( 1996 ) makino , j. : * 369 * , 200 ( 1991 ) margot , j.l . ,
nicholson , p.d .
: * 35 * , 1039 ( 2003 ) mikkola , s. , innanen , k. , wiegert , p. , connors , m. , brasser , r. : * 369 * , 15 ( 2006 ) miku , b. , et al . :
mpec 2015-s69 ( 2015 ) morais , m.h.m . ,
morbidelli , a. : * 160 * , 1 ( 2002 ) murray , c.d . ,
dermott , s.f . : solar system dynamics , p. 97 .
cambridge university press , cambridge ( 1999 ) namouni , f. : * 137 * , 293 ( 1999 ) namouni , f. , murray , c.d . :
astron . *
76 * , 131 ( 2000 ) namouni , f. , christou , a.a . ,
murray , c.d .
lett . * 83 * , 2506 ( 1999 ) sitarski , g. : * 48 * , 547 ( 1998 ) sitarski , g. : * 49 * , 421 ( 1999 ) sitarski , g. : * 56 * , 283 ( 2006 ) stacey , r.g . ,
connors , m. : * 57 * , 822 ( 2009 ) standish , e.m . : jpl planetary and lunar ephemerides , de405/le405 , interoffice memo .
312.f-98 - 048 , jet propulsion laboratory , pasadena , ca , usa ( 1998 ) tanikawa , k. , ito , t. : publ .
59 * , 989 ( 2007 ) wajer , p. : * 209
* , 488 ( 2010 ) wiegert , p.a . ,
innanen , k.a . ,
mikkola , s. : * 387 * , 685 ( 1997 ) wiegert , p.a . ,
innanen , k.a .
, mikkola , s. : * 115 * , 2604 ( 1998 ) wiegert , p. , connors , m. , chodas , p. , veillet , c. , mikkola , s. , innanen , k. : american geophysical union , fall meeting 2002 , p11a-0352 ( 2002 ) wiegert , p.a . , deboer , r. , brasser , r. , connors , m. : * 102 * , 52 ( 2008 ) yeomans , d.k .
: * 38 * , 1057 ( 2006 ) | earth co - orbitals of the horseshoe type are interesting objects to study for practical reasons .
they are relatively easy to access from our planet and that makes them attractive targets for sample return missions . here
, we show that near - earth asteroid ( nea ) 2015 so@xmath0 is a transient co - orbital to the earth that experiences a rather peculiar orbital evolution characterised by recurrent , alternating horseshoe and quasi - satellite episodes .
it is currently following a horseshoe trajectory , the ninth asteroid known to do so . besides moving inside the 1:1 mean motion resonance with the earth , it is subjected to a kozai resonance with the value of the argument of perihelion librating around 270 .
contrary to other neas , asteroid 2015 so@xmath0 may have remained in the vicinity of earth s co - orbital region for a few hundreds of thousands of years . |
in the statistical analysis of spatial processes , modelling and estimating the spatial dependence structure is fundamental . it is used by prediction techniques like kriging or conditional simulations .
its description is commonly carried out using statistical tools such as the variogram or covariogram calculated on the entire domain of interest and considered under the stationarity assumption , for reasons of parsimony or mathematical convenience .
the complexity of the spatial component of the analyzed process is therefore limited .
the assumption that the spatial dependence structure is translation invariant over the whole domain of interest may be appropriate , when the latter is small in size , when there is not enough data to justify the use of a complex model , or simply because there is no other reasonable alternative . although justified and leading to a reasonable analysis , this assumption is often inappropriate and unrealistic given certain spatial data collected in practice .
non - stationarity can occur due to many factors , including specific landscape and topographic features of the region of interest or other localized effects .
these local influences can be observed computing local variograms , whose characteristics may vary across the domain of observations . for this type of non - stationarity structures ,
making spatial predictions using conventional stationary methods is not appropriate . indeed
, applying stationary approaches in such cases would be liable to produce less accurate predictions , including an incorrect assessment of the estimation error@xcite .
several approaches have been proposed to deal with non - stationarity through second order moments ( see @xcite , for a review ) .
one of the most popular methods of introducing non - stationarity is the space deformation of @xcite and other @xcite .
it consists in starting with a stationary random function , and then transforming the distance in some smooth way to construct a non - stationary random function .
maximum likelihood and bayesian variants of this approach have been developed by @xcite , @xcite , @xcite , @xcite .
@xcite , @xcite , @xcite established some theoretical properties about uniqueness and richness of this class of non - stationary models .
some adaptations have been proposed recently by @xcite , @xcite , @xcite , @xcite .
a fundamental limitation of all estimation methodologies presented so far is the fact that implementation requires multiple independent realizations of the random function in order to obtain an estimated sample covariance or variogram matrix .
the idea of having several independent realizations of the natural field is unrealistic because there are not multiple parallel physical worlds . in practice ,
the approach is feasible when a time series is collected at each location as this gives the necessary , albeit dependent , replications . in general , we would prefer to incorporate a temporal aspect in the modelling rather than attempting repairs ( e.g. , differencing and detrending ) to achieve approximatively independent realizations .
however , many geostatistical applications involved only one measurement at each site or equivalently , only one realization of a random function . @xcite , @xcite are the first authors to address the estimation of space deformation model in the case of a single realization of a random function , obtained as the transformation of a gaussian and isotropic random function .
they exhibit a methodology based on quasi - conformal mappings and approximate likelihood estimation of the local parameters that characterize the deformation derived from partitioning densely observed data into subregions and assuming independence of the random function across partitions .
however , this approach has not been applied to real datasets and requires very dense data . in this work ,
we follow the pioneering work of @xcite , while freeing the strong assumption of replication and do not make any distributional assumptions . in addition , we take into account other shortcomings associated with this approach that are : the required property of the deformation to be bijective and the computational challenge to fit the model for moderate and large datasets .
to do so , we propose an estimation procedure based on the inclusion of spatial constraints and the use of a set of representative points referred to as anchor points rather than all data points to find the deformation .
the proposed method provides a non - parametric estimation of the deformation through a step by step approach : first a dissimilarity matrix is built by combining a non - parametric kernel variogram estimator and euclidean distance between two points in the geographical space , second the estimation of the deformation at anchor points is considered by weighted non - metric multi - dimensional scaling and finally the deformation is interpolated over the whole domain by thin - plate spline radial basis functions .
the proposed method also provides a rational and automatic estimation of the spatial dependence structure in the deformed space .
we illustrate our estimation procedure on two simulated examples and apply it to soil dataset .
this paper is organized as follows : section [ sec2 ] describes the space deformation model through its basic ingredients and main properties . in sections [ sec3 ] and [ sec4 ]
, we address the problem of estimating the non - stationary spatial dependence structure and how spatial predictions and conditional simulations should proceed .
two simulated data in section [ sec5 ] and a real dataset in section [ sec6 ] are used to illustrate the performance of the new approach and its potential . finally , section [ sec7 ] outlines concluding remarks and further work .
the main idea behind the space deformation approach is the euclidean embedding of a non - stationary random function into a new space of equal or greater dimension , where it can be easily described and modeled , that is to say where both stationarity and isotropy hold .
let @xmath0 be a constant mean random function defined on a fixed continuous domain of interest @xmath1 of the euclidean space @xmath2 and reflecting the underlying studied phenomenon .
we consider that @xmath3 is governed by the following model : @xmath4 which can be written equivalently : @xmath5 where @xmath6 represents a stationary and isotropic random function and @xmath7 represents a deterministic non - linear smooth bijective function of the @xmath1-space onto the @xmath8-space . in principle , we can allow @xmath9 , although most frequently @xmath10 . from now on , without loss of generality , we assume @xmath10 . the model specification in leads to model the variogram of @xmath3 in the form : @xmath11 where @xmath12 represents the euclidean norm in @xmath2 and @xmath13 the stationary and isotropic variogram of @xmath14 which depends only on the euclidean distance between locations in @xmath8-space . the second order structure model obtained in leads to a valid variogram , i.e. conditionally non - positive definite @xcite .
its validity is assessed by the following proposition inspired by @xcite and straightforward to establish ( see [ appendix1 ] ) .
[ prop1 ] if @xmath15 is a valid variogram , then @xmath16 is a valid variogram on @xmath17 , for any function @xmath18 .
it is then even possible to rely on a valid variogram in a different space , through a function linking the two spaces .
consequently , instead of working on the support @xmath1 of the non - stationary random function @xmath3 , the variogram of @xmath3 is defined with respect to the latent space @xmath8 where stationarity and isotropy are assumed .
any problem involving the observed random function @xmath3 is transposed by the deformation @xmath19 to the stationary and isotropic random function @xmath14 .
standard geostatistical techniques , such as kriging and conditional simulations can be apply directly to the latter . the results obtained on @xmath14 will then transpose to @xmath3 by the inverse deformation @xmath20 . as described by @xcite ,
when @xmath13 is non - decreasing , the spatial deformation @xmath19 operates as follows : the deformation effectively stretches the @xmath1-space in regions of relatively lower spatial correlation , while contracting it in regions of relatively higher spatial correlation , so that a stationary and isotropic variogram can model the spatial dependence structure as a function of the distance in the @xmath8-space representation .
it is also important to note that the spatial deformation model defined in is identifiable up to a scaling for @xmath13 and up to a homothetic euclidean motion for @xmath19 .
all isometries and homotheties are observationally equivalent .
this result is based on the following proposition equivalent to that established by @xcite for deformation based non - stationary spatial correlation model . the proof is given in [ appendix1 ] .
[ prop2 ] if @xmath21 is a solution to ( [ eq3 ] ) , then for any regular square matrix @xmath22 and any vector @xmath23 , @xmath24 with @xmath25 and @xmath26 is a solution as well .
let @xmath27 be a @xmath28 vector of observations from a unique realization of the random function @xmath3 , associated to known locations @xmath29 . in the second order structure model defined in ( [ eq3 ] ) , the functions @xmath19 and @xmath13 are unknown and need therefore to be estimated .
the estimation workflow involves four main steps .
first , we define a non - stationary variogram non - parametric kernel estimator which serves as a dissimilarity measure between two points in the geographical space @xmath1 .
second , we construct the deformed space @xmath8 using the procedure of weighted non - metric multi - dimensional scaling applied to a dissimilarity matrix built from the non - stationary variogram estimator .
third , we estimate @xmath19 by interpolating between a configuration of points in the @xmath1-space and the estimations of their deformations in the @xmath8-space using the class of thin - plate spline radial basis functions .
fourth , the estimation of @xmath13 is carried out by calculating the experimental variogram on transformed data in the deformed space @xmath8 and using a mixture of basic variogram models , providing wide flexibility to adapt to the observed structure .
we begin with an estimate of the non - stationary variogram at arbitrary locations using a kernel weighted local average of squared increments .
our proposal is to use the variogram cloud @xmath30 which is an unbiased estimator of @xmath31 @xcite as the input data for the non - parametric kernel estimator of the non - stationary variogram .
an intuitive empirical estimator of the non - stationary variogram at any two locations is given by the non - parametric kernel estimator defined as follows : @xmath32 where @xmath33 is a kernel defined as : @xmath34 , with @xmath35 a non - negative , symmetric kernel on @xmath36 with bandwidth @xmath37 .
the expression defines the spatial dissimilarity between two arbitrary points in the geographical space @xmath1 .
the purpose of the kernel function is to weight observations with respect to a reference point so that nearby observations get more weight while remote sites receive less .
the denominator of is a standardization factor that ensures @xmath38 is unbiased when the expectation of the squared difference between the observations is spatially constant .
if several realizations are available , we can simply take the average of over the different realizations .
regarding the kernel @xmath35 defined in , his choice is less important than the choice of its bandwidth parameter @xmath39 , a reasonable choice of @xmath35 will generally lead to reasonable results @xcite .
we use the quadratic kernel ( epanechnikov kernel ) which is an isotropic kernel with compact support , showing optimality properties in density estimation @xcite .
indeed , the computational cost of is greatly reduced by using a compactly supported kernel , as it reduces the number of terms to compute .
the selection of the bandwidth parameter @xmath39 is adresssed in section [ ssec4 ] .
the transformation of the geographical space @xmath1 into the deformed space @xmath8 can be seen as the reallocation of a set of points in a euclidean space ( in a given dimensionality ) .
the approach of non - metric multi - dimensional scaling ( nmds ) provides a solution to this problem @xcite .
the aim is to find a representation of points in a given dimension space , such that the ordering of the euclidean distances between points matches exactly the ordering of the dissimilarities .
till now , the fitting of a space deformation model based on nmds procedure is a challenging numerical problem where the dimensionality is roughly proportional to the number of observations . a sample of size @xmath40 requires a @xmath41 dissimilarity matrix to be stored and a @xmath42 matrix of coordinates to be inferred .
the search of the deformed space based on this dissimilarity matrix requires considerable computing time even when @xmath40 is moderately large . to reduce the computational burden ,
we can avoid mapping directly all data locations .
indeed , spatial dissimilarities calculated between pairs of close points can be unnecessary and redundant because of their high correlation .
the idea consists in obtaining the deformed space using only a representative set of @xmath43 points referred to as anchor points over the geographical space @xmath1 .
these will be rather mapped with nmds .
interpolating the anchor points in the @xmath1-space and the estimations of their deformations in the @xmath8-space , produces an estimate @xmath44 of the deformation function .
then , the location of all sampling data points in the deformed space is obtained through @xmath44 .
representative points may be chosen as a sparse grid over the domain or a subset of data points .
they allow to reduce the computation time and to give reliable results for the nmds task .
the sampling density which may be vary over the domain , can be accounted by non - uniform distribution of the anchor points .
we consider a set of @xmath45 @xmath46-dimensional anchor points @xmath47}^t$ ] of the geographical space @xmath1 . the non - stationary variogram estimator @xmath48 defined in and calculated for each pair of anchor points allows to build a @xmath49 symmetric dissimilarity matrix @xmath50 $ ] , with @xmath51 .
the specification of the dissimilarity matrix is required for the nmds procedure .
the latter is usually applied in the absence of the euclidean coordinates of the points which produce the dissimilarities . however , in our context , the points of the original space @xmath1 are already identifiable by their euclidean coordinates , hence the necessity to ensure the spatial consistency of the transformation .
moreover , @xmath48 reflects the local spatial dissimilarity , that is to say , for close pairs .
indeed , this estimator can be quite accurate for short and moderate distances , as in the stationary framework , but very imprecise , although well defined , for large distances .
it is therefore necessary to penalize the importance given to large distances compared to short distances in the search of the deformation .
therefore , given the distance matrix associated with support points @xmath52 , we build a composite @xmath49 dissimilarity matrix @xmath53 $ ] : @xmath54 \ \mbox{and } \mathbf{\tilde{d}}=[\tilde{d}_{ij}],\ ] ] where @xmath55 represents the scaled ( transformed to the interval [ 0,1 ] ) non - stationary variogram estimator at locations @xmath56 and @xmath57 ; @xmath58 is the scaled euclidean distance between @xmath56 and @xmath57 ; @xmath59 $ ] is a mixing parameter .
the composite dissimilarity matrix @xmath60 is a linear combination of a scaled dissimilarity matrix and a distance matrix .
the idea is to build a hybrid spatial dissimilarity measure which takes into account both the dissimilarities observed in the regionalized variable and the spatial proximity to ensure that the deformation function does not fold , i.e. is bijective .
thus , the parameter @xmath61 controls the non - folding .
the setting of @xmath61 is adressed in section [ ssec4 ] . given the @xmath49 symmetric matrix @xmath53 $ ] of dissimilarities between the set of @xmath45 @xmath46-dimensional anchor points @xmath47}^t$ ] ,
the objective is to represent @xmath52 as a configuration of @xmath62-dimensional points @xmath63}^t$ ] such that the following relations are satisfied as much as possible : @xmath64 where @xmath65 is a monotonic function that preserves the rank order of the dissimilarities . in other words ,
given the @xmath66 dissimilarities , we look for a representation of the anchor points in a given dimension space , so that the rank order of the configuration distances accord with the rank order of the dissimilarities .
the representation of anchor points @xmath67 in the deformed space @xmath8 is determined such that the distances between these points in the @xmath8-space minimize the loss function defined by a _
measure : @xmath68}^2}{\sum_{i < j}{p_{ij}(\lambda)h_{ij}^2(\mathbf{u})}}}\right]}^\frac{1}{2},\ ] ] where the minimization is performed over all monotonic functions @xmath65 and @xmath69 , for @xmath70 .
@xmath33 is the kernel function used in the computing of the non - stationary variogram estimator @xmath71 defined in .
the estimator @xmath72 represents the weighted least squares monotonic regression of the @xmath73 on the @xmath74 .
@xmath75 are non - negative dissimilarity weights used to weight the contribution of the corresponding elements of the dissimilarity matrix @xmath60 in computing and minimizing _ stress_. we introduce such a weighting scheme to take into account the fact that local information available in the neighborhood of anchor points may fluctuate from one anchor point pair to another . hence highly reliable dissimilarities have more impact in the loss function than unreliable ones . such weights also allow to emphasize dissimalirities at short distances .
the _ stress _ provides a measure of the degree to which the rank order of the configuration distances concurs with the rank order of the dissimilarities to be mapped .
it is invariant by translation , rotation or rescaling of the configuration .
the problem defined in is solved by the shepard - kruskal nmds iterative algorithm as described , for example in , @xcite , with the starting configuration taken as the configuration of anchor points @xmath52 . in this paragraph
, we are interested in the estimation of the functional parameters @xmath19 and @xmath13 which govern the second order structure model . the transformation of anchor points is extended to a smooth function @xmath76 over the entire geographical space @xmath1 , using the class of thin - plate spline radial basis functions . using the estimated deformation @xmath76
, an isotropic stationary variogram model is adjusted to the empirical variogram in the deformed space .
finally , the estimated deformation function , in conjunction with the estimated isotropic stationary variogram model , are used to provide estimates of the spatial dependency between any two locations in the geographical space @xmath1 .
we want to know the mapping of any point @xmath77 in the deformed space @xmath8 , as well as that in @xmath1 of any point @xmath78 . especially , a representation of all data points in the deformed space @xmath8 is required .
given the configuration of anchor points @xmath52 and the estimation of their deformation @xmath67 , we seek a @xmath62-variate function @xmath79 that satisfies the interpolation condition @xmath80 .
while a wide range of options exist , we follow @xcite in using the class of thin - plate spline radial basis functions to estimate @xmath19 .
specifically , the thin - plate spline estimator of @xmath19 takes the following form @xcite : @xmath81 where @xmath82 , @xmath83 are the anchor points seen as the centers of the radial basis function @xmath84
. the parameters @xmath85 and @xmath86 in the expression ( [ eq8 ] ) are unknown and therefore have to be determined .
this is achieved by resolving the system of equations @xmath87 under the constraints @xmath88 ( see e.g. @xcite ) .
the performance of the spatial deformation approach also depends on its ability to properly adjust an isotropic stationary variogram model in the deformed space .
thus , the modelling and estimating of the spatial dependence structure that characterizes the spatial behavior of the variable of interest is a key step .
so far , highlighted by @xcite , @xcite , @xcite , no rational choice has been developed yet for the isotropic stationary variogram model in the spatial deformation model . this choice is often made by visual inspection . here , the estimation of @xmath13 in the deformed space @xmath8 is carried out using a mixture of basic variogram models ( nugget effect , gaussian , exponential , spherical , ) thereby enriching the set of theoretical variograms and offering wide flexibility to match the observed structure : @xmath89 where @xmath90 are basic valid isotropic stationary variogram structures and @xmath91 positive real numbers .
to estimate @xmath13 , the isotropic stationary variogram model in the deformed space @xmath8 , we use a robust method recently developed by @xcite .
it consists in automatically finding a model that fits the experimental variogram . from a linear combination of some authorized basic structures a numerical algorithm
is used to estimate a parsimonious model that minimizes a weighted distance between the model and the experimental variogram .
the proposed approach relies on two hyper - parameters @xmath92 used in the computation of the composite dissimilarity matrix @xmath60 .
the method used to select these hyper - parameters is data driven .
we combine two forms of cross - validation to select their optimal value .
the first form is the cross validation based on @xmath48 .
it consists of computing the cross - validation variogram error criterion over a grid of bandwidth values to determine a small set of bandwidths with low scores : @xmath93 where @xmath94 is the non - stationary variogram estimator at locations @xmath95 and @xmath96 computed with all observations except @xmath97 .
@xmath30 is an unbiased estimator of @xmath31 . because the estimation of the spatial dependence structure is rarely a goal per se but an intermediate step before kriging , we want to choose the hyper - parameters that give the best cross - validation mean square error ( mse ) .
more explicitly , for each pair of hyper - parameters @xmath92 , we compute the leave - one - out cross - validation score : @xmath98 where @xmath99 denotes the spatial predictor computed at location @xmath95 using all observations except @xmath100 .
the prediction method is described in detail in section [ ssec5 ] .
the first form of cross - validation ( [ eq11 ] ) allows to select a small grid of bandwidth values which are used in the second form of cross - validation ( [ eq12 ] ) to search the optimal values of @xmath92 .
thus , the first coarse bandwidth search reduces computational burden of the second step .
this selection of hyper - parameters is illustrated with some examples in sections [ sec5 ] and [ sec6 ] .
the main purposes of modelling and estimating the spatial dependence structure is to spatially interpolate data and perform simulations .
we focus here on the description of the kriging and conditional simulations with the spatial deformation model .
the estimation of the space deformation model can be used in kriging - type estimation equations to provide more reliable spatial estimates and prediction standard errors than those based on an inadequate stationary variogram structure . given observations @xmath101 at @xmath102 ,
the point predictor for the unknown value of @xmath3 at unsampled location @xmath103 is given by the ordinary kriging estimator : @xmath104 which minimizes the mean square error @xmath105 under the constraint@xmath106 @xmath107 .
@xmath108 represents the transformed points .
the prediction problem of the non - stationary random function @xmath3 is then transposed to the isotropic stationary random function @xmath14 .
then , the kriging weights @xmath109 are calculated by solving the well known ordinary kriging system @xcite in the stationary framework .
thus , using the space deformation approach to estimate the regionalized variable on target grid points , we can proceed as follows : 1 .
obtain the image of the target grid points and data points through @xmath110 ; 2 .
krige the transformed target grid points using @xmath111 and transformed data points ; 3 .
obtain the kriging on the target grid points by simple correspondence .
there are a number of situations where conventional prediction techniques like kriging not properly fulfill their objectives .
this is when it comes to integrating data , to predict quantities such as extrema or quantiles .
an attractive approach to form consistent estimators of any entity of interest or quantifying uncertainties is through conditional simulations of the random function . here , the goal is to simulate a realization @xmath112 of a gaussian random function @xmath3 with constant mean @xmath45 and second order structure model : @xmath113 , with respect to the data @xmath114 .
perform conditional simulation of the non - stationary gaussian random function @xmath3 defined on the @xmath1-space means do the same for the gaussian random function @xmath14 defined on the @xmath8-space , with isotropic stationary variogram @xmath13 .
this can be accomplished based on a unconditional simulation @xmath115 of the random function @xmath14 with same mean and variogram structure .
for the latter , standard geostatistical simulation techniques already exist @xcite ( e.g. spectral method , turning band method ) .
hence an algorithm of conditional simulation of the random function @xmath3 is as follows : 1 .
calculate the kriged estimates @xmath116 $ ] for each @xmath117 ; 2 . simulate a gaussian random function @xmath115 with constant mean @xmath45 and isotropic stationary variogram @xmath13 in the domain @xmath8 and at the transformed conditioning points .
let @xmath118 and @xmath119 be the generated values ; 3 .
calculate the kriged estimates @xmath120 $ ] for each @xmath117 ; 4 . return @xmath121 .
to check whether our proposed estimation allows to find the correct deformation function , we apply it on one - dimensional example ( @xmath122 ) and two - dimensional example ( @xmath123 ) .
we start with a one - dimensional simulation for which the results can be visualized easily .
we generate at @xmath124 irregularly spaced locations a zero mean , unit variance , gaussian random function @xmath3 over the domain @xmath125 $ ] , based on the following second order structure model : @xmath126 , where @xmath13 is an isotropic stationary exponential variogram model with range parameter @xmath127 and @xmath128 .
our goal is to estimate @xmath19 .
the deformation @xmath19 leaves the domain @xmath1 globally unchanged but shrinks the left part of the segment whereas the right side is stretched .
thus , points of the segment which are located near the origin are highly correlated with their neighbors and points of the segment which are located near the extremity are slightly correlated with their neighbors .
a set of @xmath129 regularly spaced anchor points are used to build the dissimilarity matrix .
figures [ fig1a ] and [ fig1b ] show a realization of the non - stationary random function @xmath3 in the geographical space @xmath1 and in the estimated deformed space @xmath8 .
the non - stationary process is transformed into stationary one simply by stretching and compressing the @xmath130-axe as shown in figure [ fig1b ] .
the optimal value of hyper - parameters correspond to @xmath131 and @xmath132 . from the figures [ fig1c ] and [ fig1d ] we see that the estimated deformation is very close to the true deformation .
this simple example illustrates the ability of the proposed approach to find the right deformation .
0.27 0.27 0.27 0.27 we consider a standardized gaussian random function @xmath3 over a domain @xmath133}^2 $ ] , with a variogram : @xmath134 , where @xmath135 is a radial basis function with center point @xmath136 and @xmath13 an isotropic stationary cubic variogram model with range @xmath137 .
we simulate @xmath3 at @xmath138 points on a regular grid of @xmath1 .
points of the domain which are located near the center are highly correlated with their neighbors while points which are far from the center are slightly correlated with their neighbors ( figure [ fig2a ] ) .
@xmath139 points are sampled and splitted into a @xmath140 points training sample and a @xmath141 points validation sample .
we use a @xmath142 regular grid as a configuration of anchor points ( red cross points in figure [ fig2b ] ) .
figure [ fig2d ] shows the estimated deformed space which looks similar as the true deformed space ( figure [ fig2c ] ) .
our estimation method effectively stretches the domain in regions of relatively lower spatial correlation ( at extremities ) while contracting it in regions of relatively higher spatial correlation ( at the center ) so that an isotropic stationary variogram can model the spatial dependence structure in the deformed space .
0.28 0.28 0.28 0.28 figure [ fig3a ] gives the cross - validation variogram error function @xmath143 .
we see that moderate and large values of @xmath39 tend to give low scores .
figure [ fig3b ] shows the second form cross - validation function @xmath144 .
its minimization leads to @xmath145 and @xmath146 .
0.28 and @xmath144.,title="fig:",scaledwidth=100.0% ] 0.28 and @xmath144.,title="fig:",scaledwidth=100.0% ] a visualization of the variogram in a few points for estimated stationary model , estimated non - stationary model and non - stationary reference model is shown in figure [ fig4 ] .
from one site to another change in the non - stationary spatial dependence structure is clearly apparent . 0.25 ( black ) , @xmath147 ( red ) and @xmath148 ( green).,title="fig:",scaledwidth=100.0% ] 0.25 0.25 ( black ) , @xmath147 ( red ) and @xmath148 ( green).,title="fig:",scaledwidth=100.0% ] to assess the predictive performance of our approach , the regionalized variable is predicted at the 1024 validation data points .
table [ tab1 ] provides a comparative performance of the estimated stationary model , the estimated non - stationary model and the reference non - stationary model .
some well - known discrepancy measures are used , namely the mean absolute error ( mae ) , the root mean square error ( rmse ) , the normalized mean square error ( nmse ) , the logarithmic score ( logs ) and the continued probability score ( crps ) . for rmse , logs and crps , the smaller the better
; for mae , the nearer to zero the better ; for nmse the nearer to one the better . table [ tab1 ] summarizes the results for the predictive performance statistics computed on the validation data set .
the stationary model is worse than the other two models for example , in terms of both rmse , logs and crps .
the estimated non - stationary model performs comparably well as the reference model
. the cost of not using a non - stationary model can be substantial .
for example , the estimated stationary model is @xmath149 worse than the estimated non - stationary model in terms of rmse .
.predictive performance statistics on a test set of 1024 locations .
[ cols="<,^,^,^",options="header " , ] 0.30 0.30 0.30 0.30
the proposed approach appears to be an efficient tool to estimate a non - stationary spatial dependence structure from a single realization .
it improves the estimation of a random function with non - stationary spatial dependence structure , as illustrated by the simulated examples and the real application .
additionally , the non - stationary approach leads to better prediction than the stationary model as measured by predictive scores .
one advantage of the proposed approach is that instead of finding the deformation using all data points , we can do it with a reduced number of points , compared to existing methods .
this is a major improvement for large datasets .
indeed , the use of anchor points makes it possible to run the nmds step even for very large datasets .
moreover , the approach incorporates spatial constraints which guarantee the bijection property of the deformation .
this point was actually a major drawback while trying to fit a deformation function .
the proposed approach is easy to implement and allows to use all developments already made in the stationary framework through the deformation .
indeed , prediction and simulation under stationarity is a well understood subject , for which fast and robust techniques are available .
the approach also provides an exploratory analysis tool for the non - stationarity .
in fact , the spatial deformation encodes the non - stationarity and the representation of the deformed space allows to identify the regions of strong and weak continuity .
one future direction of research would be to investigate the statistical properties ( such as consistency ) of the non - parametric kernel estimator of the non - stationary variogram .
this would have to be done in an asymptotic context by following the work of @xcite in the stationary framework .
the deformation mapping need not to be restricted to the class of functions that are currently use . from our experience , thin - plate spline radial basis functions work well but a different interpolation method may prove useful
. it would be also interesting to account for covariate information in the space deformation model , in order to improve the estimation of the deformation .
this would have to be done by following the work of @xcite .
one challenge remaining is the selection of hyper - parameters .
indeed , the cross - validation procedures presented here remains computationally demanding .
the authors would like to thank dr budiman minasny at the faculty of agriculture & environment at the university of sydney in australia , for providing the data used in this paper .
[ [ appendix1 ] ] _ proof of proposition [ prop1 ] _ consider a set of points @xmath150 and a set of real numbers @xmath151 such that @xmath152 .
let @xmath153 .
we have @xmath154 hence @xmath155 is a valid variogram on @xmath1 .
_ proof of proposition [ prop2 ] _
consider @xmath21 a solution to .
let @xmath24 such that @xmath25 and @xmath26 .
we have @xmath156 hence @xmath24 is a solution as well . | stationary random functions have been successfully applied in geostatistical applications for decades . in some instances ,
the assumption of a homogeneous spatial dependence structure across the entire domain of interest is unrealistic .
a practical approach for modelling and estimating non - stationary spatial dependence structure is considered .
this consists in transforming a non - stationary random function into a stationary and isotropic one via a bijective continuous deformation of the index space .
so far , this approach has been successfully applied in the context of data from several independent realizations of a random function . in this work ,
we propose an approach for non - stationary geostatistical modelling using space deformation in the context of a single realization with possibly irregularly spaced data .
the estimation method is based on a non - stationary variogram kernel estimator which serves as a dissimilarity measure between two locations in the geographical space .
the proposed procedure combines aspects of kernel smoothing , weighted non - metric multi - dimensional scaling and thin - plate spline radial basis functions . on a simulated data ,
the method is able to retrieve the true deformation .
performances are assessed on both synthetic and real datasets .
it is shown in particular that our approach outperforms the stationary approach . beyond the prediction ,
the proposed method can also serve as a tool for exploratory analysis of the non - stationarity .
non - stationarity , variogram , deformation , kernel smoothing , kriging , simulation . |
a large number of papers have been published lately on the determination of the distance to the large magellanic cloud ( lmc ) .
this increased interest is attributed to the extreme values published after the analyses of the hipparcos data on cepheids ( feast & catchpole 1997 ) and to the application of the red clump stars of the lmc through the same type of galactic stars calibrated by hipparcos ( stanek et al . 2000 ) . along this line , there seems to exist a general belief that cepheids yield usually longer distances ( distance moduli of about @xmath3 mag and above ) , whereas rr lyrae stars give shorter ones ( @xmath4 mag ) .
this latter belief is based on statistical parallaxes ( popowski & gould 1998 ) , baade - wesselink ( bw ) analyses ( see clementini et al .
1995 , and references therein , hereafter c95 ) and the application of the hipparcos data ( fernley et al .
1998 ; see however groenewegen & salaris 1999 ) .
it is important to note that distances derived from the statistical parallax method employ rather small sample of stars ( @xmath5 objects ) , whereas the hipparcos parallaxes for rr lyrae stars are very inaccurate . actually , more precise hst parallaxes yield distances in good agreement with the `` canonical '' lmc value of @xmath3 mag derived from both prototype variables rr lyr and @xmath6 cep ( benedict et al .
2002a , b ) .
it is also important to mention that all results based on the pulsation analyses of double - mode variables ( rr lyrae , cepheids , both fundamental & first overtone and first overtone & second overtone ones ) give basically the canonical value for the lmc distance ( see kovcs 2002 and references therein ) . in the light of the above ,
it is important to address the question of consistency between the bw analyses of cepheids and rr lyrae stars .
previous works have been performed by different groups of researchers , often using different bw parameters ( e.g. , @xmath7 factors , the ratio of the observed and surface radial velocities ) and effective temperature relations ( surface brightness for cepheids
e.g. , laney & stobie 1994 , stellar atmosphere models for rr lyrae stars
e.g. , liu & janes 1990 ) .
in addition , there is a role of the somewhat delicate nature of the numerical method employed in the actual implementation of the bw analysis ( e.g. , computation of the surface acceleration ) .
therefore , we decided to visit the problem by using the same bw algorithm for both groups of stars . in order to compute the lmc distance modulus ,
the derived absolute magnitudes will be used in conjunction with currently published period - luminosity - color ( plc ) relations ( udalski et al .
1999 ; kovcs & walker 2001 , hereafter u99 and kw01 , respectively ) .
here we basically follow the standard methodology of the bw analysis as usually implemented in rr lyrae studies ( e.g. , liu & janes 1990 ) . in brief
, we use the observed @xmath8 light , @xmath9 color and @xmath10 radial velocity curves to compute the luminosity ( @xmath11 ) , effective temperature ( @xmath12 ) and radius ( @xmath13 ) variations . for any assumed value of the equilibrium test radius
@xmath14 we compute the following quantities : * _ temporal radius : _ * * @xmath15 , * * @xmath16d\tau$ ] , * * @xmath17 , + where @xmath18 denotes the period , @xmath7 is the velocity projection factor , that we assume to be constant and set equal to @xmath19 .
( we note that gieren , fouqu & gmez 1997 used a period - dependent @xmath7 factor .
their values span the range of 1.351.37 in the period regime of our sample .
the effect of using lower @xmath7 is minute , because an increase of @xmath20 in @xmath7 yields an absolute magnitude that is brighter only by @xmath20 mag . ) * _ temporal @xmath12 : _ * * @xmath21 } , \log g]$ ] , + where @xmath22 $ ] denotes the interpolation polynomial of the stellar atmosphere models of castelli , gratton & kurucz ( 1997 ) with the zero point determined by the scale of blackwell & lynas - gray , ( 1994 ) ( hereafter blg94 ) .
the models have been computed without convective overshooting .
the microturbulent velocity has been fixed at 2 kms@xmath23 .
the distribution of heavy elements follows that of the sun . in order to find the temperature zero point offset between the model values of castelli et al .
( 1997 ) and the ones given by the infrared flux method of blg94 , the compilation of the observational data on non - variable stars by c95 is used .
we get @xmath24(model)@xmath25(blg94)@xmath26 . +
it is interesting to compare this temperature scale with the one implied by the surface brightness ( sb ) method of fouqu & gieren ( 1997 ) . rewriting their formula given for the apparent angular diameter , we get @xmath27 . by using the above mentioned compilation of c95
, we get an average difference of @xmath24(sb)@xmath25(model & blg94)@xmath28 . by performing the same test on the bw samples ( see table 1 ) , the differences are positive , but less than @xmath29 for cepheids . for rr lyrae stars , the differences are even smaller , with changing signs .
the larger negative difference on the sample of c95 is due in part to the larger weight of the higher temperature stars . in conclusion
, we may state that the present temperature scale in the temperature range of interest is very similar to the ones used in earlier bw studies of cepheids .
+ for rr lyrae stars , the iron abundance [ fe / h ] is computed from the fourier decompositions of the @xmath8 light curves as given by jurcsik & kovcs ( 1996 ) . for cepheids , we set [ fe / h]@xmath300 , which seems to be a reasonable first approximation ( see luck et al .
2003 for the metallicity distribution of cepheids ) .
the temporal gravity @xmath31 is computed by @xmath32 , where @xmath33 is the gravitational constant , @xmath34 is the stellar mass , which is fixed at @xmath35 and @xmath36 for rr lyrae stars and cepheids , respectively .
since the contribution of @xmath37 to @xmath24 is relatively small when the @xmath9 color index is used , the exact value of @xmath34 is not crucial . in principle
, we can compute a consistent value for @xmath34 by using the pulsation equation ( e.g. , van albada & baker 1973 ) .
however , both the weak dependence of @xmath24 on @xmath37 , and the difference between the equilibrium and static stellar parameters ( see bono , caputo & stellingwerf 1995 ) make this procedure less straightforward and therefore , is left for the task of future studies .
the dereddened color index @xmath38 is computed from the observed one by using @xmath39 reddening values either as given in the galactic cepheid database of the david dunlap observatory or by the corresponding formula of kw01 for rr lyrae stars .
because of the large reddenings of the galactic cepheids , we use the formulae for selective absorption ratios as given by the bw studies of cepheids ( see , e.g. , gieren , fouqu & gmez 1998 ) : * * @xmath40 , * * @xmath41 , * * @xmath42 , * * @xmath43 , @xmath44 , @xmath45 , + where the @xmath46 and @xmath47 colors are in the kron - cousins and standard johnson systems , respectively . for compatibility , these reddening relations are used also for rr lyrae stars . *
_ temporal absolute @xmath8 magnitude : _ * * @xmath48},\log g)$ ] , * * @xmath49 , + where @xmath50 and @xmath51 are the solar values . the latter one and the solar bolometric magnitude @xmath52
are set equal to @xmath53 k and @xmath54 mag , respectively .
the temporal bolometric correction @xmath55 is computed from stellar atmosphere models in the same way as @xmath24 above .
the zero point of @xmath55 is set to a value which yields @xmath56 . * _ point - by - point and average distance moduli : _ * * @xmath57 , * * @xmath58 , * * @xmath59 ^ 2 $ ] , + where @xmath60 is the number of data points on the light curve .
the best fit is obtained at the test radius @xmath14 which yields the smallest @xmath61 .
usually the fits are fairly accurate , with @xmath62@xmath63 .
it is also worth mentioning the following technical details : * when straightforward fourier fitting does not work , smooth light and velocity curves are obtained through fourier decompositions of the high order polynomials fitted by least squares to the observed data points of the folded light curves . *
tables of stellar atmosphere models are employed through quadratic interpolation . *
all pulsation phases are used , including also the possibly shock - disturbed quick expansion phase . * in spite of the careful treatment of the numerical derivation of the radial velocity curve , the total gravitational acceleration occasionally reaches values which lie outside the tabulated values . in these brief moments of pulsation the closest extreme values as given in the tables are used .
the same method is employed for filtering out the non - tabulated values of @xmath12 .
we note that these out - of - range errors occur almost always out of the minimum of @xmath61 , where the assumed unphysical stellar radius further enhances the numerical problems related to @xmath37 .
once the optimum equilibrium stellar parameters are computed by the above bw technique , we can evaluate the zero points of the corresponding plc relations . both for cepheids and for rr
lyrae stars the plc relations have been derived from stellar populations different from the galactic bw samples . in the case of rr
lyrae stars there is a reasonable large overlap between the metallicities of the present bw sample and the cluster variables used by kw01 to derive the empirical formula .
therefore , we may assume that the formula may also be valid for most of the field variables used in the bw analysis .
there is also a dispute on the possible metal dependence of the zero point of the cepheid plc relation .
however , we think that there are more theoretical and empirical results which indicate the non - existence of such a dependence ( alibert et al . 1999 ; bono et al . 2002 ; turner & burke 2002 ) . in the case of cepheids
we need to evaluate @xmath64 in the following equation @xmath65 where @xmath66 denotes the kron - cousins colors , @xmath67 means intensity - averaged magnitudes .
the above formula has been derived by u99 for lmc with @xmath68 . since the left - hand - side of eq .
( 1 ) is reddening - free , the difference of @xmath64(lmc ) and the average of this constant for the bw stars will yield the true distance modulus for the lmc .
the plc relation for the fundamental mode rr lyrae stars is given by kw01 @xmath69 where the bars denote simple magnitude averages .
the constant for lmc could be evaluated from the current @xmath8 , @xmath70 observations of fundamental mode rr lyrae stars in the lmc by clementini et al .
( 2003 , hereafter c03 ) . because they give intensity - averaged magnitudes , to retain compatibility with eq .
( 2 ) , we transform these values to simple magnitude averages with the aid of the formulae of kovcs ( 2002 ) .
the average of the @xmath71 fourier amplitude is set equal to @xmath72 , in accordance with the value obtained from the analysis of the rrab stars in the corresponding macho fields ( alcock et al .
2003 ) . in this way
we get @xmath73 and @xmath74 . with the observed average @xmath18 , @xmath75 and @xmath76 of @xmath77 , @xmath78 and @xmath79 , respectively ,
we get @xmath80 . however , the average magnitudes computed by c03 for their two fields @xmath81 & @xmath70 differ by as much as @xmath82 mag in @xmath8 , and @xmath83 mag in @xmath70 .
they attribute this difference to the effect of differential extinction .
we checked this difference by computing the averages in macho fields # 6 & # 13 , overlapping with fields @xmath81 & @xmath70 of c03 . with a @xmath84 clipping we got @xmath85 ( field # 6 , 317 stars ) and @xmath86 ( field # 13 , 262 stars ) . with these small standard deviations of the means
, we can state that a difference in the order of @xmath87 mag is not probable between the two fields . in order to get another estimate on @xmath88 for the lmc , we use the data of soszyski et al .
( 2002 ) , which cover a large sample of rr lyrae stars of the small magellanic cloud ( smc ) from the ogle database . by proceeding as above , from their average magnitudes and period of @xmath89 , @xmath90 , @xmath91 , @xmath92 and @xmath93 ,
we get @xmath94 and @xmath95 , depending on whether reddened or dereddened magnitudes are used .
although eqs .
( 1 ) and ( 2 ) should , in principle be reddening - free , the above difference shows that the field - by - field reddening corrections applied by soszyski et al .
( 2002 ) may not be completely consistent with our standard extinction ratio . since the method of their differential reddening correction
has already proved to be successful ( at least in statistical sense , see u99 ) , we adopt the value obtained by the use of their dereddened magnitudes . in computing the value of @xmath88 for the lmc , we have to subtract the relative distance modulus between the lmc and smc . from the plc relations derived for cepheids in the two clouds , u99 got a difference of @xmath96 mag , which , with the above preliminaries , yields our finally adopted value of @xmath97 for the lmc .
the necessary data on rr lyrae stars have been gathered from the literature during our earlier investigations ( see kw01 and also c95 , for a full list of variables subjected to bw analyses in earlier studies ) . for cepheids we turned to the database of the mcmaster university and also to the radial velocity database of the moscow university ( see gorynya et al . 1998 ) .
we utilized only the best data available on fundamental mode variables . when it was necessary , the cepheid @xmath46 light curves published in the johnson system were transformed to the kron - cousins system with the aid of the formula given by caldwell & coulson ( 1987 ) .
occasionally , periods have been improved by using the program package mufran ( see kollth 1990 ) .
the variables used in the present study together with their derived physical parameters are given in table 1 .
( effective temperatures have been computed from the cycle - averaged @xmath9 and @xmath37 values . )
three rr lyrae stars ( w crt , ss leo and av peg ) have been excluded from the above sample in the subsequent analysis , because of their discrepant positions in the plc relation . for similar reasons , we also excluded x lac from the cepheid sample .
lrrrrr + & & & & & + sw and & @xmath98 & 6702 & @xmath99 & @xmath100 & @xmath101 + wy ant & @xmath102 & 6296 & @xmath103 & @xmath104 & @xmath105 + x ari & @xmath106 & 6111 & @xmath107 & @xmath108 & @xmath109 + rr cet & @xmath110 & 6420 & @xmath111 & @xmath112 & @xmath113 + uu cet & @xmath114 & 6258 & @xmath115 & @xmath116 & @xmath117 + w crt & @xmath118 & 6788 & @xmath119 & @xmath120 & @xmath121 + dx del & @xmath122 & 6691 & @xmath123 & @xmath112 & @xmath124 + su dra & @xmath125 & 6293 & @xmath126 & @xmath127 & @xmath128 + sw dra & @xmath129 & 6393 & @xmath130 & @xmath131 & @xmath132 + rx eri & @xmath133 & 6383 & @xmath134 & @xmath135 & @xmath113 + rr gem & @xmath136 & 6721 & @xmath137 & @xmath138 & @xmath139 + tw her & @xmath140 & 6732 & @xmath141 & @xmath142 & @xmath143 + rr leo & @xmath144 & 6458 & @xmath145 & @xmath146 & @xmath147 + ss leo & @xmath148 & 6431 & @xmath149 & @xmath150 & @xmath151 + tt lyn & @xmath152 & 6283 & @xmath153 & @xmath154 & @xmath155 + v445 oph & @xmath136 & 6792 & @xmath156 & @xmath157 & @xmath132 + av peg & @xmath158 & 6836 & @xmath159 & @xmath160 & @xmath161 + ar per & @xmath162 & 6833 & @xmath163 & @xmath164 & @xmath117 + bb pup & @xmath165 & 6712 & @xmath166 & @xmath146 & @xmath167 + w tuc & @xmath168 & 6316 & @xmath169 & @xmath170 & @xmath171 + tu uma & @xmath172 & 6369 & @xmath173 & @xmath174 & @xmath175 + uu vir & @xmath176 & 6543 & @xmath177 & @xmath112 & @xmath178 + + & & & & & + @xmath179 aql & @xmath180 & 5675 & @xmath181 & @xmath182 & @xmath183 + tt aql & @xmath184 & 5589 & @xmath185 & @xmath186 & @xmath187 + vy car & @xmath188 & 5257 & @xmath189 & @xmath190 & @xmath191 + wz car & @xmath192 & 5208 & @xmath193 & @xmath194 & @xmath195 + v cen & @xmath196 & 5684 & @xmath197 & @xmath198 & @xmath199 + vw cen & @xmath200 & 4998 & @xmath201 & @xmath202 & @xmath203 + xx cen & @xmath204 & 5483 & @xmath205 & @xmath206 & @xmath207 + @xmath6 cep & @xmath208 & 5765 & @xmath209 & @xmath210 & @xmath211 + x cyg & @xmath212 & 5291 & @xmath213 & @xmath214 & @xmath215 + x lac & @xmath216 & 5815 & @xmath217 & @xmath218 & @xmath219 + cv mon & @xmath220 & 5689 & @xmath221 & @xmath222 & @xmath223 + t mon & @xmath224 & 5096 & @xmath225 & @xmath226 & @xmath227 + uu mus & @xmath228 & 5512 & @xmath229 & @xmath230 & @xmath231 + u nor & @xmath232 & 5696 & @xmath233 & @xmath234 & @xmath235 + bf oph & @xmath236 & 5903 & @xmath237 & @xmath238 & @xmath215 + bn pup & @xmath239 & 5475 & @xmath240 & @xmath241 & @xmath242 + ry sco & @xmath243 & 5312 & @xmath244 & @xmath214 & @xmath245 + bb sgr & @xmath246 & 5645 & @xmath247 & @xmath248 & @xmath249 + u sgr & @xmath250 & 5741 & @xmath251 & @xmath252 & @xmath253 + wz sgr & @xmath254 & 5064 & @xmath255 & @xmath256 & @xmath257 + ry vel & @xmath258 & 5363 & @xmath259 & @xmath260 & @xmath261 + rz vel & @xmath262 & 5210 &
@xmath263 & @xmath264 & @xmath265 + sw vel & @xmath266 & 5231 & @xmath267 & @xmath268 & @xmath242 + t vul & @xmath269 & 5986 & @xmath270 & @xmath271 & @xmath272 + u vul & @xmath273 & 5964 & @xmath274 & @xmath275 & @xmath276 + the corresponding plc relations are shown in figs . 1 and 2 . the constants in eqs .
( 1 ) and ( 2 ) have been derived from the bw absolute magnitudes and reddening - free colors .
the straight lines represent these calibrated relations .
it is seen that the empirical slopes of the plc relations fit perfectly the bw data ( please note the highly different scales of the two figures ) .
this strong correlation is very comforting , since two , completely separate methods and datasets are compared ( bw results , based on galactic variables on one hand , and empirical results , based on lmc and globular cluster variables on the other ) .
the standard deviations of the bw points around the straight lines are @xmath277 and @xmath278 for rr lyraes and cepheids , respectively .
we recall that the corresponding empirical values are @xmath279 and @xmath280 ( see kw01 and u99 ) .
the much smaller values of these empirical standard deviations show that the bw luminosities still suffer from large random errors as it follows from the delicate nature of the bw method .
computation of the lmc distance modulus is straightforward from the above results .
it is obtained simply by the subtraction of the corresponding constants in the bw and empirical relations . by using the results of sect . 3 , from the rr lyrae sample we get @xmath281 , whereas from the cepheid sample we obtain @xmath282 .
although these true distance moduli are magically close to each other , we have to keep in mind that they both have a statistical errors of @xmath283 mag .
nevertheless , it is clear that the often cited @xmath284@xmath285 mag difference between the bw distances derived from the two groups of variables no longer exists .
they both yield the same distance within a relatively small statistical uncertainty .
cephei stars listed in table 1
. the straight line corresponds to eq .
( 1 ) with the adjusted constant given in the upper right box . ]
it is interesting to compare the absolute magnitudes obtained in the present study with those derived earlier in other bw works . with the 18 cepheids common with the sample of gieren at al .
( 1998 ) , we get and average difference ( ours - theirs ) of @xmath286 mag , with a standard deviation @xmath287 of @xmath288 mag . a similar comparison with the absolute magnitudes given for @xmath289 in table 21 of c95 , yields @xmath290 with @xmath291 .
this considerable difference results in a proper shift in the rr lyrae distance modulus and leads to a result which is consistent with the one derived from cepheids .
the fact that gieren et al .
( 1998 ) derives a @xmath292 mag lower distance for the lmc , in spite of the same average luminosity , is suspected to be caused by the difference between the plc we use and the pl they employ to compute the lmc distance .
we find also a good agreement if we compare our radii with those of gieren et al .
for the 18 common variables we get an average difference ( ours - theirs ) of @xmath293 @xmath50 with @xmath294 @xmath50 . in terms of the relative radii
these figures become @xmath295% and @xmath296% , respectively .
we can also make a comparison with the results of long - baseline interferometric measurements for cepheids .
for the two variables @xmath6 cep and @xmath179 aql common with the samples of nordgren et al .
( 2002 ) and lane , greech - eakman & nordgren ( 2002 ) , we find a rather close agreement between the radii . by applying the necessary corrections due to the different @xmath7 factors , from the interferometric / radial velocity results we obtain the radii of @xmath297 and @xmath298 for @xmath6 cep and @xmath179 aql , respectively .
these values are very close to our corresponding figures ( see table 1 ) .
it is important to note that although all direct estimates based on double - mode variables yield lmc distance modulus in agreement with the present result ( kovcs 2002 ) , the indirect distances derived from globular cluster double - mode rr lyrae ( rrd ) stars become higher by @xmath299 mag , if we use the same plc for the lmc as in the present paper .
nevertheless , these revised distance moduli computed from the rrd populations of m15 , m68 and ic4499 are still in the range of @xmath300@xmath301 , which range is consistent with the value based on the present bw analysis .
we mention the recent effort of cacciari et al .
( 2000 ) in revisiting the bw analysis of rr lyrae stars with updated physics . from their analysis of two stars they concluded that the derived luminosities were in agreement with the earlier results ( meaning the continuance of the old discrepancy , however ,
see c03 for a different conclusion from the same work ) .
we have shown that by using exactly the same method and baade - wesselink parameters ( e.g. , velocity projection factor ) , fundamental mode rr lyrae and cepheid variables yield the same distance modulus for the large magellanic cloud .
this is in variance with the earlier common belief that the two groups of stars lead to significantly different distance estimates for the lmc .
the higher luminosity obtained for the rr lyrae stars in the present study is attributed to : ( a ) the more accurate consideration of phase - dependent bolometric correction , ( b ) the more precise computation of the temporal gravity , ( c ) the better numerical treatment of both the grids of stellar atmosphere models and the observed light and velocity curves .
the good agreement between the cepheid luminosities obtained in this and in the earlier works suggests that the above conditions have already been met in those studies .
although clarification of important theoretical problems ( such as the accurate conversion of the projected velocity to the photospheric value , or the application of dynamical atmosphere models rather than static ones ) are still needed , and better agreement between various photometric data on lmc is still demanded , we think that the present result is encouraging and indicates the possibility of making progress either by increasing the sample size by more precise data , or by building dynamical atmosphere models .
we are grateful to fiorella castelli for consulting on the stellar atmosphere models and to andrzej udalski for clarifying notation in the plc relation .
thanks are also due to lszl szabados for useful discussions on cepheids , to szabolcs barcza for careful reading of the manuscript and to the anonymous referee for the constructive comments . the support of otka grant t@xmath302038437 is acknowledged .
alcock , c. et al .
, the macho collab . , 2003 , submitted to apj alibert , y. baraffe , i. , hauschildt , p. & allard , f. , 1999 , a&a , 344 , 551 benedict , g. f. et al . , 2002a , aj , 123 , 473 benedict , g. f. et al . , 2002b , aj , 124 , 1695 blackwell d. e. , lynas - gray a. e. , 1994 , a&a , 282 , 899 ( blg94 ) bono , g. , caputo , f. & stellingwerf , r. f. , 1995 , apjs , 99 , 263 bono , g. , groenewegen , m. a. t. , marconi , m. & caputo , f. 2002 , apj , 574 , l33 cacciari , c. , clementini , g. , castelli , f. & melandri , f. , 2000 , asp conf . ser . , 203 , 176 caldwell , j. a. r. & coulson , i. m. , 1987 , aj , 93 , 1090 castelli f. , gratton r. g. & kurucz r. l. , 1997 , a&a , 318 , 841 ( see also http://kurucz.harvard.edu ) clementini , g. et al . , 1995 , aj , 110 , 2319 ( c95 ) clementini , g. et al . , 2003 , aj , 125 , 1309 ( c03 ) feast , m. w. & catchpole , r. m. , 1997 , mnras , 286 , l1 fernley , j. et al . , 1998 , a&a , 330 , 515 fouqu , p. & gieren , w. p. , 1997 ,
a&a , 320 , 799 gieren , w. p. , fouqu , p. & gmez , m. , 1997 , apj , 488 , 74 gieren , w. p. , fouqu , p. & gmez , m. , 1998 , apj , 496 , 17 gorynya , n.a . , et al . , 1998 , astronomy letters , vol .
24 , no . 11
( see also http://www.sai.msu.su/groups/cluster/gcvs/ ) .
groenewegen , m. a. t. & salaris , m. , 1999 , a&a , 348 , l33 jurcsik j. , kovcs , g. , 1996 , a&a , 312 , 111 kollth , z. , 1990 , occ .
notes konkoly obs . , 1 , http://www.konkoly.hu/staff/kollath/mufran.html kovcs , g. , 2002 , asp conf .
, 265 , 163 kovcs g. , walker a. r. , 2001 , a&a , 371 , 579 ( see also erratum : a&a , 374 , 264 ) ( kw01 ) lane , b. f. , greech - eakman , m. j. & nordgren , t. e. , 2002 , apj , 573 , 330 laney , c. d. & stobie , r. s. , 1994 , mnras , 266 , 441 liu , t. & janes k. a. , 1990 , apj , 354 , 273 luck , r. e. et al . , 2003 , a&a , 401 , 939 nordgren , t. e. , lane , b. f. , hindsley , r. b. & kervella , p. , 2002 ,
aj , 123 , 3380 popowski , p. & gould , a. , 1998 , apj , 506 , 259 soszyski i. et al . , 2002 , acta astron .
52 , 369 stanek , k. z. , kaluzny , j. , wysocka , a. & thompson , i. , 2000 , acta astr .
, 50 , 191 turner , d. g. & burke , j. f. , 2002 , aj , 124 , 2931 udalski , a. et al . , 1999 ,
acta astron .
, 49 , 201 ( see also erratum at http://www.astrouw.edu.pl/@xmath303ogle ) ( u99 ) van albada , t. s. & baker , n. , 1973 , apj , 185 , 477 | by using the same algorithm in the baade - wesselink analyses of galactic rr lyrae and cepheid variables , it is shown that , within @xmath0 mag @xmath1 statistical error , they yield the same distance modulus for the large magellanic cloud . by fixing the zero point of the color temperature calibration to those of the current infrared flux methods and using updated period luminosity color relations , we get an average value of @xmath2 for the true distance modulus of the lmc .
[ firstpage ] galaxies : magellanic clouds stars : distances stars : horizontal branch stars : variables : cepheids |
precision measurements in neutron and nuclear decay offer a sensitive window to search for new physics beyond the standard electroweak model and allow also the determination of the fundamental weak vector coupling .
recent analyses based on the effective field theory performed in e.g. @xcite show that in processes involving the lightest quarks the neutron and nuclear decay will compete with experiments at highest energy accelerators .
for instance , data taken at the lhc is currently probing these interactions at the @xmath0 level ( relative to the standard weak interactions ) , with the potential to reach the @xmath1 level . in some of the @xmath2 decay correlation measurements there are prospects to reach experimental sensitivities between @xmath3 and @xmath4 making these observables interesting probes for searches of new physics originating at tev scale .
the most direct access to the exotic tensor interaction in @xmath2 decay is to measure the fierz term ( coefficient @xmath5 ) or the beta - neutrino correlation coefficient @xmath6 in a pure gamow - teller transition @xcite .
the @xmath5 coefficient shows up as a tiny energy dependent @xmath7 departure of the @xmath2 spectrum from its v - a ( standard model ) shape .
the smallness of the potential @xmath5 contribution requires that other corrections to the spectrum shape of the same order are included in the analysis . indeed , according to @xcite
the recoil terms also affect the spectrum shape with their main contribution being proportional to @xmath8 . in order to disentangle these effects the detector efficiency for @xmath2 particle as a function of energy
must be known with the precision better than @xmath3 @xcite .
the dominating contribution in the systematic uncertainty comes from back - scattering and out - scattering of electrons from the detector .
monte carlo simulation of this effect is helpful , however , it introduces its own uncertainty as the input parameters are known with limited accuracy .
monte carlo simulation would reflect the real situation better after it is adjusted to real experimental data of a particular measurement setup .
the described in this article electronics was designed for a spectrometer capable of direct registration of the back - scattering events , thus providing reference data for the monte carlo calculation of the detector efficiency .
the spectrometer itself is still in an r&d phase undergoing detailed tests and tuning .
it will be a subject of a separate paper together with the performance benchmark @xcite . in this paper
, its concept will be described only in a minimum extent at the beginning of section 2 to explain the requirements imposed onto the front - end electronics and daq .
the rest of section 2 is devoted to the electronic system architecture .
the test results were obtained with a help of a signal generator and are presented in section 3 . therein the resulting time spectrum and the charge asymmetry distribution representing the typical performance
are shown .
the asymmetry spectra were obtained using a dedicated tester to simulate the hit position on the wire with adjustable resistance division ( potentiometer ) .
conducting the electronic benchmark tests without a detector was chosen by purpose .
the detector itself is still not fully understood .
therefore it was important to assess the purely electronic contribution to the performance parameters of the spectrometer .
the paper ends with a short summary and outlook for the future experiment .
in order to facilitate the identification of the electrons impinging onto and scattered from the energy detector ( e.g. si detector , scintillator ) a low - z and low - mass tracker must be applied .
one of the attractive options is a low pressure multi - wire drift chamber with minimum number of necessary wires in order to maximize the detector transparency .
this condition can be fulfilled by a hexagonal wire geometry and the charge division technique allowing for a 3d track reconstruction without major distortion of the electron energy measurement .
the hexagonal wire geometry is not the only one considered in the project .
the rectangular ( planar ) wire configuration is the next suitable alternative .
the multi - wire drift chamber is based on the small prototype described in ref .
@xcite and will be operated with he / isobutan gas mixture ( ranging from 70%/30% to 90%/10% ) at lowered pressure ( down to 300 mbar ) .
it consists of 10 sense wire planes ( 8 wires for each plane ) separated with 24 field wire planes forming the double f - f - s - f - f - s - f - f - s - f - f - s - f - f - s - f - f structure ( f- denotes a field wire plane , s- denotes a signal wire plane ) .
the distance between neighboring signal planes is 15 mm , with wires within a plane being separated by 17.32 mm .
this wire plane structure leads to the hexagonal cell geometry as shown in fig .
[ cells ] .
each cell consists of a very thin anode wire ( nicr alloy , 25 @xmath9 m diameter ) with resistance of about 20 ohms / cm surrounded by 6 cathode field wires forming a hexagonal cuboid .
all wires are soldered to pads of the printed circuit board ( pcb ) frames .
the chamber is equipped with a two - dimensional positioning system for a beta source installed in the central region of the detector , between both parts of the mwdc structure . in the initial configuration , the electrons detected in two plastic scintillators installed at both sides of the mwdc
provide the time reference signal for the drift time measurement .
the pmt signals are also used as a trigger for the mwdc and electron energy detector readout .
acquiring the drift time and the pulse height asymmetry at both ends of the responding signal wire one can establish the electron path across the chamber cell .
the system provides charge - division position sensing in the direction parallel to the wires as well as precise drift time measurement .
the expected position resolution across wires is limited by angular straggling of primary electrons travelling in the gas and accounts to about 450 @xmath9 m as shown in ref
. @xcite .
in this situation , the precision of the drift time measurement of about 200 ps is more than needed as it corresponds to about 50 @xmath9 m for the operating conditions in view .
the mwdc will be operated in homogenous magnetic field oriented parallel to wires providing a rough electron transverse momentum filter .
the position information along the wires will be used for the identification of the sequence of the cells passed by the electrons and for distinguishing between the electrons impinging onto and scattered from the energy detector .
this is why the modest position resolution of a few mm is sufficient for that direction .
crucial in the design is the analog part of the system . in principle ,
commercial digitizers ( tdc , adc ) could be applied at the end of the acquisition chain .
however , it has been decided to equip the data channels with custom digital boards with adjusted specifications such that the whole system is handy , cost effective and scalable .
the applied on board processing ( fpga ) allows for acquiring up to 15 000 triggered events per second which gives a comfortable factor of 10 reserve as compared to the application in view .
the described modular electronic system consists of three main parts : ( i ) preamplifiers , ( ii ) analog cards containing the peak detector and constant fraction discriminator ( cfd ) , and ( iii ) the digital boards containing analog to digital converters , adc and tdc . the data logging software on the pc constitutes the data receiver .
the signal from each signal plane is processed by means of one electronics module , which consists of two analog cards and one digital board providing 16 adc and 8 tdc channels . a corresponding block diagram is shown in fig . [ daq ] .
the signals from both ends of a signal wire are fed to inputs of preamplifiers located directly on the detector frames ( see fig . [ pcbpre ] ) in order to minimize the input noise and protect the signal from em interferences .
the signals from the preamplifiers are received by the analog cards which drive the adc inputs and produce the tdc stop signals .
the digital data is transmitted via a lan port to the back - end computer where the process of receiving , sorting and formatting to a complete physical event is accomplished by the data logging software .
the card configuration settings and control is done via a rs485 port .
detailed description of each part of the system is presented in the following sections .
signal readout from both ends of the relatively low resistance wire requires application of dedicated preamplifiers .
low resistance wires force the use of fast preamplifiers with low input impedance since the change of signal amplitudes and thus position resolution depends on the input impedance .
the lower the input impedance , the higher the voltage difference that is registered at the wire ends .
the preamplifiers work in current - mode with the input impedance below 5 ohms and the input stage bandwidth exceeding 300 mhz .
the preamplifier circuit is shown in fig .
[ preamp ] .
the first stage of the preamplifier is a current - voltage converter with two fast bipolar transistors .
the voltage pulse is fed to the high - pass filter which cuts off the dc bias and the slow varying components of the signal .
the first filter is followed by a low - pass filter integrating the pulse with 100 ns time constant .
the signal is then amplified and transmitted to the differential amplifier used to drive the transmission line .
differential signal transmission increases the external noise immunity when unshielded twisted pair ribbon cables are used to connect the preamplifier outputs to the inputs of the analog module .
an example of input and output signals from a preamplifier is shown in fig .
[ inppreamp ] .
the input current signal was reproduced with the help of qucs package @xcite .
using bipolar transistors increases the preamplifier immunity to possible discharges in the gas chamber . adding more robust input protection ( e.g. a series resistor and protection diodes ) was abandoned as it would increase the input noise and impedance thus worsening the charge division resolution
however , accidental discharges can not be avoided completely and the need for replacement of damaged preamplifiers was taken into account in the design .
the individual preamplifiers are arranged as small piggyback pcb cards mounted directly on the wire frames ( fig .
[ pcbpre ] ) .
they can be replaced easily without unplugging cables or any other major intervention in the setup .
additionally , such a solution minimizes the length of the unshielded connection between the wire end and the preamplifier input .
the block diagram of the analog circuit is presented in fig .
[ analog ] .
differential signals from the preamplifiers are processed by the analog circuit .
the signals from both wire ends are split into two branches . in the first branch ,
signals from both ends are added and fed to the cfd which produces a ttl pulse for the time - to - digital converter ( tdc ) . a delay correction
prior to the summing is not needed unless it is of the order of 1 ns ( corresponding to about 20 cm cable length ) significantly less than the rise time of real signals from the mwdc .
care is taken to assure the equal length of the connecting cables .
if the sum of the analog signals is higher than the set value of the cfd threshold , the stop signal for the tdc is generated .
the schematic of the cfd is presented in fig .
[ cfd ] . in the second branch ,
both signals are transmitted to the fast peak - hold detectors , which are responsible for detection of the pulse amplitudes in a given gate time .
peak - hold detectors are used to stretch the pulse to the length acceptable for the adc converters .
the schematic of the peak - hold detector is shown in fig .
[ phd ] .
the gate and hold signals are produced by a programmable timing circuit located on the analog board which uses the tdc start signal ( external trigger ) as a time reference .
the analog boards hosting the analog signal processing circuits are equipped with built - in controllers for setting the thresholds of the cfd as well as the gate and the hold timing .
the set values are fed to the controller over a slow - control bus ( rs485 ) .
the block diagram of the digital circuit is presented in fig .
[ digitalb ] .
the digital board consists of 16 channels of 14-bit adcs ( ad8099 chips made by analog devices ) and a 8 channel tdc ( tdc - gpx chip made by acam messelectronic gmbh ) . using 14-bit adcs was dictated by the large dynamic range of the amplitudes of the signals from a gas detector operated in the proportional mode .
the high bit adc resolution increases the overall performance of the charge division method .
the time resolution of the used tdc chip is significantly better than actually needed ( 83 ps binning resolution with the typical standard deviation of 77 ps , from the device datasheet ) .
the board measures the delay between the common start ( trigger ) pulse and the individual stop signals received from the constant fraction discriminators , as well as the amplitudes of the pulses delivered by the peak - hold detectors .
the stop signals are delayed by a 50 ns delay line in order to compensate for the delays introduced in the analog circuit generating the start signal .
the adcs sample the signals with 20 msps sampling rate .
the fpga chip ( xilinx spartan xc3s400 ) receives the data from the adcs and the tdc , buffers them , provides zero - suppression and time - stamping , and formats the data frame as described in the next paragraph .
the flow diagram of the acquisition algorithm and the data transmission algorithm are shown in fig . [ daq_alg ] .
the data frame is transmitted to the integrated ethernet / udp stack ( wiznet w5100 ) which is responsible for communication with the back - end pc hosting the acquisition and control software .
the digital board is also equipped with a built - in controller providing initialization and configuration of adc , tdc and ethernet chips .
this controller is accessible via the slow - control bus .
the data are transferred to the computer in a form of udp frames .
the data formatter creates a frame for each recorded event .
the frame length is a multiple of 96 bytes .
the first part of the frame is a fixed header used for synchronization .
next follows the time tag ( 32 bit unsigned integer value ) which allows the time synchronization between multiple modules .
the third part of the frame consists of 16 adc conversion values ( 16 bit unsigned integer values ) .
the fourth part contains the tdc values - up to eight 24-bit unsigned integer values .
only the non - zero values of the tdc data are transmitted .
the time tagging is used because the ethernet protocol does not ensure the sequence of the data packet delivery .
the time - tags are shared between modules and used by the data - logging software to restore the correct data order .
[ formatter ] shows the structure of the data frame .
the slow - control interface allows settings the cfd thresholds and timing configuration parameters of the module .
each board in the module is equipped with its own controller connected to the slow - control bus .
individual addresses of the controllers are set with dip - switches installed on the boards .
the controllers use a simple ascii protocol via rs485 interface . in order to connect the bus to the acquisition computer
a commercial rs485-usb converter is used .
an example of the initialization data record necessary for establishing the communication over the slow - control bus is presented on fig .
the slow - control parameter set is read back every second and refreshed on a computer display ( fig .
[ scview ] ) .
, @xmath10 ) and digital board timing ( @xmath11,@xmath12 ) , where @xmath13 is the gate width and @xmath10 is the pulse hold time . for the digital board
@xmath11 denotes the maximum time for the incoming signal and @xmath12 is the delay for the adc .
the second column presents the commands used for saving values to the controllers .
the save commands use the same format but the uppercase option letters . ] the data logging software provides the parallel reception of data frames sent by the modules and the event building i.e. collection of data generated by the system upon a single hardware trigger signal and uniquely identified by a time tag .
each module is identified by its ip address and uses a specific udp port corresponding to this address .
the architecture of the logging software is shown in fig .
[ dls ] .
the software is multi - threaded and consists of the main thread ( the parent ) that is responsible for sorting and recording events , and the child processes whose task is to receive data from the acquisition modules and to send them via data streams to the main thread . upon initialization
the program reads the configuration file with the addresses of the modules and opens sockets for all modules to communicate with them .
the sockets listen to the different udp ports correlated with the module address .
the main loop creates a set of pipes ( data streams ) and receives the data structures from the child processes .
the size of the frame is checked and followed by the pre - selection of cases .
if the tdc value is greater than zero , the event is pushed to the fifo queue of the child process .
the event is then read from the fifo by the parent process in which the data structures from all child processes are sorted ( using the quick - sort algorithm ) by the time tags and formatted as one complete event .
this event will be identified with a particle crossing the chamber planes and directed for physics analysis .
the received events are accessible for on - line histogramming analysis by means of a shared memory mechanism and simultaneously saved on a hard disk for later off - line analysis . the cases with empty tdc conversion value ( stop signal below cfd threshold )
are used to calculate the baseline offset for the adc . in this method
it is assumed that the lack of the tdc signal in a pair of channels ( serving a particular sense wire ) means that the cell did not fire so the corresponding adcs deliver the offset values .
this in - flight offset calculation method can be disabled and the fixed offsets can be introduced with the help of the slow - control interface .
the baseline offsets subtracted ( in flight ) from the conversion value are appended to the data structures for each event so that the subtraction is reversible . finally , the data structure consists of a time tag , module number , channel number , adc1 value , adc2 value , tdc value , adc1 offset and adc2 offset , respectively .
it is worth mentioning that the amplitude thresholds which are needed for the histogram building were offset from the adc baseline values such that only the electronic noise was eliminated .
the ultimate adjustment of the thresholds was postponed to the detector efficiency tuning phase .
it will depend on the gas mixture composition and pressure as well as on the operation voltage .
the described daq system was tested using a special tester simulating the wire chamber signals .
the wire itself is replaced by a potentiometer with the range reflecting the real wire resistance ( see fig . [ tester ] ) .
the tester utilizes two preamplifiers and signal cables connected to the selected module .
a 400 mv high and 200 ns long simulator input is synchronized with a ttl signal triggering the system .
the simulator input was fed by a triangle signal with 10 ns rise time and 200 ns fall time adjusted to the expected detector pulse shape obtained from the garfield ( ref .
@xcite ) simulation .
the hit position corresponds to a unique potentiometer setting .
the corresponding adc asymmetry distributions are shown in fig .
[ adcall ] .
the zero point defines the middle of the wire .
the difference between the individual asymmetry spectra reflects the varying relation between the signal and noise amplitudes . in fig .
[ wzorzec ] sample results from one channel are presented .
[ funadca ] shows a number of centroids the adc pulse height asymmetry distributions @xmath14 acquired at different potentiometer asymmetry settings @xmath15 where @xmath16 and @xmath17 are the adc1 , adc2 amplitudes and @xmath18 , @xmath19 are the resistances for selected potentiometer settings .
the results were obtained for varying resistance division corresponding to the charge collected at both wire ends .
the obtained adc asymmetry is a monotonic function of the resistance asymmetry .
the centroids of these distributions are drawn with 1@xmath20 error bars in fig .
[ funadca ] .
the polynomial function fitted to the centroids represents the position calibration . drawing the error band allows the extraction of the position resolution as shown in fig .
[ funadcb ] .
the uncertainty of the relative position resolution is connected with the interpolation procedure . ) as a function of the resistance asymmetry @xmath21 ( eqn .
2 ) . the insert is a zoomed part of the graph showing the error bars equal to @xmath22 of the peak distributions plotted in fig .
[ adcall ] .
dotted lines interpolate the error bar ends . ] expressed in units of length of the wire deduced from fig .
[ funadca ] .
a 500 ohm wire resistance was assumed which corresponds to a 25 @xmath9 m nicr wire of @xmath23= 240 mm length .
the uncertainty is connected with the interpolation procedure ( see text ) . ]
it shows that the charge division method determines the position with a resolution between 1.2 and 3.6 mm for a 240 mm long wire . when feeding a given preamplifier pair with the simulator signals the noise distributions on the neighboring channels
were acquired as well .
neither a baseline change nor an increase of noise were observed , meaning that the electronic crosstalk is negligible .
this does not assure there is no crosstalk when the electronic system is attached to the gas detector .
inductive crosstalk between the wires depends on the configuration and on the operating conditions and is beyond the scope of this paper . in the second test
, the response of the tdc measurement was investigated as a function of the input signal asymmetry .
[ tdcconst ] shows a typical example .
the impact of the charge division on the tdc measurement is found to be less than 1 ns in the entire charge asymmetry range .
for the application in view , a 1 ns drift time corresponds to about 250 @xmath9 m distance ( from garfield simulations ) .
the maximum throughput of the system reaches 15 khz per channel ( 120 khz per module ) and is limited by the transmission time of 60 @xmath9s needed for a single event .
the front - end electronics and daq system described here was designed to be used with a special multi - wire drift chamber for tracking low energy electrons from nuclear @xmath2 decays .
it incorporates both the drift time and charge division measurements allowing for efficient 3d determination of the electron tracks with very few and only parallel sense wires .
the performed tests show that the electronic contribution to the drift time measurement uncertainty is less than 1 ns corresponding to about 250 @xmath9 m position uncertainty at the expected electron drift velocities .
such result is satisfactory since in the planned experiment the track position resolution will be dominated by the electron angular straggling effects in the gas as shown in monte carlo simulations and confirmed in the small prototype test described in ref .
. the necessary spatial resolution of the electron track position determined from the drift time need not to be better than 500 @xmath9 m .
the uncertainty of the track position obtained from the charge division measurement varies from 0.5% at wire ends to 1.5% in the center corresponding to 1.2 and 3.6 mm , respectively , for 24 cm long wires ( 25 @xmath9 m nicr ) .
this result is sufficient for the application in view : identification of the cell sequence passed by an electron spiraling in an axial magnetic field and resolving possible double track ambiguities appearing in the 2d projection .
00 t. bhattacharya , v. cirigliano , s. d. cohen , a. filipuzzi , m. gonzalez - alonso et al .
, phys . rev .
d 85 , ( 2012 ) 054512 .
o. naviliat - cuncic , m. gonzalez - alonso , ann .
( berlin ) 525 , no .
89 , ( 2013 ) 600619 .
j. d. jackson , s. b. treiman , h. w. wyld , phys .
106 , ( 1957 ) 517
. b. r. holstein , rev .
46 , ( 1974 ) 789 .
n. severijns , j. phys .
g. , nucl .
41 , ( 2014 ) 114006 .
f. wauters , et al .
c 82 ( 2010 ) 055502 .
m. perkowski et al . , in preparation ; g. soti , _ the minibeta spectrometer for the determination of weak magnetism and the fierz interference term _ , psi2013 , switzerland ; p. finlay , _ minibeta : a multiwire drift chamber for beta - spectrum shape measurements _ , aris2014 , japan .
k. lojek , k. bodek , m. kuzniak , nucl .
instr . and meth . a 611 ( 2009 ) 284 .
qucs - circuit simulator ( http://qucs.sourceforge.net/ ) .
r. veenhof , garfield , cern program library ( http://garfield.web.cern.ch/garfield ) . | this paper presents the design and implementation of the front - end electronics and the data acquisition ( daq ) system for readout of multi - wire drift chambers ( mwdc ) .
apart of the conventional drift time measurement the system delivers the hit position along the wire utilizing the charge division technique .
the system consists of preamplifiers , and analog and digital boards sending data to a back - end computer via an ethernet interface .
the data logging software formats the received data and enables an easy access to the data analysis software .
the use of specially designed preamplifiers and peak detectors allows the charge - division readout of the low resistance signal wire .
the implication of the charge - division circuitry onto the drift time measurement was studied and the overall performance of the electronic system was evaluated in dedicated off - line tests .
data acquisition system , hardware , data logging software 29.85.ca ; 29.40.gx ; 23.40.bw |
it has been widely accepted that astrophysical observations might favor a dark energy component with `` supernegative '' equation of state ( eos ) parameter @xmath4 , dubbed as phantom or super quintessence , where @xmath5 is the pressure and @xmath6 is the energy density @xcite .
the phantom model with born - infeld type lagrangian has been proposed @xcite and its generalization to @xmath7 symmetry has been done in ref.@xcite .
the phantom component leads to a catastropic big rip singularity that is characterized by divergences in the scale factor @xmath8 , the hubble parameter @xmath9 @xcite and its time - derivative @xmath10 @xcite at the finite future @xmath11 .
it has been studied that various types of singularity might occur in phantom scenarios @xcite .
barrow and his collaborators @xcite have investigated general characterization of sudden cosmological singularities and the classical stability of cosmological solutions containing these singularities .
some mechanisms have been proposed to avoid these catastropic singularities .
for example , it has been shown that these singularities in the future of the cosmic evolution might be avoided if we take suitable potential term @xcite , quantum effects @xcite and modified gravity @xcite .
if one considers an additional interaction between the phantom field and the background , the big rip singularity can be avoided in the late - time evolution of the universe @xcite . in brane - world models ,
a gauss - bonnet term is provided for the bulk , whereas phantom field is present on the brane .
if the dark energy is modelled by a phantom - generalized chaplygin gas , the evolution of universe will avoid the big rip @xcite . in the torsion cosmology
, the scalar mode of torsion could be considered as a phantom field which will derive the universe in an oscillating fashion with an accelerating expansion at late - time @xcite .
however , @xmath1 as the scale factor @xmath12 extends into the future is necessary condition @xcite , but not sufficient one for a future singularity .
recently , the little rip scenario has been proposed in @xcite , in which @xmath1 but @xmath13 asymptotically , such that there is no future singularity .
such model will nevertheless lead to a dissolution of bound structures at some time in the future . especially , the viscous fluid can produce a little rip cosmology as a purely viscosity effects @xcite . on the other hand , there is an interesting way to give rise to acceleration : a modification to the einstein - hilbert action involving new terms of inverse powers of the curvature scale ( in the following text we call it inverse-_r _ gravity ) , of the form @xmath14 ( with @xmath15 ) @xcite . in general @xmath16 gravity ( even without scalar field ) , the divide @xmath2 crossing behavior is shown in refs .
recently , du et al .
@xcite discussed the behavior of quintessence field in the inverse-_r _ gravity and found that the evolution of @xmath17 with cosmic time can cross the divide @xmath2 and @xmath17 will be less than @xmath18 . in this paper , we put forward a new mechanism to avoid the catastropic big or little rip by means of the investigation of phantom field in the inverse-_r _ gravity .
it is worth noting that crossing @xmath2 divide happens approximately in a few e - fold expansion when the inverse-_r _ term is still sufficiently large and the decrease of the energy density resulting from the crossing of @xmath17 will make the catastrophic fate avoided .
this paper is organized as follows . in sec .
2 , we give a brief review of inverse-_r _ gravity and its relation to the phantom field in einstein s general relativity ( gr ) .
an exact solution is shown in sec .
3 . in sec . 4
, we investigate the mechanism for a single phantom field to avoid big or little rip by an analytical method . in sec .
5 , we present the numerical results to avoid big or little rip of phantom depending on different potentials in frame of inverse-_r _ gravity . finally , we give a brief summary and a further discussion in sec .
there are two types of modification of einstein - hilbert action : one is ultraviolet modification which is important in the early universe ( high curvature region ) and the other is infrared modification which plays significant role in the late universe ( low curvature region ) .
the inverse-_r _ gravity is one of the current models with infrared modifications .
the inverse-_r _ gravity seems unable to pass the solar system observations @xcite , but this problem has been solved by adding a scalar curvature squared term @xmath3 to the inverse-_r _ gravity @xcite . in ref .
@xcite , the following @xmath16 theory was proposed : @xmath19 where @xmath20 and @xmath21 are constants .
this model leads to an acceptable cosmic speed - up and is consistent with some of the solar system observations .
this general case returns to the inverse-_r _ gravity when the curvature is low and @xmath22 . in this paper
, we consider the following action @xmath23 in the frw universe with scale factor @xmath8 .
the metric is @xmath24 where @xmath25 is the metric of a 3-dimensional maximally symmetric space . @xmath26
is the lagrangian of matter and @xmath27 is the lagrangian of phantom field with the potential @xmath28 , @xmath29 the friedmann equation is given by @xcite @xmath30 where @xmath31 is the hubble parameter and the dot denotes the derivative with respect to @xmath32 , and @xmath33 the equation of motion for phantom field is @xmath34 in this system the phantom field together with the extra geometric term takes effect of dark energy . in the inverse-_r _ gravity , there is an additional @xmath35-term in the modified friedmann equation ( 5 ) . to expound the observed evolving eos of the effective dark energy
, one can introduce the concept `` equivalent dark energy '' @xcite .
we derive the density of equivalent dark energy by comparing ( 5 ) with the standard friedmann equation in gr .
the latter can be written as @xmath36 comparing ( 8) with ( 5 ) , one has @xmath37 where @xmath38 .
it is worth noting that @xmath35-term makes a key difference from the standard model of gr . in the next section
, we shall study analytically the phantom dynamics of universe in inverse-_r _ gravity .
to find exact solutions is an important but difficult topic in such a highly non - linear system ( 7 ) and ( 10 ) .
the most interesting exact solution is a power - law one because it has the extra advantage that the equations for the generation of density perturbations can also be solved exactly in the cosmological inflation @xcite . in a spatially flat universe , the friedmann equation ( 5 )
reduces to @xmath39 during the phantom dominated stage . note that crossing @xmath2 divide happens approximately in a few e - fold expansion when the inverse-_r _ term is still sufficiently large and the @xmath40-term is still sufficiently small , therefore , the @xmath21-term can be neglected in rhs of ( 6 ) at the late - time of evolution .
we have @xmath41 in this case , the power - law solution arises when the potential is chosen to take the following form @xmath42},\ ] ] where @xmath43 , @xmath44 and @xmath45 are constants , and @xmath46 the equations of motion ( 7 ) and ( 10 ) then have the particular solution @xmath47 @xmath48 where @xmath49 provided that @xmath50 when @xmath43 , this solution satisfies the condition for accelerating expansion . if we take another value of @xmath44 in the potential ( 12 ) @xmath51 the system ( 7 ) and ( 10 ) have the particular solution ( 14 ) and ( 15 ) , but @xmath52 provided that @xmath53 when @xmath54 , this solution also satisfies the accelerating condition .
it is not surprising that there is a power - law solution for the dynamical system of phantom because @xmath35-term makes a key difference from the standard model of gr .
the power - law solution makes a message known to us : the big rip can be avoided in the inverse-_r _ gravity .
before studying the avoiding mechanism in this dynamical system , we discuss briefly the relation between density evolution and eos .
since the dust matter obeys the continuity equation and the bianchi identity keeps valid , equivalent dark energy itself also satisfies the continuity equation @xmath55 where @xmath56 denotes the effective pressure of the equivalent dark energy . from ( 9 ) and ( 19 ) , we can express eos as @xmath57 where @xmath58\ ] ] and @xmath59 depend on the rate of expansion .
if @xmath60 ( @xmath61 ) the equivalent dark energy behaves as quintessence ( phantom ) in gr , i.e. , @xmath62 decreases ( increases ) with expansion of universe .
next we consider the approach of ref.@xcite where the pressure is expressed as a function of the density in the form @xmath63 . taking a power law for @xmath64 , i. e. , @xmath65 ( @xmath66 and @xmath67),we
see that the fate of universe will depend on @xmath44 : ( i ) there is the possibility of little rip but no finite future singularity for @xmath68 ; ( ii ) there is a type i ( big rip ) singularity for @xmath69 ; ( iii ) there exists a type iii singularity for @xmath70 . here
, the singularities are classified according to ref . [ 7 ] .
we have @xmath71^{\frac{1}{1 - 2\gamma}},&\textrm{$\gamma>\frac{1}{2}$ and $ \gamma\neq1 $ } \end{array}\right.\end{aligned}\ ] ] where @xmath72 is taken for the sake of clarity . substituting ( 22 ) into ( 21 ) , we obtain @xmath73 ^ 4},&\textrm{$\gamma=\frac{1}{2}$},\\ -\frac{3m(t)n(t)}{2l(t)},&\textrm{$\gamma>\frac{1}{2}$ and $ \gamma\neq1 $ } \end{array}\right.\end{aligned}\ ] ] where @xmath74^{\frac{1}{1 - 2\gamma}})^4,\\\nonumber m(t)&=&[a(2\beta-1)(t_s - t)]^{\frac{1}{2\gamma-1}},\\\nonumber n(t)&=&2\sqrt{3}(1 + 2\beta)\beta+(2\beta-1)(t_s - t)[a(2\beta-1)(t_s - t)]^{\frac{1}{1 - 2\gamma}}(46\beta-16\beta^2 - 3)\\\nonumber & + & 4\sqrt{3}(2\beta-1)^2(t_s - t)^2[a(2\beta-1)(t_s - t)]^{\frac{2}{1 - 2\gamma}}\\\nonumber & + & 4(2\beta-1)^3(t_s - t)^3[a(2\beta-1)(t_s - t)]^{\frac{3}{1 - 2\gamma}}.\end{aligned}\ ] ] it is easy to prove @xmath75 ( @xmath76 is a negative definite function ) for @xmath77 , @xmath78 , which brings a possibility to avoid the big and little rip , and some type iii singularity for @xmath77 . in the inverse-_r _ gravity , the @xmath35-term makes a key difference from the standard model of gr .
in the @xmath79 case , @xmath80 will monotonically increase since @xmath76 is a negative difinite function for @xmath77 . in the @xmath81 case
, we need an analysis in detail . combining ( 7 ) and ( 10 ) ,
we find that @xmath82 @xmath83 if @xmath81 from the current observation , we must demand that eos parameter cross @xmath2 divide for the avoidance of big or little rip in the future .
the universal argument is that the crossing phenomenon happens at @xmath84 _ iff _ @xmath85 and @xmath86 . in the phantom case of gr
, we have @xmath87 such that @xmath85 implies @xmath88 .
therefore , the crossing behavior never happens . in other words ,
the big or little rip is inescapable in the einstein s gr . in inverse-_r _
gravity , @xmath85 and @xmath86 _ iff _ @xmath89 and @xmath90 at @xmath84 .
thus , we have proved the following theorems .
+ theorem .
if @xmath81 , the avoidance of big or little rip demands @xmath89 and @xmath90 at @xmath84 in the inverse-_r _ gravity where @xmath91 is the crossing time . +
for a minimally coupled phantom field @xmath0 in gr , the necessary conditions have been given for the refrainment of big rip or little rip@xcite .
that is , when @xmath92 and @xmath93 , @xmath17 approaches -1 sufficiently rapidly that big rip is refrained .
furthermore , the boundedness of the potential @xmath28 indicates whether or not the model corresponds to a little rip expansion .
thus , we examine a specific model with @xmath94 as @xmath95 in more detail . in sec .
4 , we have investigated analytically the evolution of universe in the phantom dominated epoch in inverse-_r _ gravity . for a realistic universe ,
the pressureless matter is an essential component .
however , it is difficult to study analytically the evolution of universe when we consider matter component in inverse-_r _ gravity with @xmath40 correction .
the behaviors of phantom field with different potentials are investigated by numerical calculation in this section .
the results show that the big or little rip can be avoided for these models .
a spatially flat frw universe has been assumed in all calculations .
the form of exponential potential is @xmath96 where @xmath97 is a parameter with dimension of energy density .
let us introduce the four dimensionless parameters for the numerical calculations , @xmath98 and @xmath99 .
thus , ( 7 ) and ( 10 ) can be rewritten as @xmath100 @xmath101 where @xmath102 @xmath103 @xmath104 @xmath105 here the prime denotes the derivative with respect to the so called e - folds @xmath106 and @xmath107 , @xmath108 is the current hubble parameter and @xmath109 is the current density of pressureless matter . generally speaking , @xmath110 is a little tiny number since the inflation scale @xmath21 is much higher than @xmath108 .
if we fix the inflation scale as a gut energy scale of @xmath111 gev , @xmath112 . the deceleration parameter can be expressed as @xmath113 and we introduce the dimensionless density of equivalent dark energy @xmath114 for convenience @xmath115 using numerical calculation , we plot @xmath116 and @xmath117 as functions of e - folds @xmath118 for the scalar phantom in inverse-_r _ gravity in fig .
1 . in fig .
2 , we show the evolutions of @xmath17 for the scalar phantom model with an exponential potential in gr and in inverse-_r _ gravity . obviously , the eos of equivalent dark energy changes from less than @xmath18 to larger than @xmath18 such that the big rip singularity is avoidable in the inverse-_r _ gravity whereas @xmath17 is always less than @xmath18 and asymptotically tends to a constant which leads to the big rip singularity in gr .
it is clear that @xmath119-term plays an essential role for the avoidance of big rip .
the result of numerical calculation is consistent with that of exact power law solution eqs .
( 14 ) and ( 15 ) for exponential type potential .
the quadratic potential is widely investigated since any potential around its minimum ( if it has a minimum ) can be treated as the form of quadratic .
we take the potential as @xmath120 where @xmath121 is a parameter with dimension of mass . in this case , ( 7 ) and ( 10 ) are still reduced to ( 28 ) and ( 29 ) , but @xmath122
@xmath123 in gr , phantom model with quadratic potential may lead to the little rip in the future of cosmic evolution . however , we see that the eos of equivalent dark energy changes from less than @xmath18 to larger than @xmath18 in fig .
it means that the little rip is avoidable for phantom field with a quadratic potential in the inverse-_r _ gravity . in this subsection , we study other potentials for phantom field including cubic , quantic and logarithmic potentials . in gr , the scalar phantom field with quadratic ,
cubic and quantic potentials might lead to little rip whereas the big rip may occur for exponential and logarithmic potentials@xcite . in the numerical calculation
, we only need to revise the expressions of @xmath124 and @xmath125 in ( 28 ) , ( 29 ) and ( 35 ) for different potentials , which are shown in table 1 .
.the expressions of @xmath124 and @xmath125 for various potentials [ cols="^,^,^",options="header " , ] in figs . 4 - 6
, we show the numerical results for the cubic , quantic and logarithmic potentials , respectively .
we find that the results are analogous to the case of exponential or quadratic potential , i.e. , the big or little rip is avoidable in the inverse-_r _ gravity .
in gr , the phantom field with an exponential or a quadratic potential leads to big or little rip that is characterized by the destruction of all bound structures .
we show the possibility of avoiding the big or little rip naturally in the inverse-_r _ gravity . for a special exponential type potential
, we find an exact power - law solution that makes a message known to us : the big rip can be avoided in the inverse-_r _ gravity .
we also find the sufficient condition for a universe in which the equation of state parameter crosses @xmath2 divide in the inverse-_r _ gravity . in gr , the scalar phantom field with quadratic
, cubic and quantic potentials might lead to little rip whereas the big rip may occur for exponential and logarithmic potentials .
and then we investigate these potentials for phantom field @xmath0 in the inverse-_r _ gravity with @xmath40 correction .
the big or little rip is avoidable under all these potentials .
therefore , we conclude that the avoidance of big or little rip is a robust property of inverse-_r _ gravity with @xmath40 corretion , not rigidly adhered to some special potential .
it is worth noting that crossing @xmath2 divide happens approximately in a few e - fold expansion when the inverse-_r _ term is still sufficiently large and the decrease of the energy density resulting from the crossing of @xmath17 will make the catastrophic fate avoided . finally , we give a brief discussion .
it is not a shortcoming that we discuss only the background evolution in this paper but not consider the cosmological perturbation theory .
in fact , nojiri and odintsov@xcite have pointed out that modified gravity equations of motion are higher - derivative differential equations .
the cosmological perturbations in modified gravity approaches to those in gr , since the corresponding equations are reduced to second - order differential equations . * acknowledgments * + this work is supported by national education foundation of china under grant no
. 200931271104 , national science foundation of china under grant .
11075106 and 10671128 , innovation program of shanghai municipal education commission ( 11zz123 ) and key project of chinese ministry of education ( no . | depending on the choice of its potential , the scalar phantom field @xmath0 ( the equation of state parameter @xmath1 ) leads to various catastrophic fates of the universe including big rip , little rip and other future singularity .
for example , big rip results from the evolution of the phantom field with an exponential potential and little rip stems from a quadratic potential in general relativity ( gr ) . by choosing the same potential as in gr , we suggest a new mechanism to avoid these unexpected fates ( big and little rip ) in the inverse-_r _ gravity . as a pedagogical illustration
, we give an exact solution where phantom field leads to a power - law evolution of the scale factor in an exponential type potential .
we also find the sufficient condition for a universe in which the equation of state parameter crosses @xmath2 divide . the phantom field with different potentials , including quadratic , cubic , quantic , exponential and logarithmic potentials
are studied via numerical calculation in the inverse-_r _ gravity with @xmath3 correction .
the singularity is avoidable under all these potentials .
hence , we conclude that the avoidance of big or little rip is hardly dependent on special potential . |
consider the graph @xmath1 with vertex set consisting of all maximal chains in @xmath2 , the non - crossing partition lattice of type @xmath3 , where two chains are adjacent if they differ in exactly one element .
this graph may be identified with the graph of all reduced words of a given coxeter element ( long cycle ) in the symmetric group @xmath4 , where the alphabet consists of all reflections ( transpositions ) and two words are adjacent if they agree in all but two adjacent letters , which are @xmath5 in one word and either @xmath6 or @xmath7 in the other ; here @xmath8 .
this graph is known to be connected see , e.g. , ( * ? ? ?
1.6.1 ) ; we are interested in calculating its radius .
this is motivated by @xcite and @xcite , where the analogous question for simple reflections was studied , and by @xcite which evaluated the radius of a related graph on labeled trees .
recall that a classical result of hurwitz @xcite implies ( see @xcite ) that the number of maximal chains in the non - crossing partition lattice of type @xmath3 is equal to the number of labeled trees on @xmath9 vertices .
our approach is to consider a @xmath0-hecke algebra action on the set of maximal chains .
this allows us to define a well - behaved natural weak order on this set .
each maximal interval is isomorphic to the weak order on the symmetric group .
the number of maximal elements , refined by a generalized inversion number , is the carlitz - riordan @xmath10-catalan number .
the resulting undirected hasse diagram spans the graph @xmath1 of maximal chains , implying an evaluation of the radius and an approximation of the diameter up to a factor of @xmath11 .
the non - crossing partition lattice @xmath2 , first introduced by kreweras @xcite , may be defined as follows .
let @xmath12 be the set of all reflections ( transpositions ) in the symmetric group @xmath4 , and let @xmath13 be the corresponding length function : @xmath14 is the minimal number of factors in an expression of @xmath15 as a product of reflections .
let @xmath16 be a coxeter element in this group ( i.e. , a cycle of length @xmath9 ) ; for concreteness , take @xmath17 .
then @xmath2 is the set @xmath18 ordered by @xmath19 let @xmath20 be the set of all maximal chains in the non - crossing partition lattice @xmath2 .
clearly , an element of @xmath20 corresponds to a factorization of @xmath16 into a minimal number ( @xmath21 ) of transpositions .
it can thus be written in the form @xmath22 , where @xmath23 @xmath24 and @xmath25 .
[ d.hurwitz_graph ] the _ hurwitz graph _
@xmath1 is the ( undirected ) graph with vertex set @xmath20 , where two chains are adjacent if they differ in exactly one element .
the hurwitz graph @xmath26 is drawn in figure [ figure1-tik ] ; note that @xmath27 , and the midpoint of the figure is not a vertex .
( -4,1.66 ) circle ( 0.1 ) node[left]12,23,34 ; ( -1.66,4 ) circle ( 0.1 ) node[above]12,24,23 ; ( 1.66,4 ) circle ( 0.1 ) node[above]14,12,23 ; ( 4,1.66 ) circle ( 0.1 ) node[right]14,13,12 ; ( 4,-1.66 ) circle ( 0.1 ) node[right]34,14,12 ; ( 1.66,-4 ) circle ( 0.1 ) node[below]34,24,14 ; ( -1.66,-4 ) circle ( 0.1 ) node[below]23,34,14 ; ( -4,-1.66 ) circle ( 0.1 ) node[left]23,13,34 ; ( -2,0 ) circle ( 0.1 ) node[left]13,12,34 ; ( -1.4,1.4 ) circle ( 0.1 ) node[below]12,34,24 ; ( 0,2 ) circle ( 0.1 ) node[below]24,14,23 ; ( 1.4,1.4 ) circle ( 0.1 ) node[below]14,23,13 ; ( 2,0 ) circle ( 0.1 ) node[right]13,34,12 ; ( 1.4,-1.4 ) circle ( 0.1 ) node[above]34,12,24 ; ( 0,-2 ) circle ( 0.1 ) node[above]24,23,14 ; ( -1.4,-1.4 ) circle ( 0.1 ) node[above]23,14,13 ; ( -4,1.66)(-1.66,4)(1.66,4)(4,1.66)(4,-1.66)(1.66,-4)(-1.66,-4)(-4,-1.66)(-4 , 1.66 ) ; ( -2,0)(2,0 ) ; ( 0,-2)(0,2 ) ; ( -1.4,-1.4)(1.4,1.4 ) ; ( -1.4,1.4)(1.4,-1.4 ) ; ( -4,1.66)(-1.4,1.4 ) ; ( -1.66,4)(-1.4,1.4 ) ; ( -1.66,4)(0,2 ) ; ( 1.66,4)(0,2 ) ; ( 4,1.66)(1.4,1.4 ) ; ( 1.66,4)(1.4,1.4 ) ; ( 4,1.66)(2,0 ) ; ( 4,-1.66)(2,0 ) ; ( -4,1.66)(-2,0 ) ; ( -4,-1.66)(-2,0 ) ; ( -1.66,-4)(0,-2 ) ; ( 1.66,-4)(0,-2 ) ; ( 4,-1.66)(1.4,-1.4 ) ; ( 1.66,-4)(1.4,-1.4 ) ; ( -4,-1.66)(-1.4,-1.4 ) ; ( -1.66,-4)(-1.4,-1.4 ) ; for each @xmath28 and @xmath29 let @xmath30 and @xmath31 where @xmath32 .
@xmath33 and @xmath34 @xmath35 are operators on @xmath20 , where @xmath36 slides the @xmath37-th factor to the right while @xmath38 slides the @xmath39-st factor to the left .
clearly @xmath40 and , for every @xmath41 , either @xmath42 ( if @xmath43 and @xmath44 commute ) or @xmath45 ( otherwise ) .
furthermore , the following is true .
[ t.arnold]@xcite for every @xmath28 , @xmath46 this clearly defines a braid group action on @xmath20 , a distinguished example of the hurwitz action on ordered tuples of group elements . for @xmath47
, @xmath48 is an edge of @xmath1 if and only if there exists an @xmath28 such that either @xmath49 or @xmath50 . the classical tits graph of a coxeter group is the graph
whose vertices are all the reduced words for the longest element of the group , in terms of the set of _ simple _ reflections @xmath51 , and where two words are adjacent if one is obtained from the other by a braid relation . in other words
, vertices correspond to maximal chains in the weak order on the group , and two chains are adjacent if they differ in a `` minimal cycle '' . replacing the set @xmath51 of simple reflections by the set @xmath12 of all reflections , one obtains the hurwitz graph .
the diameters of the tits graphs of the coxeter groups of types @xmath52 and @xmath53 were evaluated in @xcite .
bessis proved that the hurwitz action on @xmath20 is transitive ( * ? ? ? * prop .
1.6.1 ) , thus that the hurwitz graph @xmath1 is connected . its radius and bounds on its diameter will be given in section [ section : diameter ] .
[ d.order ] for @xmath47 , denote @xmath54 if @xmath55 belongs to a geodesic in @xmath1 from @xmath56 to @xmath57 .
this will be called _ the weak order on @xmath20 _ , and the resulting poset will be denoted by @xmath58 .
the hasse diagram of @xmath59 is drawn in figure [ figure2-tik ] .
this poset is neither a meet semilattice nor a join semilattice , as shown by the two pairs of elements @xmath60 and @xmath61 .
( 0,0 ) circle ( 0.1 ) node[below]e=12,23,34 ; ( -3,2.5 ) circle ( 0.1 ) node[left]23,13,34 ; ( -1,2.5 ) circle ( 0.1 ) node[left]13,12,34 ; ( 1,2.5 ) circle ( 0.1 ) node[right]12,34,24 ; ( 3,2.5 ) circle ( 0.1 ) node[right]12,24,23 ; ( -5,5 ) circle ( 0.1 ) node[left]23,14,13 ; ( -3,5 ) circle ( 0.1 ) node[left]23,34,14 ; ( -1,5 ) circle ( 0.1 ) node[left]13,34,12 ; ( 1,5 ) circle ( 0.1 ) node[right]34,12,24 ; ( 3,5 ) circle ( 0.1 ) node[right]14,12,23 ; ( 5,5 ) circle ( 0.1 ) node[right]24,14,23 ; ( -4,7.5 ) circle ( 0.1 ) node[above]14,23,13 ; ( -2,7.5 ) circle ( 0.1 ) node[above]14,13,12 ; ( 0,7.5 ) circle ( 0.1 ) node[above]34,14,12 ; ( 2,7.5 ) circle ( 0.1 ) node[above]34,24,14 ; ( 4,7.5 ) circle ( 0.1 ) node[above]24,23,14 ; ( 0,0)(-3,2.5)(-5,5)(-4,7.5)(3,5 ) ; ( 0,0)(3,2.5)(5,5)(4,7.5)(-3,5 ) ; ( 0,0)(-1,2.5)(-1,5)(0,7.5)(1,5)(1,2.5)(0,0 ) ; ( -3,2.5)(-3,5 ) ; ( 3,2.5)(3,5 ) ; ( -1,5)(-2,7.5)(3,5 ) ; ( 1,5)(2,7.5)(-3,5 ) ; [ weak - remark ] to motivate the name `` weak order '' consider a coxeter group @xmath62 , with standard generating set @xmath51 , acting transitively on a set @xmath63 .
choose an element @xmath64 in @xmath63 .
for every @xmath65 , let @xmath66 be the distance from @xmath64 to @xmath67 in the associated schreier graph @xmath68 ; and for every @xmath69 , define @xmath70 to _ cover _ @xmath67 if there exists some @xmath71 such that @xmath72 and @xmath73 .
this defines a natural `` weak order '' on @xmath63 , with @xmath64 as a minimal element ; analogous concepts of strong and absolute orders on @xmath62-sets were studied in @xcite . in our case , by proposition [ t.arnold ] @xmath20 carries a transitive action of the braid group of type @xmath52 which covers the coxeter group @xmath74 , and can thus be equipped with a similar order .
furthermore , it will be shown later that maximal intervals in this order are isomorphic to the weak order on @xmath74 .
the poset @xmath58 has several nice properties , such as a @xmath0-hecke algebra action , the above mentioned structure of maximal intervals , and a catalan number of maximal elements .
it will be used to evaluate the radius and bound the diameter of the graph @xmath1 .
let @xmath20 be as in section [ section : intro ] . define a map @xmath75 as follows .
[ d.sigma ] for @xmath76 define the partial products @xmath77 and the empty product @xmath78 by definition , @xmath79 for every @xmath80 .
thus @xmath81 for exactly two values of @xmath82 , and @xmath83 for exactly one such value .
ignoring @xmath84 , the set @xmath85 thus satisfies @xmath86 on the other hand , @xmath87 while @xmath88 , so that @xmath89 for every @xmath90 .
it follows that @xmath91 combining ( [ e.1 ] ) with ( [ e.2 ] ) one concludes that @xmath92 and @xmath93 it follows that the map @xmath94 defined by @xmath95 is a permutation in @xmath74 .
define @xmath96 by @xmath97 [ obs.phi ] for every @xmath98 and @xmath80 , if @xmath99 with @xmath100 then @xmath101 let @xmath102 . then @xmath103 , @xmath104 , @xmath105 , @xmath106 and @xmath107 , where permutations are written in cycle notation .
thus @xmath108 , @xmath109 , @xmath110 and @xmath111 . for a sequence @xmath112 of @xmath113 transpositions in @xmath4 ,
let @xmath114 be the geometric graph whose vertices are @xmath115 drawn on a circle and whose edges are @xmath116 .
the following characterization of the words in @xmath20 , due to goulden and yong , is very useful .
[ gy22 ] ( * ? ? ?
* theorem 2.2 ) + for @xmath117 , the word @xmath118 belongs to @xmath20 if and only if the following three conditions hold : * @xmath114 is a tree .
* @xmath114 is non - crossing ; namely , two edges may intersect only in a common vertex . * cyclically decreasing neighbors : for every @xmath119 and @xmath120 , if @xmath121 and @xmath122 then @xmath123 . here
@xmath124 is the linear order @xmath125 .
proposition [ gy22](iii ) may be stated in the following equivalent form .
[ t.avoid ] for @xmath126 and @xmath127 , if @xmath43 and @xmath128 do not commute then the pair @xmath129 is either @xmath130 , @xmath131 or @xmath132 , for some @xmath133 .
[ phi.local-range ] for @xmath126 and @xmath134 , if @xmath122 with @xmath135 then @xmath136 satisfies : @xmath137 .
the discussion leading to the definition of @xmath138 and @xmath139 above shows that , for any @xmath140 , the sequence @xmath141 is weakly decreasing except for a single step @xmath142 for which , if @xmath143 , it follows that @xmath122 and @xmath144 . in particular
, it follows that @xmath145 and @xmath146 .
[ phi.local-range-2 ] for @xmath126 and @xmath147 , if @xmath148 and @xmath149 with @xmath150 then @xmath136 satisfies : * either @xmath137 or @xmath151 . * deleting the edge @xmath152 from the tree @xmath114 leaves two connected components , @xmath153 ( containing @xmath154 ) and @xmath155 ( containing @xmath156 ) ; only one of them contains the edge @xmath157 . if this is @xmath155 then @xmath137 , otherwise @xmath151 . by lemma [ phi.local-range ] , @xmath158 . by the proof of that lemma , the directed path in @xmath114 leading from @xmath37 to @xmath159 consists of a single increasing step ( edge )
@xmath160 , together with some decreasing edges .
assuming @xmath161 , this path can not contain the step @xmath162 , and also not @xmath163 ( since @xmath147 implies @xmath164 ) .
however , since @xmath114 is non - crossing and @xmath165 , this path must pass through @xmath166 and @xmath16 .
we thus get two distinct paths in @xmath114 from @xmath166 to @xmath16 , contradicting proposition [ gy22](i ) . thus either @xmath137 or @xmath151 .
the rest of the claim now follows from the observation that @xmath37 and @xmath159 belong to distinct connected components of @xmath114 minus the edge @xmath167 . if @xmath155 contains the edge @xmath157 then , by the non - crossing property , it contains all the vertices from @xmath166 up to @xmath156 , and therefore @xmath137 .
the complementary case leads , similarly , to the complementary conclusion .
[ t.inversions ] ( characterization of inversions in @xmath168 ) + for @xmath126 and @xmath169 , @xmath170 if and only if one of the following holds : * @xmath171 such that either : * * @xmath172 and @xmath173 ; or : * * @xmath174 and @xmath175 . * @xmath176 such that @xmath177 and @xmath173 . *
@xmath176 such that either : * * @xmath148 , @xmath149 and , in the notation of lemma [ phi.local-range-2 ] , the edge @xmath157 belongs to @xmath153 ; or * * @xmath174 , @xmath178 and , in the notation of lemma [ phi.local-range-2 ] , the edge @xmath157 belongs to @xmath155 .
assume that @xmath169 and @xmath170 .
if @xmath128 and @xmath179 do not commute then , by corollary [ t.avoid ] , there are 3 options for the pair @xmath180 .
two of them appear in @xmath181 above while the third , @xmath182 , is ruled out since then , by lemma [ phi.local-range ] , @xmath183 .
if @xmath128 and @xmath179 commute then there are 4 options for the pair @xmath180 : @xmath184 , @xmath185 , @xmath186 and @xmath187 ( for @xmath188 ) .
the first option is ruled out by lemma [ phi.local-range ] , the second appears in @xmath189 above , and the third and fourth appear in @xmath190 above in the restricted versions allowed by lemmas [ phi.local-range ] and [ phi.local-range-2 ] .
conversely , if @xmath169 and one of the options @xmath181 , @xmath189 and @xmath190 holds , then lemmas [ phi.local-range ] and [ phi.local-range-2 ] show that indeed @xmath170 .
recall the operators @xmath191 and @xmath192 from section [ section : basic_concepts ] . by definition [ d.sigma ] , for all @xmath193 and @xmath194 , @xmath195 and @xmath196 [ t.main-lemma ] for every @xmath98 and @xmath197 , @xmath198 and @xmath199
we prove the formula for @xmath191 ; the proof of the other formula is similar . since @xmath200 for @xmath201 and @xmath202 for @xmath203 it follows , by observation [ obs.phi ] , that @xmath204 for all @xmath201 .
thus either @xmath205 or @xmath206 . by observation [ obs.phi ] ( for @xmath207 ) , @xmath205 if and only if @xmath208 and @xmath209 for some @xmath154 , @xmath166 and @xmath16 such that @xmath100 and @xmath210 . these @xmath128 and @xmath211
do not commute , and by corollary [ t.avoid ] necessarily @xmath212 as well .
recall the graded poset @xmath58 from section [ section : basic_concepts ] .
note that , by definition , [ t.minimum ] the word @xmath213 is the unique minimal element in @xmath58 .
[ t.unique_id ] @xmath214 clearly @xmath215 , since for @xmath216 one has @xmath217 , @xmath218 and thus @xmath219 by observation [ obs.phi ] .
conversely , assume that @xmath220 has @xmath221 .
then @xmath222 implies that @xmath223 , so that @xmath224 and @xmath225 .
it follows that @xmath226 , so that necessarily @xmath227 and @xmath228 belongs to @xmath229 , with @xmath230 .
induction on @xmath9 completes the proof . for any @xmath231 ,
let @xmath232 be its rank in the poset @xmath58 .
for @xmath233 , let @xmath234 be its inversion number .
[ t.rank_eq_inv ] for any @xmath98 , @xmath235 by induction on @xmath236 . if @xmath237 then @xmath221 . by lemma [ t.unique_id ] @xmath238 ,
so that indeed @xmath239 . for the induction step let @xmath240 , and assume that the claim holds for @xmath241 whenever @xmath242 .
let @xmath243 .
since @xmath244 there exists @xmath245 such that @xmath246 . by lemma [ t.main-lemma ] , @xmath247 unless @xmath248 such that @xmath209 and @xmath208 ( in @xmath249 ) ;
whereas @xmath250 unless @xmath248 such that @xmath251 and @xmath252 . these two exceptions can not hold simultaneously , and thus @xmath253 for either @xmath254 or @xmath255 . for this @xmath55 , @xmath256 , so by the induction hypothesis @xmath257 .
since @xmath55 and @xmath249 are connected by an arc in the graph @xmath1 , it follows that @xmath258 . on the other hand , in a geodesic from @xmath64 to @xmath249 in @xmath1 the value of @xmath259 changes by @xmath0 or @xmath260 on each arc ( again by lemma [ t.main-lemma ] ) , so that @xmath261 . thus @xmath262 .
[ t.hasse ] the ( undirected ) hasse diagram of @xmath58 is obtained from the graph @xmath1 by deleting all edges connecting a vertex @xmath263 with a vertex @xmath264 for some @xmath265 . by lemma [ t.main-lemma ] and lemma [ t.rank_eq_inv ] .
[ t.forbidden-wedge ] for every @xmath231 and @xmath193 , if @xmath266 and @xmath267 in @xmath58 then @xmath268 .
we can assume that @xmath128 and @xmath211 do not commute , since otherwise @xmath268 .
then , by corollary [ t.avoid ] , @xmath269 for some @xmath133 .
the same holds for the @xmath142-th and @xmath270-st factors of @xmath271 , as well as for those of @xmath272 . by lemma [ t.rank_eq_inv ]
, @xmath266 implies that @xmath273 , which is equivalent , by lemma [ t.main-lemma ] , to @xmath274 . by lemma [ t.inversions ]
this rules out @xmath275 and @xmath276 , and leaves only @xmath277 as an option for the @xmath142-th and @xmath270-st factors of @xmath271 .
a similar conclusion holds for @xmath272 , and this implies that @xmath278 and simultaneously @xmath279 , a contradiction .
[ t.rl-cover ] lemma [ t.forbidden-wedge ] may be stated in the following two equivalent forms .
* for every @xmath231 and @xmath193 , @xmath280 if and only if @xmath281 .
* each orbit in @xmath20 of each @xmath191 ( or @xmath192 ) consists of either a pair @xmath282 with @xmath283 or a triple @xmath284 with @xmath283 and @xmath285 .
the _ @xmath0-hecke algebra _ @xmath286 is an algebra over @xmath287 generated by @xmath288 with defining relations @xmath289 @xmath290 and @xmath291 @xmath286 can be viewed as the specialization @xmath292 of the iwahori - hecke algebra @xmath293 , deforming the group algebra of @xmath4 , or as the semigroup algebra of a suitable monoid .
note that our notation is slightly non - standard , since equation ( [ e.0_hecke_relation ] ) is usually stated as @xmath294 , which amounts to replacing @xmath295 by @xmath296 ( for all @xmath37 ) .
a faithful action of the @xmath0-hecke algebra @xmath297 on @xmath20 is introduced in this section .
this action will be applied to show that every maximal interval in @xmath58 is isomorphic to the weak order on the symmetric group @xmath74 .
recall the descent set of a permutation @xmath298 , @xmath299 [ def - down ] for @xmath300 define the _ @xmath37-th down operator _
@xmath301 by @xmath302 the down operators are well defined . by lemma [ t.main-lemma ] , for any @xmath231 and @xmath28 , either @xmath303 or @xmath304 ( or both ) . if @xmath305 then @xmath306 .
thus , by lemma [ t.rank_eq_inv ] , either @xmath307 or @xmath308 , namely : either @xmath309 or @xmath310 in @xmath58 . if both hold then , by lemma [ t.forbidden-wedge ] , @xmath311 .
[ obs.down ] for every @xmath312 and @xmath98 , @xmath313 [ t.down-relations ] the down operators @xmath314 satisfy the defining relations of the @xmath0-hecke algebra @xmath297 .
if @xmath315 then , by definition [ def - down ] , @xmath316 .
if @xmath305 then , by observation [ obs.down ] , @xmath317 and therefore , by definition [ def - down ] , @xmath318 .
thus @xmath319 for every @xmath28 .
if @xmath320 then , by definition [ def - down ] , @xmath321 . we have to verify that @xmath322 for every @xmath323 .
if @xmath324 then , by definition [ def - down ] , @xmath325 .
if @xmath305 but @xmath326 then , by definition [ def - down ] , @xmath327 and therefore @xmath328 . by observation [ obs.down ] , the permutation @xmath329 is the same as @xmath330 except that @xmath331 , which is not necessarily the case for @xmath330 . the assumption @xmath305 but
@xmath326 implies that @xmath332 .
denoting @xmath333 we get @xmath334 and therefore @xmath335 .
hence @xmath336 and , by definition [ def - down ] , @xmath337 as well . for similar reasons , if @xmath315 and @xmath338 then @xmath339 and @xmath340 , so that @xmath341 .
it remains to verify the braid relation @xmath322 when @xmath342 . by assumption @xmath343 , so that @xmath249 , @xmath344 , @xmath345 and @xmath346 are all distinct : @xmath347 in @xmath58 .
similarly with @xmath348 and @xmath349 interchanged .
for a word @xmath350 let @xmath351 . obviously , if @xmath352 then @xmath353 .
it thus suffices to verify that @xmath354 for @xmath355 . for a subset @xmath356 let @xmath357 . by proposition [ gy22](i ) , @xmath358 .
we shall consider these @xmath359 cases separately .
* @xmath360 .
+ in this case @xmath361 and @xmath362 commute , so that @xmath363 etc .
thus , by proposition [ t.arnold ] , @xmath364 * case 2 . * @xmath365 .
+ in this case one of the transpositions @xmath361 or @xmath362 commutes with other two . if @xmath43 commutes with @xmath44 and @xmath362 then , since @xmath338 , if @xmath366 with @xmath265 then , by lemma [ t.inversions](1 ) , either @xmath367 or @xmath368 and in both cases , after application of @xmath349 , this becomes @xmath277 . thus @xmath369 and
also @xmath370 similarly , if @xmath362 commutes with @xmath43 and @xmath44 then , if @xmath371 with @xmath265 , @xmath372 if @xmath44 commutes with @xmath43 and @xmath362 then , if @xmath373 with @xmath265 , @xmath374 * case 3 . * @xmath375 .
+ let @xmath376 with @xmath377 . by proposition [ gy22 ] and lemma [ t.inversions](1 ) there are exactly @xmath378 possible options for @xmath379 , which correspond to the @xmath378 maximal elements of @xmath380 , as in figure [ figure2-tik ] . inspecting this figure , it is easy to verify that the braid relation hold in all cases . for a permutation @xmath233 let @xmath381 be a reduced word for @xmath15 in the alphabet of simple reflections .
define the operator @xmath382 by @xmath383 the operator @xmath384 is well - defined .
since any two reduced words for the same permutation are connected by a sequence of braid relations , the result follows from lemma [ t.down-relations ] .
[ interval1 ] for every element @xmath231 , the lower interval @xmath385 $ ] is exactly @xmath386 . by definition , @xmath387 for any @xmath28 and therefore @xmath388 for any @xmath233 .
in the other direction , if @xmath389 then @xmath57 belongs to a geodesic from @xmath249 to @xmath64 in @xmath1 , part of which is a saturated chain @xmath390 in @xmath58 .
clearly each step is @xmath391 for some @xmath392 ( @xmath393 ) .
if the word @xmath394 is not reduced then it can be brought , by relations as in lemma [ t.down-relations ] , to a reduced word for some @xmath15 , so that @xmath395 .
[ interval2 ] let @xmath396 be any maximal element in @xmath58 .
then , for any @xmath298 , @xmath397 where @xmath398 $ ] is the maximal element in the weak order on @xmath74 . by induction on @xmath234 . if @xmath399 then necessarily @xmath400 , since otherwise there exists @xmath193 such that @xmath401 ; and then , by lemma [ t.main-lemma ] and lemma [ t.rank_eq_inv ] , either @xmath402 or @xmath403 has @xmath404 , contradicting the maximality of @xmath396 . if @xmath405 then there exists @xmath193 such that @xmath406 has @xmath407
. then @xmath408 has @xmath409 with @xmath410 . by the induction hypothesis @xmath411 , thus @xmath412 .
combining this with observation [ obs.down ] we obtain @xmath413 we conclude [ t.main2 ] each maximal interval in @xmath58 is isomorphic , as a poset , to the weak order on the symmetric group @xmath74 . by lemmas [ interval1 ] and [ interval2 ]
, @xmath139 is onto @xmath74 even when restricted to a maximal interval @xmath414 $ ] . to prove that @xmath139 is also one - to - one when restricted to @xmath414 $ ] , it suffices to show that the cardinality of the interval @xmath414 $ ] does not exceed the cardinality of @xmath74 ; and this follows again from lemma [ interval1 ] . finally , to prove that @xmath139 is a poset isomorphism notice that , for @xmath415 $ ] , if @xmath416 then either @xmath417 or @xmath418 , and therefore either @xmath419 or @xmath420 .
the converse implications are clear .
thus , for @xmath415 $ ] , @xmath421 we can now summarize .
* the @xmath0-hecke algebra @xmath297 acts faithfully on @xmath58 .
* each lower interval is invariant under this action . by lemma [ t.down-relations ]
, the down operators @xmath422 satisfy the defining relations of @xmath297 .
furthermore , by lemma [ interval1 ] , lower intervals are invariant under these operators . by theorem [ t.main2 ] , the size of each maximal interval is equal to the dimension of the @xmath0-hecke algebra , so that the action is faithful .
in this section we prove [ t.main3 ] the number of maximal elements in @xmath58 is the catalan number @xmath423 for a @xmath10-analogue of this result see corollary [ t.main3-q ] below .
recall the map @xmath75 defined in section [ section : map ] .
for @xmath298 , let @xmath424 for a subset @xmath51 of @xmath425 let @xmath426 be the subsequence of @xmath427 consisting only of the values in @xmath51 , and let @xmath428 be the pattern ( permutation ) obtained by mapping @xmath51 monotonically to @xmath429 , namely : the sequence obtained from @xmath426 by writing @xmath260 instead of the smallest number in @xmath51 , @xmath430 instead of the second smallest number , etc .
for example , if @xmath431 and @xmath432 then @xmath433 and @xmath434 .
also , for @xmath435 , let @xmath436 ( of size @xmath437 ) and @xmath438 ( of size @xmath439 ) .
thus @xmath440 , a disjoint union .
[ t.cover-number ] let @xmath298 such that @xmath441 @xmath442
. then @xmath443 for @xmath444 , @xmath445 .
thus , summing over @xmath446 : @xmath447 assume that @xmath98 satisfies @xmath448 .
then the first transposition in @xmath249 has the form @xmath449 for some @xmath450 , and the corresponding partial product @xmath451 is @xmath452 . in general , for any @xmath453 , @xmath454 is a product of @xmath455 disjoint cycles , each of which forming a ( not necessarily consecutive ) cyclic subsequence of @xmath456 , and the transition from @xmath457 to @xmath454 amounts to merging two of the cycles of @xmath457 into one cycle in which the original cycles form complementary cyclic intervals .
it follows that @xmath458 is the product of those @xmath179 with @xmath459 which permute two elements of the set @xmath460 , and thus have @xmath461 in this set but of course @xmath462 ; and , similarly , @xmath463 is the product of those @xmath179 with @xmath459 which permute two elements of the set @xmath464 , and thus have @xmath461 in this set but of course @xmath465 . the word @xmath249 is a shuffle of these two sequences of transpositions , with the addition of @xmath449 .
the @xmath139 values at these two sequences compose the ranges @xmath466 and @xmath467 , respectively . mapping these ranges monotonically to the ranges @xmath468 and @xmath469 we get @xmath470 and @xmath471 such that , for @xmath472 , @xmath473 and @xmath474 .
conversely , for any @xmath475 , @xmath470 and @xmath471 we can construct @xmath231 by reversing the above process .
this completes the proof of the formula .
_ proof of theorem [ t.main3 ] .
_ let @xmath476 be the longest element in @xmath74 , i.e. , @xmath477 @xmath24 ; denote @xmath478 .
if @xmath98 has @xmath479 then @xmath480 and therefore , for each @xmath481 , @xmath482 and @xmath483 . thus @xmath484 and @xmath485
are decreasing sequences , and the corresponding permutations @xmath486 and @xmath487 are the longest elements in @xmath488 and @xmath489 , respectively .
it follows , by lemma [ t.cover-number ] , that @xmath490 and clearly @xmath491 .
we recognize the recursion formula for the catalan numbers @xmath492 , and therefore @xmath493 as claimed .
the current section presents a bijection between the maximal elements in @xmath58 and alternating non - crossing trees , following a comment by an anonymous referee . recall that a tree with integer - labeled vertices is _ alternating _ if the vertices along any path in it form an alternating sequence : @xmath494 ; equivalently , if the neighbors of each vertex @xmath37 are all larger , or all smaller , than @xmath37 .
[ t.max_to_alt ] if @xmath57 is a maximal element in @xmath58 then the corresponding non - crossing tree @xmath495 ( as in proposition [ gy22 ] ) is alternating .
if @xmath57 is maximal in @xmath58 then , by lemma [ interval2 ] for @xmath496 , @xmath497 $ ] and thus all pairs of letters in @xmath498 are inversions . for @xmath133 ,
the pair @xmath277 does _ not _ correspond to an inversion in @xmath498 , by the characterization in lemma [ t.inversions ] .
this implies that @xmath495 is alternating .
the number of all alternating non - crossing trees on the set of vertices @xmath499 = \{1 , \ldots , n\}$ ] was shown ( * ? ? ? * theorem 6.4 ) to be the catalan number @xmath492 .
this motivates the following result .
[ t.max_eq_alt ] the mapping @xmath500 is a bijection between the set of all maximal elements in @xmath58 and the set of all alternating non - crossing trees on the set of vertices @xmath499 $ ] . by observation [ t.max_to_alt ] , this mapping , to be denoted @xmath501 , is well defined .
let us construct a map @xmath502 in the other direction . for an alternating non - crossing tree @xmath503 on
@xmath499 $ ] , defining @xmath504 amounts to imposing a linear order on the @xmath505 edges of @xmath503 .
this can be done recursively as follows : if @xmath506 , there are no edges and the order is empty . for @xmath507 note that @xmath508 must be an edge of @xmath503 ( since the path from @xmath260 to @xmath9 in @xmath503 is monotone increasing , because of the non - crossing property , and thus is also alternating if and only if it consists of a single edge ) .
the graph obtained from @xmath503 by deleting the edge @xmath508 has two connected components , @xmath509 ( containing the vertex @xmath260 ) and @xmath510 ( containing the vertex @xmath9 ) .
they are alternating non - crossing trees on complementary subsets @xmath511 $ ] and @xmath512 $ ] of @xmath499 $ ] , for a suitable @xmath453 .
they define , by recursion , corresponding words @xmath513 and @xmath514 ( using the same letter @xmath502 for the two corresponding maps ) .
define the word @xmath504 as the word @xmath515 , followed by the letter @xmath516 , followed by the word @xmath517 .
we need to show that @xmath57 is a maximal element of @xmath58 .
first , @xmath518 since @xmath519 satisfies the goulden - yong conditions of proposition [ gy22 ] , the cyclic order on neighbors following recursively from the construction . maximality of @xmath57 in @xmath58 follows from the fact that all pairs in the permutation @xmath498 are inversions , by a careful application of lemma [ t.inversions ] .
having defined the map @xmath502 , it is clear that the composition @xmath520 is the identity map on the set of all alternating non - crossing trees .
thus @xmath501 is surjective , and since both sets have the same ( catalan ) size it follows that @xmath501 is a bijection . [ 1n ]
every maximal element in @xmath58 contains @xmath516 as a factor .
this follows from theorem [ t.max_eq_alt ] together with the observation ( mentioned in this theorem s proof ) that every alternating non - crossing tree of order @xmath521 contains the edge @xmath508 .
recall that the weak order on @xmath4 is characterized by inversion sets , namely : @xmath522 in the ( right ) weak order on @xmath4 if and only if @xmath523 , where @xmath524 ; see , e.g. , ( * ? ? ? * cor .
1.5.2 and prop .
3.1.3 ) .
the goal of this section is to provide a similar characterization for @xmath58 .
[ defn - lr - inversions ] the set of _ inversions _ of @xmath525 is @xmath526 .
a pair @xmath527 is * a _ right inversion _
if there exist @xmath265 such that @xmath528 and @xmath529 , or there exist @xmath530 such that @xmath531 and @xmath532 ; * a _ left inversion _ if there exist @xmath265 such that @xmath533 and @xmath534 , or there exist @xmath530 such that @xmath533 and @xmath535 ; * a _ neutral inversion _ if there exist @xmath530 such that @xmath536 and @xmath529 .
[ left - right - explanation ] if @xmath529 , @xmath100 , write @xmath537 for the interval @xmath538\subseteq [ 1,n]$ ]
. then @xmath527 is * a right inversion if and only if @xmath539 ; * a left inversion if and only if @xmath540 ; * a neutral inversion if and only if @xmath541 .
denote the set of right , left and neutral inversions of @xmath98 by @xmath542 , @xmath543 and @xmath544 , respectively .
lemma [ t.inversions ] implies [ t.ln ] for every @xmath98 , @xmath545 a disjoint union . for @xmath546 , @xmath547 $ ] .
then @xmath548 is a disjoint union of @xmath549 , @xmath550 and @xmath551 .
define the involution @xmath552 by @xmath553 where @xmath32 and @xmath554 \in s_n$ ] . in figure [ figure2-tik ] , for example
, @xmath555 corresponds to a reflection of the hasse diagram of @xmath59 in a vertical line through @xmath64 .
[ iota ] let @xmath398 \in s_{n-1}$ ] . * for every @xmath98 : @xmath556 and also @xmath557 , @xmath558 and @xmath559 .
* @xmath555 is a rank preserving automorphism of the poset @xmath58 . [
determine ] + each of the pairs @xmath560 and @xmath561 determines the word @xmath562 uniquely ; namely , for @xmath47 , @xmath563 if @xmath564 then , of course , all the other equalities hold .
let us prove that if @xmath565 and @xmath566 then @xmath564 .
we shall use the shorthand notation @xmath567 for @xmath568 .
of course , the inversion set @xmath569 determines @xmath570 uniquely since it determines , for any @xmath571 , whether or not @xmath572 .
we shall show , by induction on @xmath9 , that @xmath573 and @xmath574 determine @xmath575 for all @xmath576 .
this trivially holds for @xmath577 .
for the induction step assume that @xmath578 with @xmath450 .
by observation [ obs.phi ] , @xmath579
. we can recover @xmath142 from @xmath574 once we show that there are exactly @xmath580 left inversions of @xmath570 involving @xmath581 .
indeed , the inversions involving @xmath581 are exactly the ordered pairs @xmath582 such that @xmath583 ( automatically satisfied ) and @xmath584 . by remark [ left - right - explanation ]
, such a pair is a _ left _
inversion if and only if the @xmath455-th edge of @xmath585 has both endpoints in the interval @xmath586 $ ] , and is not the edge @xmath587 itself . by proposition [ gy22 ] , there are exactly @xmath580 such edges .
thus we have recovered both @xmath37 and @xmath142 .
now note that @xmath588 is a product of two disjoint cycles , @xmath589 and @xmath590 , of corresponding lengths @xmath591 and @xmath592 .
denote @xmath593 and @xmath594 , and let @xmath595 and @xmath596 be monotone increasing bijections .
the word @xmath597 is a shuffle of two subwords , one supported in @xmath598 and the other in @xmath599 .
for @xmath600 we do nt yet know @xmath179 , but we do know whether it is supported in @xmath598 or in @xmath599 depending on whether @xmath601 belongs to @xmath598 or to @xmath599 .
let @xmath602 and @xmath603 be the sets of indices @xmath455 for which @xmath179 is supported in @xmath598 and in @xmath599 , respectively . denoting @xmath604 ,
the word @xmath605 is in @xmath606 , and a similar definition gives @xmath607 . by induction
, it suffices to show that @xmath573 and @xmath574 determine @xmath608 and @xmath609 ( since @xmath610 can then be retrieved from @xmath611 ) , and similarly for @xmath612 . actually , since @xmath613 is monotone increasing , @xmath614 for any @xmath615 .
since @xmath616 and @xmath617 , it follows that @xmath618 and therefore @xmath619 . thus there is a bijection between the inversions of @xmath620 and the relevant inversions of @xmath55 , which may be written ( with a slight abuse of notation ) as @xmath621
the situation for @xmath612 is only slightly different , since indeed @xmath622 but an application of @xmath623 , exchanging @xmath37 with @xmath142 , does not change its relative position in the set .
therefore @xmath624 .
remark [ left - right - explanation ] now shows that similar relations hold for @xmath625 , completing the induction step .
thus @xmath573 and @xmath574 uniquely determine @xmath55 . finally , by observation
[ iota].1 , @xmath573 and @xmath626 determine @xmath627 and @xmath628 , thus @xmath629 and @xmath55 .
@xmath574 and @xmath626 alone do not suffice to uniquely determine @xmath562 .
here is an example from @xmath380 ( see figure [ figure2-tik ] ) : @xmath630 and @xmath631 are distinct , but have @xmath632 as well as @xmath633 ( while @xmath634 ) . [ criterion ] for @xmath635 , if @xmath636 in @xmath58 then @xmath637 , @xmath638 and @xmath639 .
it suffices to prove the claim when @xmath57 covers @xmath55 in @xmath58 .
thus @xmath640 in @xmath74 , for some @xmath28 , and consequently @xmath641 for the pair @xmath642 .
there are three possible cases : @xmath643 , @xmath644 , and @xmath645 .
we want to show that @xmath638 and @xmath639 .
consider an arbitrary inversion @xmath646 of @xmath55 .
it is also an inversion of @xmath57 . if @xmath647 then @xmath648 , where @xmath649 and similarly for @xmath57 . by remark [ left - right - explanation ] ,
the type of @xmath650 ( left , right or neutral ) is then the same in @xmath55 as in @xmath57 .
if @xmath643 , namely if @xmath651 and @xmath652 commute , then @xmath653 and @xmath654 , and therefore _ all _ inversions keep their type upon transforming @xmath55 to @xmath57 : @xmath655 and @xmath656 .
assume now that @xmath644 .
then @xmath651 and @xmath652 do not commute , and by lemma [ t.inversions ] there exist @xmath133 such that @xmath657 and @xmath658 .
since @xmath654 , the inversions involving @xmath659 do not change their type upon transforming @xmath55 to @xmath57 .
it remains to consider the inversions involving @xmath660 .
consider the graphs @xmath495 and @xmath585 , which are identical except that the former has an edge @xmath661 and the latter an edge @xmath662 ( and both have an edge @xmath157 ) .
for any other edge of these graphs , the non - crossing property implies that the corresponding interval @xmath663 belongs to one of the following four classes : contained in @xmath664 $ ] ; contained in @xmath538 $ ] ; containing @xmath665 $ ] ; and disjoint from @xmath665 $ ] .
we consider only edges corresponding to inversions with @xmath660 .
using remark [ left - right - explanation ] again we see that , for the last three classes , transforming @xmath55 to @xmath57 ( namely replacing @xmath662 by @xmath661 ) does not change the type of inversion with @xmath660 .
for the first class , the type changes from neutral to either right or left .
thus we conclude that @xmath638 and @xmath639 .
the treatment of the case @xmath645 , for which @xmath666 and @xmath658 , is similar .
various natural stronger versions of proposition [ criterion ] can be shown to be false by examples from @xmath380 .
for example , @xmath667 , @xmath668 and @xmath669 are not even weakly monotone increasing functions of @xmath55 ; and inclusions of any two of the sets @xmath670 , @xmath671 and @xmath625 does not imply @xmath636 in @xmath58
. it would be interesting to know whether the converse of proposition [ criterion ] holds .
a @xmath10-analogue of theorem [ t.main3 ] is given in this section .
carlitz and riordan @xcite defined a @xmath10-catalan number @xmath672 using the recursion @xmath673 with @xmath674 .
for combinatorial interpretations of this number and further information see , e.g. , @xcite and sequence a138158 in @xcite .
in particular , @xmath675 counts dyck paths of length @xmath9 with respect to the area above the path ; see , e.g. , @xcite .
[ def : qt - catalan ] define _
@xmath676-catalan numbers _ by @xmath677 with @xmath678 [ obs.qt2 ] for every @xmath679 , @xmath680 and @xmath681 for @xmath98 , denote by @xmath682 the cardinality of @xmath542 and by @xmath683 the cardinality of @xmath543 .
let @xmath684 denote the set of all maximal elements in @xmath20 .
[ qt - prop ] for every positive integer @xmath9 , @xmath685 by induction on @xmath9 . for @xmath686
the claim obviously holds . for the induction step recall that , by corollary [ 1n ] ,
any maximal element in @xmath249 in @xmath687 contains the factor @xmath688 .
denote by @xmath689 the set of maximal elements in @xmath687 whose @xmath690-st factor is @xmath688 . by the proof of theorem [ t.main3
] , this set is characterized as follows : @xmath691 if and only if @xmath692 is a reduced word for the cycle @xmath693 , where @xmath694 $ ] , and @xmath695 is a reduced word for the cycle @xmath696 , where @xmath697 $ ] . for every @xmath698 the following holds : for every @xmath699 there exist @xmath700 such that @xmath701 ; for every @xmath702 there exist @xmath703 such that @xmath99 .
thus , by definition [ defn - lr - inversions ] , all pairs @xmath704 , with @xmath705 and @xmath706 , are neutral inversions ; all pairs @xmath707 , @xmath699 , are left inversions and all pairs @xmath708 , @xmath709 , are right inversions .
this implies that , for every @xmath710 , @xmath711 the last equality follows from the induction hypothesis .
thus @xmath712 the last equality follows from the defining recursion of @xmath713 together with ( [ qt - eq2 ] ) .
[ t.main3-q ] for every positive integer @xmath9 , @xmath714 or equivalently @xmath715 proposition [ qt - prop ] together with ( [ qt - eq1 ] ) .
denote the set of alternating non - crossing trees of order @xmath9 by @xmath716 .
[ edge_pairs ] a pair of distinct edges @xmath717 in a non - crossing alternating tree @xmath12 is * _ neutral _ if @xmath718 and @xmath719 for some @xmath188 ; * _ right _ if @xmath720 and @xmath721 for some @xmath722 and the edge @xmath723 is in the component of @xmath154 in @xmath724 ( for example , whenever @xmath725 ) ; and * _ left _
if @xmath726 and @xmath727 for some @xmath728 and the edge @xmath723 is in the component of @xmath156 in @xmath724 ( for example , whenever @xmath729 ) .
denote the cardinality of right and left and pairs in an alternating non - crossing tree @xmath730 by @xmath731 and @xmath732 , respectively .
[ qt - prop - trees ] for every positive integer @xmath9 , @xmath733 each pair of distinct edges is either neutral , right or left .
moreover , the edge @xmath64 precedes @xmath501 in the canonical linear order on @xmath12 defined by the map @xmath502 as in the proof of theorem [ t.max_eq_alt ] .
hence , a pair of edges in an alternating non - crossing tree is a right ( left ) pair if and only if the corresponding letters in the word @xmath504 form a right ( left ) inversion .
it follows that @xmath734 proposition [ qt - prop ] completes the proof .
consider the set @xmath735 of dyck paths of length @xmath736 .
each @xmath737 is a sequence @xmath738 , with @xmath739 such that @xmath740 and @xmath741 geometrically , @xmath650 corresponds to a lattice path from @xmath742 to @xmath743 , with admissible steps @xmath744 and @xmath745 , which never goes below the horizontal line ( _ baseline _ , in the sequel ) connecting its endpoints . with each @xmath737 associate
the following two statistics : @xmath746 geometrically , @xmath747 is the area between the path @xmath650 and the `` horizontal '' path @xmath748 , where the square with vertices @xmath742 , @xmath744 , @xmath749 and @xmath745 is considered to have area @xmath260 ; while @xmath750 is the sum , over all internal local minima of the path @xmath650 , of half the length of the maximal horizontal segment starting at the minimum point and stretching to the right without crossing ( but possibly touching ) @xmath650 .
note that @xmath747 is ( a linear function of ) the inversion number of the sequence @xmath650 .
[ dyck ] for every positive integer @xmath9
@xmath751 note that for @xmath752 this is a classical result @xcite .
it suffices to show that the lhs , to be denoted @xmath753 , satisfies the recursion in definition [ def : qt - catalan ] ; it clearly satisfies @xmath754 .
let @xmath755 , and define @xmath756 , so that @xmath757 is the first time that @xmath650 returns to the baseline
. clearly @xmath710 .
we can write @xmath650 as the concatenation of sequences @xmath758 where @xmath759 and @xmath760 .
now clearly @xmath761 and @xmath762 since the path @xmath650 has an internal local minimum at @xmath763 , in addition to the internal local minima of @xmath764 and @xmath765 .
thus @xmath766 and this completes the proof .
@xmath767 and @xmath768 are equidistributed : @xmath769 proposition [ dyck ] together with both parts of observation [ obs.qt2 ] .
@xmath770 proposition [ qt - prop ] together with proposition [ dyck ] .
let @xmath771 be a finite connected undirected graph , and let @xmath772 be the corresponding metric on its vertex set @xmath773 .
the _ radius _ of @xmath503 is @xmath774 and its _ diameter _ is @xmath775 recall the hurwitz graph @xmath1 ( definition [ d.hurwitz_graph ] ) , and let @xmath776 be the ( right ) cayley graph of the group @xmath74 with respect to its standard coxeter generators . [ t.phi_contracts ] the map @xmath777 is a contraction : @xmath778 it suffices to show that if @xmath779 then @xmath780 . indeed ,
if @xmath779 in @xmath1 then either @xmath419 or @xmath420 for some @xmath142 . thus , by lemma [ t.main-lemma ] , @xmath781 , so that indeed @xmath780 .
the radius of the hurwitz graph @xmath782 by definition [ d.order ] , only edges connecting vertices of the same rank are deleted during the passage from @xmath1 to the hasse diagram of @xmath58 .
it follows that the distance @xmath783 from the unique minimal element @xmath64 of @xmath58 to an arbitrary element @xmath231 is the same , whether measured in @xmath1 or in the hasse diagram .
thus , by theorem [ t.main2 ] and a well - known property of @xmath74 , @xmath784 it follows that @xmath785 consider now an arbitrary @xmath518 . by the proof of theorem [ t.main2 ]
, @xmath786 is surjective .
thus there exists @xmath231 with @xmath787 where @xmath398 \in s_{n-1}$ ] . for this @xmath249
it follows from lemma [ t.phi_contracts ] that @xmath788 and therefore @xmath789 clearly , for any graph @xmath503 , @xmath790 the diameter of @xmath1 satisfies @xmath791 the upper bound can be improved .
[ t.diameter ] the diameter of @xmath1 satisfies @xmath792 we shall use the bubble sort " algorithm to transform any two given words @xmath793 to a common word by applying various @xmath191 and @xmath192 operators . for the initial step note that , since every tree has at least one leaf ( vertex of degree @xmath260 ) , by proposition [ gy22](i ) applied to @xmath57
there is an @xmath794 which appears in exactly one of the @xmath795 factors ( transpositions ) in @xmath57 .
this unique @xmath37-factor " is within distance @xmath796 from either the first or the last factor ( end factor " ) in @xmath57 .
apply a sequence of @xmath191 and @xmath192 operators to push " this @xmath37-factor towards the closest end factor without creating any additional @xmath37-factors .
this takes at most @xmath796 steps , resulting in a word whose only @xmath37-factor is an end factor . on the other hand , consider the word @xmath249 : it may contain more than one @xmath37-factor , and we want to push all of them toward the end factor ( at the same end as in @xmath57 ) .
this can be done by pushing the @xmath37-factor which is furthest from that end factor .
at each step the furthermost @xmath37-factor gets closer to the end factor , and the total number of @xmath37-factors is unchanged or is reduced by @xmath260 .
after at most @xmath797 steps we get from @xmath249 a word whose only @xmath37-factor is the end factor .
now observe that if the only occurence of @xmath37 in a decomposition of @xmath798 ( @xmath799 $ ] for this initial step ) is in the first ( leftmost ) factor then this factor is @xmath800 ; and if the only occurence is in the last factor then this factor is @xmath801 .
this implies that the end @xmath37-factors in the two words obtained from @xmath57 and @xmath249 are equal . for the rest of the algorithm
we keep this end factor fixed , and work only on the subwords containing the other factors .
we repeat the procedure for the two words of length @xmath802 thus obtained from @xmath57 and @xmath249 . obviously , after @xmath802 such steps we get two words of length @xmath260 , which are necessarily equal .
the total number of operators used is at most @xmath803 and this is an upper bound on the diameter of @xmath1 .
an improvement of proposition [ t.diameter ] , bounding the diameter of the hurwitz graph , is desired .
[ d.hurwitz_graph-coxeter ] let @xmath62 be a coxeter group , @xmath12 its set of reflections , and @xmath16 a coxeter element in @xmath62 .
the _ hurwitz graph _
@xmath807 is the graph with vertices corresponding to all reduced words expressing @xmath16 in the alphabet @xmath12 of all reflections , where adjacency is determined by a local right or left shift : two words are adjacent in the graph if they agree in all but two adjacent letters , which are @xmath5 in one word and either @xmath6 or @xmath7 in the other .
this conjecture is supported by computer results for @xmath810 .
the computations also hint that the number of antipodes of @xmath811 ( i.e. , the number of elements of maximal rank in the associated weak order ) is given by sequence a054441 in @xcite which involves central binomial coefficients , namely the catalan numbers of type @xmath53
. * acknowledgements . *
thanks to c. athanasiadis , o. bernardi , a. kalka , c. keller , h. y. khachatryan , s. margolis , p. mcnamara , a. postnikov and v. reiner for useful discussions , comments and references . thanks also to the anonymous referees for their stimulating comments .
j. dnes , _ the representation of a permutation as the product of a minimal number of transpositions and its connection with the theory of graphs _ , publ . math .
* 4 * ( 1959 ) , 6371 .
i. m. gelfand , m. i. graev and a. postnikov , _ combinatorics of hypergeometric functions associated with positive roots _ , in : arnold - gelfand mathematical seminars : geometry and singularity theory , birkhuser , 1997 , 205-221 .
n. j. a. sloane , the on - line encyclopedia of integer sequences .
v. strehl , _ minimal transitive products of transpositions the reconstruction of a proof of a. hurwitz _ , sm .
* 37 * ( 1996 ) , art .
s37c , 12 pp .
( electronic ) . | a weak order on the set of maximal chains of the non - crossing partition lattice is introduced and studied .
a @xmath0-hecke algebra action is used to compute the radius of the graph on these chains in which two chains are adjacent if they differ in exactly one element . |
the heating and cooling processes in the interstellar medium are crucial to determining its physical state in equilibrium .
photoelectrons from dust grains or polycyclic aromatic hydrocarbons ( pahs ) are thought to dominate the heating of neutral gas ( watson 1972 ; hollenbach & tielens 1999 ) . in an indirect and inefficient mechanism ,
incident far - ultraviolet photons with energies high enough to eject electrons from dust grains ( @xmath2 ev ) heat the gas via these photoelectrons , with a typical efficiency of @xmath3 .
this efficiency is determined by microphysics of the grains , in particular the work function and the resultant photoelectric yield , and therefore the charge of the grains .
it is defined as the ratio of the gas heating rate by photoelectrons to grain heating rate by far - ultraviolet photons .
the grain heating is radiated away in the infrared .
the gas heating is also radiated away in the infrared , primarily via fine - structure lines such as [ cii ] ( 158 @xmath0 ) and [ oi ] ( 63 @xmath0 ) , with [ cii ] dominant at lower densities and temperatures . the photoelectric effect on dust
thus provides the main coupling of the far - ultraviolet radiation field to interstellar gas outside of regions , and is a major process in photodissociation regions ( hollenbach & tielens 1999 ) .
the heating efficiency of the photoelectric effect is traditionally measured by the flux ratio ( [ cii]+[oi])/fir , where fir is the dust continuum in the far - infrared .
the physical description of the photoelectric effect and its dependence on various parameters is fairly well developed . in particular ,
it has been established for some time that smaller grains are more effective at generating photoelectrons , because of their lower potential barrier to escaping electrons , and because the emerging electron is less likely to lose energy as it works its way from the ejection site within the grain to the outside . given the typical grain size distributions ,
the net result is that the smallest grains dominate the photoelectric heating ( watson 1972 , jura 1976 , de jong 1977 , draine 1978 , bakes & tielens 1994 ) .
dhendecourt & lger ( 1987 ) argued that pahs may be the key to interstellar gas heating in clouds and intercloud regions , assuming they carry a plausible 10% of interstellar carbon . on the other hand , in a standard
bakes & tielens model of the orion bar about half of the total photoelectric heating is due to small grains with radii less than 15 , typical of pahs and smaller molecules .
the importance of pahs to the ionization balance and chemistry of the interstellar medium is discussed by lepp & dalgarno ( 1988 ) .
quite independently , the hypothesis of very small grains heated stochastically to emit at wavelengths shorter than expected for their radiative environment was proposed by andriesse ( 1978 ) , and by sellgren ( 1984 ) to explain her data on reflection nebulae which showed an excess in the near infrared with abnormally elevated apparent temperatures . revealed ubiquitous dust emission at 12 @xmath0 , and very small grains were accepted as widespread ( beichman 1987 ) .
combined with rudimentary spectral data , the 12 @xmath0 emission was associated with aromatic features ( puget & leger 1989 ) .
detailed spectra with left no doubt that the mid - infrared emission from dust in all but the most extreme environments is dominated by the aromatic features , and is therefore a good tracer of pahs ( tielens et al .
helou et al .
( 2000 ) recently established that the mid - infrared spectra of normal galaxies are almost constant in shape , and dominated by the aromatic features in emission ( afe ) , which carry from a few percent up to 20% of the total infrared luminosity .
malhotra et al . ( 1997 and 2000a ) reported observations of 60 normal galaxies spanning a broad range in r(60/100 ) , fir / b - band flux and morphology , and demonstrated that [ cii]/fir decreases as r(60/100 ) increases in galaxies . since most galaxies have ratios [ cii]/fir @xmath4 , those with ratios @xmath5 were said to be `` deficient . ''
since r(60/100 ) measures average temperatures of the large dust grain population , it increases with increasing far - ultraviolet flux @xmath6 .
the lowest [ cii]/fir ratios are observed in warmer and more actively star - forming galaxies .
there have been many explanations proposed , such as optical depth effects , heating radiation too soft to ionize carbon , or the influence of an active galactic nucleus ( malhotra et al .
1997 and 2000a , luhman et al .
1998 , genzel & csarsky 2000 ) . the currently prevailing opinion , however , points to decreased efficiency of photoelectric heating due to positively charged grains . in the more actively star - forming and warmer galaxies ,
the far - ultraviolet flux is higher
. a higher ratio of far - ultraviolet flux to gas density , @xmath7 , leads to more positively charged grains and therefore to lower photoelectric efficiency .
it is possible that the decrease in [ cii]/fir , instead of indicating a decrease in the heating efficiency , might result from the gas cooling via the other principal channel , [ oi ] 63 @xmath1 m .
however , the combined flux in the cooling lines normalized to the far - infrared emission ( [ cii]+[oi])/fir still shows a clear trend towards smaller values for the more actively star - forming galaxies . [ oi ] 63 @xmath0 does become the dominant coolant for galaxies showing warmer dust temperatures , but this relative rise does not compensate for the decrease in [ cii ] relative to fir .
the `` deficiency '' seen in [ cii ] is defined by comparison to the far - infrared emission from dust grains . far - infrared emission measures the radiation emitted by large or `` classical '' grains that are in thermal equilibrium , maintaining a nearly constant temperature .
if photoelectric heating is dominated by small grains , a more appropriate normalization factor for [ cii ] line should be the emission from these grains ( helou 1999 ) .
the smallest grains briefly reach very high effective temperatures with the absorption of a single photon and therefore emit in the mid - infrared . in the pah picture , this can be viewed as a fluorescence phenomenon from a molecule at high excitation temperature . in any case , because of the invariant afe spectrum , the mid - infrared flux tracks well the total heating of this grain population .
figure 1 ( top panel ) shows the [ cii ] line strength normalized to the afe flux plotted against r(60/100 ) .
the afe flux is derived from iso - cam observations with broadband filter at 6.75 @xmath0 ( dale et al . 2000 ) , which is very well matched to measuring the afe .
the observed flux is scaled up by a factor of 1.17 to recover the integrated 510 @xmath0 flux .
the scaling factor derives from the properties of the iso - cam 6.75 @xmath0 filter and the average mid - infrared spectrum of helou et al .
we see not only a lack of any trend of decreasing [ cii]/afe with r(60/100 ) , but the scatter in the [ cii]/afe values is about half the dispersion in [ cii]/fir ( comparing top and bottom panels of figure 1 ) .
this suggests that there is indeed a special connection between mid - infrared emission and [ cii ] emission .
the measured line fluxes have 30% uncertainty which contributes substantially to the observed scatter of @xmath8 dex ( @xmath9 50% ) .
the 510 @xmath0 flux attributed to afe also includes some contribution from stellar light which is not significant except in early type galaxies ( malhotra et al .
the mid - infrared emission is from grains heated by far - ultraviolet as well as optical photons ( uchida , sellgren & werner 1998 ) , whereas only far - ultraviolet photons at energies typically greater than 6 ev are effective in heating the gas .
thus the variation of the hardness of the radiation field also introduces some scatter in the observed [ cii]/afe value .
some fraction of [ cii ] flux also arises in ionized regions where the photoelectric effect is not the dominant heating mechanism ( petuchowski & bennett 1993 , heiles 1994 , malhotra et al .
2000a ) ; extreme cases of this offset are discussed further below .
with all these terms , it is remarkable that the dispersion in [ cii]/afe is not greater than measured .
the lowest value of [ cii]/afe in figure 1 represents ngc 4418 and is more than 5@xmath10 away from the mean . as discussed by lu et al .
( 2000 ) , the mid - infrared spectrum of ngc 4418 is not typical of star - forming galaxies .
it shows none of the familiar aromatic features but instead has a very broad plateau between 6 and 8.5 @xmath0 .
this may be a partial view of the continuum along with the 9.7 @xmath0 silicate absorption feature . in any case the imaging with the iso - cam filter at 6.75 @xmath0 does not measure aromatic feature emission for ngc 4418 .
the upper limit in [ cii]/afe in figure 1 represents ic 860 , from which [ cii ] was not detected . the mid - infrared spectrum of ic 860 appears similar to that of ngc 4418 , but has a very low signal - to - noise ratio ( lu et al .
both galaxies are oh mega - masers .
two of the three highest values of [ cii]/afe in figure 1 belong to galaxies that have [ nii]/fir a factor of ten and 4@xmath10 greater than the mean value .
as discussed by malhotra et al .
( 2000a ) , this indicates a larger than average contribution from regions to the [ cii ] line emission from these galaxies , which appears to cause somewhat elevated [ cii]/afe . since [ oi](63 @xmath0 ) is also an important coolant , a more appropriate quantity to plot might be the total gas cooling with respect to the afe flux , i.e. ( [ cii]+[oi])/afe . in figure 2 this quantity is plotted against r(60/100 ) , showing a larger scatter in the ( [ cii]+[oi])/afe , but essentially the same behavior as the ratio [ cii]/afe up to r(60/100)@xmath12 . for greater values , the scatter increases quickly as does the mean ratio .
this is consistent with [ cii ] being the dominant cooling line for all but the warmest galaxies ( malhotra et al .
2000a ) . the energy level and critical density for [ oi ] ( 63 @xmath0 ) is higher than for [ cii ] , so this line is expected to become more important relative to [ cii ] in warmer and dense gas , which is associated with greater values of @xmath7 and r(60/100 ) in photodissociation region models .
the empirical results above strongly suggest that the smallest particles which produce the mid - infrared emission are a key contributor to the photoelectric effect in photodissociation regions .
theoretical studies of the orion photodissociaton region where the values of @xmath6 and @xmath13 are elevated and the ratio @xmath7 is also moderately high , show that about half the photoelectric heating is due to particles smaller than 15 (bakes & tielens 1994 ) .
the observed constancy of the [ cii]/afe ratio , in a sample where the [ cii]/fir ratio shows larger systematic variations , indicates that the role of small grains in heating the gas may be even more prominent .
the interpretation of this empirical picture however is hindered by serious uncertainties in the properties of the systems involved , as detailed below .
we present below ( [ sec : scenarios ] ) two possible scenarios to account for the empirical facts , without favoring either .
the empirical evidence can be summarized in four items : [ oi]/fir has a larger dispersion for r(60/100)@xmath14 , @xmath17 dex , and shows no convincing evidence of falling with r(60/100 ) ; there are few data points at larger r(60/100 ) values , but the dispersion appears to increase quickly . the exact composition of grains responsible for the spectral features in the mid - infrared is uncertain , but these features correspond to bending and stretching modes of c - c and c - h bonds in aromatic hydrocarbons ( duley & williams 1981 ) .
the afe have often been attributed to pahs ( lger & puget 1984 ; allamandola , tielens & barker 1985 ) ; we will adopt here the notation pah to refer to the carriers of the afe .
it is also uncertain in what proportion ionized and neutral pah might contribute to the emission , especially in light of recent observations that show little change in mid - infrared spectral shapes as the far - ultraviolet flux ranges over @xmath18 ( uchida et al .
these data appear to contradict laboratory studies which show significant changes in relative strengths of different features from ionized to neutral pahs , with the ionized pahs much stronger in the 5 to 9 @xmath0 range . in view of these uncertainties ,
we can not unambiguously associate afe with a specific stage of the photoelectric process , even though the evidence in figure 1 suggests that the afe closely track the rate of photoelectron generation into the interstellar medium .
for instance , figure 1 might be easily understood if afe derive predominantly from pahs that have just been ionized and are decaying from an excited state , or predominantly from recombining pahs , or from both in a roughly constant proportion .
however , no such association can be established , and moreover excitation by non - ionizing photons can not be ruled out ( uchida et al .
furthermore , there is currently no way to ascertain or constrain the ionization balance of the carrier population .
additional uncertainty arises from the significant variations known to occur in the concentration of afe carriers relative to large grains .
boulanger et al .
( 1990 ) studied the relative distribution of 12 and 100 @xmath0 emission in molecular clouds and concluded that the variations of r(12/100 ) required the abundance of transiently heated grains , which are mostly pahs , to vary relative to the large grain population .
in particular , they observed a systematic decrease of r(12/100 ) as the 100 @xmath0 brightness increased , which appears as limb brightening of clouds in afe with a greater contrast than explained by optical depth effects .
they proposed that this abundance variation is related to the cycling of interstellar matter between gas phase and grain surfaces . on the other hand ,
the afe contribution to the total dust luminosity from galaxies decreases as the heating increases ( helou , ryter & soifer 1992 ; lu et al .
this trend has been related to the destruction of afe carriers in the most intense heating environments near and in regions , a claim that is supported by observations of the milky way and the small magellanic cloud ( tran 1999 , contursi et al .
2000 ) , but may be linked to the effect observed by boulanger et al .
these depletion trends must then be superposed on the trends of photoelectric yield with interstellar medium parameters , resulting in the empirical evidence summarized above and in figure 1 .
the close association of [ cii ] with afe is easy to interpret at low values of r(60/100 ) , by assuming that [ cii ] represents the cooling of the gas , and afe measures the heating by photoelectrons .
however , at larger values of r(60/100 ) , reflecting an enhanced @xmath7 ( malhotra et al .
2000a ) , this association becomes puzzling since [ oi ] overtakes [ cii ] in cooling luminosity , casting doubt on the argument that [ cii]/afe is constant because it is intimately tied to the photoelectric effect . as the heating becomes more intense , the ratio of ( [ cii]+[oi])/afe rises , suggesting an apparent increase in the photoelectric efficiency of the pahs
; this however is in direct conflict with the physics , leading us to the conclusion that pahs become less significant contributors to the total photoelectric heating compared with other grain populations .
this decreased significance is consistent with the decreased energetic importance of pahs , manifested as the drop in afe / fir with increasing r(60/100 ) ( helou , ryter & soifer 1991 ) . in this picture of decreasing significance of [ cii ] as a cooling line and of afe as a dust radiator , one would have to invoke a coincidental agreement between these two rates of decrease in order to explain the unchanged ratio of [ cii]/afe .
this coincidence may reflect a certain scaling of @xmath6 with @xmath13 , similar to the @xmath19 derived by malhotra et al .
( 2000a ) .
such a scaling might result from a feedback mechanism linking the density of the medium and the intensity of the star formation , turning the coincidence into a causal connection . in any case , it is not clear what drives the systematic decrease in afe / fir , though it has been suggested that a more intense radiation field may be inducing greater ionization or de - hydrogenation of the pahs ( tielens et al .
1999 ) .
an alternate scenario to explain the observations would be a two - component model broadly similar to the helou ( 1986 ) decomposition of the far - infrared emission into an `` active '' star formation component and a `` quiescent '' or cirrus component .
the cirrus - like component would dominate at low values of r(60/100 ) , and therefore be characterized by high values of [ cii]/[oi ] , afe / fir and [ cii]/fir , just as one would expect from a photodissociation region with low density and @xmath7 . on the other hand ,
the high - density photodissociation regions at the surface of molecular clouds directly illuminated by young massive stars would provide the active component , with elevated values of @xmath20 and @xmath7 , resulting in more cooling via [ oi ] than [ cii ] , and less efficient gas heating due to grain charging . in this scenario
, one would have to assume that the afe emission from the active component is depressed compared to the diffuse component , a property marginally suggested by the observations of lower afe / fir ratios in denser molecular clouds ( boulanger et al . 1996 ) . a significant contribution to the active component by regions
would also help explain the depressed values of afe / fir .
the observed trends as a function of r(60/100 ) are then interpreted as a result of a varying proportion between the two components , with the interstellar medium falling almost bimodally into these two components , and the properties of each component varying little among galaxies . while either one of these scenarios is plausible , each requires a number of assumptions which are difficult to verify with currently available data .
this however illustrates the richness of the data at hand , and the need for a more sophisticated treatment of the interstellar medium on the scale of galaxies .
we would like to thank f. boulanger & j .- l .
puget for stimulating discussions and j. brauher for help with data processing .
this work was supported by data analysis funding from nasa , and carried out at ipac / jpl of the california institute of technology .
is an esa project with instruments funded by esa member states ( especially the pi countries : france , germany , the netherlands and the united kingdom ) , and with the participation of isas and nasa .
sm s research funding is provided by nasa through hubble fellowship grant # hf-01111.01 - 98a from the space telescope science institute , which is operated by aura under nasa contract nas5 - 26555 .
this research has made use of the nasa / ipac extragalactic database which is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa .
allamandola , l.j . ,
tielens , a.g.g.m . ,
barker , j.r .
1985 , , 290 , l25 andriesse , c.d . 1978 , , 66 , 169 bakes , e. & tielens , a. 1994 , , 427 , 822 beichman , c.a .
1987 , , 25 , 521 boulanger , f. , falgarone , e. , puget , j.l .
& helou , g. 1990 , , 364 , 136 boulanger , f. , reach , w.t . , abergel , a. et al .
1996 , , 315 , l325 contursi , a. , lequeux , j. , cesarsky , c. et al .
2000 , , in press , astro - ph/0006185 dale , d.a . , silbermann , n.a . , helou , g. et al . 2000 , , 120 , 583 de jong , t. 1977 , , 55 , 137 draine , b.t .
1978 , , 36 , 595 duley , d.w .
& williams , d.a .
1981 , , 196 , 269 genzel , r. & cesarsky , c. 2000 , , ( in press ) heiles , c. 1994 , , 436 , 720 helou , g. 1986 , , 311 , l33 helou , g. , ryter , c. & soifer , b.t .
1991 , , 376 , 505 helou , g. 1999 , in `` the universe as seen by iso '' , esa - sp 427 , eds .
p. cox & m. f. kessler ( noordwijk : esa ) , 797 helou , g. , lu , n.y . , werner , m.w . , malhotra , s. & silbermann , n.a .
2000 , , 532 , l21 hollenbach , d.j .
& tielens , a.g.g.m .
1999 , rev .
, 71 , 173 dhendecourt , l.b . & lger , a. 1987 , , 180 , l9 jura , m. 1976 , , 204 , 12 lger , a. & puget , j.l .
1984 , , 137 , l5 lepp , s. & dalgarno , a. 1988 , , 335 , 769 lu , n.y .
2000 , in preparation luhman , m.l . ,
satyapal , s. , fischer , j. , wolfire , m.g . , cox , p. , lord , s.d . ,
smith , h.a . ,
stacey , g.j .
& unger , s.j . 1998 , , 504 , l11 malhotra , s. , helou , g. , stacey , g. et al .
1997 , , 491 , l27 malhotra , s. , kaufman , m. , hollenbach , d. et al . 2000a , submitted to malhotra , s. , hollenbach , d. , helou , g. et al . 2000b , , in press petuchowski , s.j .
& bennett , c.l .
1993 , , 405 , 591 puget , j.l . & lger , a. 1989 , , 27 , 161 sellgren , k. 1984 , , 277 , 623 tielens , a.g.g.m . ,
hony , s. , van kerckhoven , c. & peeters , e. 1999 , in `` the universe as seen by iso '' , esa - sp 427 , eds .
p. cox & m. f. kessler ( noordwijk : esa ) , 579 tran , d. 1998 , ph.d .
thesis , universit de paris uchida , k.i . ,
sellgren , k. & werner , m.w .
1998 , , 493 , l109 uchida , k.i . ,
sellgren , k. , werner , m.w .
& houdashelt , m.l . 2000 , , 530 , 817 watson , w.d .
1972 , , 176 , 103 | we report a strong correlation between the [ cii ] 158 @xmath0 cooling line and the mid - infrared flux in the 5 - 10 @xmath0 range in a wide variety of star - forming galaxies .
the mid - infrared flux is dominated by aromatic feature emission ( afe ) , which is thought to arise from large polycyclic aromatic hydrocarbon molecules or `` pahs '' and is generally associated with the smallest interstellar grains . the [ cii ] line is the dominant gas coolant in most regions of atomic interstellar gas , and therefore reflects the heating input to the gas .
the ratio of these two quantities , [ cii]/afe , remains nearly constant around 1.5% against variations in the ratio of the 60 @xmath1 m band flux to the 100 @xmath1 m band flux , r(60/100 ) .
this is in contrast to the drop in the [ cii]/fir ratio with increasing r(60/100 ) , which signals higher dust temperatures and more intense radiation fields .
we interpret the stable [ cii]/afe ratio as evidence that gas heating is dominated by the pahs or small grains , which are also afe carriers , over a wide range of conditions .
the trend of decreasing [ cii]/fir and afe / fir with increasing radiation field suggests a decrease in the importance of pahs or small grains relative to large grains both in gas heating and in dust cooling .
we summarize the observed trends and suggest two plausible scenarios .
0g_0 2ciil_[cii ] ratl_[cii]/l_tir ff_7 m ff_(60 ) 100f_(100 ) # 1 |
the conventional theory for the origin of disk galaxies in cold dark matter ( cdm ) cosmologies involves the dissipational collapse of gas inside relaxed dark matter halos formed through hierarchical clustering @xcite .
this theory endows forming disk galaxies with kinematic properties inherited from the halos that host them @xcite .
early simulations of cosmological structure formation @xcite verified that during linear growth , the angular momentum of halos grows according to perturbation theory @xcite , and by incorporating these results subsequent work explained the flat rotation curves and sizes of disk galaxies through the combination of gas dissipation , halo spin , and the adiabatic response of dark matter to baryonic infall @xcite .
the initial success of the dissipational collapse model led to its establishment as the favored theory for the formation of disk galaxies .
however , numerical modeling revealed that the theory was incomplete .
while smooth , dissipational collapse may be relevant to disk galaxy formation in quiet environments , simulations and semi - analytic modeling have demonstrated that it is rare for halos to acquire most of their angular momentum via the quiescent accretion of tidally torqued material @xcite .
simulation and observational work has shown that the specific angular momentum content of dark matter halos differs from exponential disks @xcite as halos contain comparatively more low angular momentum material that might produce bulges or overly centrally concentrated disks @xcite .
furthermore , simulations of mass accretion onto galaxies indicate that the hierarchical nature of the @xmath0cdm cosmology leads to the destruction of disks when dissipative effects are neglected @xcite
. models of galaxy collisions including dissipation @xcite show that gas can loose angular momentum owing to gravitational torques during these events . in cases when the interstellar medium ( ism ) is isothermal and relatively cold and the gas fraction of the galaxies is small ( @xmath3 of the baryons ) , the subsequent inflow of gas into the centers of the merger remnants forms a roughly spherical stellar distribution through a luminous starburst @xcite , leaving objects that have essentially no extended stellar disks .
cosmological simulations of disk formation with similar physics produce galaxies that are too centrally concentrated , form too many stars , contain overly dominant bulges , and lack angular momentum compared with nearby spirals @xcite .
efforts to resolve these discrepancies in a cosmological context have mainly relied on increasing the impact of feedback from star formation @xcite .
cosmological simulations that utilize strong feedback from star formation in the form of a pressurized ism to produce disk galaxies either produce too few exponential disks compared with observations @xcite or yield galaxies that still contain a significant bulge component @xcite .
while these recent simulations have had some success , taken together the results suggest that the theory of structure formation remains faced with an apparent incompatibility between models of disk formation and the hierarchical growth of structure .
to help resolve this incompatibility , we propose a new , merger - driven theory for the formation of disk galaxies at high redshifts that supplements standard dissipational collapse . in our scenario ,
gas - dominated mergers , where the interacting systems have a gas fraction @xmath4 at the height of the merger , allow for the formation of rotationally - supported disks in remnants if energetic feedback mechanisms limit the conversion of gas into stars .
gas pressurization from a multiphase ism by feedback from star formation @xcite stabilizes gas disks and allows mergers between gas - dominated galaxies to produce large , smoothly distributed , rapidly rotating stellar disks .
our proposed scenario builds upon the results of previous simulations of remnant disk formation in mergers by @xcite and @xcite . the merger - driven scenario for disk formation transitions naturally into the merger - driven scenario for elliptical galaxy formation ( e.g. * ? ? ?
* ) as star formation in galaxies over time reduces the gas content of progenitors and prevents the further occurrence of mergers involving gas - dominated systems .
mergers between gas - rich disk galaxies at intermediate redshifts ( @xmath5 ) are still expected to produce realistic elliptical galaxies nearly universally ( see e.g. * ? ? ?
* ) , as prior quiescent star formation and interactions will decrease their gas content below the necessary @xmath6 needed to produce a disk - dominated remnant .
below we describe the simulation methodology used to demonstrate the merger - driven formation of disk galaxies ( [ section : methodology ] ) , including our methods for studying the effects of feedback ( [ subsection : methodology : feedback ] ) , orbital geometry ( [ subsection : methodology : orbits ] ) , gas fraction ( [ subsection : methodology : gas_fraction ] ) , merger mass - ratio ( [ subsection : methodology : minor_mergers ] ) , and progenitor mass ( [ subsection : methodology : small_mergers ] ) on remnant properties .
our method for analyzing the structural and kinematic properties of the remnants is detailed in [ subsection : methodology : analysis ] and
the results of the numerical experiments are presented in [ section : results ] .
we discuss and summarize our results in [ section : discussion ] and [ section : summary ] .
to demonstrate the feasibility of our scenario for merger - driven disk formation , we have performed a suite of merger simulations of isolated , gas - rich disk galaxies , with and without feedback from accretion onto supermassive black holes ( bhs ) .
the simulations use smoothed particle hydrodynamics ( sph ) @xcite to evolve the gas and incorporate a multiphase ism model @xcite where we control the pressurization @xmath7 of star - forming gas by interpolating between isothermal gas ( @xmath8 ) with a temperature @xmath9 k and a stiffer multiphase medium ( @xmath10 ) with a higher effective temperature @xmath11 k by varying the equation of state ( eos ; for details see * ? ? ?
star formation in the dense ism ( @xmath12 ) occurs on a timescale chosen to match observations @xcite .
bh growth is modeled @xcite by spherical accretion onto bh `` sink '' particles , with a fraction @xmath13 of the released gravitational potential energy of the accreted gas being deposited as thermal feedback into the gas surrounding the bh . for the simulations and accretion model considered here , @xmath14 , where @xmath15 is the fraction of rest mass energy radiated during the growth of the bh ( fixed by standard accretion theory ) and the thermal coupling efficiency @xmath16 is chosen to reproduce the local @xmath17 relation @xcite .
we consider two merger progenitors , a @xmath18 km s@xmath19 system @xcite with a rotation curve and virial mass @xmath20 similar to the milky way and a smaller , @xmath21 km s@xmath19 system with virial mass @xmath22 .
the galaxies are comprised of exponential gas and stellar disks embedded in dark matter halos with a @xcite profile having scale lengths corresponding to a navarro - frenk - white @xcite concentration of @xmath23 .
the progenitors each contain @xmath24 gas , @xmath25 stellar , and @xmath26 dark matter halo particles , and may also include a bh seed of @xmath27 that is allowed to accrete in the manner described above .
once the bh particles fall within our force resolution limit , the sink particles combine into a single supermassive bh . to characterize the conditions under which our proposed scenario for merger - driven disk formation is most successful , we simulate a variety of merger scenarios varying the ism pressurization ( @xmath7 ) , gas fraction ( @xmath28 ) , merger mass - ratio , disk orientations , and orbits .
the complete suite of simulations is described below , and a summary is provided in table [ table : models ] . for convenience
we will refer to each model with a two letter designation , with the first letter signifying the gas fraction @xmath28 and the second letter noting the ism pressurization @xmath7 .
when necessary , we add suffixes to describe e.g. orbital or merger mass - ratio variations .
the simulation designations are detailed in the following sections .
lcccccccccccc ga & 160 & 160 & 0.99 & 0.1 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + gb & 160 & 160 & 0.99 & 0.25 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + gc & 160 & 160 & 0.99 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + gd & 160 & 160 & 0.99 & 0.75 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + ge & 160 & 160 & 0.99 & 1.0 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + gan & 160 & 160 & 0.99 & 0.1 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & n + gbn & 160 & 160 & 0.99 & 0.25 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & n + gcn & 160 & 160 & 0.99 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & n + gdn & 160 & 160 & 0.99 & 0.75 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & n + gen & 160 & 160 & 0.99 & 1.0 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & n + gcra & 160 & 160 & 0.99 & 0.5 & 2.7 & 70.0 & 0 & 0 & 0 & 0 & y + gcrb & 160 & 160 & 0.99 & 0.5 & 16.0 & 70.0 & 0 & 0 & 0 & 0 & y + gcrc & 160 & 160 & 0.99 & 0.5 & 32.0 & 70.0 & 0 & 0 & 0 & 0 & y + gcrd & 160 & 160 & 0.99 & 0.5 & 48.0 & 140.0 & 0 & 0 & 0 & 0 & y + gcoc & 160 & 160 & 0.99 & 0.5 & 6.0 & 70.0 & 180 & 0 & 180 & 0 & y + gcoe & 160 & 160 & 0.99 & 0.5 & 6.0 & 70.0 & 30 & 60 & -30 & 45 & y + gcof & 160 & 160 & 0.99 & 0.5 & 6.0 & 70.0 & 60 & 60 & 150 & 0 & y + gcoo & 160 & 160 & 0.99 & 0.5 & 6.0 & 70.0 & -109 & 30 & 71 & -30 & y + gcop & 160 & 160 & 0.99 & 0.5 & 6.0 & 70.0 & -109 & 90 & 180 & 0 & y + dc & 160 & 160 & 0.4 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + ec & 160 & 160 & 0.6 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + fc & 160 & 160 & 0.8 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + dcm & 160 & 80 & 0.4 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + ecm & 160 & 80 & 0.6 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + fcm & 160 & 80 & 0.8 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + dcs & 80 & 80 & 0.4 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + ecs & 80 & 80 & 0.6 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + fcs & 80 & 80 & 0.8 & 0.5 & 6.0 & 70.0 & 0 & 0 & 0 & 0 & y + to study the impact of energetic feedback mechanisms on the formation of disks in gas - rich mergers , we perform 12 simulations of major mergers between @xmath18 km s@xmath19 galaxies with disks consisting almost entirely of gas ( @xmath29 , g " runs ) .
the ism pressurization is incrementally increased over the values @xmath30 , @xmath31 , @xmath32 , @xmath33 , and @xmath34 , stiffening the equation of state of the ism and increasing the sound speed of the gas at a given density and temperature .
the simulations are referred to as models ga - ge in table [ table : models ] .
we perform each simulation twice , once with and once without the inclusion of growing supermassive bhs in the progenitor systems .
the models without bhs are listed with the suffix n " in table [ table : models ] , and the model gcn ( @xmath29 , @xmath35 ) corresponds to the simulation performed by @xcite .
the galaxies merge on coplanar , prograde orbits with an initial separation @xmath36 kpc and pericentric distance @xmath37 kpc .
the orbit for the feedback study runs was chosen to reproduce the encounter simulated by @xcite . by using the same orbit for these simulations ,
we explore the importance of feedback for the formation of disks in mergers separate from the role of angular momentum .
we explore the role of angular momentum in producing remnant disks in mergers through additional simulations described in
[ subsection : methodology : orbits ] .
the original orbital and disk angular momentum of the progenitors will affect the ability of disks to form from mergers as angular momentum present in residual gas left over from the merger can lead to rapidly rotating structures in the remnant .
the formation of remnant stellar disks have been demonstrated in interactions with prograde , coplanar orbits @xcite , and in simulations without star formation gaseous remnant disks have been shown to form in mergers with both polar and inclined orbits @xcite . to judge the importance of orbital and disk angular momentum on a merger - driven scenario for disk galaxy formation , we perform a study of 9 additional simulations . in four models ,
we vary the pericentric passage distance @xmath38 of the merger to alter the orbital angular momentum of the interaction . typically , increasing @xmath38 will increase the angular momentum of the orbit and lead to larger , more rotationally - supported disks .
however , mergers with larger @xmath38 undergo inefficient dynamical friction , avoid strong angular momentum loss during the first passage , and therefore take longer to merge
. quiescent star formation will reduce the effective gas fraction of the progenitors before the disks merge in these cases , and may lead to larger stellar spheroids .
these two competing effects will be present in cosmological environs to some degree , and for this reason we do not attempt to adjust for the decrease in gas fraction in wide - orbit mergers .
in addition to a coplanar orbit with @xmath37 kpc , which is approximately twice the progenitor disk scale length @xmath39 , we simulate one model with a smaller pericentric passage distance ( @xmath40 , model gcra in table [ table : models ] ) and three with larger pericentric passage distances ( @xmath41 , @xmath42 , and @xmath43 , models gcrb - gcrd in table [ table : models ] ) . in the remaining five models of the orbital study , we vary the orientation of the progenitor disks . while varying the pericentric passage distance increases the orbital angular momentum , much of this angular momentum must be lost to the dark matter halo through dynamical friction before the disks can merge .
the original disk angular momenta of the progenitors may then substantially add to the angular momentum of the remnant disk .
in addition to the coplanar orbit considered by @xcite , we select five additional disk orientations from the @xcite study of equal - mass mergers of stellar disk galaxies .
these orientations range from retrograde - retrograde coplanar ( @xmath44 , @xmath45 , @xmath46 , @xmath47 , model gcoc ) to prograde - retrograde polar ( @xmath48 , @xmath49 , @xmath50 , @xmath47 , model gcof ) , with a complete list provided in table [ table : models ] .
the results of these simulations are presented in
[ subsection : results : orbits ] the production of remnant disks in mergers will depend on the gas fraction @xmath28 of the progenitor disk systems .
stellar disks present before the merger will be substantially heated during the merger and form stellar spheroids .
only galaxies with significant gas fractions before the merger will be able to re - form gaseous disks after the galaxies collide . to characterize the progenitor gas fraction necessary to form a substantial remnant disk
, we perform three simulations with growing bhs , ism pressurization @xmath35 and gas fractions @xmath51 ,
@xmath52 , and @xmath53 ( models dc , ec , and fc in table [ table : models ] ) .
the results of these simulations are presented in
[ subsection : results : gas_fraction ] in @xmath54 cosmologies , the frequency of galaxy mergers typically increases with the mass ratio of the merging pair ( see e.g. * ? ? ? * ) . as mentioned above ,
previous theoretical studies have suggested that dissipationless interactions between satellites and disk galaxies can destroy disks even if the satellite mass is relatively small ( @xmath55 a few percent of the disk mass , e.g. * ? ? ?
minor mergers with progenitor mass ratios of @xmath56 have been shown to produce small gaseous disks in simulations without star formation @xcite . in simulations including gas dissipation and star formation ,
minor mergers have been shown to induce starbursts ( e.g * ? ? ?
cosmological simulations suggest that some minor mergers may also add mass to the thick component of disk galaxy systems @xcite .
minor mergers may then greatly influence the formation and survival of disk galaxies even though the mass accretion rate from minor mergers is typically smaller than that from major mergers . to address the relevance of minor mergers for remnant disks , we rerun the gas fraction study simulations from [ subsection : methodology : gas_fraction ] substituting a @xmath57 km s@xmath19 galaxy for the second progenitor
. these mergers with mass ratios of @xmath58 will help characterize the influence of minor mergers on remnant disk properties and provide examples of models that approximate cosmologically frequent merger events .
the results of these simulations are presented in
[ subsection : results : minor_mergers ] .
the scale dependent physics of gas dissipation and star formation may influence the ability of remnant disks to form in mergers .
gas cooling is typically more efficient in smaller mass halos as the virial temperature of those systems approach the regime where hydrogen and helium recombination and collisional ionization enables large cooling rates ( see e.g. * ? ? ?
gas densities in the disks of low mass systems are typically lower than in more massive systems , leading to longer gas consumption timescales .
quiescent star formation in low mass systems during the phases of the merger when the separation between disks is large and gravitational torques are weak will be lower than for high mass systems , which may correspond to a larger gas fraction in the progenitor disks when the systems actually merge @xcite . to gauge the net impact of these effects on the formation of remnant disks in mergers of different mass scales ,
we perform three additional simulations of equal mass mergers between @xmath21 km s@xmath19 galaxies wit gas fractions of @xmath59 , @xmath52 , and @xmath53 for the progenitor systems ( models dcs - fcs in table [ table : models ] ) .
the results of these simulations are presented in
[ subsection : results : small_mergers ] each simulation is evolved over the timescale needed for the progenitor systems to merge , corresponding to @xmath60 gyr depending on the orbit . to illustrate the structure of the merging system during the collision , figure [ fig : time_sequence ]
shows the time evolution of the gas in mergers with a strongly pressurized ( @xmath61 ) ism with ( model ge , bottom panel ) and without ( model gen , top panel ) feedback from accretion onto supermassive bhs .
each panel shows a @xmath62 kpc @xmath63 @xmath62 kpc square area around the center of mass of the interacting system . at early times in the merger ( @xmath64 gyr , far - left panels ) ,
tidal interactions between the galaxies distort them , but as significant accretion onto the bhs has yet to occur , the overall gas distribution remains similar between the simulations .
as the gas in the simulation with bhs ( bottom panels ) begins to experience thermal feedback from bh accretion , a hot halo of diffuse , low - angular momentum gas expelled through a wind surrounds the interacting galaxies . by the end of the simulations at @xmath65 gyr ( far - right panels ) , rotating gas disks remain , eventually forming stellar disks .
the remnant containing a bh has cleared out some gas from the inner regions of the disk , reducing the central stellar density and producing a diffuse hot gas envelope . after the merger has completed , the structural and kinematic properties of the remnant are measured from the simulation data .
the structural components of the merger remnants , which might include a bulge , a thin or thick disk , or an extended spheroid , are comprised of stars that form at characteristically different times during the interaction .
the dynamically hot components of the remnant typically form before the final coalescence of the merger , while the stars in the colder components of the remnant mostly form after the height of the merger when the role of tidal forces has decreased .
the star formation rate ( sfr ) of the merging system ( see figure [ fig : analysis ] , upper left panel ) can be used define selection criteria for stars likely to populate separate dynamical components of the remnants based on their formation age .
the peak in the sfr that occurs during the height of the merger can be approximated by an exponential rise and decay . in
what follows , we identify `` old '' stars in the remnants as stellar particles forming earlier than before three @xmath66-folding times before the center of the sfr peak and `` new '' stars as those forming later than three @xmath66-folding times after the peak . the old and new stars in the remnant are then used to measure the separate kinematics of the dynamically hot and cold stellar structures of the remnant . typically , the old stars in the remnant have relatively low rotational velocities ( figure [ fig : analysis ] , lower left panel ) and are not rotationally supported with the ratio of rotational velocity to velocity dispersion @xmath67 ( figure [ fig : analysis ] , lower right panel ) .
new stars formed in the remnant after the height of the merger typically have high rotational velocities and are rotationally supported .
stars formed during the burst can have kinematic properties intermediate between the old and new stars , and therefore are not analyzed as a separate stellar component . the mass of the new stellar component forming in the remnant disk can be comparable or large than the old , dynamically heated bulge or extended spheroid . the mass of the stellar structures in the remnant can be measured from the surface mass density profile . after the merger the stellar angular momentum of the remnant is measured and used to define a possible `` disk''-plane perpendicular to the rotation axis of the remnant .
the plane defined normal to the stellar angular momentum vector captures the mean stellar rotation in the remnants we study , and only underestimates rotation when rotationally - supported stellar structures are misaligned with the total stellar angular momentum .
the stellar surface mass density of each remnant is then measured in logarithmically - spaced annuli and is modeled by a multi - component system ( figure [ fig : analysis ] , upper right panel ) as a function of radius @xmath68 that may include an exponential bulge @xmath69 an exponential thin disk @xmath70 an exponential thick disk @xmath71 and an extended stellar spheroid with a @xcite profile @xmath72}\ ] ] each remnant is fitted with individual thin disk , bulge thin disk , bulge thin disk thick disk , bulge thin disk spheroid , and bulge thin disk thick disk spheroid models with the requirements that the bulge scale length @xmath73 kpc , the thick disk scale length @xmath74 is larger than the thin disk scale length @xmath39 , and the spheroid scale length @xmath75 .
a simple @xmath76 estimate of the success of each model in reproducing the surface mass density profile of the remnant is used to determine the relative mass in bulge , disk , and spheroid components . when the model fits suggest that either a thick disk or an extended spheroid models provide comparable fits to the projected stellar distribution , the rotational support @xmath77 of the stars at large radii
is used to select a preferred model . in this case ,
dispersion supported systems ( @xmath67 ) are assigned extended spheroids and rotationally - supported systems ( @xmath78 ) are assigned thick disk components .
the best - fit surface mass density models are reported in table [ table : remnants ] , including the remnant bulge - disk - spheroid mass ratios . for comparison ,
table [ table : remnants ] also lists the best fit surface mass density parameters for a @xmath18 km s@xmath19 progenitor ( model ics ) with a gas fraction of @xmath59 measured using the same method .
the suite of simulations provides an extensive set of remnants that characterize the ability of extremely gas - rich mergers to produce both disk components in spheroid - dominated systems and disk - dominated systems whose structure and kinematics closely resemble large spiral galaxies . in _ nearly every _ gas - rich merger we simulated ( save model dcs , see below ) the new stars formed after the peak of the sfr produced a rotationally - supported structure .
these remnant disks are ubiquitous in gas - rich mergers , but the extent to which they dominate the stellar content and the degree of rotational support shows trends with the strength of pressurization in the ism , the presence of growing supermassive bhs , progenitor gas fraction , orbit and disk orientation , and progenitor mass ratio .
below , we consider the impact of these merger properties in turn .
lccccccccccccc ics & & & 4.9e8 & 2.73 & & & & & & 2.3e10 & & & 0.0:1.0:0.0 + ga & 6.9e10 & 0.24 & 2.1e8 & 3.50 & 8.3e7 & 4.10 & & & 2.6e10 & 1.6e10 & 3.1e10 & & 1.6:1.0:1.9 + gb & 3.9e10 & 0.41 & 2.1e8 & 4.48 & & & & & 4.2e10 & 2.6e10 & & & 1.5:1.0:0.0 + gc & 2.0e11 & 0.17 & 1.6e9 & 1.55 & & & 3.3e7 & 5.19 & 3.8e10 & 2.5e10 & & 5.7e9 & 1.2:1.0:0.0 + gd & 7.3e10 & 0.18 & 1.9e9 & 1.42 & 8.0e7 & 2.89 & & & 1.6e10 & 2.4e10 & 1.5e10 & & 1.0:1.6:1.0 + ge & 1.3e11 & 0.18 & 5.9e9 & 0.65 & & & 5.2e8 & 2.26 & 2.7e10 & 1.6e10 & & 1.6e10 & 1.0:1.2:0.0 + gan & 4.3e10 & 0.28 & & & 1.6e8 & 3.88 & & & 2.1e10 & & 5.6e10 & & 1.0:0.0:2.5 + gbn & 5.0e10 & 0.28 & 8.2e7 & 4.45 & 3.7e8 & 1.97 & & & 2.6e10 & 1.0e10 & 3.3e10 & & 2.5:1.0:3.2 + gcn & 1.8e11 & 0.18 & 1.3e9 & 1.61 & 1.8e7 & 4.15 & & & 4.1e10 & 2.1e10 & 7.3e9 & & 5.6:2.8:1.0 + gdn & 1.9e11 & 0.18 & 1.2e9 & 1.48 & & & 1.4e8 & 3.21 & 4.1e10 & 1.6e10 & & 9.6e9 & 1.5:1.0:0.0 + gen & 2.7e11 & 0.12 & 1.3e10 & 0.45 & & & 8.5e8 & 2.02 & 2.8e10 & 1.7e10 & & 2.1e10 & 1.0:1.4:0.0 + gcoc & 7.0e10 & 0.25 & 1.5e9 & 1.78 & & & 3.0e7 & 5.39 & 2.7e10 & 3.0e10 & & 5.5e9 & 1.0:1.3:0.0 + gcoe & 5.7e10 & 0.27 & 6.0e8 & 2.14 & 1.6e8 & 2.30 & & & 2.6e10 & 1.7e10 & 2.0e10 & & 1.5:1.0:1.1 + gcof & 4.7e10 & 0.25 & 1.7e9 & 1.96 & & & & & 1.9e10 & 4.2e10 & & & 1.0:2.1:0.0 + gcoo & 7.2e10 & 0.22 & 1.5e9 & 1.94 & & & 9.3e6 & 5.59 & 2.3e10 & 3.7e10 & & 1.8e9 & 1.0:1.6:0.0 + gcop & 6.2e10 & 0.24 & 1.6e9 & 1.43 & 9.0e7 & 2.90 & & & 2.3e10 & 2.1e10 & 1.7e10 & & 1.3:1.2:1.0 + gcra & 5.7e10 & 0.32 & 2.0e9 & 1.28 & 2.8e7 & 4.29 & & & 3.7e10 & 2.1e10 & 1.1e10 & & 3.1:1.8:1.0 + gcrb & 2.4e10 & 0.35 & 8.6e8 & 2.04 & & & 6.1e7 & 6.02 & 1.9e10 & 2.2e10 & & 1.3e10 & 1.0:1.8:0.0 + gcrc & 4.0e9 & 0.54 & 1.5e8 & 3.32 & 1.0e8 & 4.31 & & & 7.5e9 & 1.0e10 & 4.4e10 & & 1.0:1.4:5.9 + gcrd & & 0.67 & 4.6e7 & 5.57 & 2.0e8 & 3.42 & & & & 9.1e9 & 5.4e10 & & 0.0:1.0:5.9 + dc & 3.3e10 & 0.25 & 2.7e7 & 6.15 & 1.6e8 & 3.65 & & & 1.3e10 & 6.6e9 & 4.9e10 & & 2.0:1.0:7.5 + ec & 6.7e10 & 0.21 & 1.9e7 & 4.71 & 1.7e8 & 3.38 & & & 2.0e10 & 2.7e9 & 4.6e10 & & 7.6:1.0:17 + fc & 2.4e11 & 0.11 & 7.7e9 & 0.54 & 2.9e7 & 4.89 & 2.7e8 & 2.77 & 1.9e10 & 1.4e10 & 1.6e10 & 1.3e10 & 1.1:1.7:1.0 + dcm & 4.3e9 & 0.57 & 3.7e8 & 3.36 & & & & & 9.4e9 & 2.6e10 & & & 1.0:2.8:0.0 + ecm & 3.6e9 & 0.58 & 3.3e8 & 3.14 & 1.3e7 & 4.70 & & & 7.6e9 & 2.0e10 & 6.8e9 & & 1.1:3.0:1.0 + fcm & 7.5e9 & 0.36 & 3.7e8 & 3.06 & 1.3e7 & 3.66 & & & 6.3e9 & 2.1e10 & 4.2e9 & & 1.5:5.1:1.0 + dcs & 8.7e9 & 0.12 & 1.0e9 & 0.33 & 6.7e7 & 2.15 & & & 8.5e8 & 7.2e8 & 7.1e9 & & 1.1:1.0:9.8 + ecs & 3.3e10 & 0.10 & 9.1e8 & 0.46 & 3.5e7 & 2.43 & & & 2.1e9 & 1.2e9 & 4.8e9 & & 1.7:1.0:3.9 + fcs & 3.2e10 & 0.10 & 1.5e9 & 0.34 & 5.0e7 & 1.92 & & & 2.0e9 & 1.2e9 & 4.2e9 & & 1.7:1.0:3.4 + the kinematic properties of the remnants exhibit significant quantitative differences as the pressure support of the ism is varied .
the stellar rotation curves of the merger remnant in the @xmath79 and @xmath34 models without bhs are plotted in figure [ fig : v_feedback ] for a progenitor gas fraction of @xmath29 .
the strength of the rotating stellar component correlates with the pressurization of the ism .
the least effective ism model for producing a remnant disk is the weakly pressurized , nearly isothermal model ( @xmath30 , model gan , upper left panel ) that yields a slowly rotating , clumpy stellar remnant .
the gaseous progenitor disks in model gan are not stable against @xcite instability as the sound speed of the weakly pressurized gas can not support the massive gas disk against self - gravity .
the gaseous disk efficiently converts into stellar clumps before and during the collision , producing a remnant with very little average rotation ( @xmath80 km s@xmath19 ) .
however , the scant amount of gas remaining after the final coalescence in this merger still manages to produce a very small , rapidly rotating disk . increasing the ism pressurization to @xmath81 ( model gbn , upper middle panel ) greatly reduces the presence of stellar clumps owing to instabilities and roughly doubles the radial extent of the rotating component , but does little to improve the average rotation of the remnant .
overall , low - pressurization ism models do a poor job at producing remnants with substantial average rotation but can produce small , rapidly rotating remnant disk components .
merger - driven disk formation will require a moderately to strongly pressurized ism to provide enough kinematic stability to the gas to avoid forming too many stars before the final merger .
the moderately to strongly pressurized models ( @xmath82 , gcn - gen , upper left - lower middle panels ) produce remnants that contain rapidly rotating stellar disks in addition to bulge or extended spheroid components . during the merger ,
the pressurization from the stiffer equation of state makes the ism less compressible , reduces the central gas density , and leads to less conversion of gas into stars . a larger fraction of stars in the final remnant form after the height of the merger , increasing the mass of the disk component and the average rotational velocity of the stars . just before the final coalescence
, the moderately and strongly pressurized models have @xmath83 , much larger than the @xmath84 in the @xmath30 model and @xmath85 in the @xmath81 model .
the abundance of gas present during the final coalescence in the moderately and strongly pressurized models allows for stellar disks to form from gaseous remnant disks ( e.g. * ? ? ?
figure [ fig : smd_feedback ] shows the stellar surface mass density of two highly pressurized ( @xmath61 ) , gas - rich major merger models with ( model ge , right panel ) and without ( model gen , left panel ) supermassive bh feedback . in both cases , stellar disks form from gaseous remnant disks created during the merger .
the presence of supermassive bhs slightly decreases the mass of the central spheroid and increases its effective radius , in agreement with the findings of @xcite for mergers that produce elliptical galaxies .
combined , the velocity fields shown in figure [ fig : v_feedback ] and the stellar disks shown in figure [ fig : smd_feedback ] demonstrate that disk galaxy remnants containing rapidly rotating , stellar disk components can form from high - angular momentum , gas - rich mergers when the ism is pressurized .
while the results of [ subsection : results : feedback ] demonstrated that the pressurization of the multiphase ism from star formation and feedback can produce remnant disks in gas - rich , prograde - prograde coplanar mergers , the generality of a merger - driven scenario for disk galaxy formation would require a wider range of encounters to produce remnant disks . using simulations without star formation , @xcite demonstrated that gaseous disks could form in mergers that were not exactly coplanar , including polar orbits . here
, we examine the impact of varying a range of progenitor disk orientations and pericentric passage distances on remnant disk formation including the effects of star formation and energetic feedback mechanisms .
figure [ fig : v_orientations ] shows the rotation curves for six different fairly radial encounters , including a prograde - prograde coplanar merger ( model gc , lower right panel ) , retrograde - retrograde coplanar ( model gcoc , upper left panel ) , polar ( models gcof , upper right panel , and gcop , middle lower panel ) , and intermediate cases ( models gcoe , upper middle panel , and gcoo , lower left panel ) . in each case , the same moderately - pressurized ism ( @xmath35 ) is used and enables the new stellar component to rapidly rotate .
remarkably , some highly non - coplanar encounters produce remnants whose entire stellar remnant is rapidly rotating ( models gcof and gcoo ) and demonstrate that merger - driven disk formation is not limited to the single coplanar orbit discussed by @xcite .
certainly , some orbits lead to remnants with almost no average rotation ( model gcoe ) , though with a rapidly rotating young stellar component .
interestingly , the retrograde - retrograde polar encounter ( model gcop ) produces a remnant with a new stellar component that counter - rotates in its interior relative to the outer young stellar material .
previous studies have demonstrated the formation of counter - rotating disks to form in remnants of mergers involving progenitors with @xmath86 @xcite or in spiral galaxies via other mechanisms ( e.g. * ? ? ?
* ; * ? ? ?
we plan to examine the creation of counter - rotating cores through extremely gas - rich major mergers in subsequent work , but we note here that the relative ages of the older , central disk in model gcop compared with its outer , younger disk is consistent with the picture for the formation of kinematically decoupled cores presented by @xcite where the central disk forms primarily from strongly - torqued material during the first passage and the outer disk forms later from material that retains some of its original orbital or disk angular momentum . the rotational support of the remnants gcof and gcoo formed from non - coplanar mergers , shown in figure [ fig : vs_orientations ] with the coplanar encounter
gc , is surprising but bolsters the merger - driven scenario for disk formation . in each merger ,
the total stellar rotational support exceeds @xmath87 and the new stellar component peaks at @xmath88 .
while the mass of the prograde - prograde coplanar merger remnant is still dominated by the central bulge ( bulge - to - disk ratio @xmath89 ) , the other models are @xmath90-dominated with bulge - to - disk ratios of @xmath91 for model gcof and @xmath92 for model gcoo including thick disk components . at the time of the merger these non - coplanar mergers have @xmath93 of their baryonic component in gas , compared with @xmath94 for the coplanar merger .
the rotationally - supported remnants from non - coplanar orbits experience less star formation during their first passage than the coplanar orbit ( see also * ? ? ?
* ) , leaving more gas available to form a remnant disk after the final coalescence .
the inclined merger model gcoe ( figure [ fig : v_orientations ] , upper middle panel ) that produces a remnant with very little net rotation has a @xmath95 immediately after the final coalescence , whereas the disk remnants formed by models gcof and gcoo have @xmath94 and @xmath96 , respectively . clearly , both the amount of gas present in the interacting system before the height of merger and the fraction of gas consumed during the final coalescence may influence the formation of remnant disks by improving the rotational support of the remnant .
the simulations that explore the effects of disk orientation demonstrate that for a given amount of orbital angular momentum in a merger the contribution of disk angular momentum can affect the presence of a remnant disk , both by providing angular momentum to the remnant gaseous disk that forms the rapidly - rotating stellar component and by influencing the amount of gas consumed during the merger .
however , the orbital configurations considered above only alter the disk orientation and not the orbital angular momentum of the encounter . changing
the pericentric passage distance can influence both the angular momentum of the remnant and the amount of gas available to form a remnant disk both by changing the timescale of the merger and by affecting the gas consumption during the interaction .
figure [ fig : v_rperi ] shows the rotation curves of remnants with moderately pressurized ism models ( @xmath35 ) formed in prograde - prograde coplanar gas - rich ( @xmath29 ) mergers with pericentric passage distances increasing from @xmath97 kpc ( model gcra , upper left panel ) to @xmath98 kpc ( model gcrd , lower right panel ) .
also shown for comparison is a polar merger with @xmath99 ( upper right panel ) , the same pericentric passage distance as the prograde - prograde coplanar model gc ( upper middle panel ) .
the nearly head - on collision ( model gcra ) produces a remnant with almost no net rotation ; after the height of the merger , only a small amount of gas remains ( @xmath100 ) .
the gas available for a remnant disk increases to @xmath101 when the pericentric passage disk scale length is roughly twice the disk scale length ( models gc and gcof ) , and decreases towards larger @xmath38 with @xmath102 ( models gcrb - gcrd ) .
these orbits have a substantially longer merger timescale than the more radial orbits , allowing for more quiescent star formation and a larger stellar component present in the progenitors at the height of the merger .
the competing effects of increasing the available orbital angular momentum by increasing @xmath38 and decreasing the gas available to form a remnant disk owing to extended quiescent star formation then lead to a range of @xmath38 from which remnant disks are likely to form . while additional simulations will be needed to better characterize the best orbits for remnant disk formation , from the simulations performed here we estimate that fairly radial orbits with pericentric passage distances a few times larger than the disk scale length are most ideal for merger - driven disk formation in gas - rich mergers .
the preceding feedback and orbital studies have demonstrated that gas - rich mergers with a pressurized ism can produce remnant disks for a variety of orbital configurations .
the results of those simulations suggest that the primary influence on the presence of remnant disks is the amount of gaseous material left after the final coalescence ; increased feedback or favorable ( though diverse ) orbits reduce the sfr during the first pericentric passage and the height of the merger , leaving enough gas after the merger to form a rapidly - rotating stellar component . by altering directly the gas fraction of the progenitor systems , the role of gas fraction on the rotational support of merger remnants can be determined .
figure [ fig : vs_gas_fraction ] shows the impact of increasing the gas fraction from @xmath59 ( left panel , model dc ) to @xmath103 ( right panel , model fc ) in the progenitor disks . shown
are prograde - prograde coplanar mergers with a moderately pressurized ism model ( @xmath35 ) .
while the new stars in each model produce a rotationally - supported structure , the average rotational support of the system is well correlated with the gas fraction as judged from the increasing @xmath77 as a function of progenitor @xmath28 .
the gas fraction of the merging galaxies before and after the final coalescence are nearly in proportion to the original gas progenitor gas fractions , with @xmath104 before and @xmath105 after the final merger in model dc and roughly twice that those values for the fc model with double the original gas content . however , the original @xmath103 progenitors in the model fc merger are not gas - rich enough to produce a remnant that has a rotationally - supported total stellar component with this orbital configuration .
these simulations suggest that given the typical @xmath55 gyr timescales involved , major merger progenitors must have original gas fractions @xmath106 in order to satisfy the @xmath107 requirement during final coalescence for a disk - dominated remnant .
the simulations of [ subsection : results : feedback]-[subsection : results : gas_fraction ] have demonstrated that gas - rich mergers can lead to remnant disk systems with substantial rotational support .
however , these simulations have modeled only equal mass encounters . while major mergers on average contribute significantly to the final mass of an average galaxy at the present day , minor mergers between unequal mass galaxies are cosmologically more frequent @xcite .
the survival of disks in minor merger or infall scenarios have been readily addressed both semi - analytically @xcite and numerically @xcite .
recently , @xcite has simulated comparatively gas - poor ( @xmath108 ) mergers between galaxies with mass ratios in the range @xmath109 which produce remnants with disk - like stellar structures but elliptical - like kinematics ( see also * ? ? ?
currently , the combined results of these analyses suggest that stellar minor mergers with disks will produce remnant systems that are kinematically dispersion - dominated but the cosmological infall of satellites expected in the @xmath110 cosmology is consistent with the observed distribution of thicknesses for disk galaxies . of additional interest to a scenario for merger - driven disk formation , @xcite demonstrated in hydrodynamical simulations without star formation that gas - rich minor mergers can produce small gaseous disk remnants . a cosmological scenario for merger - driven disk formation would likely involve such minor mergers given their abundance in the mass - accretion histories of galaxies .
figure [ fig : vs_minor ] shows the rotational support of remnants formed in prograde - prograde coplanar mergers of the @xmath18 km s@xmath19 galaxy model considered in [ subsection : results : feedback]-[subsection : results : gas_fraction ] with the @xmath21 km s@xmath19 galaxy described in [ subsection : methodology : minor_mergers ] .
these minor mergers with mass ratio @xmath111 use a moderately pressurized ism ( @xmath35 ) and progenitor gas fractions increasing from @xmath59 ( model dcm , left panel ) to @xmath112 ( model fcm , right panel ) .
the remnants are disk - dominated and rotationally - supported , with @xmath113 at @xmath114 kpc .
the smaller system is still massive enough to foment central star formation by driving gas inwards in the larger galaxy , increasing the velocity dispersion by a factor of @xmath115 to @xmath116 km s@xmath19 at the center each remnant .
these results are qualitatively consistent with previous work simulating the response of galaxy disks to minor mergers at lower gas fractions ( @xmath86 , e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , but in our substantially more gas - rich systems the additional supply of gas allows the larger galaxy to retain comparatively more mass - weighted stellar rotation by the end of the simulation through continued star formation in the extended gaseous disk .
the velocity dispersion in the outer regions of our simulated remnants increases from @xmath117 km s@xmath19 in the larger progenitor at @xmath114 kpc before the merger to @xmath118 km s@xmath19 in the remnant disk , decreasing with progenitor gas fraction . while the remnant properties are not strong functions of the gas fraction for our very gas - rich mergers , with each remnant displaying similar rotational support and bulge - to - disk ratios ( @xmath119 , see table [ table : remnants ] ) , the @xmath120 throughout most of the remnant disks substantially exceeds the @xmath121 for the comparatively gas - poor mergers reported by @xcite for mergers of mass ratio @xmath122 ( see their figure 10 ) .
we therefore infer that increasing the gas fraction of the disk progenitors of minor mergers beyond the maximum @xmath123 used by @xcite will improve the rotational support of remnant disk galaxies , but we can not fully detail the strength of the effect with the simulations presented here . however , based upon the success of the gas - rich minor mergers in retaining significant rotational support for the remnant disks we suggest that gas - rich minor mergers may be an arbiter of merger - driven disk formation and should be studied further .
the scale - dependent physics of gas cooling and star formation may influence merger - driven disk formation for a mass - sequence of galaxies . to examine this possibility , we repeat the gas fraction study presented in [ subsection : results : gas_fraction ] substituting smaller , @xmath21 galaxies for the progenitor systems while maintaining the same orbit ( i.e. the same physical @xmath38 and @xmath124 ) .
figure [ fig : vs_small ] shows the @xmath77 as a function of radius for mergers with moderately pressurized ism models ( @xmath35 ) and progenitor gas fractions increasing from @xmath59 ( model dcs , left panel ) to @xmath103 ( model fcs , right panel ) .
the remnants produced in these lower - mass mergers are rotationally - supported only in their outer regions .
the merger with @xmath59 progenitors ( model dcs ) does not contain a significant rotating component comprised from new stars , the only such system in our suite of simulations .
the higher gas fraction progenitors ( models ecs and fcs ) do produce rotating structures from their young stellar components , but the average rotation of these systems is dominated by rotation in their large spheroid components .
as discovered for more massive progenitors in [ subsection : results : gas_fraction ] , the gas fractions of the merging systems both before and after the final coalescence scale roughly with the initial progenitor gas fraction for the same orbit and ism pressurization .
these systems each have gas fractions @xmath125 just before the final coalescence and @xmath126 immediately afterwards , below the gas fractions measured in mergers that produce remnant disks . while the gas densities in the disks of these systems are lower than in the more massive systems considered in [ subsection : results : gas_fraction ] and therefore produce relatively less star formation in quiescence , the merger timescale for the less massive systems is longer and leads to similar gas fractions at the final merger .
if instead of using a physically identical orbit for both the lower - mass and higher - mass progenitors the orbits for the lower - mass progenitors were scaled with the disk scale lengths , or similarly the virial radii , the less massive systems would have merged on a shorter timescale and may have produced more rotationally - supported remnants . given the increased rotational support in the remnants of lower - mass progenitors relative to the higher - mass progenitors , we speculate here that lower - mass systems may be more conducive to merger - driven disk formation than can be inferred from the simulations we have performed .
we leave explorations of these suspicions to future work .
the suite of mergers simulated in this work indicates that merger - driven disk formation may viable in the gas - rich mergers that build high - redshift galaxies .
the requirements that the merging system is gas - dominated ( @xmath107 ) during the final coalescence to provide enough gas immediately after the merger to sustain a remnant disk is likely a necessary but not sufficient condition for merger - driven disk formation .
however , physical effects that increase the gas fraction during the final merger also improve the structure and rotational support of remnant disks . increased pressurization of the ism , favorable orbits with less - violent first passages or shorter merger timescales , small mass ratio mergers that decrease the interaction - induced star formation in the more massive progenitors , and simply increased gas fractions in the progenitors all improve the rotational support of the remnants and in some cases lead to disk - dominated , rapidly rotating stellar remnants .
the importance of gas - rich mergers for a variety of galaxy properties has been increasingly realized ( for a discussion see e.g. , * ? ? ?
* ) , and the results of our simulations of gas - rich progenitors bear on hierarchical models for the mass assembly of disk galaxies .
@xcite report that a substantial fraction of the thick disk stars of a simulated disk galaxy formed in a cosmological setting originate as tidal debris from satellite systems accreted after the thin disk forms .
@xcite present a different scenario where the thick disk forms from multiple early gas - rich mergers ( @xmath94 ; c. brook , private communication ) and the thin disk forms later from the gaseous debris of these accretion events .
a recent photometric survey of the thick and thin components of edge - on disk galaxies by @xcite and kinematic decompositions of the thick and thin disk components of two galaxies by @xcite suggest that thick disks may be kinematically disjoint , with lower rotational velocities than galactic thin disks .
this observational evidence supports the thick disk accretion scenario forwarded by @xcite .
however , if substantial thick disk mass accreted after the thin disk becomes predominately stellar then heating of the thin disk will occur anyway ( e.g. * ? ? ?
* ) . instead ,
if the epoch of thick disk accretion occurs when the large thin disk is gas - rich , then our simulations of minor mergers suggest that the thick disk may be safely accreted without unreasonably increasing the thin disk velocity dispersion .
our study of the formation of remnant disks in gas rich mergers provides a compliment to studies of elliptical galaxy formation in gas - rich merging ( e.g. * ? ? ?
major mergers between disk galaxies with progenitor gas fractions @xmath127 almost uniformly produce spheroid - dominated systems as the systems are not gas - dominated at the time of final coalescence .
such spheroid - dominated remnants are similar to our lower gas fraction models ec - fc .
as @xcite demonstrate , these spheroid dominated systems obey elliptical galaxy scaling relations such as the fundamental plane @xcite .
moreover , other apparently separate considerations make the case that if ellipticals did form through mergers , the progenitors must have been gas - rich ( though not gas - dominated as the simulations here show that these events may leave behind systems with large disks ) .
for example , @xcite showed that gas fractions larger than or of order @xmath128 are required if mergers between disk galaxies are to account for the relatively high phase space densities of local ellipticals .
moreover , this same criterion ensures that the remnants obey observed scaling relations ( e.g. * ? ? ?
* ) and match kinematic properties ( e.g. * ? ? ?
* ) of ellipticals .
the even more gas - rich mergers with progenitor gas fractions @xmath129 with disk - dominated stellar components will not satisfy these elliptical galaxy scaling relations owing to their rotational support .
these gas - rich merging scenarios for elliptical galaxy formation and remnant disk galaxy formation are fully consistent , as the remnant stellar disks form in systems with gas fractions large enough to sustain remnant gaseous disks whereas elliptical galaxies are generated in highly dissipative events where the vast majority of the gas is consumed and a large gaseous remnant disk can not form owing to the nearly complete depletion of gas . in gas - rich merger events that lead to elliptical galaxy formation , almost all of the stars form before or during the final coalescence of the galaxies .
the production of remnant disks in mergers requires that progenitors are comparably even more gas - rich to enable the formation of a remnant gaseous disk massive enough to eventually form a rapidly - rotating stellar disk that can dominate the stellar structure of the remnant .
naturally , this requires that the gas fraction before the final coalescence is @xmath130 , in agreement with the results of this study .
effects that reduce the consumption of gas before the final coalescence assuage the merger - driven production of disks , and while our work concentrates on feedback and orbital effects as well as the gas fraction of the progenitors we note that additional structural properties of the progenitors that reduce gas consumption during the merger ( e.g. bulges ) may also play a role in the formation of remnant disks .
lastly , the range of hubble types observed for spiral galaxies spans a considerable range in both rotational support ( @xmath77 ) and bulge - to - disk ratio . the merger - driven scenario for disk galaxy formation here may account for systems with bulge components but likely can not explain very late - type systems ( e.g. sd galaxies ) , which presumably must form almost entirely through dissipative processes that do not produce hot stellar components .
we emphasize that feedback associated with black hole growth , while essential for accounting for the observed properties of quasars @xcite and the red colors of ellipticals ( e.g. * ? ? ?
* ; * ? ? ?
* ) , has little impact on the global structural properties of the remnants studied here .
this is consistent with the work of @xcite and @xcite who found that black hole feedback likewise has negligible consequence for ellipticals on scales of order the effective radius .
the signatures of black hole growth on the stellar distribution of merger remnants thus appears subtle and may be restricted to modifying the characteristics of the central starburst population formed during the merger ( e.g. * ? ? ?
* ; * ? ? ?
to address the often destructive impact of the frequent merging in the @xmath110 cosmology on spiral galaxies , we propose a new , merger - driven scenario for disk galaxy formation at high redshifts that supplements the standard picture based on dissipational collapse @xcite . in this scenario , gas - rich mergers can form rotationally - supported gaseous structures from residual angular momentum after the final coalescence .
this rapidly - rotating material will form stars , and if enough material is available to form a substantial rotationally - supported component the remnant can resemble a disk galaxy both structurally and kinematically .
we perform a suite of twenty - eight merger simulations to address the feasibility of such a scenario for disk galaxy formation and demonstrate that remnant disks can form from sufficiently gas rich mergers with a variety of orbits , mass ratios , and physical models for the ism or the inclusion of growing supermassive black holes .
we provide a detailed summary of these results below . * we verify the results of by @xcite and @xcite in demonstrating that gas - rich mergers can produce remnant disks . nearly every gas - rich merger produces a rapidly - rotating structure from stars formed in remnant gaseous disks after the merger .
furthermore , we demonstrate that certain gas - rich mergers can produce remnants whose entire stellar structure is rotationally - supported . a necessary condition for the formation of these disk - dominated remnants
is inferred to be a gas fraction of @xmath131 just prior to the final coalescence , which we have referred to as being `` gas - dominated '' .
merger - driven disk formation is shown to occur under a variety of gas - dominated mergers and is not limited to e.g. idealized prograde - prograde coplanar mergers .
favorable equal mass encounters for merger - driven disk formation include certain polar and inclined orbits with limited star formation during the first pericentric passage .
a successful merger - driven disk formation scenario would requires extremely gas - rich progenitors with gas fractions @xmath132 , and is therefore likely limited to high redshift galaxy assembly .
* the rotational support of remnants is shown to correlate with the pressurization of the ism . using the multiphase ism model of @xcite modified to allow for an adjustable pressurization in the form of an effective equation of state @xcite ,
the merger simulations demonstrate that pressurized ism models limit the amount of gas consumed before the final coalescence of the progenitors .
the reduced star formation increases the gaseous material available to form a rapidly - rotating remnant disk and improves the rotational support of the remnant system .
weakly pressurized ism models either produce unstable progenitors owing to the high gas fractions or have efficient star formation . in both cases
the merging systems experience a mostly stellar collision that simply kinematically heats the existing stars without producing a newly - formed disk to contribute to the rotation of the remnant .
* gas - rich minor mergers can produce disk galaxies with substantial rotational support ( @xmath133 ) .
the mass ratio @xmath134 merger simulated for a range of gas fractions produces a remnant disk galaxy rotationally - supported at all measured radii @xmath135 kpc .
the central velocity dispersion during the merger roughly doubles and the old stellar disk is heated .
the rotational support of the remnant disks remaining from gas - rich minor mergers is considerably larger than that measured for comparable gas - poor minor mergers by @xcite , who calculate @xmath121 for @xmath136 mass - ratio gas - poor mergers .
the simulations confirm that merger - driven disk formation from gas - rich encounters at high redshift may be possible and highlight the role of energetic feedback on the rotational support of extremely gas - rich merger remnants . finally , to confirm the merger - driven scenario for disk galaxy formation as a viable supplement to the dissipational collapse model ,
cosmological simulations of galaxy formation including pressurized multiphase ism physics should be performed at sufficient resolution to determine the importance of early , gas - dominated mergers to the cosmological frequency of disk galaxies .
while cosmological simulations of galaxy formation with multiphase ism physics appear promising @xcite , simulations with sufficient resolution to track early interactions between gas - rich systems may provide a more comprehensive picture of cosmological disk galaxy formation .
this work was supported in part by nsf grants aci 96 - 19019 , ast 00 - 71019 , ast 02 - 06299 , and ast 03 - 07690 , and nasa atp grants nag5 - 12140 , nag5 - 13292 , and nag5 - 13381 .
the simulations were performed at the center for parallel astrophysical computing at harvard - smithsonian center for astrophysics . | the violent hierarchical nature of the @xmath0-cold dark matter cosmology poses serious difficulties for the formation of disk galaxies . to help resolve these issues
, we describe a new , merger - driven scenario for the cosmological formation of disk galaxies at high redshifts that supplements the standard model based on dissipational collapse . in this picture ,
large gaseous disks may be produced from high - angular momentum mergers of systems that are gas - dominated , i.e. @xmath1 at the height of the merger .
pressurization from the multiphase structure of the interstellar medium prevents the complete conversion of gas into stars during the merger , and if enough gas remains to form a disk , the remnant eventually resembles a disk galaxy .
we perform numerical simulations of galaxy mergers to study how supernovae feedback strength , supermassive black hole growth and feedback , progenitor gas fraction , merger mass - ratio , and orbital geometry impact the formation of remnant disks .
we find that disks can build angular momentum through mergers and the degree of rotational support of the baryons in the merger remnant is primarily related to feedback processes associated with star formation .
nearly every simulated gas - rich merger remnant contains rapidly - rotating stellar substructure , while disk - dominated remnants are restricted to form in mergers that are gas - dominated at the time of final coalescence .
typically , gas - dominated mergers require extreme progenitor gas fractions ( @xmath2 ) .
we also show that the formation of rotationally - supported stellar systems in mergers is not restricted to idealized orbits , and both gas - rich major and minor mergers can produce disk - dominated stellar remnants .
we suggest that the hierarchical nature of the @xmath0-cold dark matter cosmology and the physics of the interstellar gas may act together to form spiral galaxies by building the angular momentum of disks through early , gas - dominated mergers
. our proposed scenario may be especially important for galaxy formation at high redshifts , where gas - dominated mergers are believed to be more common than in the local universe . |
filament channels are defined as regions in the chromosphere surrounding a polarity inversion line ( pil ) where the chromospheric fibrils are aligned with the pil , indicating the presence of an axial magnetic field @xcite . in this paper
we study filament channels on the quiet sun .
the studied channels are divided into two types : type i channels are identified based on observed long and continuous h@xmath0 filaments ; while type ii channels are identified according to the cavities at the limb observed by xrt .
we first select 10 type i filament channels during november 2006 and december 2008 .
these channels are identified based on the h@xmath0 observations ( mainly provided by kanzelhhe solar observatory or mauna loa solar observatory ) of long and continuous filaments .
type i channels are usually located in low - latitude active region remnants .
a list of type i channels is shown in table 1 .
these channels can be divided into 5 groups , and the same channel at different solar rotations is classified into one group .
only five possible corresponding cavities are observed by xrt or trace , while the filament channels are at the limb .
the visibility of cavity may be affected by the direction of the filament channels or some bright active regions close by .
ccc|ccc channel & & cavity & channel & & cavity + date / hemisphere & group & date & date / hemisphere & group & date + 2006 - 11 - 01/s & 1 & 10 - 27 & 2008 - 03 - 02/n+s & 3 & no + 2006 - 12 - 26/s & 1 & 12 - 19 & 2008 - 03 - 27/s & 3 & no + 2007 - 01 - 19/s & 1 & no & 2008 - 04 - 24/s & 3 & 05 - 02 + 2007 - 04 - 30/n & 2 & no & 2008 - 11 - 29/n & 4 & 11 - 24 + 2008 - 02 - 02/n+s & 3 & no & 2008 - 12 - 09/n & 5 & 12 - 03 + filament channels on the quiet sun are often observed as dark channels in x - rays and euv , and an example is shown in figure 1 .
sometimes , sheared loops ( indicated by white arrows in figure 1a ) within the filament channel are observed in x - rays , but not in euv .
some sheared loops are stable , while others are transient structures .
this result is consistent with the findings by @xcite using yohkoh / sxt observations .
we find that the structure on the two sides of the channels is asymmetric in both x - rays and euv .
an example of a filament channel observed on 2008 march 2 is shown in figure 2 .
this figure shows that the eastern side ( black arrow ) has curved bright features , while the western side ( white arrows ) has straight faint features .
this asymmetric structure is found for all of the type i filament channels . filament . ]
we then identify 70 type ii filament channels based on cavities in xrt synoptic observations , which are taken twice a day normally .
first we look for cavities from these synoptic images . then the filament channels are identified as the disk counterpart from the images taken 7 days earlier and later , for the cavities on the west and east limb , respectively .
the h@xmath0 filaments in these channels are shown as dotted or short segments .
the emission on the two sides of the channels is weaker than that in type i channels .
based on the structure on the two sides of the channel , the 16 well observed type ii channels can be divided into two types : curved on the disk side and straight or unclear on the polar side ( figure 3a ) ; straight or unclear on the disk side and straight on the polar side ( figure 3b ) .
only one out of the 16 well observed channels appears to show asymmetric structure on the two sides of the channel ( see figure 3a ) .
the asymmetric structure of the type i channels is likely due to the fact that these channels are inclined with respect to the lines of constant latitude , and the fields are stronger at the equatorial end of the channel .
we suggest that the channel contains a weakly twisted flux rope , and the axial flux of the flux rope increases towards the equator , so that new flux is drawn into the flux rope from one side of the pil .
the bright curved features correspond to the lower parts of the field lines that turn into the flux rope , while the straight faint features are the lower parts of the overlying coronal arcade .
the type ii channels are usually east - west oriented , and the axial flux may be relatively constant along such channels , so asymmetry is rarely observed . we construct a nonlinear force - free field ( nlfff ) model of a type i channel using the flux rope insertion method @xcite .
several flux strands with different fluxes and lengths along the pil are inserted into a potential field model , and magneto - frictional relaxation is applied to produce a nlfff model for the channel on 2008 march 1 .
front ( a ) and side ( b ) views of selected field lines are shown in figure 4 .
note that field lines are drawn into the flux rope from the north - east side where the bright curved euv features are located . in summary ,
we have observed the fine structure of filament channels on the quiet sun , and we find that for type i channels the euv and x - ray emission is asymmetric on the two sides of the channel .
the results can be interpreted in terms of a flux rope model with the magnitude of the axial flux increasing along the channel as we approach the equator .
the same effect is rarely found for east - west oriented polar crown channels .
acknowledgements : the authors wish to thank the team of hinode / xrt , stereo , trace , solis , mlso , kso for providing the valuable data .
us members of the xrt team are supported by nasa contract nnm07ab07c to smithsonian astrophysical observatory ( sao ) .
is a japanese mission developed and launched by isas / jaxa , with naoj as domestic partner and nasa and stfc ( uk ) as international partners .
it is operated by these agencies in co - operation with esa and the nsc ( norway ) . | we present a study on the structure of quiescent filament channels observed by hinode / xrt and stereo / euvi from december 2006 to february 2009 . for 10 channels identified on the solar disk , we find that the emission on the two sides of the channel is asymmetric in both x - rays and euv : one side has curved bright features while the other side has straight faint features .
we interpret the results in terms of a magnetic flux rope model .
the asymmetry in the emission is due to the variation in axial magnetic flux along the channel , which causes one polarity to turn into the flux rope , while the field lines from the other polarity are open or connected to very distant sources . for 70 channels identified by cavities at the limb
, the asymmetry can not be clearly identified . |
measurement - based quantum computation ( mqc ) is a new computing paradigm @xcite . of particular interest are universal resources of one - way quantum computation , a mqc scheme that requires only local measurements @xcite . in the original scheme of one - way quantum computing , one
initially creates a many - qubit cluster state by applying phase gates or equivalent gate operations which can be realized using the ising interaction between qubits .
many promising methods to generate cluster states using solid - state qubits have been proposed @xcite . however , since these states are not the ground states of spin hamiltonians with typical qubit - qubit interactions of ising , @xmath0 , and heisenberg form @xcite , preserving them against the time evolution generated by these spin hamiltonians remains a critical issue .
one of the established universal resources are two - dimensional ( 2d ) cluster states .
another promising candidate is the affleck - kennedy - lieb - tasaki ( aklt ) state on the honeycomb lattice @xcite , a resonance valence bond ( rvb ) type state which is a special projected entangled pair state ( peps ) @xcite .
yet , the aklt state requires non - trivial hamiltonians with spin greater than 1/2 , which are not easy to realize in solid - state systems . in this paper
, we present a new method for preserving initially prepared cluster states .
our approach relies on manipulating a two - body hamiltonian using pulse - sequence techniques developed in the nuclear magnetic resonance ( nmr ) context @xcite .
we show that , starting from the ising and @xmath0 models , one can induce an effective dynamics described by a stabilizer hamiltonian @xcite @xmath1 where @xmath2 are the correlation operators and the direct product runs over all nearest neighbors of the lattice site @xmath3 ( @xmath4 and @xmath5 are the pauli matrices ) .
combined with cluster state generation methods @xcite , our scheme facilitates stable one - way quantum computing .
we assume the original hamiltonian to be of the form @xmath6 where @xmath7 is a single - qubit part and @xmath8 the interaction part .
we take @xmath8 to be of ising @xmath9 , @xmath0 @xmath10_{ij}$ ] , and heisenberg form @xmath11_{ij}$ ] . in this paper , @xmath12 if @xmath3 and @xmath13 are nearest neighbors and @xmath14 otherwise .
we use the shorthands @xmath15_{ij}\equiv \sigma_i^x\sigma_j^x+\sigma_i^y\sigma_j^y$ ] and @xmath16_{ij}\equiv \sigma_i^x\sigma_j^x + \sigma_i^y\sigma_j^y+\sigma_i^z\sigma_j^z$ ] , and set @xmath17 .
note that a single correlation operator can be obtained using a single - qubit hamiltonian .
for example in a one - dimensional ( 1d ) qubit array , @xmath18 can be generated by the time evolution operator @xmath19 .
however , it is not evident how to obtain a sum like @xmath20 from the single - qubit hamiltonian .
most fabricated solid - state qubit systems are nano - devices , because a smaller size makes them more robust to decoherence .
an example are quantum dot systems where smaller dots have larger energy - level spacing .
since with diminishing size it becomes difficult to address these devices individually , it is of interest to consider switching on / off @xmath21 and @xmath8 independently .
we will show that , by using appropriate pulse sequences , this is possible even if we start from an @xmath22-@xmath23 hamiltonian @xcite .
desirable quantum states are preserved by a pulse sequence that is familiar in the nmr context @xcite .
we assume that each pulse is sufficiently strong such that interactions between qubits can be neglected during the pulse sequences .
the time evolution of the system is described by the density operator @xmath24 whose time dependence is given by @xmath25 for time - independent @xmath26 .
it is convenient to use the following schematic notation for this evolution : @xmath27 .
then the process @xmath28 for @xmath29 corresponds to @xmath30 where @xmath31 and @xmath32 can be chosen arbitrarily .
note that at the physical time @xmath33 the state of the system is obtained from the initial one by the time - evolution operator @xmath34 thus , as illustrated by fig .
[ fig1](a ) , @xmath35 becomes the effective system hamiltonian .
its ground state is the originally prepared cluster state , which is therefore preserved .
we now show how to construct the stabilizer hamiltonian using the relation @xmath36 an important consequence of these equations is that , for @xmath37 , we can increase the order of the pauli - matrix terms as in @xmath38 and @xmath39 . for a 1d @xmath40-qubit chain ,
the starting single - qubit hamiltonian is given by @xmath41 by switching the interaction , the 1d stabilizer hamiltonian is realized according to @xmath42 the equivalent of eq .
at the hamiltonian level .
as an example , for @xmath43 qubits , starting from @xmath44 we obtain @xmath45 .
if the system of @xmath40 qubits has periodic boundary conditions , we start from the hamiltonian @xmath46 . since ising - type interaction terms commute and the time - evolution operator @xmath47 in ( [ main_result2 ] ) factorizes , this process can be straightforwardly extended to 2d and 3d qubit systems , thus realizing the universal resource discussed in the introduction .
consequently , for ising interactions , we can construct the stabilizer hamiltonian by switching on @xmath48 only once .
and @xmath49 of the hamiltonian @xmath50 gives rise to an effective time evolution described by the stabilizer hamiltonian @xmath35 [ see eq .
( [ main_result1 ] ) ] whose ground state is a cluster state .
( b ) graph representation of a one - dimensional 4-qubit ( squares ) cluster state stabilized by the ising interaction .
( c ) two - dimensional @xmath51 twisted cluster state stabilized by the @xmath0 interaction , a universal resource for one - way quantum computation .
the dashed lines and pauli operators in each direction illustrate the twistedness of the state and the corresponding stabilizer hamiltonian.,title="fig:",width=113 ] and @xmath49 of the hamiltonian @xmath50 gives rise to an effective time evolution described by the stabilizer hamiltonian @xmath35 [ see eq .
( [ main_result1 ] ) ] whose ground state is a cluster state .
( b ) graph representation of a one - dimensional 4-qubit ( squares ) cluster state stabilized by the ising interaction .
( c ) two - dimensional @xmath51 twisted cluster state stabilized by the @xmath0 interaction , a universal resource for one - way quantum computation .
the dashed lines and pauli operators in each direction illustrate the twistedness of the state and the corresponding stabilizer hamiltonian.,title="fig:",width=170 ] next , we show how to generate the stabilizer hamiltonian using the @xmath0 interaction , assuming that @xmath21 and @xmath49 can be switched on / off independently . the stabilizer hamiltonian is formed step by step by bonding the nearest - neighbor operators .
this is because the @xmath0 interactions do not commute , @xmath52_{i-1,i } , [ xy]_{i , i+1}]\neq 0 $ ] .
we start from @xmath53_{12 } } \sigma^{x}_{1 } e^{i\theta [ xy]_{12 } } & = & \cos ( 2\theta ) \sigma^{x}_{1 } -\sin ( 2\theta ) \sigma^{z}_{1 } \sigma^{y}_{2 } , \nonumber\\ e^{-i\theta [ xy]_{12 } } \sigma^{y}_{1 } e^{i\theta [ xy]_{12 } } & = & \cos ( 2\theta ) \sigma^{y}_{1 } + \sin ( 2\theta)\sigma^{z}_{1}\sigma^{x}_{2 } , \nonumber\\ e^{-i\theta [ xy]_{12 } } \sigma^{z}_{1 } e^{i\theta [ xy]_{12 } } & = & \cos^2 ( 2\theta ) \sigma^{z}_{1 } + \sin^2 ( 2\theta ) \sigma^{z}_{2 } \nonumber\\ & + & \frac{1}{2}\sin ( 4\theta ) [ \sigma^{x}_{1}\sigma^{y}_{2}-\sigma^{y}_{1}\sigma^{x}_{2}]\ : . \label{xy3 } \end{aligned}\ ] ] for @xmath37 , these transformations increase the order of the pauli - matrix terms as @xmath54 and @xmath55 . for @xmath56
one obtains @xmath57 .
we now show how to construct a 2d stabilizer hamiltonian .
first we construct the 1d stabilizer hamiltonian , starting from @xmath58 in the specific case of six qubits in 1d , by applying eq .
( [ xy3 ] ) to @xmath15_{12}$ ] , @xmath15_{34}$ ] , and @xmath15_{56}$ ] , we obtain : @xmath59 where @xmath60_{\{2l-1,2l\}}$ ] . repeating this step with @xmath61_{\{2l,2l+1\}}$ ] , we get the 1d stabilizer hamiltonian @xmath62 that reads explicitly @xmath63 this hamiltonian
is twisted in the sense of @xcite , i.e. , the site indices of the corresponding cluster state are obtained by the permutation @xmath64 ( cyclic notation ) , for a chain of @xmath40 qubits where @xmath40 is even , see fig .
[ fig1](b , c ) .
the next step in the construction of the 2d stabilizer hamiltonian is to construct a ladder hamiltonian by bonding nearest - neighbor sites on adjacent chains @xmath65 and @xmath66 , in which all the bondings between qubits @xmath67 and @xmath68 are carried out simultaneously : @xmath69 a 2d stabilizer hamiltonian is produced by connecting the above two ladder hamiltonians with the interaction between the two ladders .
for example , when we prepare two ladders of length 4 such as in fig . [ fig1](c ) and
connect them vertically , we obtain a @xmath70 stabilizer hamiltonian . for the heisenberg interaction , we can construct only a two - qubit stabilizer hamiltonian ( note that the same is true for the @xmath71 interaction ) .
the basic relation is @xmath72_{12 } } \sigma^{y}_{1 } e^{i\theta [ xyz]_{12 } } = \cos^2 ( 2\theta ) \sigma^{y}_{1 } \nonumber \\ & & + \sin^2 ( 2\theta ) \sigma^{y}_{2 } + \frac{1}{2 } \sin ( 4\theta ) ( \sigma^{x}_{1}\sigma^{y}_{2}- \sigma^{y}_{1}\sigma^{x}_{2})\ : .
\label{heisenberg1}\end{aligned}\ ] ] for the ising and @xmath0 models , we can eliminate the single pauli matrix terms leaving the interaction terms [ see eqs .
( [ ising2 ] ) and ( [ xy3 ] ) ] .
however , in eq .
( [ heisenberg1 ] ) , if we set @xmath73 or @xmath74 , we also eliminate the @xmath75 term .
this is because the heisenberg interaction contains terms in all three spatial directions @xcite . in the case of two qubits ,
we obtain @xmath76 from the initial hamiltonian @xmath77 by using eq .
( [ heisenberg1 ] ) for @xmath78 . by applying a @xmath79-rotation
, we obtain the two - qubit stabilizer hamiltonian @xmath80 .
the scheme discussed up to now relies on switching on / off the single - qubit hamiltonian @xmath81 [ see eq .
( [ h0 ] ) ] and the ising or @xmath0 interaction part @xmath8 separately .
there is a number of schemes for switching on / off interactions between qubits ( see , e.g. , @xcite ) .
however , they make the system more complicated and require additional overhead . here
, we solve this problem by demonstrating how to extract @xmath21 and @xmath8 by using appropriate pulse sequences .
we illustrate the idea using the standard nmr hamiltonian @xmath82 which has the property that @xmath83=0 $ ] . in this case ,
@xmath21 and @xmath8 can be switched on / off by using a simple pulse sequence .
the interaction part @xmath48 can be extracted by using two sandwiched @xmath79-pulses such as @xmath84 . on the other hand ,
two steps are required to obtain @xmath21 .
let us consider a 1d qubit chain . by applying a @xmath79-pulse about the @xmath85-axis ( denoted by @xmath86 ) to all the qubits on the even sites , we obtain @xmath87 .
similarly , we obtain @xmath88 by applying a @xmath86-pulse to all the qubits on the odd sites . combining these two processes yields @xmath21 .
this method is easily generalized to 2d or 3d qubit arrays .
if @xmath89\neq 0 $ ] , this nmr method can not be used . even in this case ,
@xmath21 and @xmath8 can be extracted separately .
the idea follows from average hamiltonian theory which is based on the baker - campbell - hausdorff ( bch ) formula for the expansion of @xmath90 @xcite . a stroboscopic application of the hamiltonian designed by a series of short pulses can reduce or eliminate unwanted terms , if @xmath91 .
first we extract @xmath8 by setting @xmath92 and @xmath93 in the bch formula , where @xmath94 and @xmath95 .
@xmath96 is realized by applying a @xmath97-pulse on every qubit . from the bch formula ,
we obtain @xmath98+\frac{1}{3 } [ h_0 , [ h_0,h_1]])\;. \label{ab}\ ] ] the exponent corresponds to a third - order expansion in @xmath99 for @xmath100 . if we repeat this operation @xmath101 times like @xmath102 such that @xmath103 , the @xmath104-th term is of order @xmath105^k$ ] . therefore , @xmath21 is canceled , and we obtain only @xmath8 in this order .
when we apply @xmath106+\frac{1}{3 } [ h_0 , [ h_0,h_1]])\:,\ ] ] we can eliminate the second term in eq .
( [ ab ] ) . in the limit @xmath107 under the condition of @xmath108 ,
@xmath21 is exactly eliminated .
the extraction of @xmath21 can be achieved analogously . moreover , as shown in @xcite , if the @xmath104th - order term is the first nonvanishing correction , the decay rate @xmath109 of the qubit system is enlarged according to @xmath110 as long as @xmath111 where @xmath112 is the time required for each single step [ @xmath113 and @xmath96 in eq .
( [ ab ] ) ] . for the @xmath0 model , we have to switch off subsets of @xmath8 corresponding to @xmath114 and @xmath115 , as discussed after eq .
( [ s_1 ] ) .
this is equivalent to choosing @xmath113 and @xmath96 appropriately : e.g. , for the 1d chain , @xmath116 and @xmath117 where @xmath118_{23}+[xy]_{45}+[xy]_{67}+ .. )$ ] and @xmath119_{12}+[xy]_{34}+[xy]_{56}+ .. )$ ] .
that means , @xmath96 is generated by applying @xmath79 pulses to qubits @xmath120 .
let us consider the 1d @xmath0 model with @xmath121 in eq .
( [ h0 ] ) . the following operation can be used to obtain @xmath21 .
( i ) applying a @xmath86-pulse to all the qubits on the even sites changes the sign of @xmath122 .
( ii ) by further applying a @xmath97-pulse to the same subset of qubits , we obtain @xmath123_{i , i+1}$ ] . as a result , we obtain @xmath124 . repeating the same operations with the qubits on the odd sites
, we obtain @xmath125 .
for the ising model , the process ( i ) is not required . for the heisenberg model , the same procedure as in the case of the @xmath0 model does not eliminate the term @xmath126 .
thus , additional similar steps are required to eliminate this term .
since a practical realization of these pulse sequences will not be free of imperfections , we now analyze the effect of pulse duration errors @xmath127
. a central quantity will be the cluster - state fidelity @xmath128 . in the ising case , @xmath35 is given by eq .
( [ main_result2 ] ) , and an analogous equation in the @xmath0 case .
let us consider the 1d ising case . for @xmath129 , the first - order correction to eq .
( [ main_result2 ] ) reads @xmath130\ : .
\label{hprime}\ ] ] the effect of these terms is calculated from perturbation theory using the expressions for @xmath131 , where @xmath132 is the initial cluster state , generated e.g. as proposed in ref .
@xcite , and @xmath133 @xcite .
the lowest - order expression of the cluster - state fidelity reads @xmath134^{-2}$ ] , and the correction scales with @xmath135 which is a signature of the robustness of our method . the simplest and most powerful method to further reduce the effect of pulse imperfections is the symmetrization of the pulse sequence frequently used in nmr @xcite .
we first note that eq .
( [ rho1 ] ) is equivalent to @xmath136 where , as before , @xmath29 . the second half of this pulse sequence results in a perturbation term that has the opposite sign as compared to eq .
( [ hprime ] ) . applying eq .
( [ ab ] ) leads to a cancellation of the first - order perturbation term .
if the original interval length @xmath32 is divided into an even number @xmath101 of subintervals , @xmath137 , the perturbation term is replaced by @xmath138 i\tau\delta/(4n)$ ] , and the fidelity @xmath139 is given by @xmath140^{-2}$ ] .
hence , the fidelity is improved , @xmath141 , if @xmath142 .
these perturbative results are complemented by exact numerical calculations of @xmath143 and the gate fidelity @xmath144 for @xmath145 in systems with up to @xmath146 qubits . here , @xmath147
, @xmath148 corresponds to the qubit pair @xmath149 . in both the ising and @xmath0 case , we averaged these fidelities over 2000 random realizations of the @xmath147 taken from a gaussian distribution with varying width @xmath150 .
the results indicate that the method is rather robust even outside the regime where @xmath150 is much smaller than @xmath151 .
for instance , the @xmath0-model calculation shows that @xmath152 is bigger than @xmath153 for @xmath154 , while @xmath155 can be bigger than @xmath153 even for @xmath150 as large as 0.07 .
the comparison between the two models shows that the robustness in the @xmath0 case ( see fig . [ fig2 ] ) is somewhat better than in the ising case . on random errors in the pulse duration for one - dimensional @xmath0 qubit arrays of length @xmath156 , @xmath157 , and @xmath146 .
the width of the gaussian distribution is denoted by @xmath150 ; the error bars indicate the standard deviation . ]
the distinguishing feature of the stabilizer hamiltonians discussed here is that their ground states are directly related to the universal resource of measurement - based quantum computation . more generally , the stabilizer formalism has further applications in quantum information theory @xcite .
for instance , in quantum error - correcting codes @xcite , the stabilizer formalism is used to express codewords . the method illustrated in eq .
( [ rho1 ] ) can be used to obtain eigenvalues for the syndrome measurements in the process of detecting errors .
moreover , relations ( [ ising2 ] ) and ( [ xy3 ] ) can be used to effectively generate codewords in solid - state qubits .
for example , the three - qubit greenberger - horne - zeilinger ( ghz ) state @xmath158 , which is used for the nine - qubit code , is effective generated by @xmath159 , where @xmath160 is applied to the three - qubit array .
the five - qubit code @xcite can also be generated by using this method .
to conclude , we have shown that initially prepared cluster states , e.g. , 2d cluster states that are universal resources of measurement - based quantum computation , can be preserved with high fidelity .
this is achieved by inducing the effective dynamics of 2d stabilizer hamiltonians by means of specially tailored pulse sequences , starting from natural qubit - qubit interactions .
we have also shown how this procedure can be implemented in the case of always - on interactions .
our work will facilitate implementations of one - way quantum computing .
we would like to thank j. koga and f. nori for discussions .
this work was financially supported by the eu project solid , the swiss snf , the nccr nanoscience , and the nccr quantum science and technology .
r. heule , c. bruder , d. burgarth , and v.m .
stojanovi phys .
a * 82 * , 052333 ( 2010 ) ; eur .
j. d * 63 * , 41 ( 2011 ) ; v.m .
stojanovi , a. fedorov , a. wallraff , and c. bruder , phys .
b * 85 * , 054504 ( 2012 ) . | the common spin hamiltonians such as the ising , @xmath0 , or heisenberg model do not have eigenstates that are suitable resources for measurement - based quantum computation .
various highly - entangled many - body states have been suggested as a universal resource for this type of computation , however , it is not easy to preserve these states in solid - state systems due to their short coherence times . to solve this problem , we propose a scheme for generating a hamiltonian that has a cluster state as ground state .
our approach employs a series of pulse sequences inspired by established nmr techniques and holds promise for applications in many areas of quantum information processing . |
one of the consequences of the rapid development of the internet and growing presence of information communication technologies is that large part of individuals daily activities , both off and online , is regularly recorded and stored .
this newly available data granted us with a substantial insight into activities of a large number of individuals for a long period of time and led to development of new methods and tools which would enable better insight into dynamics of social groups @xcite .
the structure and features of social connections have both strong influence and depend on social processes such as cooperation @xcite , diffusion of innovations @xcite and collective knowledge building @xcite .
therefore , it is not surprising that complex network theory has proven to be very successful in uncovering mechanisms governing the behaviour of individuals and social groups @xcite .
human activity patterns as well as the structure of social networks and the emergence of collective behaviour in different online communities have been extensively studied in a last decade @xcite . on the other hand , dynamics of offline social groups , where the activities take place trough offline meetings ( events ) ,
have drawn relatively little attention given their importance .
these groups , both professional and leisure ones , have large benefits and influence on everyday lives of individuals , their broader communities and society in general : they provide a social support for vulnerable individuals @xcite , can be used for political champagnes and movements @xcite , or can have an important role in career development @xcite . as they have different purpose of their existence they also vary in structure of participants , dynamics of meetings and organisation .
some groups , such as cancer support groups or scientific conference communities , are intended for a narrow circle of people while others , leisure groups for instance , bring together people of all professions and ages . in pre - internet era
these groups have been , by their organisation and means of communication between their members , strictly offline , while today we are witnessing the appearance of growing number of hybrid groups which combine both online and offline communication @xcite .
although inherently different all these social groups have two main characteristics in common : they do nt have formal organisation , although their members follow certain written and non - written rules , and the membership in them is on a voluntary basis .
bearing this in mind it is clear that the function , dynamics and longevity of these self - organised communities depend primarily on their ability to attract new and retain old members active in the group activities .
understanding the reasons and detection of key factors which influence members to remain active in social group dynamics are thus important , especially having in mind their relevance for the broader social communities and society . in previous work
@xcite we have shown that the scientists participation patterns in conference series are not random and that they exhibit an universal behaviour independent of conference subject , size or location .
using the empirical analysis and theoretical modelling we have shown that the scientist s conference attendance depends on the balance between the number of previous attendance and non - attendance and argued that this repetitiveness is driven by her association with the conference community , i.e. with the number and strength of social ties with other conference community members .
we also argued that similar behaviour when it comes to member s participation patterns in organised group events can be expected in other social communities .
here we provide the empirical evidence supporting these claims and further investigate the relationship between dynamics of individuals participation in social group activities and structure of its social network .
meetup portal , whose group dynamics we are studying , is an event - based social network .
meetup members use the online communication for the organisation of offline gatherings .
the online availability of the event attendance lists and group membership enables us to examine the event participation dynamics of the meetup groups and its influence on the structure of social networks between group members .
the diversity of meetup groups in terms of the type of activity and size allows us to further examine and confirm the universality of member s participation patterns .
the previous works using meetup source of data have been mostly focused on event recommendation problem @xcite and the structural properties of social networks and relationships between event participants @xcite by disregarding evolutionary behaviour of meetup groups . in this work
, we examine the event induced evolution of social networks for four large meetup groups from different categories . like in the case of conference participation , we study the probability distribution of total number of meetup attendance and show that it also exhibits a truncated power law for all four groups , like in the case of conference participation dynamics @xcite .
this finding suggests that event participation dynamics of meetup groups is characterised by positive feedback mechanism , which is of social origins and is a directly related to member s association with the social community of the specific meetup group . using complex network theory we examine in more details the correlation between member s decision to participate at an event and her association with other members of that meetup group .
specifically , we track how member s connectedness with community changes with the number of attendance by measuring the change in clustering coefficient and relation between degree and strength in evolving weighted social network , where only statistically significant connections are considered .
our results indicate that greater involvement in group activities is more associated with the strengthening of the existing than to creation of new ties .
this is consistent with previous research on meetup which has shown that repeated event attendance leads to increase of bonding and decrease of bridging social capital @xcite .
furthermore , in view of the fact that people interact and network evolves through events , we examine how particular event affects the network size and structure .
we investigate effect of event size and time ordering on social network organization by studying the change in network topology , number of distinctive links and clustering , caused by the removal of specific event and find that the purpose of large events is to facilitate new connections , while during the small events already acquainted members strengthen their interpersonal ties .
similar behaviour was observed at the level of communities , where small communities are typically closed for new members , while contrary to this , the change of membership in large communities is favourable @xcite .
this paper is organized as follows : we first study the distribution of the total number of participations for four meetup groups from different categories .
next we introduce filtered weighted social network to characterize significant social connections between members and discuss its structural properties .
specifically , we study how the local topological properties evolve with the growth of the number of participations in order to derive relationships between the member association with the group and activity patterns . in order to analyse impact of the particular event on the network organization ,
we remove events using different strategies and show how it influences social structure .
meetup is the online social networking platform which enables people with common interest to start a group with purpose of arranging the offline meetings ( events , meetups ) all over the world .
the groups have various topics and are sorted into @xmath0 different categories , such as careers , hobbies , socializing , health , etc .
these groups are of various sizes , have different event dynamics and hierarchical organisation .
they also differ in type of activity members engage , ranging from socialising events , like parties and clubbing , to professional trainings , such as seminars and lectures .
common to all groups is the way they organise offline events : each member of the group gets an invitation to event to which she replays with yes / no , creating in that way a record of attendance for each event .
we use this information to analyse event participation patterns and to study the evolution of social network . here
, we analyse four large groups , each having more than three thousand organized events , ( see methods and table [ dataset ] ) , from four different categories .
we chose these four groups because of their convenience for statistical analysis , large number of members and organised events , and also for the fact that they are different when it comes to the type of activity and interest their members share .
the _ geamclt _
( geam ) group is made of _ foodie thrill - seekers _ who mostly meet in the restaurants and bars in order to try out new exciting foods and drinks , while people in _
vegashiker _ ( lvhk ) are hikers who seek excitement trough physical activity .
the _ pittsburgh - free
_ ( pghf ) is a group which invites its members to free , or almost free , social events , and the _ techlife columbus _ ( tech ) is about social events that focus on tech community networking , entrepreneurship , environmental sustainability , and professional development .
figure [ total ] shows that the probability distributions of the total number of members attendance of group events for all four groups exhibits truncated power law behaviour , with power law exponent larger than one .
power law and truncated power - law behaviour of the probability distributions can be observed for the number of and the time lag between two successive participations in group organised events , fig [ fig_a ] and fig [ fig_b ] in si .
in fact , we find that the similar participation patterns ( they differ in the value of exponents ) can be observed for all meetup groups , regardless of their size , number of events or category .
as in the case of conference participation dynamics @xcite , this indicates that the probability to participate in the next event depends exclusively on the balance of the number of previous participations and non - participations .
we argued in @xcite that forces behind conference participation dynamics are of social origin , and it follows from the fig [ total ] that the same can be argued in the case of meetup group participation dynamics .
the more participations in group activities member has , the stronger and more numerous are her connections to the other group members , and thus her association with the community .
we further explore this assumptions by investigating the event driven evolution of the social networks of the four different meetup groups . .
] we construct the social network between group members , for each separate groups , as a network of co - occurrence on the same event ( see methods for more details ) . by definition
these networks are weighted networks where the link weight between two members is equal to the number of events they participated together .
these networks are very dense , which is a direct result of construction method , with broad distribution of link weights ( see fig [ fig_c ] in si ) .
co - occurrence at the same event does nt necessarily imply a relationship between two members .
for instance , a member of the group that attends many events , or big events , has large number of acquaintances , and thus large number of social connections , which are not of equal importance when it comes to her association with the community .
similarly , two members that attend large number of events can have relatively large number of co - occurrences which can be the result of coincidence and not an indicator of their strong relationship . in order to filter out these less important connections we use filtering technique based on configuration model for bipartite networks @xcite ( see methods ) . by applying this technique to weighted networks
we reduce their density and put more emphasis on the links which are less likely to be the result of coincidence .
this way we put more emphasis on the links of higher weight without the removal of all links with weight smaller then certain threshold ( see fig [ fig_c ] in si ) , a standard procedure for network pruning .
we explore the evolution of these social networks of significant relationships between meetup group members by studying how the local characteristics of the nodes ( members ) are changing with their growing number of participations in group activities .
+ association with the community of the specific meetup group can be quantitatively expressed trough several local and global topological measures of weighted networks .
specifically , we explore how the number of significant connections ( member s degree ) and their strength ( member s strength ) , as well as how member s embeddedness in the group ( non and weighted clustering coefficient ) are changing with the number of attended group events .
figure [ deg_str ] shows how average strength of a node depends on its degree in filtered networks of four selected meetup groups . while the degree equals to the number of member s significant social relationships , the strength measures how strongly she is connected to the rest of the group @xcite . in all considered meetup groups members with small and medium number of acquaintances ( @xmath1 )
have similar value of strength and degree , i.e. their association with the community is described by the number of people they know , not from the strength of their connections ( see fig [ deg_str ] ) .
having in mind that the average size of the event in these four groups is less than @xmath2 , we can conclude that majority of member with degree less than @xmath3 are the ones that attended only few group meetups .
previous study @xcite has found that the probability for a member to attend the group events strongly depends on weather her friends will also attend .
the non - linear relationship between degree and average strength for @xmath4 shows that event participation of already engaged members ( ones who already attended few meetings ) is more linked to strength of social relations than to their number .
this means that at the beginning of their engagement in group activities , when the associations is relatively small , the participation is conditioned with the number of members a person knows , while later , when the associations becomes stronger , the intensity of relations with already known members becomes more important .
+ this finding is further supported by the change of average degree and strength with the number of participations .
figure [ deg_str_part ] shows how the degree and strength evolve with the number of participations in group events averaged over all members . at the beginning
, the degree and strength have the same value and grow at the same rate , but after only few participations the strength becomes larger than degree , and starts to grow much faster for the members of all four meetup communities .
after @xmath5 attended events the average strength of a member is up to ten times larger than her degree ( see fig [ fig_d ] in si ) .
this indicates that the event participation dynamics is mostly governed by the need of the member to maintain and strengthen her relationships with the already known members of the community . as a matter of fact
, our analysis of member s embeddedness in social network shows that it is not about maintaining the strong relations with single members of community , but rather with the small subgroups of members .
the relatively high value average clustering coefficient , @xmath6 , shown in fig [ c_cw_part ] indicates that there is a high probability ( more than @xmath7 on average ) that friends of member also form significant relationships .
the slow decay of @xmath6 with the number of participations and the fact that it remains relatively large ( above @xmath8 ) even for participants with thousand attended meetups , fig [ c_cw_part ] , shows that personal networks of members have tendency to remain clustered , i.e. have relatively high number of closed triplets compared to random networks .
we further examine the structure of these triplets and its change with the number of participations by calculating the averaged weighted clustering coefficient .
the weighted clustering coefficient , @xmath9 measures the local cohesiveness of the personal networks by taking into account the intensity of interactions between local triplets @xcite .
this measure does not just take into account the number of closed triplets of the node @xmath10 but also their total relative weight with respect to the total strength of the nodes ( see methods ) .
we examine how its value , averaged over all participants that have attended @xmath11 events , is changing with the number of attended events @xmath12 .
as it is shown in fig [ c_cw_part ] the member s network of personal contacts shows high level of cohesiveness , on the average . like its non - weighted counterpart , the value of @xmath13 only slightly declines during member s early involvement in group activities , while later it remains constant and independent of the number of participations .
the comparison of weighted and non - weighted clustering coefficients reveals an information about the role of strong relationships in the local network , i.e. weather they form triplets or bridges between different cohesive groups @xcite . at the beginning of member
s involvement in the group , these two coefficients have similar value , fig [ c_cw_part ] , which indicates that the cohesiveness of subgroup of personal contacts is not that important for early participation dynamics .
as the number of attended events grows , as well as the number and strength of personal contacts , the weighted clustering coefficient becomes larger than its non - weighted counterpart , indicating that member s strongest ties are with the other members who are also friends .
the fact that in latter engagement weighted clustering coefficient is larger than its non - weighted counterpart indicates that clustering has an important role in network organisation of meetup groups and thus in the group participation dynamics @xcite . + and weighted clustering coefficient @xmath13 , with the number of attended events . ] in the previous work @xcite we have shown that the conference participation dynamics is independent of conference topic , type and size .
the same holds true for meetup participation dynamics , i.e. the member s participation patterns in the meetup group activities do not depend on the group size , category , location or type of activity .
however , the size of group events and their time - order may influence the structure of network and thus group dynamics .
we explore how topological properties of the networks , specifically the number of acquaintances and network cohesion , change after the removal of events according to certain order ( see methods for details ) .
firstly , we study how the removal of events according to certain order influence the number of overall acquaintances in the network .
for this purpose we define measure @xmath14 ( see methods ) which we use to quantify the percentage of the remaining significant acquaintances after the removal of event .
figure [ eta_event ] shows the change of measure @xmath14 after the removal of @xmath15 fraction of events according to chosen strategy .
we see that the most of the new significant connections are usually made on the largest events .
the importance of large events for the creation of new acquaintances is especially striking for three groups geam , pghf , and tech , where about @xmath16 of acquaintances only met at the @xmath17 of the largest events .
for lvhk the decrease is slower , probably due to a difference in event size fluctuations ( see fig [ fig_e ] in si ) , but still more than @xmath18 of acquaintances disappear if we remove @xmath19 of the largest events , which is still much higher percentage of contacts compared to random removal of events ( see fig [ eta_event ] ( right ) ) .
similar results are observed when we remove events from the smallest to largest , fig [ eta_event ] ( left ) .
only @xmath17 of acquaintances are being destroyed by removing @xmath16 of the smallest events , for all four groups .
this indicates that the new , weak , connections ties are usually formed during large events , while these weak , already existing , acquaintances are further strengthen during smaller meetups . on the other hand , the removal of events according to their temporal order , fig [ eta_event ] , has very similar effect as random removal , i.e. the value of parameter @xmath14 decreases gradually as we remove events .
+ with the removal of events according to their size ( left ) and time order and random ( right ) .
abbreviations indicate the order in which we remove events : * b * -from the largest to the smallest , * s * from the smallest to the largest , * f * from the first to the last and * r * random . ]
similar conclusions can be drawn from change of average weighted clustering coefficient with the removal of events , fig [ cw_event ] .
removal of events in ordering from the smallest to the largest , does nt result in the significant change of @xmath20 ( now averaged over all nodes in the network ) .
the same value of weighted clustering coefficient , even after removal of @xmath16 of events , shows that small events are not attended by _ a pair of _ but rather _ by a group _ of old friends . on the other hand , the removal of events from the largest to the smallest results in gradual decrease of @xmath20 .
a certain fraction of triads in the network are made by at least one low weight link , and they are being the first one destroyed by the removal of the largest events , leading to gradual decrease of @xmath21 .
removal of events according to their temporal order results in the change of @xmath20 similar to one obtained with random removal of events confirming further that the time - ordering of events does nt have an influence on the network structure .
in this article we explore event participation dynamics and underlying social mechanism of meetup groups . the motivation behind this was to further explore event - driven dynamics , work we started by exploring participation patterns of scientists at scientific conferences @xcite , and to better examine the social origins behind the repeated attendance of the group events , which was not feasible with the conference data .
the results in this manuscript are based on empirical analysis of participation patterns and topological characteristics of networks for four different meetup groups made up of people with different motives and readiness to participate in group activities : geam , pghf , tech , lvhk .
although these four groups differ in category and type of activity , we have shown that all four of them are characterised with similar participation patterns : the probability distribution of total number of participations , number of successive participations and time lag between two successive participations follow the power - law and truncated power law behaviour with the value of power law exponents between @xmath22 and @xmath23 .
the resemblance of these patterns with the ones observed for conference participation @xcite indicates that these two , seemingly different , social system dynamics are governed by similar mechanism .
this means that the probability for a member to participate at the future events depends non - linearly on the balance between the number of previous participations and non - participations . as in the case of conferences
@xcite this behaviour is independent of group category , size , or location meaning that members association with the community of meetup group strongly influence their event participations patterns and thus the frequency and longevity of their engagement in group activities .
the member s association with the community is primarily manifested trough her interconnectedness with other members of the specific meetup group community , i.e. in the structure of her personal social network .
we have examined topological properties of filtered weighted social network constructed from the members event co - occurrence . by filtering the network we emphasized the importance of significant links , ones which are not the result of coincidence but rather an indicator of existing social relations .
the analysis of local topological properties of these networks has revealed that strength of connectedness with the community for the members with small number of participations is predominantly the consequence of the width of their social circles .
the average strength and degree of members with @xmath24 , which on average corresponds to only a few participations , are equal , while the strength of members who know more than @xmath3 people and have participated in more than a few events , is several times higher then their degree .
this means that strengthening of existing ties becomes more important than meeting new people after a few participations .
these arguments are further extended with our observation of the evolution of average strength and degree with the growth of number of participations . both , average degree and strength grow , but the growth rate of strength is higher than degree for all four meetup groups .
all four groups are characterised with very high cohesiveness of their social communities .
the evolution of clustering coefficients , non and weighted one , and their ratio show that bonding with community becomes more important as the members engagement in group activity progress . as in the case of conference participation ,
frequent attendees of group activities tend to form a core whose stability grows with the number of participations @xcite .
the need of frequent attendees to maintain and increase their bonding with the rest of the community influences their probability to attend future meetings and thus governs the event participation dynamics of the meetup groups .
while group category , type of activity and size do nt significantly affect the participation dynamics in group activities and network structure , the size of separate events organised in groups does have an influence on the evolution of social networks .
large events represent an opportunity for members to make new acquaintances , i.e. to establish new connections , while small meetings are typically the gatherings of members with preexisting connections and their main purpose is to facilitate the stronger bonding among group members .
while the size of the event influence structure of social networks , it turns out that their time - order is irrelevant for group dynamics .
the universality of members event participation patterns , shown in this and previous work @xcite , and its socially driven nature give us a better insight not only about the dynamics of studied social communities but also about others which are organised on very similar principles : communities that bring together people with the similar interests and where the participation is voluntary .
having in mind that these type of groups constitute a large part of human life , including all life aspects , understanding their functioning and dynamics is of great importance .
our results not only contribute to the corpus of increasing knowledge but also indicate the key factor which influences the group longevity and successful functioning : the association of group members with the community .
this and recent success stories @xcite suggest that complex network theory can be an extremely useful tool in creating successful communities .
future studies will be conducted towards further confirmation of universality of event - participation patterns and better understanding of how social association and contacts can be used for creating conditions for successful functioning of learning and health support groups .
there are more than @xmath25 groups in @xmath26 countries classified into @xmath0 categories . for each of selected four groups ,
we have collected list of events organized by the group and information on members who confirmed their participation in the given event since the group s beginnings .
each member has a unique i d which enables us to follow her activity in group events during the time .
more details about the group sizes and the number of events is given in table [ dataset ] . * network construction * we start with a bipartite member - event network represented with participation matrix @xmath27 .
let @xmath28 denotes total number of members in the group and @xmath29 is the total number of the events organized by the group .
if the member @xmath10 participated in the event @xmath30 element of matrix @xmath31 takes a value @xmath22 , otherwise @xmath32 . in the bipartite network created in this way , the member s degree is equal to the total number of events member participated in , while the event s degree is defined as the total number of members that have attended that event .
the social network , which is the result of members interactions during the meetup events and is represented by weighted matrix @xmath33 , is created from the weighted projection of bipartite network to the member partition @xcite . in the obtained weighted network nodes
correspond to individual members while the value of the element of weighted matrix , @xmath34 , corresponds to the number of common events two members have attended together .
* network filtering * the observed weighted network is the dense network where some of the non - zero edges can be the result of coincidence . for instance , these edges can be found between members who attended large number of events or events with many participants , and therefore they do not necessarily indicate social connections between members .
the pruning of these type of networks and separation of significant edges from the non - significant ones is not a trivial task @xcite . for this reasons we start from the bipartite network and use the method that determines the significance of @xmath34 link based on configuration model of random bipartite network @xcite . in this model of random networks ,
the event size and the number of events a member attended are fixed , while all other correlations are destroyed ( see si for further explanations ) .
based on this model , for each link in bipartite network , @xmath31 , we determine the probability @xmath35 that user @xmath10 has attended the event @xmath30 .
the assumption of uncorrelated network enables us to also estimate the probability that two member , @xmath10 and @xmath36 , have attended the same event , which is equal to @xmath37 .
the probability that two members have attended the same @xmath38 events is then given by poisson binomial distribution @xmath39 where @xmath40 is the subset of @xmath38 events that can be chosen from given @xmath41 events @xcite .
we define @xmath42-value as the probability that two members @xmath10 and @xmath36 has co - occurred on at least @xmath43 events , i.e. that the link weight between these two members is @xmath43 or higher @xmath44 the relationship between users @xmath10 and @xmath36 will be considered statistically significant if @xmath45 . in our case , threshold @xmath46 .
all links with @xmath47 are the consequence of chance and are considered as non - significant and thus removed from the network .
this way we obtain weighted social network of significant relations between members of the meetup group @xmath48 .
the details on how we estimate @xmath35 and @xmath49 for each link are given in si .
* topological measures * all topological measures considered in this work are calculated for weighted social network of significant relations @xmath48 .
we consider the following topological measures of the nodes : * the node degree @xmath50 , where @xmath51 is heaviside function ( @xmath52 if @xmath53 otherwise @xmath54 ) ; * the node strength @xmath55 @xcite ; * non - weighted clustering coefficient of the node @xmath56 @xcite . * weighted clustering coefficient of the node @xmath57 @xcite .
the weighted clustering coefficient of the network , @xmath20 , and its non - weighted counterpart , @xmath58 , are values averaged over all nodes in the network . in order to explore the relevance of event size and time ordering in evolution of social network topology we analyse how removal of events , according to specific ordering , influence the number of acquaintance and network cohesion .
specifically , we observe the change of measure @xmath14 , which represents the fraction of the remaining acquaintances , and weighted clustering coefficient , @xmath20 , after the removal of certain fraction of events .
the removal of event results in the change of the link weights between group members .
for instance , if two members , @xmath10 and @xmath36 , have participated in the event @xmath30 , the removal of this event will result in the decrease of the link weight @xmath48 by one .
the further removal of events in which these two members have co - occurred will eventually lead to termination of their social connection , i.e. @xmath59 . if @xmath60 is the matrix of link weights after the removal of fraction of @xmath15 events and @xmath61 is the original matrix of significant relations , then the value of parameter @xmath14 after the removal of @xmath15 events is calculated as @xmath62 the value of weighted clustering coefficient , @xmath20 , after the removal of fraction of @xmath15 events is calculated using the same formula as for the @xmath20 just using the value of @xmath60 instead of @xmath63 .
we remove events according to several different strategies : * we sort events by the number of participants .
then , we remove sorted events in both , descending and ascending order .
* we sort events by arrangement in time .
we remove sorted events in direct order .
* we remove events in random order .
we perform this procedure for each list of events @xmath5 times .
numerical simulations were run on the paradox supercomputing facility at the scientific computing laboratory of the institute of physics belgrade .
10 castellano c , fortunato s , loreto v. statistical physics of social dynamics .
rev mod phys .
2009;81:591646 .
five rules for the evolution of cooperation . science .
2006;314(5805):15601563 .
fowler jh , christakis na .
cooperative behavior cascades in human social networks .
proc natl acad sci usa .
2010;107(12):53345338 .
granovetter ms .
the strength of weak ties .
am j sociol
. 1973;78(6):13601380 .
pastor - satorras r , castellano c , van mieghem p , vespignani a. epidemic processes in complex networks .
rev mod phys .
2015;87:925979 .
mitrovi dankulov m , melnik r , tadi b. the dynamics of meaningful social interactions and the emergence of collective knowledge .
2015;5:12197 .
boccaletti s , latora v , moreno y , chavez m , hwang du . .
2006;424(45):175 308 .
holme p , saramki j. .
2012;519(3):97 125 .
aral s , walker d. identifying influential and susceptible members of social networks . science .
2012;337(6092):337341 .
gonzlez - bailn s , borge - holthoefer j , moreno y. broadcasters and hidden influentials in online protest diffusion .
am behav sci
. 2013;57(7):943965 .
lin yr , chi y , zhu s , sundaram h , tseng bl .
facetnet : a framework for analyzing communities and their evolutions in dynamic networks . in : proceedings of the 17th international conference on world wide web .
www 08 ; 2008 .
p. 685694 .
mitrovi m , paltoglou g , tadi b. quantitative analysis of bloggers collective behavior powered by emotions .
j stat mech .
2011;2011(02):p02005 . garas a , garcia d , skowron m , schweitzer f. .
2012;2:402 .
trk j , iiguez g , yasseri t , san miguel m , kaski k , kertsz j. .
phys rev lett
. 2013;110:088701 .
yasseri t , sumi r , rung a , kornai a , kertsz j. .
plos one . 2012;7(6):112 .
montazeri a , jarvandi s soghraand haghighat , vahdani a mariamand sajadian , ebrahimi m , haji - mahmoodi m. .
patient educ couns .
2001;45:195198 .
davison kp , pennebaker jw , dickerson ss . .
. 2000;55:205217 .
tam cho wk , gimpel jg , shaw dr . .
q j polit sci . 2012;7:105133 .
weinberg bd , , williams cb . .
j direct data digit mark pract .
2006;8(1):4657 .
smiljani j , chatterjee a , kauppinen t , mitrovi dankulov m. a theoretical model for the associative nature of conference participation .
plos one . 2016;11(2):112 .
qiao z , zhang p , zhou c , cao y , guo l , zhang y. event recommendation in event - based social networks . in : proceedings of the twenty - eighth aaai conference on artificial intelligence .
aaai14 ; 2014 .
p. 31303131 .
zhang w , wang j , feng w. combining latent factor model with location features for event - based group recommendation . in : proceedings of the 19th acm sigkdd international conference on knowledge discovery and data mining .
kdd 13 ; 2013 .
p. 910918 .
pham tan , li x , cong g , zhang z. a general graph - based model for recommendation in event - based social networks . in : 2015 ieee 31st international conference on data engineering
p. 567578 .
macedo aq , marinho lb , santos rlt . .
in : proceedings of the 9th acm conference on recommender systems .
recsys 15 ; 2015 .
p. 123130 .
liu x , he q , tian y , lee wc , mcpherson j , han j. . in : proceedings of the 18th acm sigkdd international conference on knowledge discovery and data mining .
kdd 12 ; 2012 .
p. 10321040 .
jiang jy , li ct . .
in : international aaai conference on web and social media ; 2016 .
p. 599602 .
sessions lf . .
information , communication & society . 2010;13(3):375395 .
mccully w , lampe c , sarkar c , velasquez a , sreevinasan a. . in : proceedings of the 7th international symposium on wikis and open collaboration .
wikisym 11 ; 2011 .
p. 3948 .
palla g , barabsi al , vicsek t. quantifying social group evolution .
2007;446(7136):664667 .
backstrom l , huttenlocher d , kleinberg j , lan x. group formation in large social networks : membership , growth , and evolution . in : proceedings of the 12th acm sigkdd international conference on knowledge discovery and data mining .
kdd 06 ; 2006 .
p. 4454 .
arxiv e - prints .
2016;. f , di clemente r , gabrielli a , squartini t. .
arxiv e - prints .
2016;. barrat a , barthlemy m , pastor - satorras r , vespignani a. the architecture of complex weighted networks .
proc natl acad sci usa
. van dijk j , maier g. pap reg sci .
2006;85(4):483504 .
cosgrave p. engineering serendipity : the story of web summit s growth ; 2014 .
available from : https://goo.gl/h3awmi .
mitrovi m , tadi b. .
eur phys j b. 2010;73(2):293301 .
mitrovi m , paltoglou g , tadi b. .
eur phys j b. 2010;77(4):597609 .
dianati n. unwinding the hairball graph : pruning algorithms for weighted complex networks .
phys rev e. 2016;93:012304 .
saracco f , di clemente r , gabrielli a , squartini t. randomizing bipartite networks : the case of the world trade web .
2015;5:10595 .
cellai d , bianconi g. multiplex networks with heterogeneous activities of the nodes .
phys rev e. 2016;93:032302 .
liebig j , rao a. fast extraction of the backbone of projected bipartite networks to aid community detection .
europhys lett .
2016;113(2):28003 .
hong y. on computing the distribution function for the poisson binomial distribution .
computational statistics and data analysis , 2013;59:4151 .
the meetup dataset , containing information on organised events by certain meetup group and members of that group that confirmed attendance at an event , allows us to construct member - event bipartite network with adjacency matrix @xmath27 . for each member @xmath64 and event @xmath65 , matrix element @xmath66 if member @xmath10 participated event @xmath30 , or @xmath32 , otherwise .
the degree of member @xmath10 is defined as the total number of events member @xmath10 participated in , @xmath67 , and similarly , the degree of event @xmath30 is defined as the total number of members attended the event , @xmath68 .
given the matrix @xmath27 , social relations between meetup members can be analysed using the projected unipartite member - member weighted network , where the weight of the link between two members is equal to the number of events they both attended .
the observed weighted network is the dense network where some of the non - zero edges can be a matter of coincidence .
for instance , two frequent attendees can meet several times due to a chance not due to the fact that there is some relation between them , which means that the connection between them is not significant for our analysis .
also , the connections between members that meet at big events and never again can not be regarded as social relations and thus they need to be excluded from our analysis . to make the distinction between significant and non - significant edges is nontrivial task @xcite .
here we use the method which enables us to calculate the significance of the link between two members based on the probability for that link to occur in random network . as a null model we use configuration model of bipartite network @xcite .
first we describe general framework for constructing randomized network ensemble @xmath69 with given structural constraints @xmath70 .
the maximum - entropy probability of the graph in the ensemble , @xmath71 , is given by @xmath72 where the @xmath73 are lagrangian multipliers and the partition function of these network ensembles is defined as @xmath74 the ensemble average of a graph property @xmath75 can be expressed as @xmath76 then the constants @xmath77 could be determined from ( [ constraints ] ) .
let us now consider configuration model of the member - event bipartite network with given degree sequence @xmath78 and @xmath79 . in this case
the partition function can be written as @xmath80 the lagrangian multipliers @xmath81 and @xmath82 are determined from @xmath83 @xmath84 finally , we can calculate the probability @xmath35 that a member @xmath10 attended event @xmath30 .
if we define coupling parameter @xmath85 and write partition function in the form @xmath86 then , it holds @xmath87 now , when the probability @xmath35 is given , the probability that members @xmath10 and @xmath36 both participated in event @xmath30 is @xmath88 .
the probability @xmath49 of having an edge of the weight @xmath38 between the nodes @xmath10 and @xmath36 is given by poisson binomial distribution @xmath89 where @xmath40 is the subset of @xmath38 events that can be chosen from given @xmath41 events @xcite .
we use dft - cf method ( discrete fourier transform of characteristic function ) , proposed in @xcite , to compute poisson binomial distribution . on the basis of @xmath49 ,
we define @xmath42-value as the probability that edge @xmath90 has weight higher or equal than @xmath43 @xmath44 the edge @xmath90 will be considered statistically significant if @xmath91 . in our case ,
threshold @xmath92 .
if @xmath93 , the edge @xmath90 should be removed as spurious statistical connection between members ( set @xmath94 ) . | affiliation with various social groups can be a critical factor when it comes to quality of life of every individual , making these groups an essential element of every society .
the group dynamics , longevity and effectiveness strongly depend on group s ability to attract new members and keep them engaged in group activities .
it was shown that high heterogeneity of scientist s engagement in conference activities of the specific scientific community depends on the balance between the number of previous attendance and non - attendance and is directly related to scientist s association with that community . here
we show that the same holds for leisure groups of meetup website and further quantify member s association with the group .
we examine how structure of personal social networks is evolving with event attendance .
our results show that member s increasing engagement in group activities is primarily associated with the strengthening of already existing ties and increase of bonding social capital .
we also show that meetup social networks grow trough big events while small events contribute to the group s cohesiveness . |
in the setting of local quantum field theory , all information on the spectrum resides in the green functions of the theory , as they build up the scattering @xmath0-matrix .
a special class of green functions are the two - point functions .
we should not only consider the elementary propagators of the theory , but also the two - point functions of local composite operators , which correspond to propagating bound states , are relevant .
scattering amplitudes are in general analytic functions of the lorentz invariants constructed from the external momenta characterizing a given process . in the present work we shall only be interested in two - point functions which can be considered as functions of the external momentum @xmath1 , analytically continued to the complex @xmath2-plane with a cut on the real axis starting at @xmath3 , which is identified as the threshold for multi - particle production .
such a two - point correlation function can be cast in the form of a dispersion relation , that is , an integral representation written in terms of the function s discontinuities in the complex plane .
for the two - point function of an arbitrary operator @xmath4 we can write the general expression @xmath5 the imaginary part in the argument of the integral stands for the discontinuity of the function @xmath6 across the cut .
this ought to be a positive definite quantity proportional to the total cross - section , as demanded by the optical theorem .
the representation is known as the klln - lehmann spectral representation . for a concise treatment ,
let us refer to e.g. @xcite .
this klln - lehmann spectral representation can be constructed , for the case of real masses in minkowski space , through the use of the cutkosky cut rules @xcite , which state that , in order to calculate the discontinuity associated with a feynman diagram , we just need to cut propagator lines in the diagram in all possible ways , thereby replacing the corresponding propagators by @xmath7-functions localizing to the physical phase space , and then to sum over all these contributions .
physically , we place the particles running in these lines on - shell . specifying for a moment to the strong interaction , it is well - known that qcd exhibits confinement , i.e. the color - charged elementary degrees of freedom ( quarks and gluons ) do not propagate in the physical world .
the observables are the color - neutral bound states , corresponding to the gauge invariant operators constructed from the gluon and quarks fields , being the mesons , baryons , glueballs and quark - gluon hybrids .
a big contemporary challenge is to extract information on that nonperturbative spectrum using theoretical , phenomenological and/or numerical lattice tools . in the so - called sum rules approach to the qcd spectrum ,
spectral representations play a pivotal role , we refer to the comprehensive tome @xcite , to @xcite for the seminal works , or to @xcite for a pedestrian s introduction . in many approaches to qcd , including the sum rules ,
an important role is played by the elementary propagators , in particular the gluon propagator , which is usually studied in the landau gauge , @xmath8 .
the elementary propagators are the fundamental building blocks of more complicated green functions , including the ones corresponding to bound states @xcite .
for some recent advancements on the gluon propagator , including numerical lattice results , let us refer to the non - extensive list of @xcite .
it is apparent that the landau gauge gluon propagator exhibits a violation of positivity @xcite , a fact which also received direct lattice confirmation @xcite .
this violation of positivity means nothing more than that there can not exist a klln - lehmann representation @xcite , hence the landau gauge gluon is unphysical , and it can be interpreted as a reflection of confinement . in @xcite , it was discussed how @xmath9 can qualitatively and quantitatively describe the lattice predictions for the ( infrared ) landau gauge gluon propagator , and this up to @xmath10 .
the expression can be seen as a dynamically improved version of the so - called gribov propagator , @xmath11 which attracted a lot of attention as a prototype of a confining gluon propagator , and which is related to the dealing with the issue of gauge ( gribov ) copies in the landau gauge @xcite .
given that works out well , it might subsequently be used in explicit computations , for example when one would like to learn something on the glueball or meson spectrum .
we notice that the propagator can be decomposed into 2 yukawa propagators , with either 2 real , or 2 complex conjugate masses , depending on the relative size of the scales present .
the case of 2 complex conjugate masses seems to be realized @xcite , and a ( tree level ) positivity violation is then guaranteed .
the unphysical degrees of freedom corresponding to those complex conjugate masses were called @xmath12-particles @xcite .
also the so - called stingl propagator @xcite , @xmath13 can exhibit a pair of complex conjugate poles , as a special case of .
if we are interested in a spectral function inspired approach to e.g. bound states in theories with propagators of the type , we of course would need to know how to compute with such a propagator .
research in this direction was initiated with the work on a toy model @xcite , where the necessary spectral densities were computed by clever manipulations , which unfortunately do not have general applicability , or , more realistically , for gauge theories in @xcite by using the cutkosky cut rules and assuming that the results can be continued to euclidean space and the case of complex conjugate masses .
the main purpose of this paper is to show that the cut rules results , which we summarized in section 2 , are actually correct .
we shall therefore bring to attention in section 3 the stieltjes integral transformation , as well as its inverse , as studied by widder @xcite , which allows to compute directly in euclidean space the spectral density , even in cases when there is no cut rules prescription .
we shall work out an introductory example in section 4 .
in section 5 , we comment on the use of subtracted spectral representations to avoid divergences .
we then present in section 6 in full detail the most interesting case of 2 @xmath12-particle propagators with complex conjugate masses , wherefore a completely analytical treatment can also be given . in our concluding remarks ,
section 7 , we point out that the widder result to compute the spectral function could perhaps also be used in a numerical context when more complicated propagators than e.g. are used , which are completely inadequate for fully analytical computations .
we assume that @xmath14 is a generic ( composite ) operator containing 2 fields .
we are then interested in the study of the following @xmath15-loop expression @xmath16 where @xmath17 is a function of the lorentz invariants constructed from the momenta flowing in the loop . for simplicity
, we may assume @xmath18 to be a scalar function , thus a fortiori also @xmath17 .
if @xmath18 would carry lorentz indices , say @xmath19 , the eventual correlation function @xmath20 would be a tensor object .
one can always decompose this tensor into a suitably constructed basis ( using the vector @xmath21 and trivial unit tensors ) , and by projecting on the basic tensor structures , one can construct the spectral densities associated to each structure . the latter problem can then be reduced to that of the scalar case by contractions .
for example , assume that the two - point function is a rank 2 tensor , @xmath22 .
each rank two tensor can be decomposed into its transversal and longitudinal part , according to @xmath23 so that knowledge of @xmath24 , resp .
@xmath25 , follows from the ( scalar ) quantity @xmath26 , resp .
@xmath27 . using these scalar objects
, one can the construct the spectral densities corresponding to the transversal and longitudinal piece of the two - point functionb .
the here described procedure can be generalized to more complicated tensorial structures .
we thus wish to find the spectral representation of the integral , and we shall therefore use the cutkosky cut rules . in order to do so , we must assume that @xmath28 and @xmath29 are real , and that we are in minkowski space time .
by cutting the lines comprising this @xmath15-loop diagram and , according to the rules , performing the replacement @xcite @xmath30 the spectral function is proportional to the discontinuity , @xmath31 , and it can be determined from @xmath32\,.\end{aligned}\ ] ] in order to evaluate ( [ cutk - cutrules ] ) we will work in a frame where @xmath33 .
we have then @xmath34\,,\end{aligned}\ ] ] where @xmath35 . in the evaluation of this
integral one must observe that the function @xmath17 comes from possible derivatives in the operator @xmath36 and that it is a lorentz scalar .
it can thus only be a function of @xmath37 , @xmath38 and @xmath39 , which comes from the scalars @xmath40 , @xmath41 and @xmath42 .
after the evaluation we can write the lorentz invariant spectral function @xmath43 by going to an arbitrary frame , whilst replacing @xmath44 . we are ultimately interested , however , in the case where the masses involved are complex and the momenta are in euclidean space .
we then consider expressions like ( [ cutk - eval ] ) as functions of real masses and minkowski momenta and will analytically continue and @xmath29 . ]
it to complex mass values in euclidean momentum space .
we want to draw attention here that , if we would immediately assume @xmath28 and @xmath29 to be complex , we can not apply the cut rules .
we can not put an @xmath12-particle physically on - shell , or mathematically spoken , does not make much sense for @xmath45 while @xmath46 .
we will now illustrate the procedure just described by computing spectral functions in different dimensionalities for the two - point function with @xmath47 .
this corresponds to the toy model introduced in @xcite .
we are thus interested in finding a spectral representation of the form @xmath48 for the following euclidean two - point function @xmath49 with conjugate masses parametrized as @xmath50 we assume that , this is immediately clear . in the context of the model of @xcite , @xmath51 itself also corresponds to the mass of another degree of freedom , as such @xmath51 is supposed to be positive . ] @xmath52 , and without loss of generality we may take @xmath53 .
we shall therefore first use the minkowski formula , and integrate over @xmath54 in ( [ cutk - eval ] ) , obtaining @xmath55\,.\end{aligned}\ ] ] the integrand depends only on @xmath56 so that we can write @xmath57 where we also took into account the theta function .
the remaining delta function can be cast in a more convenient form using the property @xmath58 where @xmath59 is such that @xmath60 . in our case
, @xmath59 satisfies @xmath61 the integral can now be readily evaluated @xmath62 where @xmath63 we obtain @xmath64 we eventually conclude that @xmath65 using @xmath44 and the equivalence @xmath66 .
analogously , for @xmath67 and @xmath68 , we find @xmath69 the threshold @xmath70 is in all cases given by @xmath71
in this section , we shall first refresh the concept of the stieltjes integral transformation , and then discuss how the inverse transformation can be found .
let us assume a function @xmath72 , defined for @xmath73 , and implicitly define another function @xmath74 by means of @xmath75 where we assume that the latter integral exists .
this operation defines the stieltjes integral transformation @xcite . upon comparing and ,
it is clear that a stieltjes representation of the function @xmath72 is nothing else than the klln - lehmann representation if @xmath72 would be a two - point function in euclidean space . if @xmath76 for @xmath77 , @xmath78 sets the threshold @xmath70 .
the spectral density @xmath74 can be reconstructed from @xcite @xmath79\,.\ ] ] let us first prove this statement along the lines of @xcite , as the tools of the proof shall turn out to be useful for our later analysis too .
we start with @xmath80&=&(-1)^{n+1 } \frac{1}{(n!)^2 } { \partial}_t^{n}\left[t^{2n+1}{\partial}_t^{n+1}\int_0^{+\infty}du\frac{\rho(u)}{t+u}\right]\nonumber\\ & = & \frac{n+1}{n!}{\partial}_t^{n}\left[t^{2n+1}\int_0^{+\infty}du\frac{\rho(u)}{(t+u)^{n+2}}\right]\nonumber\\ & = & \frac{n+1}{n ! } \int_0^{+\infty } du \rho(u)u^n { \partial}_t^{n}\left[\frac{t^{2n+1}}{u^n(t+u)^{n+2}}\right]\,.\end{aligned}\ ] ] it seems to be more involved to compute the latter derivative , but we can use a nice trick to do so @xcite . defining @xmath81
we notice that this is a homogenous function of order @xmath82 , as for any @xmath83 , @xmath84 deriving w.r.t . @xmath85 and setting @xmath86 leads to the euler characterization of homogenous functions .
specifically , @xmath87 or for @xmath88 , @xmath89 if @xmath90 is homogenous of order @xmath82 , so is @xmath91 , hence we can iteratively employ to compute @xmath92= ( -1)^n t^{n+1 } { \partial}_u^{n}\frac{1}{(t+u)^{n+2}}=t^{n+1}\frac{(2n+1)!}{(n+1)!}\frac{1}{(t+u)^{2n+2}}\,.\end{aligned}\ ] ] we are thus lead to @xmath93 with @xmath94 in the limit @xmath95 , the latter integral is ideally suited for a steepest descent evaluation , since @xmath96 doing so , we find @xmath97 with @xmath98 and , for @xmath99 , @xmath100 recalling stirling s formula @xmath101 we can write that @xmath102 we then find @xmath103 hereby proving
. once having determined @xmath74 via , one can then define @xmath104 for @xmath105 , which is analytic , with the exception of a branch cut on the negative real axis , @xmath106 @xcite .
obviously , there will be no branch cut for @xmath107 $ ] if @xmath76 for @xmath108 $ ] , with @xmath109 .
using cauchy s formula , one can then see that it actually holds that @xmath110\,,\ ] ] which corresponds to the discontinuity of @xmath111 along the negative real axis .
en route , this explains the dispersion relation . apparently , would provide us a with a much easier way to compute the spectral density @xmath74 , but the difficulty is that in most cases , we do not know how to evaluate @xmath111 from its original ( integral ) definition for @xmath112 .
in contrast with this , only needs @xmath111 for @xmath113 .
this is exactly what we need , since we can always evaluate the euclidean momentum integrals defining a two - point function @xmath114 , in which case we can assume @xmath115 .
afterwards , @xmath111 is _ defined _ by means of for all @xmath105 . in principle , also makes sense when @xmath74 is a distribution .
a basic example is @xmath116 , which leads to the familiar yukawa - propagator @xmath117 , which evidently has no branch cut .
as a warming up exercise , we shall treat here a well - known textbook example , and compute the spectral density in the case of 2 real masses .
this was already treated in @xcite , albeit in minkowski space .
we shall work in euclidean space , and we start from .
we introduce a feynman parameter @xmath118 , yielding @xmath119 ^ 2}\,.\ ] ] the substitution @xmath120 gives @xmath121 with @xmath122 usage of the standard euclidean space formula @xmath123 leads to @xmath124 for the rest of this section , we shall mainly concentrate ourselves on the @xmath125 case .
we are thus interested in @xmath126 setting @xmath127 and dropping the irrelevant prefactor , we may focus on @xmath128 subsequently , @xmath129 where we temporarily set @xmath130 . doing so , we have @xmath131 we already computed the derivative appearing in the r.h.s . of ,
reusing the result yields @xmath132 we notice the great power of this formulation , as the spectral density can now be obtained from a steepest descent evaluation of the feynman parameter integral .
it is useful to perform the substitution @xmath133 to rewrite as @xmath134 with @xmath135 we are now ready to search for the maxima of @xmath136 for @xmath137 $ ] .
solving @xmath138 gives @xmath139 we can discriminate between a few cases .
1 . @xmath140 + in this case , the root appearing in exists .
however , only @xmath141 .
we have @xmath142 we can then look at the behaviour of the integral for @xmath99 , @xmath143 using @xmath144 we find @xmath145 using , we conclude that @xmath76 as @xmath146 .
2 . @xmath147 + in this interval , the roots appearing in are not real , so again only @xmath148 is relevant .
the rest of the reasoning remains valid , so we conclude again that @xmath76 .
3 . @xmath149 + in this case , the roots are again real - valued .
looking at , we notice that @xmath148 now corresponds to a minimum , so we can disregard it .
only @xmath150 and @xmath151 are relevant , as both are positive and correspond to local maxima .
we shall not write down some intermediate results , as they are quite lengthy .
we set @xmath152 it then suffices to notice that @xmath153 which is exactly the value we need to find a nonvanishing finite contribution to the integral . in particular ,
@xmath154 which simplifies to @xmath155 after a bit of algebra .
at the end of the day , we find that @xmath156 it is easily verified that @xmath157 for @xmath158 .
these results are in full agreement with those obtained from the cutkosky rules , see also @xcite , given that we transform from minkowski to euclidean space .
setting @xmath68 in , we clearly stumble upon a divergence @xmath159 . in order to deal with finite quantities
, one usually turns to a suitably subtracted spectral representation .
in particular , if we formally set @xmath160 for a divergent - behaviour of @xmath74 , something which is usually the case . ] quantity @xmath161 , it is clear that a suitable number of derivatives w.r.t .
@xmath118 will render us with a finite result , @xmath162 leading to a finite subtracted spectral representation after @xmath163 integrations from @xmath164 to @xmath118 , @xmath165 in order to know the spectral representation corresponding to @xmath161 , it looks preferential to study the finite function @xmath166 , , and bring it into a spectral representation of the form @xmath167 from which we can also read off the desired spectral density @xmath74 of the original function @xmath161 . using the techniques of section 2 , it is not difficult to show that . ] @xmath168\,,\ ] ] a formula which allows to compute the spectral density @xmath74 from knowledge of the quantity @xmath166 , and this without encountering any infinities at any time . in practice , glancing again at , we would have @xmath169 and upon comparing with , we notice that these are almost identical .
only the `` prefactor function '' of the @xmath170 will be different .
in particular , this means that the saddle point equation etc for @xmath68 remains the same as for @xmath125 , just as almost all the rest of the derivation . only during the last step in computing the spectral density , we shall notice a difference , as only there the `` prefactor function '' plays a role .
the case @xmath67 would need to be worked out from the beginning , but similar tricks as for @xmath125 or @xmath68 can be applied .
this can be easily appreciated as a systematics in the derivatives of @xmath171 exists , the latter being the basic form that will appear in the @xmath67 case .
we now come to the most interesting application of this paper .
we reconsider the integral , rewritten in the form , but now we immediately assume that the occurring mass scales are complex conjugate , and given by
. it can be checked that the `` trick - inspired '' approaches of @xcite to construct the spectral density , do not longer work out in the case that @xmath172 .
we wish to confirm that the results and obtained via a blind trust in the cutkosky approach , are actually correct .
we start our analysis from the first line of .
we have @xmath173 now with @xmath174 in principle , one might try to do a steepest descent evaluation of the integral , keeping in mind that this time one would need to deform the integration contour to follow a line of steepest descent through the equivalent of the saddle points in the complex @xmath118-plane .
this looks as a very complicated exercise , instead we shall follow a somewhat different reasoning . let us first notice that defines a real function , as @xmath175+\int_0^{1/2}\frac{dx}{x(1-x)}\left[\frac{1}{\alpha^*}\left(\frac{\alpha^ * t}{(t+\alpha^*)^{2}}\right)^{n+1}\right]\,,\end{aligned}\ ] ] via the substitution @xmath176 in the 2nd integral .
we can thus say @xmath177\,,\end{aligned}\ ] ] and it suffices to study @xmath178 it would be beneficial if we could use @xmath179 as ( complex ) integration variable , whereby we integrate along the contour @xmath180 , which has the parameter representation @xmath181\to \alpha(x)=\frac{a}{x(1-x)}+ib\frac{2x-1}{x(1-x)}\,.\ ] ] this @xmath180 will start from a point at complex infinity in the right lower half plane and will end on the real axis at @xmath182 .
inverting gives @xmath183 a careful examination learns that the minus sign is the appropriate choice , given that @xmath184 lives in the lower half @xmath179-plane for @xmath185 $ ] .
we can subsequently reexpress as @xmath186 after a little algebra . for the further analysis ,
it is easier to switch the orientation of the contour @xmath180 , so than we can write @xmath187 with @xmath188 as shown in figure 1 , and @xmath189 given by .
[ fig1 ] the presence of the square root in complicates the situation a bit .
setting @xmath190 learns there is a `` crucifix - cut '' given by @xmath191\,,\alpha_y=0\ , .
\end{array}\right.\ ] ] our goal is now to compute by means of a steepest descent approach .
we can use previously gained knowledge , from which we learn that @xmath189 has a saddle point at @xmath192 . in the vicinity of the saddle point
, we may write @xmath193 we can now consider a few cases . 1 . @xmath194 + since everything is analytic in the considered region of the complex plane is of no concern for the analyticity , since @xmath195 . ] , we can deform our contour @xmath188 into @xmath196 , which shares its begin and end point with @xmath188 , such that @xmath196 passes through the saddle point .
we have to choose the orientation of the contour @xmath196 in the vicinity of the saddle point in such a way that the imaginary part of @xmath197 remains constant . from
, it is easily seen that this is the case when we take @xmath179 real for a while until we have passed @xmath198 , and then we can let @xmath196 bend over to let it flow to the correct end point at infinity .
doing so , we find @xmath199 where we made use of . combining this result with gives @xmath200 hence , @xmath201 2 .
@xmath202 + in this region , we can still deform the contour @xmath188 into an appropriate @xmath196 in a completely similar way , as we do not have to cross any cut in order to do so .
since we are below the cut , we find after a completely analogous reasoning as in the first case @xmath203 so that now @xmath204 3 . @xmath205 + at first sight , we can repeat the analysis of subcase 2 .
this reasoning is however flawed
. a basic ingredient of the steepest descent approach is that @xmath206 reaches its maximum along the ( deformed ) contour at the value of the saddle point , in this case @xmath192 .
if we have to first stretch our contour @xmath188 into the left direction to pick up the saddle point , we are violating this assumption , as it can be checked .
the grey region in figure 2 displays the region in the complex @xmath179-plane we can not cross with our contour .
+ [ fig2 ] + the boundaries of these curves are defined by @xmath207 if @xmath208 , the begin point of @xmath188 lies within the white blob , and we can ( only ) follow the @xmath209-axis , pick up the saddle point at @xmath210 , leave this region and go to infinity . if however we start outside this region , there is no way that we can go through the saddle point and go back to infinity without self - crossing the contour .
said otherwise , the steepest descent approach is inapplicable in this case .
4 . @xmath211 + in this region , we seem to be in even more serious trouble , as we can even no longer deform our contour in a useful way , due to the vertical branch cut at @xmath212 . 5 . @xmath213 + let us now present a combined study of the two previous cases , based on a different approach . using the triangle inequality for integrals ,
we have @xmath214 we consequently observe that the ( positive ) function @xmath215 has maxima at the solutions of @xmath216 namely , @xmath217 with @xmath218 on the interval @xmath219\times[0,2a]$ ] , this implies that , for a certain @xmath220 , @xmath221 and using this we can majorate further as @xmath222 , and thus also @xmath76 . combining all information gathered so far
, we have shown that @xmath223 whereby we can observe that @xmath224 for @xmath225 . as a check on the final result
, we notice that its @xmath226 limit coincides with the @xmath227 limit of the earlier obtained result . for the general case , we could of course explicitly compute the spectral integral of , as well as the original feynman parameter integral , and verify if both results are the same .
we find it however more instructive for the reader to display in figure 3 both results for the explicit example @xmath228 , @xmath229 , making their equivalence clearly visible .
+ similar conclusions can be reached for @xmath67 or @xmath68 . in all cases ,
we thus find perfect agreement between the results obtained by application of the cutkosky cut rules , and those obtained by using the widder stieltjes formalism .
we have demonstrated the usefulness of the stieltjes integral transform to compute the klln - lehmann spectral density of ( euclidean ) two - point functions , even in cases where the validity of the usually employed cutkosky rules is not clear .
the main recipes are given in sections 4 and 5 , in particular in expression which gives the spectral density @xmath74 in terms of the two - point function @xmath114 , when the latter is known for euclidean momenta @xmath230 .
we have paid particular attention to the case of gribov - like propagators , which entails the presence of propagators with 2 complex conjugate masses .
it turns out that , at least at one loop , this case can we worked out analytically to the end using a steepest descent approach .
our main conclusion is that the results , obtained by the cut rules for real masses and in minkowski space , may be continued to euclidean space , while also the two real masses can formally be replaced by two complex conjugate values .
this kind of results already found use in recent works like @xcite , and have now been proven to be correct . however , the inversion formula is also valid in cases where the cut rules are unavailable .
we recall that many works have been devoted to the study of the quantum equations of motion ( schwinger - dyson formalism ) , giving also estimates for the gluon propagator , amongst other quantities @xcite . in the deep infrared and ultraviolet
, an analytical approach is possible . in the intermediate momentum regime , however , numerical solutions are in order due to the complexity of the system .
the numerical outcome can then fitted by well - guessed functional forms .
let us mention two major examples @xmath231^{-3/5}\,,\nonumber\\ f(k^2,m_0 ^ 2)&=&\rho_1 m_0 ^ 2+\rho_2\frac{m_0 ^ 4}{q^2+m_0 ^ 2}\,,\quad \rho_1=-\frac{1}{2}\,,\quad\rho_2=\frac{5}{2}\,,\quad m_0=612\,\text{mev}\ , , \quad\lambda=645\,\text{mev}\,,\end{aligned}\ ] ] which represents the so - called `` massive '' @xcite ( also known as `` decoupling '' @xcite ) solution , or @xmath232^{\frac{13}{22}}\,,\nonumber\\ \beta_0&=&\frac{11n}{3}\,,\quad\alpha(0)=\frac{8.915}{n}\,,\quad \lambda=710\,\text{mev}\,,\end{aligned}\ ] ] with @xmath233 for the so - called `` scaling '' solution @xcite .
it is evidently out of the question to apply the standard cut rules with the foregoing propagators , since they are not even of the form @xmath234 , so the replacement rule loses its meaning .
the theory of the ( stieltjes ) inversion could however be applied , albeit perhaps rather in a numerical fashion , employing stable approximations for the infinite numbers of derivatives appearing in formula .
let us also mention here that in many cases , the concrete application of the cutkosky rules become intractable beyond one loop , due to the highly complicated phase space integrals . in the particular case of bound states , like a glueball or meson , which can crucially depend on a viable input for the gluon propagator
, one might imagine to use propagators of the type or to compute , in one approximation or another , suitable two - point functions , to obtain consequently an estimate for the spectral density , and then eventually use the latter as input for whatever method one likes to employ to find estimates for e.g. bound state masses
. one might think about employing laplacian sum rules as in @xcite or techniques based on pad approximants as in @xcite , to name only a few approaches to the bound state problem whereby the spectral density enters .
we end by mentioning that or are quite different in the deep infrared as tends to a strictly positive constant for @xmath235 , while vanishes in the same limit .
the former scenario is recovered on the lattice @xcite .
one could nevertheless wonder if both solutions could reproduce more or less the same physics , as preliminary investigated in @xcite in a specific example and approximation .
we hope to come back to this issue in the future , using the tools developed in this paper .
the content of section 5 should be of particular interest , as also the propagators or will lead to divergent spectral integrals .
we wish to thank s. p. sorella , n. temme and n. vandersickel for useful discussions .
d. dudal is supported by the research - foundation flanders ( fwo vlaanderen ) , while m. s. guimares acknowledges financial support by faperj , fundao de amparo pesquisa do estado do rio de janeiro .
d. dudal acknowledges the hospitality at the uerj where this work was initiated and finished . | we present a steepest descent calculation of the klln - lehmann spectral density of two - point functions involving complex conjugate masses in euclidean space .
this problem occurs in studies of ( gauge ) theories with gribov - like propagators .
as the presence of complex masses and the use of euclidean space brings the theory outside of the strict validity of the cutkosky cut rules , we discuss an alternative method based on the widder inversion operator of the stieltjes transformation .
it turns out that the results coincide with those obtained by naively applying the cut rules .
we also point out the potential usefulness of the stieltjes ( inversion ) formalism when non - standard propagators are used , in which case cut rules are not available at all . |
the work of h.c.k has been supported by the research fund of pusan national university .
the work of m.p . has been supported by polish kbn grant pb 2 p03b 019 17 .
the work of k.g . has been supported by the bmbf , the dfg , and the cosy
project(jlich ) .
kim , m. praszaowicz , and k. goeke , phys . rev .
* d61 * ( 2000 ) 114006 .
kim , m. praszalowicz and k. goeke , acta phys .
* b31 * ( 2000 ) 1767 ; h .- ch .
kim , m. praszalowicz and k. goeke , hep - ph/0007022 . | we present the recent investigation on the effect of the su(3 ) symmetry breaking in the spin structure of the octet baryons .
-3.5 in 3.3 in the spin structure of the nucleon has been a hot issue since the measurement of the first moment of the proton spin structure function @xmath0 @xcite by the european muon collaboration ( emc ) .
the emc result indicates that the strange quark is polarized in the opposite direction to other light quarks .
it implies that the spin of the nucleon is not carried by quarks only . while it is now known that a large fraction of the nucleon spin is provided by gluons and their orbital angular momenta , it is still very important to understand the mechanism of how the quarks carry the nucleon spin .
the information on the nucleon spin carried by quarks comes from two different sources : the first moment of the spin structure function @xmath1 of the proton and the experimental data of hyperon semileptonic decay ( hsd ) . with su(3 )
symmetry assumed , one obtains two reduced matrix elements @xmath2 and @xmath3 .
taking for @xmath4 and for @xmath5 , one gets for the proton : @xmath6 , @xmath7 and @xmath8 , which implies @xmath9 , quite a small number as compared with the naive expectation from the quark model : @xmath10 .
recently , the spin content of the @xmath11 hyperon has drawn attention to us .
in contrast to the nucleon , the nonrelativistic quark model predicts that the spin of the @xmath11 comes solely from the strange quark , since the up and down quarks constitute the spin singlet .
furthermore , the @xmath12 is deeply related to the fragmentation funtion which can be measured experimentally @xcite
. however , note that @xmath13 and @xmath14 are _ not directly measured_. the standard way to extract @xmath15 is to assume the su(3 ) symmetry for hsd . in this case
it is enough to take _ any _ two decays and data of deep inelastic scattering as input .
with su(3 ) symmetry assumed , one can pick out two different values of the constants of hsd among the existing data .
one normally uses neutron and @xmath16 @xmath17 decays as an input .
however , if the su(3 ) symmetry breaking were not important , any pair out of the six known hsd constants would give roughly the same number for @xmath15 .
this is , however , _ not _ the case .
in fact , the effect of su(3 ) symmetry breaking in hsd is not at all small .
furthermore , the uncertainty of the hsd constant for @xmath18 is noticeably larger than those for the other processes .
thus , it is still of great significance to understand the effect of su(3 ) symmetry breaking on the quark - spin content of the nucleon . in order to analyze the symmetry breaking in weak decays , it is inevitable to resort to some specific model @xcite .
we employ the chiral quark - soliton model(@xmath19qsm ) : see ref.@xcite and references therein ) to incorporate the su(3 ) symmetry breaking into the semileptonic decays .
the @xmath19qsm satisfactorily describes the axial - vector properties of hyperons ( see ref .
@xcite and references therein . ) .
since the symmetry breaking pattern of the @xmath19qsm is identical to the one derived in large @xmath20 qcd @xcite , our analysis is in fact much more general than the model itself .
in addition , we employ the experimental data for hsd in order to fix the dynamical variables in the model , so that we can minimize the model - dependence of the present study . however , since @xmath21b@xmath22 does not correspond to the su(3 ) octet axial - vector current , a further assumption is required to connect it to the octet axial - vector constants . the @xmath19qsm ( as most of the hedgehog models ) has a remarkable virtue of connecting the @xmath21b@xmath22 with @xmath23 and @xmath24 , and the semileptonic decay constants in a direct manner .
this connection introduces a model dependence into our analysis . however , there is no significant numerical difference between the results obtained with and without this model dependent ingredient @xcite .
thus , we would like to report in brief the recent investigation on the effect of the su(3 ) symmetry breaking , based on refs .
@xcite , to which we refer for details . figure 1 shows the dependence of the integrated quark densities @xmath25 of the nucleon on the strange quark mass @xmath26 , of which definitions can be found in ref .
@xcite : while the up and down quark densities have smaller errors as @xmath27 increases , the uncertainty of the @xmath28 becomes much larger . with @xmath29 mev
, we obtain @xmath30 .
it indicates that @xmath28 is in practice undetermined .
the reason is that the error of the semileptonic decay constant for the process @xmath31 propagates strongly in the @xmath28 .
figure 2 plots the flavor decomposition of the quark - spin content of the @xmath11 .
as in the case of the nucleon , the error of the @xmath32 is noticeable with @xmath27 increased .
figure 3 depicts the splitting of the @xmath33s due to the strange quark mass .
the result is according to the following equation : @xmath34 where @xmath35 is the third component of the spin operator .
the @xmath27 stands for the strange quark mass which plays a role of breaking su(3 ) symmetry .
@xmath36 is the matrix element of the su(3 ) wigner function .
the @xmath37 denote the dynamical variables from the model which are fixed by the hsd data .
they appear also in the octet axial - vector constants , which makes it possible to link the octet axial - vector constants to the octet one . actually , the simplest version of the sykrme model without vector mesons gives @xmath38 . in the case of the @xmath19qsm provides nonzero @xmath39 . in the limit of the large soliton
, the ratio of @xmath40 goes to zero , which is in agreement with ref .
@xcite .
when the size of the soliton becomes closer to zero , which corresponds to the limit of the nonrelativistic quark model , @xmath41 turns out be one , as it should be .
the removal of the degeneracy in @xmath42s comes from the second term in eq.([ga0 ] ) .
while the @xmath43 and @xmath14 show relatively mild dependence on @xmath27 , the @xmath44 and @xmath45 depend quite strongly on @xmath27 .
large error bars with @xmath27 is due to the propagation of the uncertainty in @xmath46 decay to the variables @xmath37 . from figure 3
, we find a very interesting thing .
with the effect of the su(3 ) symmetry breaking , @xmath14 and @xmath44 get larger and finally quite closer to one .
it indicates that the spins of @xmath11 and @xmath46 are almost carried by quarks .
that is , @xmath11 and @xmath46 behave more `` nonrelativistically '' .
hence , extracting the @xmath47 from the experimental data as precisely as possible and comparing it with that of the nucleon , one would see the effect of the su(3 ) symmetry breaking , explicitly . in summary ,
we have presented in this talk a recent investigation on the effect of the su(3 ) symmetry breaking within the framework of the chiral quark - soliton model , incorporating the experimental data of hyperon semileptonic decay . because of the large uncertainty in the data for the @xmath46 decay , one is not able to determine the @xmath48 , which means that one can not tell for sure about how much the strange quark carries the nucleon spin .
the su(3 ) symmetry breaking removes the degeneracy of the @xmath49s .
therefore , knowing the difference between the @xmath43 and @xmath47 will bring out the effect of the su(3 ) symmetry breaking .
the @xmath11 and @xmath46 behave more `` nonrelativistically '' than the nucleon . in order to determine the spin content of the su(3 ) baryons carried by quarks
, it is highly required to reduce the experimental data of @xmath46 semileptonic decay . |
speckles are high - contrast fine - scale granular patterns occurring whenever radiation is scattered from a surface characterized by some roughness on the scale of the radiation wavelength .
originally discovered for laser light in the early 1960s , the speckle phenomenon plays an important role not only in optics , but also in other fields , where it is used for several applications , including , for example , radar and ultrasound medical imagery @xcite .
in recent years , optical speckles have been employed in combination with cold atoms @xcite and especially bose - einstein condensates ( becs ) @xcite in order to investigate the behavior of matter waves in the presence of disordered potentials @xcite .
many interesting features of becs in one - dimensional ( 1d ) speckle potentials have been addressed from both the experimental and the theoretical sides .
these include classical localization and fragmentation effects , frequency shifts and damping of collective excitations , and inhibition of transport properties @xcite , and have culminated with the observation of anderson localization for a bec @xcite .
also , the superfluid - insulator transition @xcite and the transport of coherent matter waves have been recently investigated from the theoretical point of view for speckles in higher dimensions @xcite . despite this intense research activity ,
the properties of the single particle spectrum of the speckle potential have been addressed only partially , even in the 1d case . in @xcite
it has been argued that for low energies the density of states ( dos ) of _ blue - detuned _ repulsive speckles is characterized by a lifshitz tail @xcite , whereas the presence for higher energies of an effective mobility edge , has been discussed in @xcite . in addition , though there is a vast literature about random 1d systems @xcite , most of the general theorems apply strictly to the case of uncorrelated disorder , whereas the speckles , being a correlated disordered potential , deserve an explicit treatment . in this article
we discuss the properties of dos for a 1d speckle potential , by comparing analytical predictions , based on the statistical properties of the speckles , with explicit calculations for the spectrum of numerically generated speckle patterns .
we consider both _
blue - detuned _ ( repulsive ) and _ red - detuned _ ( attractive ) cases .
the speckles are characterized by their intensity @xmath0 and a correlation length @xmath1 , which represents the length scale over which the modification of the potential occurs .
therefore , for dimensional reasons , the single - particle properties are determined by a single dimensionless parameter @xmath2 , with @xmath3 being the mass of the particle . in our conventions , the plus and the minus signs refer to the blue - detuned and red - detuned speckle , respectively . in the presence of disorder ,
the asymptotic behavior of the dos near the edges of the spectrum is dominated by rare large fluctuations of the potential . for the red - detuned potential , the low - energy tail of the dos originates from the deep wells at negative energies .
the latter is closely related to the distribution of intensity maxima in the speckle patterns . in the case of a blue - detuned speckle the spectrum is bounded from below .
the dos near the edge results from the large regions with very small intensity .
for both cases we analyze the behavior of the dos throughout the @xmath4 range , from the semiclassical limit @xmath5 , down to the quantum regime @xmath6 , where smoothing effects become important .
the article is organized as follows . in secs .
[ sec : speckle ] and [ sec : weinrib ] we introduce the speckle potential and its statistical properties . then in sec .
[ sec : red ] we discuss the dos for the attractive _ red - detuned _ speckles , for different regimes of @xmath4 .
the case of the _ blue - detuned _ repulsive speckles is considered in sec .
[ sec : blue ] .
our results are summarized in sec .
[ sec : sum ] .
the electric field @xmath7 created by a laser speckle on a 1d euclidean line at the image plane is , to a very good approximation , a realization of a complex gaussian variable . in general , @xmath7 can be taken as statistically isotropic in the complex @xmath8 plane and statistically homogeneous and isotropic on the @xmath9 line .
thus , @xmath10 and @xmath11 vanish while the autocorrelation function is given by @xcite @xmath12,\ ] ] where @xmath13 and @xmath14 is the aperture width .
the probability distribution of intensity @xmath15 across a speckle pattern follows a negative - exponential ( or rayleigh ) distribution @xcite , @xmath16=\exp{[-i / i_0]}/i_0\ ] ] where @xmath17 is the mean intensity while the most probable intensity is zero .
the intensity autocorrelation function is given by @xmath18 $ ] .
the ( auto ) correlation length @xmath1 of the speckle potential is defined through the equation @xmath19 , which gives the width at the half maximum of the autocorrelation function .
it is related to the aperture width by @xmath20 .
when the speckle electric field is shined on a sample of atoms , this results in a disordered potential felt by the atoms , that is proportional to the local intensity of the speckles , @xmath21 with @xmath15 . in general ,
if the wavelength of the light is far detuned from the atomic resonance , the constant @xmath22 is proportional to the inverse of the detuning @xmath23 , and can be either positive or negative @xcite .
this corresponds to a disordered potential bounded from below and composed by a series of barriers ( @xmath24 , _ blue detuning _ ) or bounded from above and made of potential wells ( @xmath25 , _ red detuning _ ) . in what follows
, we include @xmath26 in redefinition of @xmath27 , which is positive as well as @xmath28 .
then we use @xmath29 and @xmath30 for the blue - detuned and red - detuned speckles , respectively .
the quantum properties of a particle of mass @xmath3 in the speckle potential are determined by the solutions of the schrdinger equation , @xmath31\psi(x)=e{\psi(x)},\ ] ] with @xmath32 being the particle wave function .
note that there are two relevant length scales in the problem @xcite .
the first one is the correlation length @xmath1 of the disorder which introduces the corresponding energy scale , @xmath33 additionally to this length scale , the particle of mass @xmath3 moving in the random potential with typical strength @xmath27 introduces a new length scale @xmath34 defined by @xmath35 by comparing @xmath27 with @xmath36 the dimensionless parameter @xmath4 can be written as @xmath37 for large values of @xmath38 the disorder is strongly correlated while for @xmath39 the limit of uncorrelated disorder is approached . expressing the intensity of the speckle potential in units of @xmath27 so that @xmath40 and the length in units of @xmath1 , we can rewrite the hamiltonian in a dimensionless form as @xmath41 with a single parameter @xmath4 . and the coordinate in units of @xmath1 . ]
once the electric field correlation function ( [ eq : sinc_one_d ] ) is known , the statistical properties of the intensity are completely determined .
this is discussed in the next section , where we consider the statistics of extrema of this correlated and non - gaussian potential , relevant for the motion of a particle in a random landscape @xcite . by using this description ,
we derive the dos and compare the analytical predictions with the spectrum computed numerically for a randomly generated speckle pattern . to generate the speckle potential profile we use the mathematical counterpart of the mechanism that gives rise to optical speckles .
the constructive or destructive interference of randomly phased elementary components of the radiation field caused by the roughness of the scattering surface occurs also in the discrete fourier transform of a sample function of a random process @xcite .
this effect can be exploited to generate numerically a speckle pattern with the statistical properties discussed earlier @xcite .
then , the spectrum of the system can be computed numerically by mapping the stationary schrdinger equation @xmath42 on a discretized grid @xcite .
all the results presented in this article have been obtained for a system of length @xmath43 and are averaged over a number of realizations ranging from @xmath44 to @xmath45 .
[ sec : weinrib ] the typical shape of the potential profile created by a blue - detuned speckle pattern is shown in fig . [
fig : speckle ] .
the potential profile can be seen as series of peaks ( maxima ) statistically distributed around the typical value of the order of @xmath27 .
the peaks are separated by valleys of typical width @xmath1 with minima situated most probably near the lower - boundary of the potential . for the red - detuned case
the picture is analogous , with the roles of maxima and minima exchanged . in other words , the statistics of the potential minima for the red - detuned case
is that of the maxima of the blue - detuned one .
we are interested in the density of minima @xmath46 and the density of maxima @xmath47 , at specified intensity @xmath48 , in the spatial distribution @xmath28 of intensities along the image line . using the method of weinrib and halperin @xcite ,
the number of minima or maxima can be calculated from the integrated joint probability , @xmath49 where the plus ( minus ) sign refers to minima ( maxima ) and we have used the conventions @xmath50 and @xmath51 .
it follows that the density of minima ( maxima ) per unit length and intensity can be written as @xmath52 in order to evaluate the statistical average of eq .
( [ eq : dens_min_one_d ] ) , we need to calculate the joint probability of @xmath53 , @xmath54 and @xmath55 ( all at the same point ) starting from the joint probability of the gaussian random variables @xmath8 , @xmath56 , and @xmath57 . because @xmath7 is statistically homogeneous along the @xmath9 coordinate , the density in eq .
( [ eq : dens_min_one_d ] ) will be independent of position .
of minima ( top ) and maxima ( bottom ) for a blue - detuned speckle ( in log scale in the insets ) .
the solid lines represent the analytical prediction of eq .
( [ eq : rhomin ] ) , while the dots have been obtained by averaging over @xmath58 numerical speckle realizations . ] the statistics of a set of gaussian variables is completely determined by the two - point correlations @xmath59 between them .
the nonzero correlators evaluated at the same point are calculated by considering the derivatives of @xmath59 .
one finds @xmath60 with @xmath61 where we have introduced the length @xmath62 for convenience .
the dimensionless parameter @xmath63 is restricted in one dimension to the range @xmath64 .
it depends only on the the shape of the correlation function of the electric field but not on the length scale @xmath1 . for the correlator of the laser speckle given by eq .
( [ eq : sinc_one_d ] ) , we have @xmath65 . on dimensional grounds ,
it is clear that the density of minima ( maxima ) of eq .
( [ eq : dens_min_one_d ] ) scales as @xmath66 evaluating the statistical average in eq .
( [ eq : dens_min_one_d ] ) and rewriting in terms of dimensionless intensity measured in units of @xmath27 , we obtain @xmath67 where @xmath68 $ ] , + @xmath69 /\left(\kappa^{-1}-1\right)\right\}}$ ] , and the plus ( minus ) sign refers to the minima ( maxima ) .
the details of the derivation are given in the appendix .
the densities ( [ eq : rhomin ] ) are plotted in fig .
[ fig : min_max ] .
the agreement with the numerical simulation is excellent . the asymptotic behavior of the integrals in eq .
( [ eq : rhomin ] ) can be also evaluated analytically .
this is useful because the distribution of the deep negative - energy wells of the red - detuned speckle potential is related to the large @xmath53 asymptotic tail of the maxima distribution of the blue - detuned speckle .
this latter behaves for @xmath65 as @xmath70}{\sqrt{\tilde{i}}}\right ) \nonumber \\ & \approx & { \sqrt{\frac{\tilde{i}}{\pi } } } e^{-\tilde{i}}.\end{aligned}\ ] ] near the boundary of the blue - detuned speckle potential the density of minima diverges as @xmath71 however , in the thermodynamic limit , the dos of a particle in disordered potential bounded from below is expected to vanish at the boundary .
hence , the correlation between the dos and the number of minima has to be rather weak . between neighboring maxima ( dotted line ) and between neighboring minima ( solid line ) .
] more likely , the dos near the boundary of the spectrum is associated with the distribution of distance between neighboring minima @xmath72 and neighboring maxima @xmath73 .
it is known that the presence of lifshitz tails in the dos for a bounded - from - below potential is related to the existence of large regions free of disorder for the binary distribution or large regions with negligible potential for continuous distributions . in the case of _ blue - detunded _ speckle potential
, the zero intensity is the most probable so that one would expect that such regions are not so rare .
some idea about their statistical properties can be gained through the distributions of distances between minima and maxima , which can be computed numerically .
the normalized distributions of the distance between maxima and between minima are shown in fig .
[ fig : distance ] .
the average distance for the both distribution is @xmath74 .
this value can be easily obtained by integrating the density in eq .
( [ eq : rhomin ] ) over all the intensities and by assuming that the points of minima ( or maxima ) are homogeneously distributed .
the right tails of both distributions can be well fitted in the window @xmath75 $ ] by simple exponentials : @xmath76 the exponentially small probability of existence of a region of size @xmath77 with relatively small intensity accounts for the lifshitz tail in the dos of the blue detuned speckle potential .
we consider first the dos for a particle in the red - detuned speckle potential , which corresponds to @xmath78 .
we discuss separately three different regimes , ranging from the semiclassical limit , for @xmath79 , down to the quantum regime , when @xmath80 .
by varying the parameter @xmath4 the characteristic extension of the localized states differs considerably . in fig .
[ fig : eigenstates ] we present the lowest - lying eigenstates for three different intensities of the same speckle profile , corresponding to @xmath81 .
this picture shows that for shallow potentials ( the _ quantum _ regime ) the eigenstates extend over several potential wells .
the lowest - lying eigenstates localize inside deep wells .
however , the ground state is not necessarily localized in the deepest well since the width of the well plays also a crucial role . as the speckle intensity
is increased , the eigenstate width shrinks and their positions may change . eventually , for large enough @xmath38 , the eigenstates tend to stack up as bound states inside isolated wells , with the ground state being the lowest eigenstate of the deepest well ( _ semiclassical _ regime ) . .
only a window of length @xmath82 of the whole system is shown .
the figure shows the ground state ( blue solid line ) and several first excited states ( black dashed line ) .
note that , as the speckle intensity changes , the position of the eigenstates may change and some may move in or out from the selected window . ] in the limit @xmath79 the low - energy tail of the dos is determined by deep isolated states occupying single wells of typical size @xmath1 .
therefore , for @xmath83 , the dos can be calculated in the semiclassical approximation @xcite @xmath84 where @xmath85 is the density of state of the continuum in one dimension .
integrating over the intensity in eq .
( [ eq : tf ] ) we find @xmath86 the dos in the negative - energy tail predicted by eq .
( [ eq : tf_d_less ] ) is in excellent agreement with the curve obtained in the numerical simulation as shown in fig .
[ fig : semiclassical ] .
note that this result differs from the dos of a long - range correlated gaussian potential that , sufficiently deep in the tail , has the form @xmath87 @xcite .
the semiclassical result of eq .
( [ eq : tf_d_less ] ) can be also understood in terms of the density of minima of the red - detuned speckle potential . since the latter is the blue - detuned speckle potential taken with negative sign , the distribution of minima is nothing but @xmath88 , namely , the distribution of maxima of eq .
( [ eq : dens_min_one_d ] ) studied in sec .
[ sec : weinrib ] .
we assume that each deep state occupies an isolated single well associated with the corresponding minimum of the potential .
this approximation is not true in general but is valid for deep enough states .
the potential inside of the well can be completely described by derivatives taken at the bottom of the well , denoted by @xmath89 .
each well is characterized by @xmath90 $ ] bound states with energies @xmath91 , @xmath92 , which can be found from the solution of the schrdinger equation for the particle in the potential @xmath93 .
the dos can be then calculated by summing over all levels in each well and over all wells as follows @xmath94 } n_{\rm min}^{\rm red}\left(\{i\}\right ) \delta\left(e - e_{n}\left[\{i\}\right]\right).\ ] ] for large @xmath1 the typical wells which dominate the dos at large negative energies can be considered within a harmonic approximation by setting @xmath95 , with @xmath96 so that the resulting spectrum corresponds to a quantum oscillator with the frequency @xmath97 .
the summation over @xmath98 can be taken to infinity because in the limit @xmath99 the levels of the harmonic oscillator approach a continuum .
therefore , for @xmath83 we can approximate ( [ eq : harm ] ) as @xmath100 we have computed the statistical average and carried out the summation over @xmath98 in eq .
( [ eq : dens_max_one_d_mod _ ] ) by summing over large numbers of states .
the obtained curve is found to be exactly on the top of the semiclassical curve of eq .
( [ eq : tf_d_less ] ) shown in fig .
[ fig : semiclassical ] .
corrections beyond the harmonic approximation and/or correlations between different points could be taken into account , including higher - order derivatives of @xmath28 at the bottoms of wells . in the presence of a red - detuned speckle potential with @xmath101 as a function of the rescaled energy @xmath102 in linear scale and in logarithmic scale ( inset ) .
the dots represent the numerical data ( averaged over @xmath58 realizations ) , while the solid line indicates the analytical semiclassical curve ( [ eq : tf_d_less ] ) . ] in a red - detuned speckle potential as a function of @xmath102 for different values @xmath78 : @xmath101 ( red squares ) , @xmath103 ( blue circles ) , @xmath104 ( open circles ) and @xmath105 ( green triangles ) .
the black solid line refers to the analytical semiclassical result ( [ eq : tf_d_less ] ) valid at large @xmath38 . ] upon decreasing @xmath38 the quasiclassical approximation becomes inadequate , as shown in fig .
[ fig : red_univ_semic ] . in general , at sufficiently large negative energies @xmath106 , the dos is determined by the deep wells of size @xmath107 no matter how short the correlation length @xmath1 is @xcite . at higher energies
@xmath108 , the wave function of the localized state is spread over more than one typical fluctuation of the disorder potential , that is , @xmath109 .
this picture is supported by the numerical analysis of the spatial extension of the localized states in the low - energy tail .
as one can see from fig .
[ fig : eigenstates ] the wave functions for @xmath105 corresponding to the energy range shown in fig .
[ fig : red_univ_semic ] spread over a region @xmath110 .
the same analysis for @xmath104 reveals that the typical extension of the wave function for the low - energy states is of the same order as @xmath1 .
note that , since the condition @xmath106 corresponds to @xmath111 , it is not possible to see the transition to the deep localized levels at @xmath106 from the data of fig .
[ fig : red_univ_semic ] for @xmath6 . here
we study the low - energy tail of the dos for moderate values of @xmath112 .
this value occurs typically in the experiments with cold atoms in optical speckles @xcite . in this regime
the semiclassical approximation breaks down due to quantum - mechanical effects despite that the interaction between different wells remains negligible .
the decrease of @xmath38 results in a depletion of the dos seen in fig .
[ fig : red_univ_semic ] .
this effect can simply be explained by the reduction of the number of levels in each potential well . in the harmonic approximation ( [ eq : harm ] ) , one does indeed find that decreasing @xmath1 at fixed @xmath34 leads to an increase of the distance between levels .
alternatively , reducing @xmath27 at fixed @xmath1 causes a shrinking of the interlevel distance which is counterbalanced by a larger decreasing of the wells depth .
thus , the decrease of @xmath4 always corresponds to a reduction of the number of bound states in each well . in this case
we can not rely on the infinite - sum approximation .
taking into account only a ground state in each well severely underestimates the dos . keeping any constant finite upper limit
@xmath113 $ ] in the sum over levels remains inadequate due to a quite broad distribution of the number of bound states in different wells .
a different approach has to be considered .
one method is to extend the semiclassical approximation given by eq .
( [ eq : tf ] ) in order to include quantum effects . to this end
it is useful to introduce the 1d cumulated dos @xcite @xmath114 which is related to the total number of states in a well of size @xmath115 created at the minimum @xmath116 of the potential @xmath117 , @xmath118 the condition to have at least one state requires that this number is larger than 1 , that is , @xmath119 .
this can be incorporated in the cumulative dos by shifting the energy zero @xcite : @xmath120 in logarithmic scale and in linear scale ( inset ) .
the solid line describes the modified semiclassical approximation of eqs .
( [ eq : spectr_dens_lloyd])-([eq : fit ] ) with @xmath121 valid in the low - energy tail . at positive energies , the dos approaches asymptotically the 1d continuum @xmath122 , indicated by the dashed line . ] changing from the cumulated dos to the dos we arrive at @xmath123}{\sqrt{e - v-\frac{\hbar^2}{2 m \lambda^2 } } } ~d v,\ ] ] where @xmath124 $ ] is the normalized probability distribution function of minima of the red - detuned speckle potential , so that @xmath125 $ ] is the similar distribution of maxima of the blue - detuned speckle : @xmath126=\frac{n_{\rm max}(i,\xi,\kappa ) } { \int_0^{\infty } n_{\rm max}(i',\xi,\kappa)~d{i^{'}}}.\ ] ] moreover , because we know that the extension of the wave function of these states does not exceed the typical size of the potential wells , we can take the energy of the localized state proportional to the zero - point energy inside the well , that is , @xmath127 where the parameter @xmath128 can not be determined exactly within this method .
however the result is not extremely sensitive to the choice of @xmath128 in the limits @xmath129 . choosing the value @xmath130 the modified semiclassical method based on eqs .
( [ eq : spectr_dens_lloyd ] ) and ( [ eq : prob_min ] ) reproduces very well the data for @xmath104 , as shown in fig .
[ fig : intermediate ] .
we also tried to fit the computed dos by a stretched exponential .
we found that it can not be done well in the whole window of numerical data .
however , in smaller windows it can be fitted by a stretch exponential with the exponent varying between 1 and 1.5 and weakly depending on @xmath128 . when the correlation length of the disorder @xmath1 is reduced to @xmath131 , the wave function of the localized states in the low - energy tail @xmath132 is spread over a distance @xmath115 larger than the typical distance between two minima ( see fig .
[ fig : eigenstates ] ) .
therefore , the probability of constructing a local wave function is not directly related to the probability distribution of the bare potential @xmath133 $ ] or that of its extrema @xmath134 $ ] .
the interplay between different wells has to be taken into account as well .
this can be done by considering the probability @xmath135 $ ] of the potential integrated over the characteristic extension @xmath136 of the state of energy @xmath137 @xcite . for finite @xmath138
we can divide the region occupied by the wave function into @xmath139 parts , which are approximately uncorrelated .
assuming that the correlation inside each part is strong enough we can approximately replace the distribution of potential integrated over the correlation length @xmath140 $ ] by the bare distribution ( [ eq : ray ] ) .
then the distribution integrated over the width of wave function @xmath135 $ ] can be computed using the characteristic function method .
the characteristic function is defined as @xmath141 so that @xmath142^n$ ] .
this gives @xmath143 = \frac1{i_0 } \frac{n^n}{\gamma(n ) } \left(\frac{i}{i_0}\right)^{n-1 } e^{-n i / i_0},\ ] ] with @xmath139 .
it can be shown that for very large @xmath115 the integrated exponential distribution ( [ eq : fin - n ] ) approaches asymptotically a gaussian law : @xmath144\stackrel{\lambda\gg 1}{\sim}\frac{\sqrt\lambda}{\sqrt{2\pi}i_0\sqrt{\xi } } \exp{\left(-\frac{\lambda(i - i_0)^2}{2i_0 ^ 2\xi}\right)}.\ ] ] by reversing the argument , we can say that for a given energy there is an optimal well of size @xmath115 created by the average potential @xmath145 , which may bind a state whose zero - point energy should not be larger than @xmath146 .
therefore , the negative low - energy tail of the dos can be calculated with a variational principle from @xcite @xmath147}{\sqrt{e - v-\frac{\hbar^2}{2 m \lambda^2 } } } ~d v.\ ] ] by making the variation with respect to @xmath115 , for large negative energies @xmath148 , we find that the dos of eq .
( [ eq : spectr_dens_lifs ] ) goes asymptotically as @xmath149 the limit @xmath150 of very short disorder correlation length corresponds thus to the white - noise potential @xmath151 where @xmath152 has to remain finite as @xmath153 .
therefore , we see that the asymptotic dos ( [ eq : spectr_dens_lifs_bis ] ) reproduces the typical lifshitz tail law for unbounded delta - correlated potential @xmath154 where @xmath155 is nothing but the 1d larkin energy @xcite . .
the black solid line corresponds to the variational result of eq .
( [ eq : spectr_dens_lifs ] ) with @xmath135 $ ] given by eq . ( [ eq : fin - n ] ) .
the inset shows the number of wells occupied by the states as a function of energy .
the green squares and red circles are computed for the first few states in a particular speckle sample using the attenuation factor of 10 and 100 , respectively .
the solid line is the prediction of the variational calculations . ] in fig .
[ fig : red_univ_semic ] we show some numerical results for the systems with values of @xmath4 down to @xmath105 .
for such a small value of @xmath38 , the typical wave - functions of the states in the low - energy tail are spread over a distance a few times larger than the typical distance between neighboring wells .
these states are not yet sufficiently extended in order to reach for the distribution of @xmath135 $ ] the asymptotic gaussian probability shape of eq .
( [ eq : gauss ] ) .
the probability distribution @xmath135 $ ] given by eq .
( [ eq : fin - n ] ) lies somewhere between the exponential law ( [ eq : ray ] ) and the gaussian curve of eq .
( [ eq : gauss ] ) . to compute the dos for moderately small @xmath4
, we substitute eq .
( [ eq : fin - n ] ) into eq .
( [ eq : spectr_dens_lifs ] ) and minimize it with respect to the number of occupied wells @xmath156 .
this gives the result which interpolates between the semiclassical result of eq .
( [ eq : spectr_dens_lloyd ] ) and the lifshitz tail ( [ eq : spectr_dens_lifs_tris ] ) . the corresponding dos computed for @xmath105 is shown in fig .
[ fig : reds01 ] together with numerical data .
the variational prediction for the number of occupied states is shown in the inset .
the dots in the inset depict the number of occupied wells for the first few states in a particular speckle sample . to estimate this number from the numerical data we calculate the width of each wave function using the criterion of the amplitude attenuation by factors of 10 and 100 .
we find a good agreement with the results of the variational calculation . .
only a window of length @xmath82 of the whole system is shown .
the figure shows the ground state ( blue solid line ) and several first excited states ( black dashed line ) . ]
the random potential created by a blue - detuned laser speckle is bounded from below so that the dos vanishes at energies lower than the natural boundary @xmath157 .
analogously to the red - detuned problem , the localized region of the spectrum is controlled by the parameter @xmath4 .
the lowest - lying eigenstates for three different intensities of the same speckle profile , corresponding to @xmath158 , and @xmath58 are depicted in fig .
[ fig : eigenstates - blue ] .
this picture shows that with decreasing @xmath4 the wave functions extend over larger numbers of potential wells .
( greed circles ) and @xmath159 ( magenta squares ) .
the black solid line is the dos of free particle .
the best simultaneous fit of the lifshitz tail of both dos by eq .
( [ lifshitz-1 ] ) with the same @xmath160 and @xmath161 . ] the dos computed numerically for @xmath162 and @xmath159 are shown in fig .
[ fig : blue - small - s ] . at large energies
the dos approaches the limit of free particle while at small energies it exhibits the lifshitz s tail behavior .
it is well known that the dos near the bottom of a bounded - from - below disorder potential is controlled by the existence of very large regions free " of disorder potential .
the probability of their appearance in the speckle pattern can be estimated as follows .
we assume the existence of a bound state with energy @xmath137 localized inside this region . in general , this region has the average intensity that is not exactly zero but well below the energy level @xmath137 .
since we are interested in the asymptotic low energy behavior , we can fix the largest intensity to @xmath163 .
however , very small intensity corresponds also to a vanishing electric field .
the probability to have a region of size @xmath77 with the electric field in the interval between @xmath164 and @xmath165 is given by @xmath166 \nonumber \\ & & = \exp \left[- \int\limits_0^{l } { \cal e}^{*}(x ) g^{-1}(x - x'){\cal e}(x')dx dx ' \right ] \nonumber \\ & & \times ( \det g)^{-1}\prod\limits_{x\in[0,l ] } \frac{d{\cal e}(x ) d{\cal e}^{*}(x)}{\pi a_0},\end{aligned}\ ] ] where @xmath167 is the uv cutoff and the inverse correlator is defined by @xmath168 the fourier transform of the electric field correlator ( [ eq : sinc_one_d ] ) reads @xmath169 where @xmath170 is the heaviside step function .
in what follows , we use @xmath171 absorbing the numerical factor of @xmath172 into the overall coefficient . for small @xmath4
the eigenstates spread over distance @xmath77 much larger than the typical well width @xmath1 . in this regime @xmath173
can be estimated using eq .
( [ g - fourier ] ) as follows : @xmath174 substituting this into eq .
( [ w-1 ] ) and putting @xmath175 in the exponential gives the probability for the region of size @xmath77 to have the electric field in the interval between @xmath176 and @xmath177 , where @xmath178 is arbitrary . integrating out @xmath178 and summing over @xmath9 we obtain @xmath179 \sim ( i_0 \xi)^{-2l/\xi } ( 2 e /a_0)^{2l / a_0}.\ ] ] since we are interested in the leading exponential behavior , we can replace in eq .
( [ w0 - 2 ] ) @xmath167 by @xmath1 and arrive at @xmath180 \sim \exp \left[-\frac{2l}{\xi}\ln\left(\frac{i_0}{e } \right)\right].\ ] ] the region being almost free of disorder has a bound state with energy @xmath181 , where @xmath182 is a constant of order @xmath183 . using the latter in eq .
( [ w0 - 3 ] ) to express @xmath77 in terms of @xmath137 , we obtain the lifshitz s tail for the energy distribution as @xmath184.\end{aligned}\ ] ] the logarithmic corrections to the lifshitz tail of the same form as in eq .
( [ lifshitz-1 ] ) was found for the uncorrelated disorder by mapping the problem to the random walk of a classical particle in a medium with traps @xcite and using a variational approach @xcite .
we fit using eq .
( [ lifshitz-1 ] ) the lower energy tail of the dos computed numerically for @xmath162 and @xmath159 .
the corresponding curves are shown in the inset of fig .
[ fig : blue - small - s ] .
we find that taking into account the logarithmic corrections is necessary in order to fit the data in the energy interval that is accessible numerically .
the dos computed numerically for @xmath185 , and @xmath186 are shown in fig .
[ fig : blue - large - s ] .
the dos exhibits a quite flat crossover region between the narrow lifshitz s tail and the large energy asymptotic behavior corresponding to the free particle . for general @xmath4 the lifshitz tail of the dos
is expected to have the form @xmath187,\ ] ] where @xmath188 is a universal function depending only on the form of @xmath189 .
fitting the dos computed numerically ( see inset of fig . [
fig : blue - large - s ] ) we find that for @xmath190 $ ] this function can be approximated by @xmath191 .
thus , the correlations in the speckle potentials growing with @xmath4 renormalize the effective exponent of @xmath4 in the exponential of the lifshitz tail .
( red diamond ) , @xmath192 ( blue circles ) and @xmath193 ( green squares ) .
the black solid line refers to the analytical semiclassical result ( [ eq : tf_d_less_blue ] ) valid at large @xmath4 and @xmath194 .
the best simultaneous fit of the lifshitz tail of three dos by @xmath195 $ ] with the same @xmath196 and @xmath197 . ] to compute the dos in the crossover regime , we use the semiclassical approximation and obtain @xmath198,\end{aligned}\ ] ] where @xmath199 is the imaginary error function . the dos
( [ eq : tf_d_less_blue ] ) goes asymptotic as @xmath200 for large positive energies . near the boundary of the spectrum
@xmath201 it behaves as @xmath202 the comparison between eq .
( [ eq : tf_d_less_blue ] ) and the numerical data data in fig .
[ fig : blue - large - s ] shows that there is a range of energies where the two results agree quite well .
the possible existence of the effective mobility edge in 1d systems with certain correlated disorder has been considered in several works @xcite .
it is expected , for example , for the potentials with correlations having a finite support in fourier space . in particular as follows from eq .
( [ g - fourier ] ) the speckle potential has no fourier components with @xmath203 . as a result ,
in the born approximation there is no backscattering for particles with momentum larger than @xmath204 that defines the effective mobility edge .
the localization length above this edge is determined by corrections to the born approximations and larger by one order of magnitude . in our unities the effective mobility edge
is expected at @xmath205,\ ] ] which gives @xmath206 for @xmath159 and @xmath207 for @xmath192 . at these energies
the dos is already very close to the dos of a free particle .
one can not see any signature of the effective mobility edge like it happens with the dos and the real mobility edge in higher dimensions . to probe the effective mobility edge
, one has to study the asymptotic behavior of the wave functions at this energy @xcite which is beyond of the scope of this article .
in this work we have studied the dos of a quantum particle in a laser - speckle potential using the statistics of the speckle potential profile and numerical simulations .
the dos in the speckle potential can be characterized by a single parameter @xmath4 , which depends on the correlation length of the disorder , the average value of the laser intensity and the mass of the particles .
we have considered both cases of red - detuned and blue - detuned potentials for which the random potential is bounded from above and from below , respectively . for the negative - energy region of dos in the _ red - detuned _ speckle
there are three different regimes : semiclassical ( @xmath208 ) , intermediate ( @xmath209 ) , and quantum ( @xmath6 ) .
for strongly correlated disorder , @xmath208 , the tail of the dos is related to the statistical distribution of the deep minima of the disorder potential . in the quantum limit with @xmath6 ,
the dos is controlled by smoothing effects leading to the lifshitz tail of uncorrelated gaussian potential . in the crossover regime , @xmath209 ,
the dos can be described within the semiclassical approach by including quantum corrections . in the case of a _ blue - detuned _ speckle potential ,
the dos near the bottom of the spectrum is given by the lifshitz tail related to the existence of large regions with negligible intensity .
we have found logarithmic corrections to the celebrated lifshitz result due to the continuous nature of the distribution of the potential .
the relevance of these corrections for typical setup parameters in current experiments is supported by our numerical results . because @xmath210 has nonzero correlations only with itself , the total joint probability in eq .
( [ eq : dens_min_one_d ] ) can be written as @xmath211 in rescaled unities of @xmath27 and @xmath212 we can immediately write @xmath213 the joint probability for the two correlated variables @xmath137 and @xmath214 can be written using matrix representation as @xmath215 with @xmath216 and @xmath217 which has an inverse matrix that reads @xmath218 equation ( [ eq : joint_prob_one_d ] ) can be rewritten as @xmath219 using the following variable transformation @xmath220 and setting @xmath221 , we obtain @xmath222 therefore , the joint probability ( [ eq : joint_prob_one_d_total ] ) can be written as @xmath223 where @xmath224 } , \nonumber\end{aligned}\ ] ] and the jacobian is @xmath225 .
integration over @xmath178 yields a factor @xmath226 .
we can also analytically integrate out @xmath53 and @xmath227 , which is a gaussian integral . as a result
, we arrive at eq .
( [ eq : rhomin ] ) . | we study the single particle density of states of a one - dimensional speckle potential , which is correlated and non - gaussian .
we consider both the repulsive and the attractive cases .
the system is controlled by a single dimensionless parameter determined by the mass of the particle , the correlation length and the average intensity of the field .
depending on the value of this parameter , the system exhibits different regimes , characterized by the localization properties of the eigenfunctions .
we calculate the corresponding density of states using the statistical properties of the speckle potential .
we find good agreement with the results of numerical simulations . |
it is well - known that fcnc processes are very sensitive tests to physics beyond the standard model ( sm ) and , in particular , to supersymmetric extensions of the sm ( ssm ) .
there are two main sources of fcnc signals in the ssm .
first , since fcnc processes occur only beyond tree - level , they are sensitive to the existence of new particles circulating in the relevant loops . in this way ,
the usual sm predictions on fcnc processes ( e.g. @xmath8 mixing or @xmath9 ) are modified by any ssm @xcite .
second , besides the kobayashi
maskawa mechanism , supersymmetry provides new direct sources of flavour violation , namely the possible ( and even natural as we will see ) presence of _ off - diagonal _ terms ( say generically @xmath0 ) in the squark and slepton mass matrices . in the present paper
we will concentrate on this second source of flavour violation , which induces not only modifications in the sm fcnc processes ( such as @xmath9 ) , but also the appearance of new fcnc processes , particularly in the leptonic sector , e.g. @xmath4 or @xmath10 . in this paper
we will focus all our attention on the constraints on @xmath0 from the @xmath4 process for the following reasons : _ i ) _ : : among all the fcnc processes ( both in the hadronic and in the leptonic sector ) @xmath4 is by far the one with higher potential to restrict the value of the off - diagonal terms , @xmath0 @xcite . _
ii ) _ : : unlike in the hadronic processes , the calculation of @xmath4 is very clean and not affected by big uncertainties . actually , we will see that the constraints on the ssm coming from @xmath4 are not only very strong , but also that its evaluation is remarkably independent of the particular details of the model under consideration .
let us briefly review the emergence of non - diagonal scalar mass matrices in the minimal supersymmetric standard model ( mssm ) .
the mssm is defined by the superpotential , @xmath11 , and the form of the soft supersymmetry breaking terms .
@xmath11 is given by [ w ] w=_i , j\ { h^u_ijq_i h_2 u_r_j + h^d_ijq_i h_1 d_r_j + h^e_ijl_i h_1 e_r_j } + h_1 h_2 , where @xmath12 are generation indices , @xmath13 ( @xmath14 ) are the scalar partners of the quark ( lepton ) su(2 ) doublets , @xmath15 ( @xmath16 ) are the quark ( lepton ) singlets and @xmath17 are the two susy higgs doublets ; the @xmath18factors are the ( matricial ) yukawa couplings and @xmath19 is the usual higgs mixing parameter . in all terms of eq .
( [ w ] ) the usual su(2 ) contraction is assumed , e.g. @xmath20 with @xmath21 . from @xmath11 the ( global ) supersymmetric part of the lagrangian , @xmath22 , is readily obtained .
[ lsusy ] l_susy&=&-_k||^2 -(_k , l_k_l + h.c . ) + d - terms + & + & gauge interactions , where @xmath23 ( @xmath24 ) run over all the scalar ( fermionic ) components of the chiral superfields .
in addition to this , the soft breaking terms ( gaugino and scalar masses , and trilinear and bilinear scalar terms ) coming from the ( unknown ) supersymmetry breaking mechanism have the form -l_soft&=&m_a_a_a + ( m_l^2)_ij |l_il_j + ( m_e_r^2)_ij |e_r_i e_r_j + & + & ( m_q^2)_ij |q_iq_j + ( m_u_r^2)_ij |u_r_i u_r_j + ( m_d_r^2)_ij |d_r_i d_r_j [ nodiag ] + & + & , where @xmath25 is a gauge group index , @xmath26 are the gauginos , and the remaining fields in the formula denote just their corresponding scalar components . in the simplest version of the mssm
the soft breaking parameters are taken as _ universal _ ( at the unification scale @xmath27 ) .
then , the independent parameters of the theory are [ inpar ] , m , m_1/2,a , b ( the rest of the parameters can be worked out demanding a correct unification of the gauge coupling constants and correct masses for all the observed particles ) .
however this simplification is not at all a general principle .
in particular there is no theoretical argument against @xmath28 , @xmath29 having a non - diagonal structure ( as reflected in eq .
( [ nodiag ] ) ) .
the ultimate reason for the specific pattern of these matrices has to be searched in the type of sugra theory from which the mssm derives ( particularly in the structure of the khler potential ) and on the mechanism of supersymmetry breaking .
both ingredients are nowadays unknown , although superstring theories offer plausible ansatzs , which , by the way , do not support in general the universality assumption ( we will turn to this point later ) . on the other hand
, it is expected that the yukawa matrices , @xmath30 , @xmath31 , @xmath32 , may be also non - diagonal ( this is actually obliged for @xmath30 or @xmath31 ) . after performing the usual rotation in the superfields to diagonalize @xmath30 , @xmath31 , @xmath32 , the fermionic mass matrices and the contribution to the scalar mass matrices coming from
, @xmath33 @xmath34 and , except for the stop mass , is much smaller than the mass terms of eq .
( [ nodiag ] ) . ] ( [ lsusy ] ) become diagonal . however ,
if the @xmath28 matrices are not universal , they will develop off - diagonal entries , even if they are initially diagonal .
a similar thing occurs with the @xmath29 matrices . in the following
we will concentrate our attention on the @xmath35 terms , setting @xmath36 .
notice here that @xmath29 always comes accompanied by the corresponding yukawa coupling ( see eq .
( [ nodiag ] ) or eqs .
( [ rgesc],[rgestr ] ) below ) .
this makes its effect negligible when studying the impact of @xmath37 on @xmath4 . on the other ,
hand @xmath38 is also constrained by fcnc processes @xcite ( and in particular by @xmath4 itself ) , but we will not be concerned by this issue in the present work . from the previous arguments , it is natural to assume that the off - diagonal entries , @xmath0 , can be _
sizeable _ , or even of the same order as the diagonal terms [ sizedelt ] m^2 ~o(m^2 ) .
certainly , there are proposed mechanisms to avoid this , e.g. the above - mentioned assumption of universality , horizontal symmetries @xcite , plastic soft terms @xcite , etc . on the other hand , in some string scenarios @xcite
@xmath0 could naturally be very small .
however one should wonder whether , in the absence of any additional assumption , the perfectly possible and even natural situation of eq .
( [ sizedelt ] ) could still be compatible with the experimental data and , more precisely , with the present experimental bound @xcite on @xmath4 [ megexp ] br(e,)510 ^ -11 , which , as stated above , gives the strongest constraints on @xmath0 , more precisely on @xmath39 this will be the main goal of this article .
the results will turn out to be surprisingly model - independent .
in fact they will depend only on two parameters : the value of @xmath0 itself and the gaugino mass , @xmath2 .
also , we will make more precise ( and relaxed ) the assumption of eq .
( [ sizedelt ] ) , namely our results will be valid for any @xmath0 satisfying [ sizedelt2 ] . in principle
it might seem that @xmath40 is absolutely incompatible with ( [ megexp ] ) due to the existence of diagrams of the type of those shown in fig . 1 ( to be discussed in detail in sect . 3 ) .
however this is not necessarily the case , since @xmath4 is a low energy process , while the assumption ( [ sizedelt ] ) naturally arises at some unification scale , @xmath41 .
hence , an ordinary renormalization group ( rg ) running is in order . as a matter of fact
, this running softens the problem , since the diagonal entries of the mass matrices increase substantially @xcite while the off diagonal ones remain almost unchanged .
this possibility has also been considered in ref .
@xcite , though the calculation was affected by very large uncertainties , as we will see afterwards . in sect .
2 the rg running from high to low energies is carried out , stressing the fact that the @xmath1 conjecture , eq .
( [ sizedelt ] ) , automatically leads to a gaugino dominance " framework ( i.e. @xmath3 ) , where all the relevant low - energy quantities apart from @xmath0 itself are determined in terms of a unique parameter , @xmath2 . in sect .
3 the complete calculation of @xmath42 is performed , indicating which are the most important diagrams and why .
the results are presented in sect .
4 . they are very general , depending only on the values of @xmath2 and @xmath43 , and have strong implications , in particular the need of a large and precise @xmath44 hierarchy in order to maintain the consistency with the experiment .
furthermore , it will be shown that the analysis and results are valid for a wide range of @xmath45 values , as anticipated in eq .
( [ sizedelt2 ] ) .
finally , the summary and conclusions are presented in sect .
the expression of @xmath42 in the mssm depends on several low - energy parameters , namely @xmath46 , @xmath19 , @xmath47 , and the spectrum of masses of sleptons and gauginos ( the details of the precise dependence are left for sect .
these are not independent parameters .
they come as a low energy result of the actual initial parameters of the theory ( see eq .
( [ inpar ] ) ) , which are assumed to be given at the unification scale , @xmath27 .
of course , the connection has to be made through the corresponding rges .
next we list some of the most relevant rges in our calculation @xcite : * gaugino masses : = -b_a _ a m_a , a= 1,2,3 [ rgegau ] where @xmath48 , @xmath49 being the scale at which we evaluate the masses , @xmath50 are the coefficients of the 1loop beta functions for the gauge couplings , @xmath51 , and @xmath52 are the gauge coupling constants . * scalar mass matrices for sleptons : & = & _ ij ( 3 _ 2 m_2 ^ 2 + _ 1 m_1 ^ 2 ) - , + & = & _ ij ( 4 _ 1 m_1 ^ 2 ) - , where @xmath53 . * matricial trilinear terms ( also for sleptons ) : & = & - _ ij ( 3 _ 2 m_2 + 3 _ 1 m_1 + ( tr ( ^e h^e^)+ 3 tr ( ^dh^d^ ) ) ) + & - & .
[ rgestr ] * @xmath19 parameter : = ( 3 _ 2 + _ 1 ) ^2 - ( 3 h^u h^u^ + 3 h^d h^d^ + h^e h^e^ ) ^2 .
[ rgemu ] * b parameter : & = & - ( 3 _ 2 m_2 + _ 1 m_1 ) - ( 3 ( h^u ^u ) .
+ & + & . 3 ( h^d ^d ) + ( h^e ^e ) ) .
[ rgeb ] the first thing to notice is that , due to the structure of the equations for scalar masses ( [ rgesc ] ) , the ratio @xmath45 will in general be small at low energies ( even if it is @xmath54 at @xmath27 ) , provided that gaugino masses are bigger and @xmath55 will be given as a result of the whole calculation in the following sections . ]
than scalar masses , @xmath56 , because of the contribution of the former in the rges of the diagonal parts of the latter , which is not the case for the off - diagonal entries ( note the @xmath57 factors in eqs .
( [ rgesc ] ) ) .
this is the reason why the rges have the potential to `` cure '' initial sizeable values of @xmath43 .
therefore , the assumption of naturally large flavour mixing at @xmath27 ( see eq .
( [ sizedelt ] ) ) leads us necessarily to the @xmath56 scenario .
moreover , this sometimes called `` gaugino dominance '' scenario @xcite presents other very interesting features , such as that _ all _ the soft breaking parameters ( @xmath58 , @xmath59 , @xmath60 , @xmath61 , @xmath62 and @xmath63 ) are essentially determined _ at low energies _ by the value of @xmath2 at @xmath27 , independently of their initial values , as can be again easily seen in eqs .
( [ rgegau ] , [ rgesc ] , [ rgestr ] , [ rgeb ] ) .
so it is perfectly sensible to set the @xmath55s , @xmath47s and @xmath63 equal to zero at @xmath27 , then also @xmath64 , @xmath65 , since @xmath47 , @xmath63 are necessarily @xmath66 in order to avoid dangerous charge and color breaking minima . besides
, the value of @xmath63 is related to the value of @xmath47 or @xmath55 in many supergravity models . for a review of these issues
see @xcite and references therein . ] , and this is what we shall do for the rest of the analysis .
note that all this does _ not _ apply to the @xmath19 parameter as it renormalises proportional to itself ( see eq .
( [ rgemu ] ) ) .
so , for a given @xmath56 , the only parameter left from the initial set ( [ inpar ] ) to determine the whole low energy susy spectrum is @xmath19 .
however there is a constraint which we have not considered up to now , namely the requirement of a correct electroweak breaking .
this fixes the value of @xmath19 , giving us the _ whole _ spectrum and other relevant low - energy quantities ( such as @xmath47 and @xmath67 ) in terms of a _
parameter @xmath2 .
from the technical point of view , we have performed the calculation along the lines already sketched in a previous work @xcite , that is , we have computed the 2loop rges for the gauge couplings and imposed their unification at @xmath27 , taking into account all the susy thresholds . also , to obtain a correct electroweak breaking
, we have included all the spectrum in the 1-loop effective potential for the higgs fields , choosing the appropriate minimization scale as was already explained in @xcite . in this task
( and also in the rest of the paper ) we work under the assumption that only the yukawa couplings of the third generation are the relevant ones , and their values are fixed by the experimental masses of the corresponding fermions ( for the top we take @xmath68 gev ) .
the results are summarized in figs .
2a , 2b . as a general feature , which is worth mentioning
, we have found that the values of @xmath69 obtained in this framework tend to be rather large ( ranging from 11 to 26 as @xmath2 increases from 150 gev to 10 tev , as can be seen from fig .
2b ) ; this fact will be very important in our calculation given the presence of a term proportional to @xmath70 in one of the contributions to @xmath42 ( see eqs .
( [ amps ] ) in sect . 3 ) .
we should comment here that previous analyses have ignored the electroweak breaking constraints , just fixing @xmath69 to a particular ( low ) value @xcite and making a certain `` average '' in some low - energy parameters .
the present results indicate that this is an important source of error and uncertainty , which can be avoided by taking into account the electroweak breaking process .
to summarize , a scenario of gaugino dominance ( which arises from the assumption of naturally large flavour mixing at high energy ) determines the low energy spectrum and other relevant low - energy quantities such as @xmath47 , @xmath19 , @xmath67 , in terms of a unique parameter , @xmath2 .
this is a very interesting fact that makes the subsequent analysis rather accurate and model independent .
we follow here a scheme of calculation along the lines of refs .
the effective lagrangian that describes the @xmath72 decay is usually parameterized as [ oper ] l_eff = |_electron _ ( b_l p_l + b_r p_r ) _
muon f^ , where @xmath73 .
then , the branching ratio @xmath74 is given by [ br ] br(e , ) = , where @xmath75 is the fermi constant and @xmath76 is the muon mass .
the @xmath77 , @xmath78 amplitudes arise from one loop diagrams that involve a flip of the leptonic flavour triggered by the slepton mixing , besides the propagation of a neutralino or chargino ( see fig . 1 ) .
note that the structure of ( [ oper ] ) implies that the helicity of the muon and the electron must be opposite .
since the electron and muon yukawa couplings are very suppressed , only the gauge part of the couplings of the charginos and neutralinos will play a role in the diagrams .
in fact , one important consequence of the gaugino dominance framework is that for large enough values of @xmath2 the neutralinos ( charginos ) are almost pure neutral ( charged ) gaugino and higgsino .
this result follows from the low - energy inequality @xmath79 , something that occurs whenever @xmath80 gev .
then , the relevant diagrams , corresponding to bino ( @xmath81 ) and wino ( @xmath82 , @xmath83 ) exchange , factor . ] are shown in fig . 1 .
some comments are in order here .
the crosses in the scalar propagators of the diagrams denote mass insertions
. these can be either of the @xmath84 type or of the @xmath85 type .
the former occur in all the diagrams and correspond to the off - diagonal entries in the soft mass matrices of the sleptons .
the latter , occurring only in diags .
4,6 , denotes a change of chirality without flavour mixing .
this arises from @xmath22 and @xmath86 ( see eqs .
( [ lsusy],[nodiag ] ) ) once @xmath87 and @xmath88 acquire non - vanishing vevs . it can be checked that @xmath89 ( note that it is proportional to the fermion mass , which makes @xmath90 negligible ) , where from now on we will drop the superindex @xmath19 from @xmath91 . it should be noted that the diagrams of fig .
1 correspond to an expansion in @xmath0 ( both of the @xmath84 type and of the @xmath85 type ) .
this is justified due to the smallness of @xmath0 ( of either type ) compared to the slepton and gaugino masses .
recall here that even though @xmath0 may be @xmath92 at @xmath27 , it must necessarily be @xmath93 at low energies in order to be consistent with the experiment ( this will become apparent in the next section ) . on the other hand
the crosses in the fermionic propagators denote a change in helicity , which gives a factor proportional to the fermion mass .
let us remark here that this emerges from the pure calculation and does not amount to any mass expansion . splitting the induced @xmath94 values as [ descbs ]
b_r & = & b^^-_r + b^^0_r + b^_r , + b_l & = & b^_l , the different diagrams contribute in the following way : [ bsds ] b^^-_r & = & d_1 , b^^0_r = d_2 , + b^ _ r & = & d_3 + d_4 , b^ _
l = d_5 + d_6 , where @xmath95 correspond to diagrams 16 of fig . 1 .
we have evaluated _ all _ of them .
the expressions given in the previous literature are either incomplete or not directly applicable to our case . some of them correspond to the limit in which the photino exchange gives the main contribution ( which is not valid in this case ) .
other estimates do not take into account the left - right mixing between charged sleptons , that will be essential , or ignore some of the diagrams .
also , it is normally assumed that left and right sleptons are degenerated in mass and a different relative sign between @xmath96 and @xmath97 ( see eqs .
( [ ds ] ) below ) . ] .
however , it is clear that a mass splitting will usually arise due to the different dependence of the rges , eq .
( [ rgesc ] ) , on the gaugino masses . in the gaugino dominance scenario
this effect tends to be very large , see fig .
the complete expressions are : [ ds ] d_1 & = & + d_2 & = & - + d_3 & = & - [ amps ] + d_4 & = & + d_5 & = & - + d_6 & = & where [ emes ] m _ _ e ^2 = m _ _ ^2 m _ ^2 , m _ _ l ^2 = m _ _ l ^2 m _ _ l ^2 , m _ _ r ^2 = m _ _ r ^2 m _ _ r ^2 and f(x ) & = & ( 1-x)^-5 + g(x ) & = & ( 1-x)^-5 + l(x ) & = & ( 1-x)^-4 [ fs ] note that the equalities of eq .
( [ emes ] ) do not come from any universality assumption , but from the mere gaugino dominance in the corresponding rges .
although in principle all the diagrams can have a similar magnitude ( e.g. if we assume @xmath98 ) , in practice diagrams 4 and 6 are the dominant ones .
this comes from the coefficient of proportionality @xmath99 , which turns out to be very important in the gaugino dominance framework due to the large @xmath67 value ( see sect . 2 ) .
finally , let us remark that the previous results on @xmath100 are in agreement with the recent paper of ref .
@xcite , where charginos and neutralinos are allowed to be non - pure gaugino or higgsino states .
the expressions obtained in the previous section for @xmath101 depend on two different sets of parameters .
first , the different masses involved in the game ( @xmath102 @xmath103 ) and certain relevant low - energy quantities @xmath104 .
second , the three independent flavour - mixing mass entries : @xmath105 . as explained in sect . 2 ,
once we are working in the framework of gaugino dominance , @xmath106 , the first set is completely determined in terms of the initial gaugino mass , @xmath2 .
recall that we were led to this framework by the mere assumption of naturally large flavour mixing at @xmath27 ( see eq .
( [ sizedelt ] ) ) .
the three flavour - mixing mass parameters , however , remain independent ( and we will consider them in that way in the following ) , although it is logical to suppose that they are of the same order .
the constraints on the mssm from @xmath101 arise by comparing the above calculations ( eqs .
( [ br][fs ] ) ) with the present experimental bound , eq .
( [ megexp ] ) .
we have illustrated this in fig . 3 , where an overall mass - mixing parameter @xmath107 has been taken for simplicity .
then we have plotted @xmath42 vs. @xmath2 for different values of @xmath43 . from this figure
we can derive the maximum allowed value of @xmath43 ( or , equivalently , the minimum allowed value of @xmath108 ) for each value of @xmath2 .
this is represented in fig . 4 for four different cases : _ a ) _ @xmath109 ; _ b ) _ only @xmath110
; _ c ) _ only @xmath111 and _ d ) _ only @xmath112 , which gives a complete picture of the results .
notice that the _ ( d ) _ case is the less restrictive one .
this is because , as explained in sect . 3 , given the large value of @xmath67 the dominant diagrams are nos . 4 and 6 of fig . 1 , which do not involve @xmath113 ,
( [ ds ] ) . on the other hand ,
the strongest constraint comes from @xmath114 .
this is because the mass of the right sleptons , @xmath115 , is smaller than that of the left ones , @xmath116 , as a consequence of their different rg running ( note from eq .
( [ rgesc ] ) the different dependence of @xmath116 , @xmath115 on the gaugino masses ) .
the constraints are in general extremely strong . for case _ ( a ) _ , which is the most representative one , the corresponding curve can be approximately fitted by the simple constraint , where @xmath67 can be well fitted by @xmath117^{1/5}$ ] , see fig .
the dependence of @xmath118 on @xmath119 can be understood from the dependence of the dominant diagrams ( nos . 4 and 6 of fig . 1 ) on @xmath67
, see eqs .
( [ ds ] ) . ] [ fit ] 34 tev ( similar equations can be written for the curves associated to the other _ ( b ) _ , _ ( c ) _ , _ ( d ) _ cases ) . under the assumption of eq .
( [ sizedelt ] ) , i.e. @xmath120 , the results of fig . 4 or eq .
( [ fit ] ) imply that , indeed , a very large hierarchy between the scalar and gaugino masses is needed in order to reconcile the theoretical and experimental results .
this gives full justification to our assumption of a gaugino dominance framework once eq .
( [ sizedelt ] ) has been conjectured .
for example , for @xmath121 gev the assumption @xmath122 demands @xmath123 . for smaller values of @xmath2
the required hierarchy is in fact more severe , whereas too large values of @xmath2 start to conflict an electroweak breaking process with no fine tuning @xcite . actually , it is hard to think of a scenario where such a dramatical hierarchy can naturally arise . consequently , we can conclude at this point that a naturally large flavour mixing , as that conjectured in eq .
( [ sizedelt ] ) , can hardly be reconciled with the experiment in a natural way . on the other hand , the results from fig . 4 and eq .
( [ fit ] ) are so extreme that we can relax the assumption @xmath124 , allowing for larger values of @xmath55 relative to @xmath43 without losing the validity of the calculation .
the latter is based on the use of the gaugino dominance framework , which is reasonably accurate for @xmath125 .
then , from ( [ fit ] ) , it follows that all our results , and eq .
( [ fit ] ) itself , are valid whenever we start with a @xmath43 at @xmath27 such that [ fit2 ] , as was foretold in eq .
( [ sizedelt2 ] ) .
this result extends enormously the scope of application of our calculation and hence its interest . in this way
we see that rather small values of @xmath45 at @xmath27 require a gaugino dominance scenario to be cured , thus leading to a complete determination of all the relevant low - energy parameters in terms of @xmath2 ( see sect . 2 and fig . 2 ) .
it is interesting to compare these results with those obtained in the initial calculation of ref . @xcite .
there , besides assuming @xmath126 , @xmath127 ( i.e. our case _ ( c ) _ )
, several additional assumptions and averages were adopted ( some of them already commented in sect . 2 ) .
the confrontation of the two calculations is made in fig . 5 , where we have represented @xmath128 vs. @xmath129 to facilitate the comparison with the plots of ref .
@xcite . obviously , there is a big difference between their results ( that we have been able to reproduce ) and ours .
this reflects the very large uncertainties in the calculation of ref .
@xcite ( acknowledged by its authors ) , which are dramatically suppressed once a correct electroweak breaking process is imposed .
in the absence of any additional assumption it is natural to conjecture that sizeable flavour - mixing mass entries , @xmath0 , may appear in the mass matrices of the scalars of the mssm , i.e. @xmath1 .
this flavour violation can still be reconciled with the experiment if the gaugino mass , @xmath2 , is large enough to yield ( through the renormalization group running ) a sufficiently small @xmath130 at low energy .
we have analyzed in detail this possibility , focussing our attention on the leptonic sector , particularly on the @xmath4 decay , which is by far the fcnc process with higher potential to restrict the value of the off - diagonal terms , @xmath0 .
the results are the following : 1 .
the @xmath1 conjecture automatically leads to a _ gaugino dominance _
framework ( i.e. @xmath3 ) , where , apart from @xmath0 itself , _ all _ the relevant low - energy quantities ( mass spectrum , @xmath47 , @xmath19 , @xmath67 ) are determined in terms of a unique parameter , @xmath2 ( see fig . 2 ) .
this makes the subsequent analysis and results remarkably model independent .
the resulting constraints in the mssm , obtained by comparing the calculated @xmath42 with the experimental bound , are very strong ( see figs
. 3 , 4 ) . more precisely , assuming @xmath131 , we arrive at the approximate constraint [ fit2p ] 34 tev ( and similar equations for other cases ) .
this makes , in our opinion , the natural flavour mixing conjecture @xmath1 extremely hard to be reconciled with the experiment in a natural way .
hence , @xmath132 should be small already at the unification scale .
3 . the required values of @xmath133 to be within the experimental limits are so large that the gaugino dominance assumption , which is an essential ingredient of our analysis , remains valid for values of @xmath6 much larger than @xmath0 , namely for @xmath134 , thus extending enormously the scope of interest and application of our results . to perform the previous calculation
we have completed the earlier evaluations @xcite of @xmath42 ( see diagrams of fig . 1 ) .
the results , summarized in eqs .
( [ br][fs ] ) , are in concordance with the recent results of ref .
@xcite .
we would like to stress the high degree of model independence of our analysis and results .
in fact , in the entire calculation we have only made , for the sake of simplicity , the assumption of universality of gaugino masses at the unification scale .
the relaxation of this assumption does not imply any essential conceptual change in the analysis , leading to straightforward modifications in the results of the paper , without affecting the main conclusions above . finally , let us comment that the need of starting with small @xmath132 can be satisfied in some theoretically well - founded scenarios , which become favoured from this point of view .
in particular , besides the proposed mechanisms of refs .
@xcite , we would like to stress that many string constructions can be consistent with that requirement . more precisely , in orbifold compactifications schemes @xcite , which are known to give an interesting phenomenology @xcite , the soft susy breaking scalar masses always consist of a universal piece plus a contribution proportional to the so - called modular weight ( @xmath135 ) associated with the field under consideration @xcite .
thus , scalar fields belonging to the same twisted sector of the theory ( or to the untwisted one ) acquire degenerate soft masses , something that can perfectly occur in many realistic scenarios .
a similar thing happens in the so - called dilaton - dominated limit @xcite .
other scenarios , however , can produce a larger non - universality of the scalar masses , with potentially dangerous contributions to fcnc processes @xcite . in any case
, these non - universality effects are to produce non - vanishing off - diagonal terms in the scalar mass matrices once the usual rotation of fields to get diagonal fermionic mass matrices is carried out .
the phenomenological viability of these physically relevant scenarios undoubtedly deserves further investigation .
work along these lines is currently in progress @xcite .
we thank s. pokorski for his help to clarify some aspects of ref . @xcite .
j. ellis and d.v .
nanopoulos , phys .
lett . b110 ( 1982 ) 44 ; + r. barbieri and r. gatto , phys .
b110 ( 1982 ) 211 ; + m. duncan , nucl .
b221 ( 1983 ) 285 ; + j. donoghue , h.p .
nilles and d. wyler , phys .
b128 ( 1983 ) 55 ; + a. bouquet , j. kaplan and c.a .
savoy , phys .
lett . b148 ( 1984 ) 69 ; + l. hall , v. kostelecky and s. raby , nucl .
b267 ( 1986 ) 415 .
a. font , l.e .
ibez , h.p .
nilles and f. quevedo , phys .
b210 ( 1988 ) 101 ; + j.a .
casas and c. muoz , phys .
b214 ( 1988 ) 63 ; + j.a .
casas , e.k .
katehou and c. muoz , nucl .
phys . b317 ( 1989 ) 171 . | in the absence of any additional assumption it is natural to conjecture that sizeable flavour - mixing mass entries , @xmath0 , may appear in the mass matrices of the scalars of the mssm , i.e. @xmath1 .
this flavour violation can still be reconciled with the experiment if the gaugino mass , @xmath2 , is large enough , leading to a _ gaugino dominance _ framework ( i.e. @xmath3 ) , which permits a remarkably model independent analysis .
we study this possibility focussing our attention on the @xmath4 decay . in this way we obtain very strong and general constraints , in particular @xmath5 . on the other hand ,
we show that our analysis and results remain valid for values of @xmath6 much larger than @xmath0 , namely for @xmath7 , thus extending enormously their scope of application . finally , we discuss the implications for superstring scenarios .
4.5 mm = -2.2 cm 4.5 mm
@=11 stequation
@=12 -23.3 cm |
qcd has been called the perfect theory @xcite ; as a renormalizable field theory whose validity could extend up to the grand unification scale , it provides the mechanism for generating nearly all the mass of the visible universe .
the current quark masses are the only external parameters in the theory .
quenched qcd , without dynamical quarks , explains the hadron spectrum to an accuracy of 10% .
these lattice results suggest that gluons play a central role in the structure of matter .
the role of glue in qcd is best understood in the asymptotic weak coupling regimes of the theory where analytical computations are feasible .
much of the discussion in perturbative qcd ( pqcd ) has been in the bjorken - feynman asymptotics where @xmath0 , @xmath1 and @xmath2 .
the machinery of precision physics in qcd such as the operator product expansion and factorization theorems are derived in this limit of the theory .
the progress in this direction has been truly remarkable @xcite . in dis for instance , both coefficient functions and splitting functions have been derived to next - to - next - to leading order ( nnlo )
. what does the hadron look like in the bjorken - feynman asymptotics ?
the dglap evolution equations tell us that the gluon distribution grows rapidly with increasing @xmath3 at small @xmath4 .
however , the phase space density ( in a particular gauge and frame ) , decreases rapidly with increasing @xmath3 .
the proton become more dilute " even though the number of partons increases ; the typical size of resolved partons decreases as @xmath5 , faster than the increase in the number through qcd evolution .
the more dilute the hadron , the cleaner will be the qcd background for new physics beyond the standard model .
much of the current focus in qcd is in quantifying this background
. it would be unfortunate however if this were the only focus in qcd studies because the theory , even in the weak coupling domain , contains rich and non - trivial dynamics .
we speak here of the regge - gribov asymptotics where @xmath6 , @xmath1 and @xmath7 .
this regime of strong color fields is responsible for the bulk of multiparticle production in qcd
. what does the hadron look like in the regge - gribov asymptotics ?
the bfkl equation , which resums the leading logarithms in @xmath8 , indicates that the gluon distributions grow even more rapidly in this asymptotics . unlike the bjorken - feynman case
, the phase space density in the hadron grows rapidly as well .
the stability of the theory requires that the phase space densities ( or more generally , the field strengths squared ) be no larger than @xmath9 . in the pqcd framework ,
mechanisms for the saturation of the growth in the phase space density are provided by
higher twist " recombination and screening contributions @xcite .
these counter the bremsstrahlung growth of soft gluons described by the dglap and bfkl equations .
the saturation scale @xmath10 generated by the dynamics demarcates the separation between the linear and non - linear regimes of the theory : for momenta @xmath11 , non - linear qcd dynamics is dominant , for momenta @xmath12 , weak coupling physics is governed by the dglap / bfkl evolution equations .
the universal properties of gluons in the non - linear regime are described by a classical effective field theory of dynamical gluon fields coupled to static , stochastic sources .
this is the color glass condensate ( cgc ) @xcite .
the evolution of multi - parton correlators with energy is described by the wilsonian jimwlk renormalization group ( rg ) equations @xcite . in the limit of large nuclei and large @xmath13
, one recovers the balitsky - kovchegov equation for the forward dipole cross - section @xcite .
a universal saturation scale arises naturally in the theory and its energy dependence is given by the jimwlk / bk equations .
the typical momentum of gluons @xmath14 ; the bulk of the contributions to high energy cross - sections may be therefore described in a weak coupling framework .
the saturation scale also grows with the nuclear size . a fast compact probe of size
@xmath15 , where @xmath16 is the proton size , will interact coherently at high energies with partons localized in nucleons all along the nuclear diameter .
the field strength squared experienced by the probe is therefore enhanced parametrically by a factor proportional to the nuclear diameter @xmath17 . as clearly illustrated in the cgc effective theory ,
the dynamics of partons at small @xmath8 is universal regardless of one speaks of hadrons or nuclei ; the latter , as we will discuss further , are therefore an efficient ( and cheaper ) amplifier of the non - linear dynamics of these gluons .
this talk is organized as follows .
we will outline our current ( limited ) understanding of the different dynamical regimes in high energy qcd from experiments at hera and rhic .
we will then discuss how experiments at the lhc and future dis experiments on nuclei can help further quantify our understanding . finally , to illustrate the scope of these studies
, we will discuss how the strong color field dynamics of partons in nuclear wavefunctions contributes to a quantitative understanding of the formation and subsequent thermalization of a strongly interacting glasma " in heavy ion collisions .
a strong hint that semi - hard scales may play a role in small @xmath8 dynamics at hera came from geometrical scaling " of the hera data @xcite .
the inclusive virtual photon+proton cross - section for @xmath18 and all available @xmath3 scales as a function of @xmath19 , where @xmath20 gev@xmath21 .
here @xmath22 is the rapidity ; @xmath23 and @xmath24 are parameters fit to the data @xcite .
further , the inclusive diffractive , vector meson and dvcs cross - sections at hera , with a slight modification , where @xmath25 denotes the mass of the diffractive / vector meson final state.]in the definition of @xmath26 , also appear to show geometrical scaling @xcite .
geometrical scaling of the e+p data is shown in fig .
[ fig : scaling ] .
a recent quality factor " statistical analysis @xcite indicates that this scaling is robust ; it is however unable to distinguish between the above fixed coupling energy dependence of @xmath27 and the running coupling @xmath28 dependence of the saturation scale .
geometrical scaling is only asymptotic in both fixed and running coupling evolution equations .
pre - asymptotic corrections have been computed previously , to good approximation , in both fixed and running coupling cases for the bk equation @xcite .
a recent nlo bk analysis @xcite suggests that the onset of the scaling asymptotics may be precocious , thereby providing a possible explanation for its manifestation in the hera data .
a caveat that has been raised is that there is a strong correlation between @xmath8 and @xmath3 in the hera data @xcite .
the scaling however persists even where there is a significant lever arm in @xmath3 for small @xmath8 .
nevertheless , geometrical scaling alone is not sufficient evidence of saturation effects and it is important to look at the data in greater detail in saturation / cgc models . all saturation models @xcite express the inclusive virtual photon+proton cross - section as @xmath29 here @xmath30 represents the probability for a virtual photon to produce a quark anti - quark pair of size @xmath31 and @xmath32 denotes the _ dipole cross section _ for this pair to scatter off the target at an impact parameter @xmath33 .
the former is well known from qed , while the latter represents the dynamics of qcd scattering at small @xmath8 . a simple saturation model ( known as the gbw model @xcite ) of the dipole cross section , parametrized as @xmath34 where @xmath35 , gives a good qualitative fit to the hera inclusive cross section data for @xmath36 and @xmath37 .
though this model captures the qualitative features of saturation , it does not contain the bremsstrahlung limit of perturbative qcd ( pqcd ) that applies to small dipoles of size @xmath38 . in the classical effective theory of the cgc
, one can derive , to leading logarithmic accuracy , the dipole cross section @xcite containing the right small @xmath39 limit .
this dipole cross section can be represented as @xcite @xmath40 , \label{eq : bekw}\ ] ] where @xmath41 is the impact parameter profile function in the proton , normalized as @xmath42 and @xmath43 is proportional to the gluon distribution @xcite @xmath44 evolved from the initial scale @xmath45 by the dglap equations .
the dipole cross section in eq .
( [ eq : bekw ] ) was implemented in the impact parameter saturation model ( ipsat ) @xcite where the parameters are fit to reproduce the hera data on the inclusive structure function @xmath46 . here
@xmath47 is defined as the solution of @xmath48 .
the ipsat dipole cross section in eq .
( [ eq : bekw ] ) is valid when leading logarithms in @xmath8 in pqcd are not dominant over leading logs in @xmath3 . at very small @xmath8 , where logs in @xmath8 dominate , quantum evolution in the cgc describes both the bfkl limit of linear small @xmath8 evolution as well as nonlinear jimwlk / bk evolution at high parton densities @xcite .
these asymptotics are combined with a more realistic @xmath49-dependence in the b - cgc model @xcite .
both the ipsat model and the b - cgc model provide excellent fits to hera data for @xmath50 @xcite .
an important caveat @xcite to the success of the saturation models is that the saturation scale , at median impact parameters , extracted from these fits is @xmath51 gev@xmath21 even at the lowest @xmath8 values at hera @xcite .
the saturation scale extracted from the fit in the ipsat model is shown in fig . [
fig : qs ] .
we should note however that the uncertainty in the magnitude of the saturation scale is significant and is a factor of 2 larger in recent cgc fits @xcite .
nlo computations in the small @xmath8 dipole framework are now becoming available @xcite ; these will provide theoretical guidance into precisely how the coupling runs as a function of @xmath47 at small @xmath8 . finally , from fig .
[ fig : qs ] , it is clear that the energy dependence of the extracted @xmath47 is significantly stronger than those predicted in non - perturbative models @xcite .
the strong field dynamics of small @xmath8 partons is universal and should be manifest in large nuclei at lower energies than in the proton . in fig .
[ fig : shad ] ( left ) , we show the well known shadowing of @xmath52 in the fixed target e+a e665 and nmc experiments
. expressed in terms of @xmath19 ( fig .
[ fig : shad ] ( right ) ) , the data show geometrical scaling @xcite . in ref .
@xcite , the @xmath53 dependence of @xmath27 is determined to be @xmath54 and not @xmath55 as suggested in a simple random walk picture . however , as we shall discuss shortly , this conclusion is a little misleading . from the nmc and e665 fixed target experiments .
right : the data scaled as a function of @xmath19 @xcite.,title="fig : " ] from the nmc and e665 fixed target experiments .
right : the data scaled as a function of @xmath19 @xcite.,title="fig : " ] a study of nuclear dis in the ipsat cgc framework was performed in ref .
the average differential dipole cross section is well approximated by @xmath56 $ ] , where @xmath57 is the well known woods saxon distribution . here
@xmath58 is determined from the ipsat fits to the e+p data ; no additional parameters are introduced for @xmath59 collisions . in fig .
[ fig : shad2 ] ( left ) , the model is compared to nmc data on carbon and calcium nuclei - the agreement is quite good . in fig .
[ fig : shad2 ] ( right ) , we show the extracted saturation scale in nuclei for both central and median impact parameters . to a good approximation
-function impact parameter dependence in the gbw model , which yields @xmath60 for @xmath61 and @xmath62 . ] , the saturation scale in nuclei scales as @xmath63 .
the factor of @xmath64 gives a huge `` oomph '' in the parton density of a nucleus relative to that of a proton at the same @xmath8 .
_ indeed , one would require a center of mass energy @xmath65 times larger in an e+p collider relative to an e+au collider to obtain the same @xmath66
. _ and @xmath8 dependence of the saturation scale in the ipsat model @xcite.,title="fig : " ] and @xmath8 dependence of the saturation scale in the ipsat model @xcite.,title="fig : " ] the reasons for the additional enhancement are two fold .
firstly , because the density profile in a nucleus is more uniform than that of the proton , @xmath67 is only @xmath68% of the value at @xmath69 ; in contrast , in gold nuclei it is 70% . because the median impact parameter dominates inclusive scattering , this effect gives a significant enhancement to the effective @xmath27 .
the second reason for the enhancement is the dglap - like growth of the gluon distribution in the ipsat nuclear dipole cross section . for two nuclei , @xmath53 and @xmath70 ( with @xmath71 ) , in a `` smooth nucleus '' approximation ( @xmath72 ) , @xmath73 , where @xmath43 was defined in eq .
[ eq : bekw_f ] .
the scaling violations in @xmath43 imply that , as observed in refs .
@xcite , the growth of @xmath47 is faster than @xmath55 . also , because the increase of @xmath43 with @xmath3 is faster for smaller @xmath8 , the @xmath53-dependence of @xmath47 is stronger for higher energies .
in contrast , the dipole cross section in the b - cgc model depends only on the combination . ]
@xmath74 without dglap scaling violations .
it therefore does not have this particular nuclear enhancement .
another interesting possibility , following from running coupling corrections to the leading logs in @xmath8 , is that qcd evolution actually depletes the nuclear enhancement of @xmath47 at very small @xmath8 @xcite .
_ precise extraction of the @xmath53 dependence of @xmath47 can therefore help distinguish between `` classical '' and `` quantum '' rg evolution at small x_. we now turn to a discussion of cgc effects in hadronic collisions .
a systematic treatment of the scattering of two strong color sources ( such as two high energy nuclei ) is discussed in section [ sec : glasma ] . to leading order
, the problem reduces to the solution of the classical yang - mills ( cym ) equations averaged over color sources for each nucleus @xcite ; the variance of this distribution of sources is proportional to @xmath75 . besides the nuclear radius , @xmath75 is the only scale in the problem , and the @xmath76 expression for the saturation scale was used in cgc models of nuclear collisions to successfully predict the multiplicity @xcite and the centrality dependence of the multiplicity @xcite dependence in gold+gold collisions at rhic .
the universality of the saturation scale also has a bearing on the hydrodynamics of the quark gluon plasma ( qgp ) ; the universal form leads to a lower eccentricity @xcite ( and therefore lower viscosity ) than a non - universal form that generates a larger eccentricity @xcite ( leaving room for a larger viscosity ) of the qgp . for asymmetric ( off - central rapidity ) nuclear collisions , or proton / deuteron+heavy nucleus collisions , @xmath77-factorization can be derived systematically for gluon production , at leading order , in the cgc framework @xcite .
the simplicity of @xmath77 factorization is convenient for phenomenology ; predictions based on this formalism describe the rapidity distributions in @xmath78 collisions @xcite and the phenomenon of limiting fragmentation " @xcite .
the latter , and deviations thereoff , are described by solutions of the bk - equation .
predictions for the multiplicity distribution in a+a collisions at the lhc @xcite for both gbw and classical cgc ( mv ) initial conditions give a charged particle multiplicity of 1000 - 1400 in central lead+lead collisions at the lhc .
the results are shown in fig .
[ fig : lf ] . in deuteron+gold collisions at rhic , the normalized ratio @xmath79 of the inclusive hadron spectrum relative to the same in proton+proton collisions shows a mild cronin " peak at mid - rapidities corresponding to multiple scattering in the classical cgc ; at forward rapidities , however , @xmath79 decreases systematically below unity . in the cgc , this reflects quantum evolution of the dipole cross - section in a large nucleus and has the same origin as the extension of the geometrical scaling regime @xcite to @xmath80 .
this effect in the dipole cross - section , as opposed to @xmath81 ( dglap ) and @xmath82 ( bfkl ) . ] , should also exist in hadronic collisions @xcite ; specifically , it was predicted this would occur in deuteron+gold collisions @xcite . in general , @xmath79 while suggestive , is not an ideal variable because it is not clear the same formalism applies to p+p collisions at the same rapidity .
data on the inclusive hadron spectrum in deuteron+gold collisions can be directly compared to model predictions @xcite .
the result is shown in fig .
[ fig : rda ] . for a comprehensive review of applications of cgc picture to rhic phenomenology
, we refer the reader to ref .
@xcite .
distributions in deuteron+gold collisions compared to theory curves for different rapidities . from @xcite . ]
there are a couple of caveats to this picture .
firstly , @xmath77 factorization is very fragile . it does not hold for quark production even at leading order in the parton density @xcite , albeit it may be a good approximation for large masses and transverse momenta @xcite . for gluon production
, it does not hold beyond leading order in the parton density @xcite .
secondly , a combined comprehensive analysis of hera and rhic data is still lacking though there have been first attempts in this direction @xcite .
the lhc is the ultimate small @xmath8 machine in terms of reach in @xmath8 for large @xmath3 .
a plot from ref .
@xcite illustrating this reach is shown in fig .
[ fig : lhc - eic ] ( left ) . for a recent review of the small @xmath8 opportunities at the lhc ,
see ref .
the lhc will provide further , more extensive tests of the hints for the cgc seen at rhic . at very high energies , a novel
diffusive scaling " regime has been proposed , which incorporates the physics of pomeron loops @xcite .
recent developments were reviewed at dis06 by iancu and at dis07 by shoshi @xcite ; possible signatures at the lhc have been proposed @xcite . however , very recent computations including running coupling effects suggest that this regime is unlikely to be accessed realistic collider energies @xcite . the universality of parton distributions is often taken for granted but factorization theorems proving this universality have been proven only for a limited number of inclusive final states . however , as we have discussed , small @xmath8 is the domain of rich multi - parton correlations .
these are more sensitive to more exclusive final states for which universality is not proven @xcite . therefore , while the lhc will have unprecedented reach in @xmath8 , precision studies of high energy qcd and clean theoretical interpretations of these motivate future dis projects .
two such projects discussed at this conference are the eic project in the united states @xcite and the lhec project in europe @xcite .
-@xmath3 reach of different final states at the lhc compared to other experiments with nuclei . from @xcite .
right : the saturation scale in the proton , calcium and gold in the kinematic acceptance of the eic.,title="fig : " ] -@xmath3 reach of different final states at the lhc compared to other experiments with nuclei . from @xcite .
right : the saturation scale in the proton , calcium and gold in the kinematic acceptance of the eic.,title="fig : " ] as we discussed previously , strong color fields may be more easily accessible in dis off nuclei relative to the proton due to the oomph " factor . in fig . [
fig : lhc - eic ] ( right ) , we show the saturation scale @xmath83 overlaid on the @xmath8-@xmath3 kinematic domain spanned by the eic . it is interesting that there is a significant kinematic domain where @xmath84 , including in particular @xmath85 gev@xmath21 . in the weak field regime where @xmath86 , we are accustomed to thinking of @xmath87 . in the strong field regime , where @xmath88 , we likely have
instead @xmath89 . as suggested by the figure , the eic ( and clearly the lhec )
will cleanly probe the cross - over regime from linear to non - linear dynamics in qcd
. a particularly striking feature of e+a dis will be diffractive scattering @xcite ; it is anticipated that @xmath90 of the cross - section corresponds to hard diffractive final states .
for further discussion of the physics of an electron ion collider , see ref . @xcite .
the word glasma " describes the strongly interacting matter in heavy ion collisions from the time when particles are produced in the shattering of two cgcs to the time when a thermalized qgp is formed @xcite .
we will discuss here a systematic approach to computing particle production in heavy ion collisions to nlo .
this approach suggests a deep connection between quantum evolution effects in the nuclear wavefunction and instabilities that may be responsible for fast thermalization of the glasma . a cartoon of multi - particle production in a heavy ion collision is shown in fig .
[ fig : cartoon_1 ] . the probability of producing @xmath91 particles , in field theories ( such as the cgc ) with strong external sources can be expressed as @xcite @xmath92 where @xmath93 denotes the sum of vacuum - to - vacuum graphs with r cuts .
this formula has remarkable features : a ) @xmath94 is non - perturbative in @xmath95 even for @xmath96-no simple power expansion in terms of @xmath95 exists .
b ) @xmath94 , for any @xmath91 , gets contributions from cut tree vacuum graphs - this would not apply for field theories in the vacuum .
c ) even at tree level , @xmath94 is _ not _ a poisson distribution which is counter to presumptions that classical field theories have only trivial " poissonian correlations .
the simple formula in eq . [ eq : mp1 ] contains many features of the well known agk calculus @xcite of multi - particle production . computing the probabilities in eq .
[ eq : mp1 ] is hopeless even for @xmath96 .
fortunately , a systematic expansion in powers of g exists for moments of the multiplicity . both the lo and nlo multiplicity can be represented in terms of solutions of equations of motion with retarded boundary conditions . at leading order ,
these are solutions @xmath97 to the yang - mills equations ; these equations , with _
boost invariant _ cgc initial conditions , were solved numerically in ref .
@xcite to compute the gauge fields at late times . a next - to - leading order
( nlo ) computation is important to understand the renormalization and factorization issues that are fundamental to any quantum field theory . as we shall discuss shortly , it is also important to understand the quantum fluctuations that generate the plasma instabilities which may speed up thermalization . remarkably , the nlo contributions can be computed by solving the initial value problem of small fluctuation equations of motion with retarded boundary conditions .
a similar algorithm has been constructed and implemented to study quark pair production in the classical cgc background field @xcite . in the glasma , the classical lo _
boost invariant _ @xmath98 and @xmath70 fields are purely longitudinal at @xmath99 .
the corresponding momentum distributions , at @xmath100 , are very unstable indeed , they lead to an instability which may be analogous to the well known weibel instability in electromagnetic plasmas . for a review and relevant references , see ref .
3 + 1-d numerical simulations demonstrate that small rapidity dependent quantum fluctuations grow exponentially and generate longitudinal pressure @xcite . the initial
seed for the simulations corresponded to white noise " gaussian random fluctuations .
the maximally unstable modes of the longitudinal pressure grow as @xmath101 with @xmath102 ; this form of the growth was previously predicted for weibel instabilities in expanding plasmas @xcite .
albeit the solutions of the yang - mills equations display similar features to the htl studies , a deeper understanding of this connection is elusive .
first quantum corrections to the classical background field of two nuclei at @xmath99 give initial conditions @xcite that are quite different from those in ref .
simulations are underway to determine whether these initial conditions speed up thermalization .
a full treatment of quantum fluctuations requires that we understand how some nlo contributions are absorbed in the evolution of the nuclear wavefunctions with energy while the rest contribute to gluon production . a proof of this high energy
factorization " is in progress @xcite . to fully understand fast thermalization in the presence of instabilities ,
one also needs a kinetic theory of the glasma that describes the decay of classical fields into particles .
a first step has been made in this direction @xcite .
, phys . rev .
* d 49 * , 2233 ( 1994 ) ; _ ibid .
_ , * d 49 * , 3352 ( 1994 ) ; _ ibid . _ * d 50 * , 2225 ( 1994 ) ; yu.v .
kovchegov , phys .
d 54 * , 5463 ( 1996 ) ; j. jalilian - marian , a. kovner , l.d .
mclerran , h. weigert , phys . rev .
* d 55 * , 5414 ( 1997 ) ; e. iancu , a. leonidov and l. d. mclerran , nucl .
a * 692 * , 583 ( 2001 ) . ,
phys . rev . *
d 55 * , 5414 ( 1997 ) ; j. jalilian - marian , a. kovner , a. leonidov , h. weigert , nucl .
* b 504 * , 415 ( 1997 ) ; _ ibid._,phys . rev . *
d 59 * , 034007 ( 1999 ) ; e. iancu , a. leonidov and l. d. mclerran , nucl . phys .
a * 692 * , 583 ( 2001 ) ; e. ferreiro , e. iancu , a. leonidov and l. mclerran , nucl .
a * 703 * , 489 ( 2002 ) .
v. s. fadin , r. fiore , a. v. grabovsky and a. papa , arxiv:0705.1885 [ hep - ph ] ; v. s. fadin , r. fiore and a. papa , phys .
b * 647 * , 179 ( 2007 ) ; nucl .
b * 769 * , 108 ( 2007 ) ; i. balitsky , phys .
d * 75 * , 014001 ( 2007 ) ; y. v. kovchegov and h. weigert , nucl .
phys . a * 789 * , 260 ( 2007 ) ; _ ibid . _ , nucl . phys .
a * 784 * , 188 ( 2007 ) ; e. gardi , j. kuokkanen , k. rummukainen and h. weigert , nucl .
a * 784 * , 282 ( 2007 ) .
a. krasnitz , y. nara and r. venugopalan , phys .
rev . lett .
* 87 * , 192302 ( 2001 ) ; nucl .
a * 717 * , 268 ( 2003 ) ; _ ibid . _ , * 727 * , 427 ( 2003 ) ; t. lappi , phys . rev .
c * 67 * , 054903 ( 2003 ) ; phys .
b * 643 * , 11 ( 2006 ) ; k. fukushima , arxiv:0704.3625 [ hep - ph ] .
y. v. kovchegov and a. h. mueller , nucl . phys .
b * 529 * , 451 ( 1998 ) ; a. dumitru and l. d. mclerran , nucl .
a * 700 * , 492 ( 2002 ) ; j. p. blaizot , f. gelis and r. venugopalan , nucl .
a * 743 * , 13 ( 2004 ) ; f. gelis and y. mehtar - tani , phys .
d * 73 * , 034019 ( 2006 ) .
blaizot , f. gelis , r. venugopalan , nucl . phys . * a 743 * , 57 ( 2004 ) ; h. fujii , f. gelis , r. venugopalan , phys .
lett . * 95 * , 162002 ( 2005 ) ; nucl .
a * 780 * , 146 ( 2006 ) ; k. tuchin , phys .
b * 593 * , 66 ( 2004 ) ; n.n .
nikolaev , w. schafer , phys . rev .
* d 71 * , 014023 ( 2005 ) ; n.n .
nikolaev , w. schafer , b.g .
zakharov , phys .
lett . * 95 * , 221803 ( 2005 ) .
n. n. nikolaev , b. g. zakharov and v. r. zoller , z. phys .
a * 351 * , 435 ( 1995 ) ; l. frankfurt , v. guzey and m. strikman , phys .
b * 586 * , 41 ( 2004 ) ; e. levin and m. lublinsky , nucl .
a * 712 * , 95 ( 2002 ) ; m. s. kugeratski , v. p. goncalves and f. s. navarra , eur .
j. c * 46 * , 413 ( 2006 ) . | we discuss the empirical evidence for a universal color glass condensate and outline prospects for further studies at future colliders . some ramifications for initial conditions in heavy ion collisions
are pointed out . |
the focusing nonlinear schrdinger equation ( nls ) @xmath2 where @xmath3 , and @xmath4 is the laplacian , has been the subject of intense study , due to its role in various areas of physics , such as nonlinear optics and bose - einstein condensates ( bec ) .
it is well - known that the nls possesses solutions that become singular in a finite time @xcite .
of special interest is the critical ( @xmath5 ) nls @xmath6 which models the collapse of intense laser beams that propagate in a bulk kerr medium . in this study , we consider the focusing biharmonic nonlinear schrdinger equation ( bnls ) @xmath7 where @xmath8 is the biharmonic operator .
equation admits waveguide ( standing - wave ) solutions of the form @xmath9 , where @xmath10 satisfies the `` standing - wave '' equation @xmath11 the bnls is called `` _ @xmath0-critical _ '' , or simply `` critical '' if @xmath12 . in this case , the @xmath0 norm ( `` _ power _ '' ) is conserved under the bnls dilation symmetry @xmath13 .
the critical bnls can be rewritten as @xmath14 correspondingly , the bnls with @xmath15 is called subcritical , and the bnls with @xmath16 is called supercritical .
this is analogous to the nls , where the critical case is @xmath17 . in @xcite
, ben - artzi , koch and saut proved that the bnls is locally well - posed in @xmath18 , when @xmath19 is in the _ @xmath18-subcritical _ regime @xmath20 global existence and scattering of bnls solutions in the @xmath18-critical case @xmath21 were studied by miao , xu and zhao @xcite and by pausader @xcite .
the latter work also showed well - posedness for small data .
the @xmath18-critical defocusing bnls was studied by miao , xu and zhao @xcite and by pausader @xcite .
the above studies focused on non - singular solutions . in this work ,
we study singular solutions of the bnls in @xmath18 , i.e. , solutions that exist in @xmath22 over some finite time interval @xmath23 , but for which @xmath24 the first study of singular bnls solutions was done by fibich , ilan and papanicolau @xcite , who proved the following results : [ thrm : ge_subcritical ] solutions of the subcritical ( @xmath15 ) focusing bnls exist globally .
[ thrm : ge_critical ] let @xmath25 , where @xmath26 , and @xmath10 is the ground state of with @xmath27 .
then , the solution of the critical focusing bnls exists globally .
the simulations in @xcite suggested that there exist singular solutions for @xmath12 and @xmath16 , and that these singularities are of the blowup type , namely , the solution becomes infinitely localized .
however , in contradistinction with nls theory , there is currently no rigorous proof that solutions of the bnls can become singular in either the critical or the supercritical case . to the best of our knowledge ,
the only work , apart from @xcite , which considered singular solutions of the bnls is by chae , hong and lee @xcite , who proved that if singular solutions of the critical bnls exist , then they have a power - concentration property .
see section [ ssec : power_conc ] for more details .
in this work , we consider singular solutions of the focusing bnls in the @xmath18-subcritical regime .
our purpose is to characterize these singular solutions : their profile , blowup rate , power concentration , et cetera . in some cases
, we assume radial symmetry , i.e. , that @xmath28 where @xmath29 in these cases , equation reduces to @xmath30 where @xmath31 is the radial biharmonic operator . specifically , the critical bnls reduces to @xmath32 the paper is organized as follows : in section [ sec : invariance ] , we use noether theorem to derive conservation laws for the bnls .
we recall that in the critical nls , the conservation law which follows from invariance of the action integral under dilation leads to the nls `` variance identity '' , which can be used to prove the existence of singular solutions . in section [ ssec : varid ]
we use a similar procedure to derive the `` variance identity '' for the critical bnls , and then generalize it to the supercritical bnls .
however , since it is not clear that the `` bnls variance '' is positive definite , this identity does not lead to a proof of the existence of singular solutions .
the ground states of the bnls standing - wave equation were previously computed only in the one - dimensional case @xcite , since they were computed using a shooting method , which can not be easily generalized to multi - dimensions . in section [ sec :
stationary ] we use the spectral renormalization method to compute the ground - states of the critical bnls for one , two and three dimensions .
the calculated ground - states provide the first numerical estimate of the critical power for collapse @xmath33 in the two - dimensional and three - dimensional cases , see section [ sec : critical_power ] .
direct simulations of the critical bnls suggest that the constant @xmath34 in theorem [ thrm : ge_critical ] is optimal . in section [ sec : rigorous ]
we use rigorous analysis to study the critical bnls .
the blowup rate is shown to be lower - bounded by a quartic root , i.e. , @xmath35 the corresponding bound for the critical nls is a square root , i.e. , @xmath36 we then prove that singular solutions converge to a self - similar profile strongly in @xmath37 , for any @xmath19 in the @xmath18-subcritical regime .
this implies that the singular solutions have the power - concentration property , whereby the amount of power that enters the singularity point is at least @xmath34 .
these rigorous results mirror those of the critical nls .
let us denote the location of the maximal amplitude of a radially - symmetric solution by @xmath38 singular solutions are called `` _ peak - type _ '' when @xmath39 for @xmath40 , and `` _ ring - type _ '' when @xmath41 for @xmath42 . in this work ,
we use asymptotic analysis and numerics to find and characterize peak - type singular solutions of the bnls equation .
ring - type singular solutions of the bnls will be studied elsewhere @xcite . in section [ sec : crit_peak ] we use asymptotics and numerics to show that peak - type singular solutions of the critical bnls collapse with the quasi self - similar profile @xmath43 where the self - similar profile is the ground state @xmath44 . the blowup rate is shown to be slightly faster than the quartic - root bound .
this is analogous to the critical nls , where the blowup rate of peak - type solutions is slightly faster than the square - root bound , due to the loglog correction ( the `` loglog law '' ) .
it is an open question whether the correction to the bnls blowup rate is also a loglog term . in section [ sec : supercrit_peak ] we consider peak - type singular solutions of the supercritical bnls . in this case ,
asymptotics and numerics show that singular solutions are of the quasi self - similar form @xmath45 the blowup rate of @xmath46 is exactly a quartic - root , and the self - similar profile @xmath47 is different from the ground - state .
rather , as in the supercritical nls , the self - similar profile @xmath48 is the zero - hamiltonian solution of a nonlinear eigenvalue problem .
although @xmath47 is not in @xmath0 , it can be the self - similar profile of a collapsing @xmath18 solution , since the collapsing solution is `` only '' quasi self - similar .
section [ sec : num_meth ] presents the numerical methods .
the computations of singular bnls solutions that focus by factors of @xmath1 and more , necessitated the usage of adaptive grids .
we develop a modified version of the static grid redistribution method @xcite , which is more convenient for the biharmonic problem . calculating the bnls standing waves in multi - dimensions
is done using the spectral renormalization method . in this study , we use rigorous theory , asymptotic theory and numerics to analyze singular solutions of the bnls .
all the results presented in this work mirror those of the nls , `` up to a change by a factor of @xmath49 '' in the blowup rate ( @xmath50 ) , in the critical value of @xmath19 ( @xmath51 ) , et cetera .
however , several key features of nls theory are still missing from bnls theory .
first , the `` variance identity '' for the bnls can not be used to prove that singular solutions exist .
second , the critical nls is invariant under the pseudo - conformal ( `` lens - transformation '' ) symmetry which can also be used to construct explicit singular solutions . at this time
, it is unknown whether an analogous identity for the critical bnls exists .
third , in critical nls theory , the self - similar profile is known to possess a quadratic radial phase term , i.e. , @xmath52 this term represents the focusing of the solution towards @xmath53 , and plays a key role in the rigorous and asymptotic theory of the critical nls .
at this time , we do not know the analogous radial phase term for the critical bnls .
finally , we note that a similar `` up to a factor of @xmath49 '' connection exists between singular solutions of the nonlinear heat equation , see @xcite , and the biharmonic nonlinear heat equation , see @xcite .
for example , the @xmath54 norm of singular solutions blows up as @xmath55 for the nonlinear heat equation , and as @xmath56 for the biharmonic heat equation .
the `` similarity up to a factor of @xmath49 '' , however , is not perfect .
for example , the self - similar spatial variable is @xmath57 for the nonlinear heat equation , and @xmath58 { { { t_{\rm c}}}-t } $ ] for the biharmonic nonlinear heat equation .
another difference between the equations is that singular solutions are _ asymptotically _ self - similar for the nonlinear heat equation , and truly self - similar for the biharmonic heat equation .
in contrast , the nls possesses self - similar singular solutions , whereas for the bnls it is unknown whether singular solutions are truly , or only asymptotically , self - similar .
the bnls is the euler - lagrange equation of the action integral @xmath59 where @xmath60 is the lagrangian density @xmath61 therefore , the conserved quantities of the bnls can be found using noether theorem , see appendix [ app : noether ] . as in the standard nls ,
invariance of the action integral under phase - multiplications @xmath62 implies conservation of the `` power '' ( @xmath0 norm ) , i.e , .
@xmath63 similarly , invariance under temporal translations @xmath64 implies conservation of the hamiltonian @xmath65 = { \left\vert{\delta\psi}\right\vert}_2 ^ 2 -\frac{1}{\sigma+1 } { \left\vert{\psi}\right\vert}_{2(\sigma+1)}^{2(\sigma+1)},\ ] ] and invariance under spatial translations @xmath66 implies conservation of the linear momentum , i.e. , @xmath67 in the critical case @xmath68 , the action integral is also invariant under the dilation transformation @xmath69 the corresponding conserved quantity is @xmath70 we recall that the action integral of the critical nls is invariant under the dilation transformation @xmath71 the corresponding conserved quantity is @xmath72 in addition , the integral term @xmath73 is the time - derivative of the variance , i.e. , @xmath74 therefore , it follows that @xmath75 in the supercritical nls , the second derivative of the variance is not related to this conservation law .
nevertheless , direct differentiation shows that @xmath76 therefore @xmath77 since @xmath78 , the variance identity shows that solutions of the critical and supercritical nls , whose hamiltonian is negative , become singular in a finite time @xcite .
we next extend the analogy between and to the non - critical case . in the case of the bnls
, direct differentiation shows that @xmath79 therefore , if we define @xmath80 where @xmath81 is a positive constant , we get the bnls variance identity @xmath82 in order to use this identity to prove singularity formation , however , one must show that @xmath83 , as defined by , has to remain positive .
direct integration by parts gives that @xmath84 while the second term on the rhs is a temporal derivative of a positive - definite quantity , the first term on the rhs is not .
therefore , it remains an open question whether @xmath83 has to be positive .
the bnls equation admits the standing - wave solutions @xmath85 the equation for the standing - wave profile is @xmath86 where @xmath87 . for example , in one dimension , equation is given by @xmath88 and in two dimensions by @xmath89 [ eqs : radial_stationary_state ] if we impose radial symmetry , eq .
reduces to @xmath90 where @xmath91 is given in . at @xmath53 ,
all the odd derivatives vanish , and so the solution of is subject to the boundary conditions @xmath92 the solution of eq .
was computed numerically in the one - dimensional case in @xcite as follows . in the @xmath93 case ,
can be integrated once , yielding an explicit relation between @xmath94 and @xmath95 .
this parameter - reduction enables the usage of a one - parameter shooting approach .
unfortunately , such a parameter reduction is not possible in higher dimensions .
therefore , in multi - dimensions we compute the ground - states using the spectral renormalization method ( srm ) , which was introduced by petviashvili in @xcite , and more recently by albowitz and musslimani in @xcite .
see section [ ssec : srm ] for further details . in figure
[ fig : srm ] we display the results in the critical case , i.e. , the ground states of @xmath96 figure [ fig : srm_1d ] displays the ground - state of eq . in the critical @xmath93 case , as calculated by the srm .
the solution is in excellent agreement with the solution computed in @xcite using the shooting method .
figure [ fig : srm_2d ] and figure [ fig : srm_3d ] display the ground state in the critical @xmath97 and @xmath98 cases .
we note that , while the srm method that we use does not enforce radial symmetry , the calculated ground states for @xmath99 and @xmath100 are radially symmetric ( data not shown ) .
as noted in @xcite , the ground - states of the bnls are non - monotonic in @xmath101 and change their sign , in contradistinction with the ground - states of the nls which are monotonically - decreasing and strictly positive .
theorem [ thrm : ge_critical ] shows that the critical power for collapse in the critical bnls is @xmath102 , when @xmath10 is the ground - state of .
the computation of @xmath10 , see section [ sec : stationary ] , allows for the numerical calculation of the critical power @xmath34 .
the case @xmath103 was found in @xcite to be @xmath104 using the calculated ground state in the two - dimensional case , see figure [ fig : srm_2d ] , we now calculate the critical power in the two - dimensional case , giving @xmath105 similarly , using the calculated ground state in the @xmath98 case , see figure [ fig : srm_3d ] , gives @xmath106 we now ask whether theorem [ thrm : ge_critical ] is sharp , in the sense that for any @xmath107 , there exists an initial condition @xmath108 such that @xmath109 and the corresponding solution of the critical bnls becomes singular . as noted , at present
there is no proof that solutions of the bnls can become singular .
therefore , in particular , it is unknown whether theorem [ thrm : ge_critical ] is indeed sharp .
hence , we will explore this issue numerically .
we recall that in the critical nls , the necessary condition for collapse @xmath110 is sharp , since for any @xmath107 the initial condition @xmath111 becomes singular in a finite time @xcite .
therefore , we now check numerically whether for @xmath112 , the solution of the critical bnls with the initial condition @xmath113 becomes singular . to do this
, we solve the one - dimensional and two - dimensional critical bnls equations with the initial condition @xmath114 , see fig [ fig : peak_ric ] . in both cases
, the solution appears to blow up , suggesting that theorem [ thrm : ge_critical ] is sharp .
see section [ ssec : sgr ] for the numerical methodology used for solving the bnls .
we also note that in the critical nls , if the initial condition is different from the ground state , the power input required for collapse is strictly larger than @xmath115 , see @xcite .
this the case also in the bnls .
for example , in the one - dimensional critical bnls with a gaussian shaped initial condition @xmath116 , the input power required for collapse is strictly larger than @xmath117 , see @xcite . in the two - dimensional critical bnls with
a gaussian shaped initial condition , the input power required is strictly larger than @xmath118 , see @xcite .
in @xcite , cazanave and weissler proved that the blowup rate for singular solutions of the critical nls is not slower than a square - root , i.e. , that @xmath119 . the analogous result for
the critical bnls is as follows : [ thrm : low - bound ] let @xmath120 be a solution of the critical bnls ( [ eq : cbnls ] ) that becomes singular at @xmath121 , and let @xmath122 . then , @xmath123 such that @xmath124 we follow the proof given by merle @xcite for the critical nls .
for a fixed @xmath125 , @xmath126 , let us define @xmath127 then , @xmath128 is defined for @xmath129 , and satisfies the bnls equation @xmath130 since @xmath131 this implies that @xmath132 , i.e. , that @xmath133 becomes singular as @xmath134 .
in addition , [ eqs : psi1_0_bounds ] @xmath135 from the definition of @xmath128 and power conservation it follows that @xmath136 using equations ( [ eq : delta_psi1_bound ] ) and ( [ eq : psi1_bound ] ) and the cauchy - schwartz inequality gives @xmath137 together , the three formulae ( [ eqs : psi1_0_bounds ] ) imply that for any fixed @xmath138 @xmath139 in other words , for each @xmath125 , the initial @xmath18 norm of @xmath128 is bounded by a function of @xmath140 .
specifically , this bound is independent of @xmath125 . from the local existence theory @xcite ,
@xmath128 exists in @xmath141 $ ] , where @xmath142 therefore , it follows from that @xmath143 depends on @xmath144 , but is independent of @xmath125 .
since @xmath128 blows up at @xmath145 we have that @xmath146 from which the result follows . in @xcite
, weinstein showed that the collapsing core of all singular solutions of the critical nls approaches a self - similar profile .
we now prove the analogous result for the critical bnls : [ thrm : self - similarity ] let @xmath147 and let @xmath148 be a solution of the radially - symmetric critical bnls ( [ eq : cbnls ] ) with initial conditions @xmath149 , that becomes singular at @xmath121 .
let @xmath150 and let @xmath151 then , for any sequence @xmath152 there is a subsequence @xmath153 such that @xmath154 strongly in @xmath155 , for all @xmath156 such that , where @xmath19 is in the @xmath18-subcritical regime .]@xmath157 in addition , @xmath158 , where @xmath10 is the ground state of equation .
let @xmath159 and define @xmath160 from the definition of @xmath161 it follows that @xmath162 = { { \mathit l } } ^4h[\psi(t_k)].\ ] ] therefore , using cauchy - schwartz , @xmath163 since @xmath164 is bounded , it follows that there exists a subsequence of @xmath161 which converges weakly in @xmath18 to a function @xmath165 . from the compactness
lemma [ lem : compactness ] , see appendix [ app : compactness ] , it follows that @xmath166 strongly in @xmath155 , for all @xmath156 given by .
next , we prove that @xmath167\leq0 $ ] .
since @xmath168 , it follows that @xmath169 , and so @xmath170 . additionally , since @xmath171 for some @xmath172 , we have that @xmath173 and so @xmath174 \leq \lim_{k\to\infty } h[\phi_k ] = \lim_{k\to\infty } { { \mathit l } } ^4h[\psi_0 ] = 0.\ ] ] in addition , since @xmath175 = \lim_{k\to\infty } \left ( 1-\frac{1}{\sigma+1}{\left\vert{\phi_k}\right\vert}_{2(\sigma+1)}^{2(\sigma+1 ) } \right),\ ] ] it follows that @xmath176 so @xmath177 .
therefore , corollary [ cor : p2h_bound ] , see appendix [ app : gagli - niren ] , implies that @xmath178 .
solutions of critical nls have the power concentration property , whereby the amount of power that collapses into the singularity is at least @xmath179 , see @xcite . in
what follows , we prove the analogous results for the critical bnls .
[ cor : power - concentration ] let @xmath147 , and let @xmath148 be a solution of the radially - symmetric critical bnls ( [ eq : radial_cbnls ] ) that becomes singular at @xmath121 .
then , @xmath180 , @xmath181 where @xmath182 the result shall follow directly from corollary [ cor : power - conc - lower - bound ] .
the following corollary shows that the rate of power - concentration is not slower than the blowup rate @xmath183 .
the nls analogue is due to tsutsumi @xcite and weinstein @xcite .
[ cor : power - conc - lower - bound ] let @xmath147 , let @xmath148 be a solution of the radially - symmetric critical bnls ( [ eq : radial_cbnls ] ) that becomes singular at @xmath121 , and let @xmath150 .
then , 1 . for any monotonically - decreasing function @xmath184
such that @xmath185 we have that @xmath186 2 .
for any @xmath187 , @xmath188 such that @xmath189 see appendix [ app : proof - power - conc - lower - bound ] . since the second part of corollary [ cor : power - conc - lower - bound ] is true for all @xmath187 and since @xmath190 , corollary [ cor : power - concentration ] follows .
the next corollary shows that the power - concentration rate has a quartic - root upper bound .
the analogue in nls theory , which is a square - root upper bound , was proved in @xcite .
[ cor : power - conc - upper - bound ] let @xmath147 and let @xmath148 be a solution of the radially - symmetric critical bnls ( [ eq : radial_cbnls ] ) that becomes singular at @xmath121 . then , 1 . for any monotonically - decreasing @xmath191
such that @xmath192 we have that @xmath186 2 . for any @xmath187 ,
@xmath188 such that @xmath193 theorem [ thrm : low - bound ] implies that @xmath194 .
therefore , the result follows immediately from corollary [ cor : power - conc - lower - bound ] .
recently , chae , hong and lee @xcite used the harmonic analysis method of bourgain @xcite to prove that singular solutions of the critical bnls for @xmath195 have the power - concentration property @xmath196 where @xmath197 is a positive constant .
this result is more general than corollaries [ cor : power - concentration],[cor : power - conc - lower - bound],[cor : power - conc - upper - bound ] in that it does not assume radial symmetry .
the proof given here , however , is considerably simpler .
more importantly , it shows that @xmath198 .
in this , we consider radially - symmetric singular solutions that are `` _ peak - type _ '' , i.e. , for which @xmath39 for @xmath40 , where @xmath199 is the location of the maximal amplitude .
the critical nls admits singular solutions that collapse with the universal @xmath200 profile , i.e. , @xmath201 , where @xmath202 the self - similar profile @xmath203 is the ground - state solution of @xmath204 the blowup rate of @xmath46 is given by the _ loglog law _
@xcite @xmath205 since the blowup rate is slightly faster than a square root @xmath206 therefore , the phase term @xmath207 in vanishes as @xmath208 .
hence , the blowup profile reduces to @xmath209 we now look for the `` corresponding '' peak - type singular solutions of the critical bnls .
theorem [ thrm : self - similarity ] suggests that the collapsing core of the singular solution approaches a self - similar form , i.e. , @xmath210 where @xmath211 and @xmath212 .
substituting into and requiring that @xmath213 \sim \left [ \delta\psi \right ] \sim \left [ |\psi|^{d/2}\psi \right ] $ ] suggests that @xmath214 let us consider the self - similar profile @xmath47 . in the singular region
@xmath215 we have that @xmath216 and @xmath217 hence , @xmath47 satisfies @xmath218 theorem [ thrm : low - bound ] shows that that blowup rate of @xmath46 is lower - bounded by a quartic - root . in the critical nls the blowup rate of peak - type solutions is slightly faster than the analogous square - root rate , due to the loglog correction .
hence , we expect that the blowup rate of peak - type critical bnls solutions is slightly faster than a quartic root , i.e. , @xmath219{{{t_{\rm c}}}-t } } \to 0.\ ] ] in that case , @xmath220 and reduces to the standing - wave equation .
since the ground - states of attain their maximal amplitudes at @xmath221 , see section [ sec : stationary ] , these are peak - type solutions .
the above informal analysis thus leads to the following conjecture : [ conj : crit_peak_rate_profile ] the critical bnls admits peak - type singular solutions such that : 1 . the collapsing core approaches the self - similar profile + [ eqs : crit_peak_qss ] @xmath222 where @xmath223 + and @xmath10 is the ground - state of equation .
the blowup rate of @xmath46 is slightly faster than a quartic - root , i.e. , @xmath224 in section [ ssec : crit_peak_simulations ] we provide numerical evidence in support of conjecture [ conj : crit_peak_rate_profile ] .
the one - dimensional critical bnls @xmath225 was solved with the gaussian initial condition @xmath226 with @xmath227 , whose power is @xmath228 .
the maximal amplitude of the solution @xmath229 as a function of time is plotted in fig [ fig : peak_amp_ls_1d ] .
the amplitude increases abruptly by a factor of @xmath230 around @xmath231 , suggesting that the solution becomes singular in a finite time .
the simulation was repeated for the radially - symmetric two - dimensional critical bnls @xmath232 with the gaussian initial condition @xmath233 with @xmath234 , whose power is @xmath235 the amplitude increases abruptly by a factor of @xmath1 . around @xmath236 , see fig [ fig : peak_amp_ls_2d ] , again suggesting that the solution becomes singular in a finite time .
we next consider the self - similar profile of the collapsing solutions from figure [ fig : peak_amp_ls ] . in order to verify that it is given by ,
we rescale the solutions as @xmath237 with @xmath238 . the rescaled solutions at focusing levels of @xmath239 and @xmath240 are indistinguishable , see fig [ fig : crit_peak_rescaled ] , indicating that the collapsing core is indeed self - similar according to .
a predicted , the self - similar profile is very close to the ground - state @xmath10 in the core region @xmath241 .
we next compute the blowup rate @xmath242 , defined by the relation@xmath243 to do that , we perform a least - squares fit of @xmath244 with @xmath245 , see figure [ fig : crit_peak_powerlaw ] , obtaining a value of @xmath246 for both @xmath103 and @xmath99 .
this value of @xmath242 is slightly above @xmath247 , implying that the quartic - root lower - bound given by theorem [ thrm : low - bound ] is close to the actual blowup - rate of peak - type singular solutions .
next , we provide two indications that the blowup rate is faster than @xmath247 . first
, if the blowup rate is exactly @xmath247 , then @xmath248 should be finite and strictly negative .
however , up to focusing level of @xmath240 , @xmath249 does not appear to converge to a negative constant , but rather to increase slowly towards @xmath250 , see figure [ fig : crit_peak_l3lt ] .
second , according to the informal analysis in section [ ssec : crit_peak_analysis ] , the blowup rate is faster than a quartic root if and only if the self - similar profile @xmath47 satisfies the standing - wave equation , which is indeed what we observed numerically in figure [ fig : crit_peak_rescaled ] .
* remark : * in the critical nls the blowup rate of peak - type solutions is slightly faster than the analogous square - root lower - bound , due to the well - known loglog - correction .
figure [ fig : crit_peak_powerlaw ] shows that the blowup rate is slightly faster than a quartic root , and figure [ fig : crit_peak_l3lt ] shows that @xmath251 very slowly .
together , this suggests that the blowup rate in the critical bnls is only slightly faster than the analogous quartic root . at present , we do not know if the blowup rate of peak - type solutions of the critical bnls is a quartic root with a @xmath252 correction .
we note , however , that the loglog correction in the critical nls can not be determined numerically @xcite , and can only be derived analytically .
therefore , we expect that the determination of the analogous correction to the @xmath247 blowup rate of the critical bnls will also have to be done analytically , and not numerically .
in contrast to the extensive theory on singularity formation in the critical nls , much less is known about the supercritical nls . numerical simulations and formal calculations ( see ,
e.g. , ( * ? ? ?
* chapter 7 ) and the references therein ) suggest that peak - type singular solutions of the supercritical nls collapse with a universal @xmath253 profile , i.e. , [ eq : intro_psiq ] @xmath254 , \\
\psi_\text{non - singular}(t , r)\quad & r\ge r_c , \end{cases}\ ] ] where @xmath255 note that the singular region @xmath256 $ ] is constant in the coordinate @xmath101 .
therefore , in the rescaled variable @xmath257 , the singular region @xmath258 $ ] becomes infinite as @xmath259 .
this is in contradistinction with the critical case , wherein the singular region @xmath260 $ ] is constant in the rescaled variable @xmath261 , but shrinks to a point in the original coordinate @xmath101 .
the self - similar profile @xmath262 is the solution of @xmath263 solutions of ( [ eq : ode4q ] ) are complex - valued , and depend on the parameter @xmath264 and on the initial condition @xmath265 .
solutions of ( [ eq : ode4q ] ) whose amplitude @xmath266 is monotonically - decreasing in @xmath261 , and which have a zero hamiltonian , are called _ admissible solutions _ @xcite . for each choice of @xmath267 , equation has a unique admissible solution ( up to a multiplication by a constant phase @xmath268 ) .
this solution is attained for specific real values of @xmath264 and @xmath269 , which we denote as @xmath270 the blowup rate of @xmath46 is a square - root , i.e. , @xmath271 numerical simulations and formal calculations suggest that : 1 . the self - similar profile of singular peak - type solutions of the nls is an admissible solution of .
since @xmath272 attains its maximal amplitude at @xmath221 , the solution is peak - type .
the constant @xmath264 of the blowup rate is equal to @xmath273 .
hence , in particular , @xmath264 is universal ( i.e. , is independent of the initial condition @xmath274 ) .
the admissible solution @xmath272 satisfies @xmath275 thus , @xmath276 .
nevertheless , @xmath272 can be the self - similar profile of @xmath277 solutions , since @xmath278 only for @xmath279 $ ] , see @xcite .
as in the supercritical nls , we expect that singular peak - type solutions of the supercritical bnls collapse as @xmath280 where @xmath281 is a self - similar profile , to be determined . as in the supercritical nls ,
the singular region @xmath256 $ ] , is constant in the coordinate @xmath101 .
therefore , in the rescaled variable @xmath257 , the singular region @xmath258 $ ] becomes infinite as @xmath259 .
this is again in contradistinction with the critical - bnls case , where the singular region @xmath260 $ ] is constant in the rescaled variable @xmath261 , but shrinks to a point in the original coordinate @xmath101 .
the bnls is invariant under the dilation symmetry @xmath282 , where @xmath283 is a constant . in the supercritical case
@xmath16 , this suggests that @xmath284 similar arguments as in section [ ssec : crit_peak_analysis ] show that @xmath285 therefore , as in the supercritical nls , we expect the collapsing part of the solution to approach the self - similar profile , hence the self - similar profile reduces to . ]
@xmath286 theorem [ thrm : low - bound ] showed that in the critical case , if @xmath287 , then @xmath288 .
the following lemma extends this result to peak - type solutions of the supercritical bnls .
[ lem : supercrit_peak_rate_profile ] let @xmath16 , and let @xmath120 be a peak - type singular solution of the bnls that collapses with the @xmath281 profile . if @xmath289 , then @xmath290 .
furthermore , @xmath291 if and only if the self - similar profile @xmath47 satisfies the equation @xmath292 if @xmath293 , then @xmath294 and @xmath295 hence , the equation for @xmath48 is @xmath296 implying that @xmath297 should be bounded as @xmath298 .
since @xmath299 , it follows that @xmath300 .
if @xmath291 , then @xmath301 , and equation reduces to .
let us consider the fourth - order nonlinear ode for the self - similar profile @xmath47 .
its solution requires four boundary conditions , and the determination of the parameter @xmath264 .
radial symmetry implies that @xmath302 .
since the solution is invariant up to rescaling @xmath303 , one can set , with no loss of generality , @xmath304 .
therefore , we require two additional constraints in order to determine @xmath264 and @xmath305 .
we recall that in the supercritical nls the `` admissible value '' of @xmath264 is determined from the requirements that @xmath262 has a zero hamiltonian and a monotonically - decreasing amplitude , see section [ ssec : supercrit_peak_nls_review ] . for the bnls
, the following informal argument suggests that the zero - hamiltonian condition should also holds .
indeed , from hamiltonian conservation it follows that @xmath306 $ ] is bounded , because otherwise the non - singular region would also have an infinite hamiltonian . calculating the hamiltonian of @xmath281
, we have @xmath307 \sim l^{-4/\sigma+d-4 } \left [ \int_{\rho=0}^{r_c / l(t ) } \left ( { \left\vert { \delta_\rho b(\rho ) } \right\vert}^2 -\frac{1}{1+\sigma}{\left\vert{b}\right\vert}^{2 + 2\sigma } \right ) \rho^{d-1 } d\rho \right].\ ] ] from @xmath18-subcriticality , see , it follows that @xmath308 as @xmath309 .
therefore , if @xmath310 $ ] remains bounded as @xmath208 then @xmath311 = \int_{\rho=0}^{\infty } \left ( { \left\vert { \delta_\rho b } \right\vert}^2 -\frac1{1+\sigma } { \left\vert { b } \right\vert}^{2 + 2\sigma } \right ) \rho^{d-1}d\rho
= 0.\ ] ] wkb analysis of shows that , see appendix [ app : wkb ] , @xmath312 where @xmath313 and @xmath314 .
equation therefore has two algebraically - decaying solutions , @xmath315 and @xmath316 , an exponentially - increasing solution @xmath317 , and an exponentially - decreasing solution @xmath318 . since @xmath16 , the exponent @xmath319 of @xmath316 is larger that the exponent @xmath320 of @xmath315 , hence @xmath321 as @xmath322 .
[ lem : admissible_b23 ] let @xmath47 be a zero - hamiltonian solution of . then , @xmath323 and @xmath324 furthermore , @xmath325 .
the exponentially increasing solution @xmath317 must vanish identically if the integrals are to converge , hence @xmath326 .
next , the @xmath327 phase term in @xmath316 implies that @xmath328 hence , the integral @xmath329 diverges in the @xmath18-subcritical regime @xmath330 , i.e. , @xmath331 . since
, in addition , @xmath332 the hamiltonian can be finite only if @xmath333 , in which case @xmath325 .
lemma [ lem : admissible_b23 ] shows that the condition @xmath334=0 $ ] imposes the two constraints @xmath333 and @xmath326 .
hence , zero - hamiltonian solutions of equation satisfy the five boundary conditions @xmath335 therefore they form a discrete set of solutions and of values of @xmath264 .
we conjecture that the additional condition of monotonicity of @xmath336 will lead to a unique admissible solution @xmath48 and a unique value of @xmath264 .
[ cor : admissible_b23_power ] let @xmath47 be a zero - hamiltonian solution of .
then , @xmath337
. nevertheless , @xmath338 .
since @xmath339 , @xmath340 following the arguments of @xcite , the profile @xmath281 satisfies @xmath341 in summary , we conjecture the following : [ conj : supercrit_peak_rate_profile ] + let @xmath120 be peak - type singular solution of the supercritical bnls .
then , [ eqs : supercrit_peak_qss ] 1 .
the collapsing core approaches the self - similar profile @xmath281 , i.e. , @xmath342 where @xmath343 2 .
the self - similar profile @xmath47 is the solution of @xmath344=0 , \end{gathered}\ ] ] where @xmath345 and @xmath334 $ ] is the hamiltonian of @xmath48 , see .
3 . in particular , @xmath346 .
4 . equation has a unique `` admissible solution '' with a unique `` admissible value '' of @xmath347 , such that @xmath348 is monotonically decreasing .
additionally , @xmath349 as @xmath322 .
the admissible solution is the self - similar profile @xmath48 of @xmath281 , see .
the blowup rate of singular peak - type solutions is exactly a quartic root , i.e.,@xmath350{{{t_{\rm c}}}-t } , \qquad \kappa>0.\ ] ] 7 .
the coefficient @xmath264 of the blowup rate of @xmath46 is equal to the value of @xmath264 of the admissible solution @xmath48 , i.e. , @xmath351{{{t_{\rm c}}}-t } } = \kappa(\sigma , d).\ ] ] in particular , @xmath264 is universal ( i.e. , it does not depend on the initial condition ) . in section [ ssec : supercrit_peak_simulations ] we provide numerical evidence in support of conjecture [ conj : supercrit_peak_rate_profile ] . [
cols="^,^,^,^,^,^",options="header " , ] the radially - symmetric bnls was solved in the supercritical case @xmath352 with the initial condition @xmath353 , and in the supercritical case @xmath354 with the initial condition @xmath355 . in both cases ,
the solutions blowup at a finite time , see figure [ fig : supercrit_peak_amp_ls ] . to check whether the solutions collapse with the self - similar profile , the solution was rescaled according to .
the rescaled solutions at the focusing levels @xmath239 and @xmath240 are indistinguishable in both the one - dimensional case ( figure [ fig : supercrit_peak_rescaled_1d ] ) and the two - dimensional case ( figure [ fig : supercrit_peak_rescaled_2d ] ) , providing numerical support that the solution collapses with the @xmath281 profile . as predicted , the self - similar profile is different than the ground - state @xmath10 . indeed , figure [ fig : supercrit_peak_rescaled_far ] shows that as @xmath356 , the self - similar profile of @xmath120 decays as @xmath357 , which is in agreement with the decay rate of @xmath358 .
we next verify that the solution converges to the asymptotic profile for @xmath256 $ ] , i.e. , for @xmath258 $ ] . to do this , we plot in figure [ fig : supercrit_peak_convergence ] the rescaled solution at focusing levels of @xmath359 , as a function of @xmath360 .
the curves are indistinguishable at @xmath361 , but bifurcate at increasing values of @xmath362 .
these `` bifurcations positions '' are marked by circles in figure [ fig : supercrit_peak_convergence ] , and their @xmath362 values are listed in table [ tab : supercrit_peak_convergence ] .
the `` bifurcation positions '' are linear in @xmath363 , indicating that the region where @xmath293 is indeed @xmath258 $ ] , which corresponds to @xmath256 $ ] . to compute the blowup rate @xmath242 , we performed a least - squares fit of @xmath244 with @xmath245 , see figure [ fig : supercrit_peak_powerlaw ] .
the resulting values are @xmath364 in the @xmath352 case and @xmath365 in the @xmath354 case .
next , we provide two indications that the blowup rate is exactly @xmath247 , i.e. , that @xmath366{{{t_{\rm c}}}-t},\qquad \kappa>0.\ ] ] first , if the blowup rate is exactly a quartic root , then @xmath367 indeed , figure [ fig : supercrit_peak_l3lt ] shows that in the case @xmath103 , @xmath368 , @xmath369 , implying that @xmath370{4\cdot0.289 } \approx 1.037\,.\ ] ] in the case @xmath99 , @xmath371 , @xmath372 , implying that @xmath373{4\cdot0.171 } \approx 0.909\,.\ ] ] since @xmath249 converges to a finite , negative constant , this shows that the blowup rate is exactly @xmath247 .
second , according to lemma [ lem : supercrit_peak_rate_profile ] , if @xmath374 the self - similar profile @xmath47 should not satisfy the standing - wave equation . in figure
[ fig : supercrit_peak_rescaled ] we saw that the rescaled self - similar bnls solutions and the corresponding ground - states differ considerably ( compare with figure [ fig : crit_peak_rescaled ] ) , indicating that the blowup rate is exactly @xmath247 .
therefore , the numerical results again support conjecture [ conj : supercrit_peak_rate_profile ] .
finally , we verified that the value of @xmath264 in the blowup rate is universal .
we solve the bnls in the case @xmath352 with the initial condition @xmath375 . in this case , the calculated value of @xmath376 is @xmath377{-4l_tl^3}\approx1.037 $ ] , which is equal , to first @xmath378 significant digits , to the previously obtained value , see , for the initial condition @xmath353 .
similarly , in the case @xmath354 , we solve the equation with the initial condition @xmath379 .
the calculated value of @xmath380 is @xmath377{-4l_tl^3}\approx0.913 $ ] , which is equal , to first @xmath49 significant digits , to the previously obtained value , see , for the initial condition @xmath381 .
in this study , we computed singular solutions of the bnls equation .
these solutions become highly - localized , so that the spatial scale - difference between the singular region @xmath382 and the exterior regions can be as large as @xmath383 . in order to resolve the solution at both the singular and non - singular regions
, we use an adaptive grid .
we generate the adaptive grids using the _ static grid redistribution _ ( sgr ) method , which was first introduced by ren and wang @xcite , and later simplified and improved by gavish and ditkowsky @xcite . using this approach ,
the solution is allowed to propagate ( self - focus ) until it becomes under - resolved . at this stage , a new grid , with the same number of grid - points ,
is generated using deboors ` equidistribution principle ' , wherein the grid points @xmath384 are spaced such that a certain weight function @xmath385 $ ] is equidistributed , i.e. , that @xmath386dr = \text{const},\ ] ] see @xcite for details .
the algorithms of @xcite keeps a recursive set of grids , the coordinates of each is given in the reference frame of the previous grid .
this enables the simulation to reach high focusing levels , where the grid - points position can not be stored at the physical reference frame due to loss of significant digits . in this study , however , we implement a simplified version of the method of @xcite , in which we dispense with the above hierarchy of grids and store the grid - points position in terms of the original reference frame . on this highly non - uniform grid ,
we use a standard non - uniform finite - difference approximation of the radial biharmonic operator .
the approximation is third - order accurate , with a seven - point stencil .
thus , we are limited by the standard machine accuracy - bit floating points , with relative machine error of @xmath387 . using @xmath388-bit floating points will ,
for all practical purposes , eliminate the above restriction . ] to focusing levels of no more than @xmath389 .
however , the resulting gain in software ( and numerical ) simplicity justifies this limitation .
the method in @xcite allows control of the fraction of grid points that migrate into the singular region , preventing under - resolution at the exterior regions .
this is done by using a weight - function @xmath390 , which penalizes large inter - grid distances .
however , we found that this numerical mechanism , while necessary , is insufficient for our purposes .
in order to understand the reason , let us consider the grid - point spacings @xmath391 . using the method of @xcite with both @xmath392 and
@xmath390 causes a very sharp bi - partition of the grid points to those inside the singular region , whose spacing is determined by @xmath392 and is @xmath393 , and to those outside the singular region , whose spacing is determined by @xmath390 and is @xmath394 , see figure [ fig : w3_effect_old ] .
inside each of these regions , the finite difference approximation we use is well conditioned .
however , at the transition between these two regions , the finite - difference stencil , seven - points in width , spans grid - spacings with @xmath395 scale - difference leading to under - resolution which completely violates the validity of the finite - difference approximation . in order to overcome this limitation ,
we improve the algorithm of @xcite by adding a third weight function @xmath396 which penalizes the second - difference @xmath397 operator of the grid locations , allowing for a smooth transition between the singular region and the non - singular region , see fig [ fig : w3_effect_new ] . on the sequence of grids ,
the equations are solved using a predictor - corrector crank - nicholson scheme , which is second - order in time . here
, we describe the adaptation of the spectral renormalization method ( srm ) to the standing - wave bnls equation .
denoting the fourier transform of @xmath398 by @xmath399({{\mathbf { k}}})$ ] , equation transforms to @xmath400({{\mathbf { k}}})=\frac{1}{k^4 + 1 } \mathcal f\left[|r|^{2\sigma}r\right],\ ] ] where @xmath401 , leading to the fixed - point iterative scheme @xmath402= \frac{1}{k^4 + 1 } \mathcal f\left[|r_m|^{2\sigma}r_m\right ] , \qquad m=0,1,\dots~.\ ] ] typically this iterative scheme diverges either to @xmath403 or to @xmath404 . in order to avoid this problem ,
we renormalize the solution as follows .
multiplying equation by @xmath399^*$ ] and integrating over @xmath405 gives the integral relation : [ eqs : srm ] @xmath406 \equiv \int{\left\vert{\mathcal f[r]}\right\vert}^2d{{\mathbf { k } } } ,
\quad sr[r ] \equiv \int \frac{1}{k^4 + 1 } \mathcal f\left[|r|^{2\sigma}r\right]\mathcal f[r]^*d{{\mathbf { k}}}.\ ] ] we now define @xmath407 such that the integral relation is satisfied by @xmath408 , i.e , that @xmath409 = c_m^2sl[r_m ] = c_m^{2\sigma+2}sr[r_m]=sr\left[r_{m+1/2}\right],\ ] ] leading to @xmath410}{sr[r_m]}\right)^{\frac{1}{2\sigma } } , $ ] and hence to @xmath411}{sr[r_m]}\right)^{1+\frac{1}{2\sigma } } { \left\vert{r_m}\right\vert}^{2\sigma}r_m.\ ] ] the spectral renormalization method is therefore given by the iterations @xmath412}{sr[r_m]}\right)^{1+\frac{1}{2\sigma } } \frac{1}{k^4 + 1 } \mathcal f\left(|r_m|^{2\sigma}r_m\right ) , \qquad m=1,2\dots~.\ ] ] in this work , we use the srm to solve for the cases @xmath413 without imposing radial symmetry .
alternatively , one might have solved the radial equation using a modified hankel - like transform instead of the fourier transform .
our main reason for not doing so is the convenience and cost - effectiveness of using the fast fourier transform .
we also note that our non - radially - symmetric method produced a radially - symmetric solution , which suggests that the ground state is radially symmetric .
we thank nir gavish for useful discussions .
this research was partially supported by grant # 123/2008 from the israel science foundation ( isf ) .
here we provide an extension of the compactness lemma for @xmath414 functions @xcite , to the case of @xmath415 : ( compactness lemma)[lem : compactness ] let @xmath147 and let @xmath416 be in the @xmath18-subcritical regime .
then , the embedding @xmath417 is compact , i.e. , every bounded sequence @xmath418 has a subsequence @xmath419 which converges strongly in @xmath420 .
if @xmath421 , then the sequence @xmath422 has a subsequence @xmath419 which converges weakly to @xmath423 in @xmath18 . since the limit of radial functions is a radial function , @xmath424 .
in addition , since for any bounded domain @xmath425 , the embedding @xmath426 is compact , there is a subsequence which converges strongly to @xmath423 in @xmath427 , i.e. , @xmath428 . from the gagliardo - nirenberg inequality on the bounded domain @xmath425 ,
see @xcite , @xmath429 and since @xmath430 in the @xmath18-subcritical case , it follows that @xmath431 strongly in @xmath432 , so that @xmath433 next , strauss radial lemma @xcite for @xmath434 functions gives that @xmath435 and @xmath436 , @xmath437 so that @xmath438 s.t .
@xmath439 @xmath440 finally , since @xmath441 the convergence in @xmath442 is obtained .
from the proof of theorem [ thrm : self - similarity ] , @xmath443 and since @xmath444 , we have that @xmath445 @xmath446 since @xmath447 , it follows that @xmath448 , and so @xmath449 this is true @xmath450 , and so @xmath451 for the second result , since @xmath452 , it follows that for all @xmath187 there exist @xmath453 such that @xmath454 therefore , since @xmath455 a similar argument as in the previous section gives @xmath456
in @xmath0-critical case @xmath12 , which is always in the @xmath18-subcritical regime , the appropriate gagliardo - nirenberg inequality in @xmath18 is @xcite : ( gagliardo - nirenberg inequality ) let @xmath12 , and let @xmath457 , then @xmath458 we note that the ground - state @xmath10 of equation is the minimizer of the gagliardo - nirenberg inequality @xcite , and that its @xmath0 norm , the critical power , satisfies @xmath459 hence , the gagliardo - nirenberg inequality implies the following corollary : [ cor : p2h_bound ] let @xmath460 and @xmath12 , then @xmath461 \geq \left [ 1-\left ( \frac{{\left\vert{f}\right\vert}_2 ^ 2}{{{p_{\rm cr } } } } \right)^{\sigma } \right ] \cdot { \left\vert{\delta f}\right\vert}_2 ^ 2,\ ] ] so that @xmath462\geq0.\ ] ]
as @xmath356 , the nonlinear term in becomes negligible , and reduces to @xmath463 where @xmath464 in order to apply the wkb method , we substitute @xmath465 , and expand @xmath466 substituting @xmath467 and balancing terms shows that @xmath468 , and that the equation for the leading - order , the @xmath469 terms , is @xmath470 therefore , @xmath471 the equation for the next order , the @xmath472 terms , is @xmath473 implying that @xmath474 the next - order terms are @xmath475 and can be neglected .
[ eqs : wkb_solutions ] we therefore obtain the three solutions @xmath476 since is a fourth order ode , another solution is required . to obtain the fourth solution , we substitute @xmath477 in and obtain that the equation for the leading - order , the @xmath478 terms , is @xmath479 and that the next order is @xmath480 and can be neglected .
the fourth solution is therefore @xmath481
the lagrangian density of the bnls is @xmath482 for example , the bnls action integral is invariant under the phase - multiplication @xmath486 in this case , @xmath487 and so theorem [ thrm : noether ] implies that the integral @xmath488 d{{\mathbf { x } } } \\ & = \int \left [ \frac i2\psi^ * \left ( \psi_t \cdot 0 + \nabla \psi \cdot 0 - i \psi
\right ) - \frac i2\psi \left ( \psi_t^ * \cdot 0 + \nabla \psi ^*\cdot 0 + i \psi^ * \right ) -\mathcal{l}\cdot 0 \right ] d{{\mathbf { x } } } \\ & = { \left\vert{\psi}\right\vert}_2 ^ 2,\end{aligned}\ ] ] i.e. , the power , is a conserved quantity .
other conservation laws can be found in a similar manner .
height 2pt depth -1.6pt width 23pt , _ global well - posedness and scattering for the focusing energy - critical nonlinear schrdinger equations of fourth order in the radial case _ ,
j. of differ .
equations , 246 ( 2009 ) , pp .
3715 3749 . | we consider singular solutions of the biharmonic nls . in the @xmath0-critical case ,
the blowup rate is bounded by a quartic - root power law , the solution approaches a self - similar profile , and a finite amount of @xmath0-norm , which is no less than the critical power , concentrates into the singularity ( `` strong collapse '' ) . in the @xmath0-critical and supercritical cases , we use asymptotic analysis and numerical simulations to characterize singular solutions with a peak - type self - similar collapsing core . in the critical case ,
the blowup rate is slightly faster than a quartic - root , and the self - similar profile is given by the standing - wave ground - state . in the supercritical case , the blowup rate is exactly a quartic - root , and the self - similar profile is a zero - hamiltonian solution of a nonlinear eigenvalue problem .
these findings are verified numerically ( up to focusing levels of @xmath1 ) using an adaptive grid method .
we also calculate the ground states of the standing - wave equations and the critical power for collapse in two and three dimensions . |
signal propagation in the form of waves is a ubiquitous feature of the functioning of neural networks .
waves transmitting electrical activity across neural structures have been observed in a large variety of situations , both in artificially grown cultures and in living brain tissues , see _
e.g _ @xcite and @xcite for an instance of each case ; many other examples can be found in the literature .
this experimental phenomenology has fostered numerous computational and analytical studies on theoretical models for wave propagation . to mention a single category of exact results
, one can cite proofs of existence of waves with context dependent shape : fronts , pulses , periodic wave trains , etc , both in full voltage / conductance models and in firing rate models , see _ e.g. _ @xcite .
in parallel , numerical studies have investigated propagation features such as firing synchrony within cortical layers , and their dependence on dynamical ingredients : feedback , surrounding noise , or external stimulus , see for instance @xcite .
our paper aims to develop a rigorous mathematical investigation of how the global dynamics of a ( simple model of a ) neural network may cause it to organize to a wave behavior , in spite of being forced by a rather unrelated signal .
given that the natural setting of neural ensembles typically features an external environment that is prone to providing an array of irregular stimuli , it seems that such forcing is in no way _ a priori _ tailored toward generating periodic patterns in layered ensembles .
nevertheless , recordings from tissues suggest that self - organization will often ensue despite this obvious and inherent mismatch . in order to get insight into the generation of periodic traveling waves through _ ad hoc _ stimulus , we consider unidirectional chains of coupled oscillators .
inspired by the propagation of synfire waves through cortical layers @xcite , such systems can be regarded as basic phase variable models of feed - forward networks featuring synchronized groups , in which each pool can be treated as a phase oscillator that repetitively alternates a refractory period with a firing burst .
in addition , chains with unidirectional coupling as in equation below are representative of some physiological systems , such as central pattern generators @xcite .
also , acyclic chains of type - i oscillators have been used as simple examples for the analysis of network reliability @xcite .
more specifically , the model under consideration deals with chains of coupled phase oscillators whose dynamics is given by the following coupled odes @xmath2 where @xmath3 and * @xmath4 .
up to a rescaling of time , we can always assume that @xmath5 .
* @xmath6 is the so - called phase response curve ( prc ) . recall that a type - i oscillator is one
for which the prc is a non - negative one - humped function @xcite .
* @xmath7 mimics incoming stimuli from the preceding node and also takes the form of a unimodal function . *
the first oscillator at site @xmath8 evolves according to some forcing signal @xmath9 ( external stimulus ) , _ i.e. _ we have @xmath10 for all @xmath11 .
the forcing is assumed to be continuous , increasing and periodic ( _ viz .
_ there exists @xmath12 such that @xmath13 for all @xmath11 ) .
numerical simulations of the system for some smooth functions @xmath6 and @xmath7 and forcing signals such as the uniform function @xmath14 , have revealed that the asymptotic dynamics settles to a highly - organized regime , independently of initial conditions @xcite . as @xmath15 , the phase @xmath16 approaches the perturbation of a periodic function - the same function @xmath17 ( @xmath18 ) at each site , up to an appropriate time shift @xmath19 - and this perturbation
is attenuated by the chain in going further and further down . in brief terms , traveling waves are typically observed in the far - downstream , large time limit ; see fig .
[ snapshots ] .
mathematically speaking , this means that , letting @xmath20 denote the solution to with forcing @xmath9 ( and say , typical initial condition ) , there exists a periodic function @xmath17 and a time shift @xmath21 such that we have @xmath22 moreover , this phenomenon occurs for arbitrary forcing period , nor to @xmath23 ; see @xcite for an illustration . ] in some range and also appears to be robust to changes in the coupling intensity @xmath24 . , @xmath25 , @xmath26 , @xmath27 , @xmath28 ) of oscillator phases @xmath29 with periodic boundary condition ( @xmath30 ) for a typical trajectory issued from a random initial condition , when forcing with a uniform signal @xmath14 .
clearly , a traveling wave with periodic shape emerges in the far - downstream large time limit ( i.e. , as @xmath31 and @xmath32 ) . ]
this behavior was somehow unanticipated because does not reveal any crucial properties usually required in the proofs of wave existence , such as the monotonicity of the profile dynamics ( analogous property to the maximum principle in parabolic pdes ) , see _ e.g. _
@xcite for lattice differential equations and @xcite for discrete time recursions .
, we show that monotonicity with respect to pointwise ordering on sequences in @xmath33 fails in this system . ] in this context , proving the existence of waves remains unsolved and so is the stability problem , not to mention any justification of the generation phenomenon when forcing with _ ad hoc _ signal .
notice however that , by assuming the existence of waves and their local stability for the single - site dynamics , a proof of stability for the whole chain has been obtained and applied to the design of numerical algorithms for the double - precision construction of wave shapes @xcite .
( our stability proof here is inspired by this one . ) in order to get mathematical insights into wave generation under _ ad hoc _ forcing , here , we analyze simple piecewise affine systems for which the dynamics can be solved explicitly .
this analysis can be viewed as an exploratory step in the endeavor of searching for full proofs in ( more general ) nonlinear systems .
hence , the functions @xmath6 and @xmath7 are both assumed to be piecewise constant on the circle , taking on only the two distinct values of 1 ( on ) or 0 ( off ) . in this setting , our analysis shows that the numerical phenomenology can be mathematically confirmed . for all parameter values
, we prove the existence of tw with arbitrary period in some interval , and their global stability with respect to initial perturbations in the phase space @xmath34 , not only when the forcing at @xmath8 is chosen to be a tw shape but also for an open set of periodic signals with identical period .
in addition , this open set is shown to contain uniform forcing @xmath14 provided that the coupling intensity @xmath24 is sufficiently small .
the paper is organized as follows .
the next section contains the accurate definition of the initial value problem , the basic properties of the associated flow and the statements of the main results .
the rest of the paper is devoted to proofs . in section [ s - exist ] ,
we prove the existence of tw by establishing an explicit expression of their shape .
we study the tw stability with respect to initial conditions in section [ s - stab ] by considering the associated stroboscopic dynamics , firstly for the first site , and then for the second site , from where the stability of the full chains is deduced .
finally , stability with respect to changes in forcing is shown in section [ s - gener ] , as a by - product of the arguments developed in the previous sections .
section [ s - concl ] offers some concluding remarks .
as mentioned before , the dynamical systems under consideration in this paper are special cases of equation in which the stimulus @xmath7 and the prc @xmath6 are ( non - negative and normalized ) square functions , namely , @xmath35 and @xmath36 denotes the floor function ( recall that @xmath37 for all @xmath38 ) . @xmath39 and @xmath40 denote elements in @xmath41 whereas @xmath42 and @xmath43 represent functions of the real positive variable with values in @xmath41 . ]
@xmath44\\ 0\ \text{otherwise } \end{array}\right .
\quad\text{and}\quad \delta(\vartheta)=\left\{\begin{array}{l } 1\ \text{if}\ \vartheta\ \text{mod}\ 1\in [ 0,a_1]\\ 0\ \text{otherwise } \end{array}\right.\quad\forall \vartheta\in\t.\ ] ] ( of note , the stimulus @xmath7 can be made arbitrarily brief by choosing @xmath45 arbitrarily close to 0 . moreover , that the two intervals @xmath46 $ ] and @xmath47 $ ] have the same left boundary is a simplifying assumption that reduces the number of parameters .
other cases are of interest , for instance , oscillators hearing poorly when transmitting , which corresponds to non - overlapping intervals . )
more formally , we shall examine the following system of coupled differential equations for semi - infinite configurations @xmath48 @xmath49 where * the forcing signal @xmath9 is assumed to be a lipschitz - continuous , @xmath50-periodic - valued @xmath9 as @xmath50-periodic provided that @xmath13 for all @xmath11 . ] ( @xmath51 ) , and increasing function with slope ( wherever defined ) at least 1 , is thought of being piecewise affine or even simply affine . ] and satisfying @xmath52 , * @xmath53 is an arbitrary initial configuration , * @xmath54 and @xmath55 are arbitrary parameters .
solutions of equation are in general denoted by @xmath56 but the notation @xmath57 is also employed when the dependence on initial condition needs to be explicitly mentioned .
the solutions of equation have a series of basic properties which we present and succinctly argue in a rather informal way .
these facts can be formally established by explicitly solving the dynamics .
the details are left to the reader . *
existence of the flow .
* given any forcing signal @xmath9 and any initial condition @xmath58 , for every @xmath59 , there exists a unique function @xmath60 which satisfies equation for all @xmath11 .
this function @xmath61 is continuous , increasing , and piecewise affine with alternating slope in @xmath62 .
moreover each piece of slope 1 must have length @xmath63 .
these facts readily follow from solving the dynamics inductively down the chain . assuming that @xmath64 is given for some @xmath59 ( or considering the forcing term @xmath9 if @xmath8 ) , the slope of the first piece of @xmath65 only depends on the relative position of @xmath66 with respect to @xmath67 and of @xmath40 with respect to @xmath68 .
the length of this piece depends on its slope , on @xmath66 and on the smallest @xmath69 such that @xmath70 ; this infimum time has to be positive . by induction
, this process generates the whole function @xmath71 by using the location of @xmath43 and @xmath72 at the end of each piece , and the next time when @xmath73 . * continuous dependence on inputs . *
endow @xmath74 with pointwise topology and , given @xmath75 , endow continuous and monotonic functions of @xmath76 $ ] with uniform topology and norm @xmath77 .
for every @xmath59 and @xmath75 , the quantity @xmath78}\|$ ] continuously depends both on the forcing signal @xmath79}$ ] and on the initial condition @xmath58 .
indeed , if two forcing signals @xmath9 and @xmath80 are close , then the lower bound on their derivatives implies that the respective times in @xmath81 $ ] at which they reach @xmath82 must be close . if , in addition , the initial conditions @xmath83 and @xmath84 are close , then the trajectories @xmath85 and @xmath86 alternate their slopes at close times ; hence @xmath87}-\theta_1^{(g)}(\xi_1,\cdot)|_{[0,t]}\|$ ] must be small . since the slopes are at least 1 , the respective times at which @xmath88}$ ] and @xmath89}$ ] reach @xmath82 are close . by repeating the argument , we conclude that @xmath90}-\theta_2^{(g)}(\xi_1,\xi_2,\cdot)|_{[0,t]}\|$ ] must be small when @xmath91 and @xmath92 are sufficiently close
. then , the result for an arbitrary @xmath59 follows by induction .
* semi - group property . * as suggested above , a large part of the analysis consists in focusing on the one - dimensional dynamics of the first oscillator @xmath93 forced by the stimulus @xmath9 , prior to extending the results to subsequent sites .
indeed , the dynamics of the oscillator @xmath31 can be regarded as a forced system with input signal @xmath72 .
for the one - dimensional forced system , letting @xmath94 for all @xmath11 denotes the time translations , we shall especially rely on the semi - group property of the flow , _ viz .
_ if @xmath95 is a solution with initial condition @xmath83 and forcing @xmath9 , then , for every @xmath96 , @xmath97 is a solution with initial condition @xmath98 and forcing @xmath99
. * monotonicity failure . * consider the following partial order on sequences in @xmath100 ( employed in typical proofs of existence of tw in lattice dynamical systems ) .
we say that @xmath101 iff @xmath102 for all @xmath31 .
clearly , we may have @xmath103 for some @xmath11 and yet @xmath104 for another @xmath105 .
for instance , it suffices to choose @xmath106 and @xmath107 with @xmath91 and @xmath92 sufficiently close .
see figure [ monotonicity_fail ] below . and
@xmath108 at site @xmath109 , and @xmath110 and @xmath111 at site @xmath112 . ]
a solution @xmath113 of the system is called a * traveling wave * ( tw ) if there exists @xmath21 such that @xmath114 in other words , a tw is a solution for which the forcing signal @xmath9 exactly repeats at every site , modulo an appropriate time shift : the phase at any given site @xmath31 and time @xmath115 mimics the phase at the previous site and time @xmath116 ( see fig . [ tw_rep - and - converge ] , left panel ) . for such solutions
, the forcing signal also plays the role of the wave shape and the quantity @xmath117 represents the wave number . in our setting , any tw shape / forcing signal obviously has to be a piecewise affine function with slopes only taking the values @xmath25 and @xmath118 . for @xmath119 and @xmath120 vs. time @xmath115 , together with the forcing signal @xmath9 ( which satisfies condition ( c ) in thm [ mainres1 ] , so as to ensure that the wave is stable see text ) .
* subsequent panels : * example of a trajectory @xmath121 for @xmath122 and disjoint consecutive time windows .
the initial phases have been chosen randomly and the forcing signal @xmath9 is the same as in the left panel .
clearly , the convergence @xmath123 occurs at the first site s=1 and then propagates down the chain , as described in the text ; the higher the value of @xmath31 is , the longer the transient time @xmath121 needs in order to get close to @xmath124 . ] as we shall see below , tw exist that are asymptotically stable , not only with respect to perturbations of the initial phases @xmath125 , but more importantly , also with respect to changes in the forcing signal . for convenience ,
we first separately state existence and uniqueness .
( existence . ) for every @xmath126 and @xmath127 , there exists a ( non - empty ) interval @xmath128 and for every @xmath129 , there exist a @xmath50-periodic forcing signal @xmath9 and @xmath130 such that @xmath131 is a tw .
( uniqueness . ) for every @xmath129 , the forcing signal @xmath9 and the shift @xmath19 as above are unique , provided that the following constraint is required \(c ) @xmath9 is a piecewise affine forcing signal whose restriction @xmath132 has slope @xmath118 only on a sub - interval whose left boundary is @xmath19 .
[ mainres1 ] for the proof , see section [ s - exist ] , in particular corollary [ existw ] . notice that constraint ( c ) has no intrinsic interest other than unambiguously identifying appropriate tw shapes for the stability statement . to identify stable waves matters because the system also possesses neutral and unstable traveling waves . for stability , we shall use notions that are appropriate to forced systems , and adapted to our setting .
in particular , since the information flow is unidirectional here , it is natural to only require that perturbations relax in pointwise topology , rather than in uniform topology .
therefore , we shall consider the dynamics on arbitrary finite collections of sites which , without loss of generality , can be chosen to be the first @xmath31 sites , for an arbitrary @xmath31 .
moreover , there is no need for local stability considerations here because we shall be concerned with tw for which the basin of attraction is as large as it can get from a topological viewpoint .
accordingly , we shall say that a tw @xmath133 is * globally asymptotically stable * if there exists a sequence @xmath134 such that , for every @xmath59 and every initial condition for which @xmath135 for all @xmath136 , we have @xmath137 in other words , a solution is globally asymptotically stable if its basin of attraction is as large as it can get from a topological viewpoint .
again , the exceptional initial conditions @xmath138 must exist for fixed - point index reasons . as can be expected , our next statement claims global asymptotic stability of waves , provided they are suitably chosen according to the above criterion
. there exists a ( non - empty ) sub - interval @xmath139 such that for every @xmath140 , the tw @xmath133 determined by constraint ( c ) is globally asymptotically stable .
[ mainres15 ] see fig .
[ tw_rep - and - converge ] for an illustration of this result .
theorem [ mainres15 ] is proved in section [ s - stab ] , see especially the concluding statement corollary [ stabtw ] .
in addition , for initial conditions not satisfying the stability condition , our proof shows that the first coordinate @xmath31 for which this condition fails asymptotically approaches an unstable periodic solution . in theorem [ mainres15 ] , we claim stability of the tw @xmath133 with respect to perturbations of initial conditions when forcing with @xmath9 .
we now consider the analogous property when the forcing signal is also perturbed . as described in the introduction , in general , for any given site @xmath31 , one may not expect relaxation to @xmath124 exactly but rather to a perturbation of this signal , which is itself attenuated as @xmath141 .
accordingly , we shall say that a tw @xmath133 is * robust with respect to perturbations of the forcing * if , for every forcing signal @xmath80 in a @xmath142-neighborhood of @xmath9 , there exists a neighborhood @xmath143 ( product topology ) of @xmath144 such that , for every @xmath145 , we have @xmath146 as indicated in the introduction .
robustness with respect to forcing perturbations is expected to hold in general smooth systems of the form ( as is the global stability of tw ) . in the current piecewise affine
setting , the phase @xmath147 actually relaxes to @xmath124 at every site @xmath31 ( a stronger result ) , as described in the following statement . for every @xmath140
, there exists a @xmath142-neighborhood @xmath148 of @xmath9 such that , for every @xmath149 , the tw @xmath133 determined by constraint ( c ) globally attracts solutions of the system with forcing @xmath80 .
that is to say , there exists a sequence @xmath150 such that for every @xmath59 and every initial condition for which @xmath151 for all @xmath136 , we have @xmath152 in addition , for @xmath24 small enough ( depending on @xmath153 ) , there exists a ( non - empty ) sub - interval @xmath154 such that , for every @xmath155 , the neighborhood @xmath148 contains the uniform forcing @xmath156 , for all @xmath11 .
[ mainres2 ] theorem [ mainres2 ] is established in section [ s - gener ] and implies in particular robustness with respect to forcing perturbations . for every @xmath140 ,
the tw determined by the constraint ( c ) is robust with respect to perturbations of the forcing .
an alternative characterization of tw solutions can be given as follows @xmath157 together with the semi - group property of the first oscillator dynamics , in order to prove the existence of tw , it suffices to find a forcing signal @xmath9 and a phase shift @xmath19 such that @xmath158 for all @xmath11 . using the translation operator @xmath159 , this is equivalent to solving the following delay - differential equation @xmath160 for the pair @xmath133 .
the purpose of the section is precisely to solve this equation . to that goal
, it is useful to begin by identifying all possible cases .
since we assume @xmath52 , the forcing signal / tw shape @xmath9 can be entirely characterized by the partition of @xmath161 $ ] into intervals where the slope is constant . without loss of generality
, we can also assume that @xmath162 .
implies @xmath163 and we must have @xmath164 . ] under these assumptions , there can only be four cases depending on the initial location @xmath165 of @xmath93 ( above or below @xmath127 ) and the location @xmath166 of @xmath167 ( above or below @xmath45 ) when @xmath93 reaches 1 . by examining each case , one easily constructs the desired partition .
a tw shape @xmath9 must fall into one of the following cases ( see figure [ exampl ] ) : * @xmath168 and @xmath169
. then there exists @xmath170 such that we have for the tw coordinate at site @xmath109 @xmath171 * @xmath172 and @xmath173
. then there exists @xmath174 such that we have @xmath175 * @xmath172 and @xmath169 .
then @xmath176 for all @xmath177 $ ] .
* @xmath168 and @xmath173 .
then there exist @xmath178 and @xmath179 such that we have @xmath180 [ specify ] for @xmath177 $ ] in each tw case of lemma [ specify ] . ] _ proof . _
the four cases are clearly mutually exclusive .
that they fully characterize the shape over the period ( and the explicit expressions as claimed ) is a consequence of the following observations . *
the assumption @xmath168 implies that @xmath93 immediately increases with speed @xmath118 when @xmath115 crosses 0 , and until either it reaches @xmath127 or @xmath167 reaches @xmath45 . then @xmath93 must increase with speed 1 until ( at least ) time @xmath181 when it reaches 1 .
the inequality @xmath169 implies that @xmath167 is above @xmath45 for all @xmath182 ; hence @xmath93 continues to grow at rate 1 until @xmath93 reaches 1 . * assuming a first phase as before , if otherwise @xmath183 , then @xmath93 increases again with speed @xmath118 after it has crossed 1 , and until either @xmath167 reaches @xmath45 or @xmath93 reaches @xmath184 .
the latter case is impossible because it would imply that @xmath93 increases by more than 1 over the fundamental time interval .
this uniquely determines case ( c ) and we have @xmath185 and @xmath186 . *
if @xmath172 , then @xmath93 can not have speed 1 before it reaches 1 .
since we assume that @xmath167 is already larger than @xmath45 when this happens , it follows that @xmath93 can never have speed @xmath118 . * assuming a first phase as in ( c ) , if @xmath183 , then @xmath93 accelerates after time @xmath181 and until either @xmath167 reaches @xmath45 or @xmath93 reaches @xmath184 .
after that , @xmath93 has speed 1 until either @xmath167 reaches 1 or @xmath93 reaches 2 .
again the latter case is impossible because total increase over one period is at most 1.@xmath187 for our purpose here , it is enough to focus on case ( a ) .
( indeed , lemma [ stabilitylem ] below shows that these are the only possibly asymptotically stable waves . ) in this case , lemma [ specify ] indicates that the tw shape is completely determined by the numbers @xmath188 and @xmath50 .
our next statement claims that @xmath189 and @xmath50 are actually entirely determined by the phase shift @xmath19 , and therefore , so is the tw shape . for every choice of parameters @xmath190 and every phase shift @xmath21
, there exists at most one tw solution in case ( a ) .
[ prouniqcas ] one can actually prove that a similar statement holds in cases ( b ) , ( c ) and ( c ) .
more importantly , this statement paves the way to uniqueness as stated in theorem [ mainres1 ] .
_ proof of the lemma .
_ together with continuity , lemma [ specify ] implies that the solution @xmath191 , on the fundamental interval @xmath161 $ ] , is given as follows is a consequence of the fact that @xmath169 - see figures in cases ( a ) and ( c ) . ] @xmath192 by translation , this determines the shape @xmath193 over the interval @xmath194 $ ] , _
i.e._@xmath195 we must have @xmath196 ( _ i.e. _ @xmath197 ) because @xmath9 has to have speed @xmath25 when it reaches 1 ( see figure [ exampl ] ) .
hence @xmath198 must growth with speed 1 for @xmath199 and hence for @xmath200 by periodicity .
consequently , on the interval @xmath161 $ ] , the tw shape writes @xmath201 where the 2-parameter family @xmath202 is defined by @xmath203 notice that @xmath204 for every @xmath205 .
moreover , we must have @xmath206 which yields @xmath207 . therefore , in order to prove the lemma , it suffices to prove uniqueness of @xmath189 given a phase shift @xmath19 .
this is the purpose of the next statement . for every choice of parameters @xmath190 and every phase shift @xmath21 , the equation @xmath208 ( resp .
@xmath209 ) has a unique positive solution denoted @xmath210 ( resp .
@xmath211 ) .
moreover , we have @xmath212 .
notice that the quantity @xmath189 in this statement satisfies the inequalities @xmath213 .
_ proof of the claim .
_ assuming @xmath214 , the equation @xmath208 has unique solution @xmath215 assuming @xmath216 , the equation @xmath209 gives @xmath217 .
now , @xmath189 is the first time in @xmath161 $ ] when @xmath93 adopts speed @xmath25 . as argued in the proof of lemma [ specify ]
, this time corresponds to the smallest of times when @xmath93 reaches @xmath127 or @xmath167 reaches @xmath45 ; _ viz .
_ @xmath212 and the claim is proved .
@xmath187 to conclude about existence of waves , it remains to investigate the conditions on @xmath19 such that a tw shape @xmath218 satisfies the conditions of lemma [ prouniqcas ] .
the result is given in the next statement .
there exists a tw solution in case ( a ) iff @xmath219 [ exist ] the constraints here define a non - empty interval of @xmath19 for every possible choice of parameters @xmath190 . by lemma [ prouniqcas ] ,
let @xmath128 be the corresponding interval of forcing periods @xmath50 .
we have proved the following statement .
equation has a unique traveling wave solution in case ( a ) , for every forcing period in @xmath128 .
[ existw ] _ proof of lemma [ exist ] .
_ we need to check the conditions @xmath220 and @xmath169 of lemma [ specify ] for a shape @xmath218 as in the proof of lemma [ prouniqcas ] .
first , we have @xmath221 .
so the condition @xmath220 is equivalent to @xmath222 . in order to check the inequality @xmath169
, we need to take into account the quantities @xmath210 and @xmath211 defined in the proof of lemma [ prouniqcas ] .
notice that the condition @xmath223 is equivalent to @xmath224 we consider the cases @xmath225 and @xmath226 separately . in the first case ,
the inequality @xmath169 is equivalent to @xmath227 .
however , we know from the proof of lemma [ prouniqcas ] that we must have @xmath228 ; hence there is nothing to prove . in the second case @xmath226
, we first observe that @xmath229 for all @xmath230 and in particular for @xmath231 .
therefore , and using also @xmath232 from the proof of lemma [ prouniqcas ] , the inequality @xmath169 is equivalent to @xmath233 , that is to say @xmath234 altogether , we conclude that the inequality @xmath169 is equivalent to @xmath235 . elementary algebra , together with the assumption @xmath236 , show that @xmath237 ; hence the desired condition follows .
this section reports the stability analysis of tw and aims to establish the first statement in theorem [ mainres2 ] .
we first focus on the stability analysis , first local and then global , of fixed points of the * stroboscopic map * @xmath238 that updates and translates back the first coordinate @xmath93 after a forcing period , namely @xmath239 by periodicity of @xmath9 and of the vector field in equation , we have @xmath240 hence , the orbits of @xmath241 indeed capture the asymptotic behavior of @xmath93 .
equation implies that @xmath241 is a lift of an endomorphism of the circle that preserves orientation . for all @xmath242 , from equation .
moreover , the continuous dependence of @xmath243 on @xmath83 implies that @xmath241 must be continuous . as for monotonicity , by contradiction ,
if we had @xmath244 and @xmath245 , then the corresponding trajectories @xmath85 and @xmath246 would have crossed for some @xmath247 .
this is impossible by uniqueness of the vector field acting on this coordinate . ]
one can show that @xmath241 is actually strictly increasing ; hence it is a homeomorphism from @xmath41 into itself .
the theory of circle maps ( see e.g. @xcite ) states the existence of a rotation number for every orbit , namely the following limit exists @xmath248 and does not depend on @xmath83 . in principle
, this number can take any value in @xmath249 .
however , a tw solution exists iff the corresponding stroboscopic map has a fixed point , that is to say iff @xmath250 in particular , the stroboscopic map associated with a tw must have vanishing rotation number . in this section ,
we study the local stability of the fixed point @xmath165 of the stroboscopic map @xmath241 associated with a tw shape .
a first statement respectively identifies the stable , unstable and neutral cases according to the decomposition in lemma [ specify ] . following the decomposition in lemma [ specify ] , we have * the fixed point @xmath165 is locally asymptotically stable if in case ( a ) with @xmath251 when @xmath252 . *
it is unstable if in case ( b ) with @xmath253 when @xmath254 , * and is neutral in any other case .
[ stabilitylem ] _ proof .
_ we want to evaluate the behavior of the difference @xmath255 , where @xmath256 is a small perturbation of the initial condition @xmath257 . _ stable case .
_ by continuous dependence of @xmath258 on @xmath83 , the time @xmath259 when @xmath260 can be made arbitrarily close to @xmath189 by choosing @xmath84 sufficiently close to @xmath165
. therefore , there exists @xmath261 such that for every @xmath262 , there exists @xmath263 such that @xmath264 when @xmath260 .
accordingly , the dynamics of the two trajectories in the interval @xmath161 $ ] can be summarized as follows : * both @xmath258 and @xmath265 have speed @xmath118 for @xmath266 . * @xmath258 has speed 1 and @xmath265 has speed @xmath118 for @xmath267 . *
both @xmath258 and @xmath265 have speed @xmath25 for @xmath268 .
it results that @xmath269 has speed @xmath118 on a longer time interval and thus @xmath270 we also have @xmath271 by strict monotonicity ; hence @xmath272 and the argument can be repeated with @xmath273 to obtain @xmath274 by induction , it follows that the sequence @xmath275 is decreasing and non - negative ; hence it converges .
a standard contradiction argument based on the contraction and on the continuity of @xmath241 concludes that the limit must be 0 .
this proves local asymptotic stability with respect to negative initial perturbations .
a similar argument applies to positive perturbations
. _ unstable case .
_ similarly to as before , let @xmath261 be sufficiently small so that for every @xmath262 , we have @xmath276 . comparing again the two trajectories , we have : * both @xmath258 and @xmath265 have speed @xmath25 for @xmath277 .
( nb : @xmath278 ) .
* @xmath258 has speed @xmath118 and @xmath265 has speed @xmath25 for @xmath279 .
* both @xmath258 and @xmath265 have speed @xmath118 for @xmath280 . *
both @xmath258 and @xmath265 have speed @xmath25 for @xmath281 . in this case , @xmath269 has speed @xmath118 on a shorter interval and thus @xmath282 for every @xmath262 , _ viz . _
the tw is unstable with respect to negative perturbations .
a similar argument applies to positive perturbations .
_ neutral case . _
the analysis is similar .
one shows that the total duration when the perturbed trajectory has speed @xmath118 is identical to that of the tw ; hence @xmath283 .
@xmath187 in the proof of lemma [ exist ] above , we have identified the condition @xmath226 ( see equation ) as sufficient to ensure being in case ( a ) with @xmath251 when @xmath252 . by cross - checking this constraint with the one in the statement of that lemma ( and using also lemma [ prouniqcas ] , _
i.e. _ that the tw is entirely determined by the parameters @xmath190 and its phase shift @xmath19 ) , we get the following conclusion . for every choice of parameters @xmath190
, there exists an interval @xmath139 such that for every @xmath140 , there exists a @xmath50-periodic tw shape @xmath9 and a phase shift @xmath130 such that @xmath165 is a locally asymptotically stable fixed point of @xmath241 .
[ globexists ] _ proof .
_ in order to be able to choose @xmath19 that simultaneously satisfies @xmath226 and the conditions of lemma [ exist ] , it suffices to prove that the inequality @xmath284 holds for all parameter values . to that goal
, we consider separately different parameter regimes . if @xmath285 , then the left hand side of the inequality is equal 0 , while the right hand side always remains non - negative ; hence the inequality holds . in order to investigate the case @xmath286
, we notice that direct calculations imply the following conclusions * the inequality @xmath287 is equivalent to @xmath288 , * the condition @xmath289 simplifies to @xmath290 and we obviously have @xmath291 when @xmath286 .
furthermore , we clearly have @xmath292 ; hence the statement also holds in the case where @xmath293 . from now on
, we can assume @xmath294 .
we have @xmath295 accordingly , one has to consider three cases * if @xmath288 then @xmath296 and the statement holds provided that @xmath297 , which is true . * if @xmath298 then @xmath299 and the inequality to check is @xmath300 which holds true .
* finally , if @xmath299 and @xmath296 , then the inequality to verify is @xmath301 which is equivalent to @xmath302
. however , the inequality @xmath296 implies @xmath303 and we have @xmath304 because of @xmath299 .
the proof of the corollary is complete .
@xmath187 once local stability has been established , a careful computation of @xmath241 allows one to show that the fixed points are actually globally stable .
when @xmath140 , the fixed point @xmath165 is globally stable .
that is to say , there exists a unique unstable fixed point @xmath305 such that we have @xmath306 [ globstab ] to be more accurate , the proof below actually shows that for every @xmath307 , there exists @xmath308 such that @xmath309 _ proof .
_ we are going to prove that , for the tw in case ( a ) with @xmath19 simultaneously satisfying @xmath226 and the conditions of lemma [ exist ] , the restriction of @xmath241 to the interval @xmath310 consists of four affine pieces ( each piece being defined on an interval ) : two pieces are rigid rotations and they are interspersed by one contracting and one expanding piece .
since @xmath241 is a lift of an endomorphism of the circle that preserves the orientation , and since it has a locally stable fixed point @xmath165 ( corollary [ globexists ] ) , the graph of the contracting piece must intersect the main diagonal of @xmath311 . as a consequence
, the graphs of the following and preceding neutral pieces can not intersect this line . by continuity and periodicity , the graph of the remaining expanding piece must intersect this line as well .
let @xmath312 be this unique unstable fixed point .
the proposition immediately follows . in order to prove the decomposition into the four desired pieces
, we are going to consider various cases . to that goal , recall first that @xmath52 and @xmath313 , _ i.e. _ @xmath191 can only have speed @xmath118 for @xmath314 ( within the interval @xmath315 ) . consider the trajectory of the coordinate @xmath93 with initial value @xmath316 .
since we have @xmath317 , the largest integer @xmath93 reaches before time @xmath210 is at most 1 , _
viz._@xmath318 we consider separately two cases : * either the last rapid phase with speed @xmath118 ( when @xmath319 ) stops when @xmath320 .
this occurs iff @xmath321 * or its stops when @xmath93 reaches @xmath322 . , _
i.e. _ we may have @xmath323 . ]
this occurs when @xmath324 see figure [ f - four_intervals ] for an illustration of the action of @xmath241 according to these two cases , when @xmath325 . throughout the proof
, we shall frequently make use of the time @xmath326 when the coordinate @xmath93 reaches the value @xmath327 , _
i.e. _ @xmath328 . here
@xmath329 are arbitrary .
_ case ( a ) .
_ by continuous and monotonic dependence on initial conditions , every trajectory starting initially with @xmath83 sufficiently close to 0 ( and @xmath330 ) , will not only experience the same number @xmath331 of rapid phases with speed @xmath118 , but the last rapid phase ( the unique rapid phase if @xmath325 ) will also stop at @xmath320 .
assume @xmath332 .
as time evolves between 0 and @xmath210 , speed changes for such trajectories occur when the level lines @xmath127 and @xmath25 are reached ( _ i.e. _ when @xmath333 and @xmath334 ) . consequently , the delays between the corresponding instants for two distinct trajectories remain constants _ i.e. _ we have @xmath335 this equality implies that the cumulated lengths of the rapid phases satisfy @xmath336 _
i.e. _ they are independent of @xmath83 .
we conclude that @xmath337 ; _ viz .
_ the map @xmath241 is a rigid rotation in some right neighborhood of 0 .
if @xmath325 , the same conclusion immediately follows from the fact that lengths of the rapid phases are simply given by @xmath210 .
moreover , the largest initial condition @xmath338 for which this property holds is defined by @xmath339 and we have @xmath340 .
therefore , @xmath241 is a rigid rotation on @xmath341 . to continue , we separate case ( a ) into 2 subcases : * either @xmath342 . * or @xmath343 assume first that case ( a1 ) holds . in the case
@xmath332 , using similar considerations as above , for trajectories now starting in @xmath344 $ ] , we obtain @xmath345 and @xmath346 and @xmath347 ( and monotonicity together with @xmath340 implies the existence of @xmath348 ) . for the cumulated lengths of rapid phases , the last equality implies @xmath349 and using that for the initial rapid phase ( which is the only rapid phase if @xmath325 )
, we have @xmath350 we obtain that @xmath351 from where it follows using that @xmath241 is piecewise affine , that the restriction of @xmath241 to the interval @xmath352 must be a contraction .
let @xmath353 be such that @xmath354 .
the dichotomy ( a1 ) _ vs. _ ( a2 ) is equivalent to @xmath355 _ vs.
_ @xmath356 . in the latter case , the same conclusion as before applies to the interval @xmath357 .
now , for @xmath358 when in case ( a1 ) and @xmath332 , the equality between change speed instants writes @xmath359 and similar calculations to those for @xmath360 show that @xmath241 is also a rigid rotation on @xmath361 . if @xmath325 , that @xmath241 is a rigid rotation on @xmath361 immediately follows from the fact that there is no rapid phase at all . in case ( a2 ) , the same conclusion applies to @xmath362 .
finally , for the interval @xmath363 in case ( a1 ) , one shows that the cumulated rapid phases for the trajectories issued from @xmath353 and @xmath25 are identical up to @xmath364 .
moreover , on the interval @xmath365 , the function @xmath366 has speed 1 and @xmath367 has speed @xmath118 ( because @xmath368 by degree 1 ) . is the fourth and last interval to consider in this analysis .
there are no more acceleration pattern and @xmath241 indeed decomposes over exactly four intervals .
] it follows that @xmath369 which implies @xmath241 must be an expansion on this interval . in case ( a2 )
, the same conclusion easily follows for the interval @xmath370 .
, as discussed in the two major cases of proposition [ globstab ] ( for @xmath325 and @xmath371 ) .
the dynamics of critical initial conditions are shown in light solid black .
the blue - colored intervals denote the subset of @xmath41 on which @xmath241 is a translation , whereas green and red denote contraction and expansion , respectively . ] _ case ( b ) .
_ the arguments are similar . for @xmath372 in the neighborhood of 0 ,
the length of the last rapid phase is now independent of @xmath83 ( @xmath332 ) and the length of the first rapid phase decreases when @xmath83 increases . hence @xmath241 is contraction on this first interval .
its upper boundary is given by @xmath373 , where as before @xmath353 is defined by @xmath374 .
for @xmath358 ( or @xmath375 depending on the case ) , the cumulated duration of the accelerated phases does not depend on @xmath83 ( in particular because the last rapid phase stops when the coordinate reaches @xmath322 ) . in the case
@xmath371 , for @xmath376 the trajectory has an extra final rapid phase when compared to @xmath353 and this holds for all @xmath377 where @xmath378 is defined by @xmath379 .
is granted from the fact that @xmath380 which follows from @xmath381 . ] finally , for @xmath382 , @xmath241 is a rigid rotation .
the case @xmath383 can be treated similarly .
@xmath187 in order to prove theorem [ mainres15 ] , we first consider the stroboscopic dynamics at the second site and then extend the results down the chain .
thanks to the structure of equation , the second site coordinate after one forcing period can be expressed in terms of the time-@xmath50 stroboscopic map @xmath384 as follows @xmath385 by induction , we have ( recall that @xmath386 represents the time translation by an amount @xmath50 , _ i.e. _ @xmath387 ) @xmath388 the semi - group property of the flow implies that @xmath389 is a globally stable fixed point of the map @xmath390 .
using similar arguments as in the stability proof in @xcite , we use this property here to show that this stable point attracts the iterates @xmath391 .
let @xmath140 , @xmath42 be the associated case ( a ) tw shape and @xmath392 be given by proposition [ globstab ] .
for every @xmath393 , there exists a unique @xmath394 such that we have @xmath395 [ stab2site ] the same comment as the one after proposition [ globstab ] applies here .
namely , the proof actually establishes that that for every @xmath396 , there exists @xmath397 such that @xmath398 _ proof .
_ let @xmath307 be fixed .
the proof relies on the following property of the stroboscopic map .
* we have @xmath399 where the uniform norm @xmath77 is extended to functions on @xmath41 . in order to prove this
fact , notice that proposition [ globstab ] states that @xmath400 converges to @xmath401 .
when regarding these points as initial conditions in @xmath41 for the one - dimensional system submitted to stimulus @xmath9 , the continuous dependence of solutions on initial conditions implies @xmath402}\|=\|r^\alpha f|_{[0,\tau]}\|.\ ] ] the convergence @xmath403 for every @xmath91 then follows from the continuous dependence of solutions on the forcing signal . finally , that the convergence is uniform in @xmath41 is a consequence of the continuity and monotonicity of the maps @xmath404 and @xmath390 .
now , the first step of the proof consists in showing that every sequence @xmath405 either converges to a point @xmath406 or to one point in @xmath407 where @xmath408 is defined by @xmath409 to that goal , we are going to prove that if a sequence does not converge to @xmath407 , _
i.e. _ if there exists @xmath261 and a diverging subsequence @xmath410 so that the following estimate for distances in @xmath41 holds @xmath411 then we have @xmath412 together with periodicity , the inequality implies the existence @xmath413 such that @xmath414 the behavior of the map @xmath390 in the interval @xmath415 around the stable fixed point @xmath389 implies the existence of @xmath416 such that @xmath417 by the property ( p ) above , let @xmath418 sufficiently large so that @xmath419 let @xmath420 be sufficiently large so that @xmath421 for all @xmath422 . using the characterization of @xmath423 above , let also @xmath397 be such that @xmath424 then by decomposing the next iterate as follows @xmath425 and using and and periodicity , we obtain @xmath426 and then by induction @xmath427 this proves that , under the assumption , every accumulation point of the sequence @xmath405 must lie in @xmath428 $ ]
. however , by continuity of @xmath390 and the property ( p ) above , the set @xmath429 of these accumulation points must be invariant under @xmath390 , _
viz._@xmath430 since the only invariant set of @xmath390 that intersects @xmath428 $ ] is the fixed point @xmath389 , we conclude @xmath431 as desired .
to conclude the proof , it remains to prove the existence of a unique @xmath432 such that @xmath433 according to the first part of the proof , it suffices to prove the existence of a unique @xmath434 such that the sequence @xmath435 eventually enters and remains inside the interval @xmath436 where @xmath390 is expanding ( because , henceforth , the sequence must approach @xmath437 ) .
let @xmath438 be the derivative of @xmath390 on @xmath436 .
we claim ( and prove below ) the existence of a closed subinterval @xmath439 , of a number @xmath440 and of a sufficiently large @xmath441 such that @xmath442 more precisely , @xmath443 is not unique and its boundaries can be chosen arbitrarily close to those of @xmath436 ( and similarly @xmath444 can be chosen arbitrarily close to @xmath445 ) .
accordingly , the threshold @xmath446 depends on @xmath443 ( and @xmath444 ) and diverges as @xmath443 approaches @xmath436 .
now , given @xmath447 , consider the intersection set @xmath448 thanks to the property , these sets form a family of nested closed non - empty intervals whose diameter tend to 0 as @xmath449 . by the nested ball theorem
, we conclude the existence of a unique point @xmath450 which depends on the whole solution @xmath93 at site 1 , such that @xmath451 furthermore , as a composition of homeomorphisms , the map @xmath452 is itself a homeomorphism of @xmath41 .
hence there exists a unique @xmath394 such that @xmath453 .
clearly , we have @xmath454 as desired . to complete the proof , it remains to establish the property .
notice first that , as @xmath455 gets large , the times at which the forcing signals @xmath456 and @xmath457 respectively cross the levels 0 , @xmath45 and 1 come close together . as a consequence , the restriction of @xmath404 to the interval @xmath436 , since close to @xmath458 by the property ( p ) , must consist of an expanding piece , possibly preceded and/or followed by a rigid rotation . moreover , the length of these ( putative ) rigid rotation pieces must be bounded above by a number that depends only on @xmath445 and on @xmath459 , and which vanishes as @xmath449 .
( indeed , if otherwise , the length(s ) of the rigid rotation piece(s ) remained bounded below by a positive number , we would have a contradiction with the property ( p ) above , because the distance between @xmath460 and @xmath461 would remain bounded below by a positive number for @xmath91 close to the boundaries of @xmath436 . )
this proves the existence of @xmath439 on which all @xmath462 for @xmath455 sufficiently large , must be expanding .
in addition , by taking @xmath455 even larger if necessary ( so that @xmath459 is even smaller ) , we can make sure that the expanding piece of @xmath404 over @xmath443 intersects the diagonal , since the limit map @xmath390 does so .
the first claim of property then easily follows .
the second claim that the expanding slope must be bounded below by @xmath444 can proved using a similar contradiction argument as above .
the proof of the proposition is complete .
@xmath187 in order to extend the stability results down the chain , proceeding similarly as when introducing @xmath463 before proposition [ stab2site ] , we consider the maps @xmath464 ( @xmath465 ) that specify the coordinates @xmath466 . in particular , for @xmath112 , a similar reasoning as the one in the proof of the property ( p ) above shows that the conclusion of proposition [ stab2site ] implies @xmath467 as soon as @xmath393 and @xmath396 .
moreover , by repeating _
mutatis mutandis _ the proof of proposition [ stab2site ] , we conclude that , under the same conditions , there exists a unique @xmath468 such that @xmath469 by induction , one obtains the following statement , from which theorem [ mainres15 ] easily follows .
given an arbitrary @xmath59 , assume that the initial phases @xmath470 have been chosen so that the solution behaves as follows @xmath471 then there exists a unique @xmath472 so that we have @xmath473 [ stabtw ]
recall that for @xmath140 we have @xmath251 for the associated tw where @xmath474 is such that @xmath475 .
hence , for every continuous increasing and @xmath50-periodic forcing @xmath80 such that @xmath476 and @xmath477 , the stroboscopic map in the neighborhood of @xmath165 remains unchanged , _ viz .
_ @xmath478 by repeating _
mutatis mutandis _ the reasoning in the proof of proposition [ globstab ] , one shows that the restriction of @xmath479 to @xmath310 also consists of four pieces ; two rigid rotations interspersed by a contracting and an expanding piece .
therefore , there must exist @xmath480 such that @xmath481 in other words , the asymptotic dynamics of @xmath93 is the same as when forcing with @xmath9 .
the conclusion for the rest of the chain @xmath482 immediately follows .
finally , in order to make sure that the conclusion applies to the uniform forcing @xmath156 , it suffices to show that @xmath483 . using the expression @xmath484 and @xmath474
, we obtain after simple algebra @xmath485 it is easy to see that this condition defines an interval of @xmath19 for all parameter values @xmath486 . moreover , and depending whether @xmath127 is smaller than @xmath487 or not , for @xmath488 , this interval either contains or coincides with the interval defined in the proof of corollary [ globexists ] . by continuity , for @xmath54
sufficiently small , there exists an interval of @xmath19 - that corresponds to periods in a subinterval @xmath154 - in which both the condition and the ones in the proof of corollary [ globexists ] hold .
the statement easily follows .
for simple feed forward chains of type - i oscillators , our analysis proved that periodic wave trains can be generated from arbitrary initial condition , even when the root node is forced using an unrelated signal .
moreover , these stable waves exist for an open ( parameter - dependent ) interval of wave number and period .
the existence of globally attracting waves for arbitrary wave number in some range is reminiscent of the inertia - free dynamics of tilted frenkel - kontorova chains , which constitute coupled oscillator models for spatially modulated structures in solid - state physics @xcite .
there are however essential differences between the two situations . instead of a uni - directional interaction ,
the coupling is of bi - directional type in frenkel - kontorova chains and involves left and right neighbors .
more importantly , the overall dynamics there is monotonic and , as mentioned in the introduction , this property is critical for the proof of existence and stability of waves @xcite .
finally , we notice that the results on asymptotic stability and on stability with respect to changes in forcing are based on hyperbolic properties of the stroboscopic dynamics .
accordingly , we believe that , using continuation methods , these results , and more generally , results on generation of traveling waves in unidirectional chains of type - i oscillators can be established in a rigorous mathematical way , in more general models with smooth prc and stimulus nonlinearities . this will be the subject of future studies .
the authors declare that they have no competing interests .
bf and sm both designed the research , proceeded to the analysis , and wrote the paper .
the work of bf was supported by eu marie curie fellowship piof - ga-2009 - 235741 and by cnrs peps _ physique thorique et ses interfaces_. r.o .
dror , c.c .
canavier , r.j .
butera , j.w .
clark and j.h .
byrne , _ a mathematical criterion based on phase response curves for stability in a ring of coupled oscillators _
cybernetics * 80 * ( 1999 ) 11 - 23 .
d. kleinfeld , k.r .
delaney , m.s .
fee , j.a .
flores , d.w .
tank and a. gelperin , _ dynamics of propagating waves in the olfactory network of a terrestrial mollusk : an electrical and optical study _ , j. neurophysiol . * 72 * ( 1994 ) 1402 - 1419 . | we investigate the dynamics of unidirectional semi - infinite chains of type - i oscillators that are periodically forced at their root node , as an archetype of wave generation in neural networks . in previous studies , numerical simulations based on uniform forcing
have revealed that trajectories approach a traveling wave in the far - downstream , large time limit .
while this phenomenon seems typical , it is hardly anticipated because the system does not exhibit any of the crucial properties employed in available proofs of existence of traveling waves in lattice dynamical systems . here
, we give a full mathematical proof of generation under uniform forcing in a simple piecewise affine setting for which the dynamics can be solved explicitly .
in particular , our analysis proves existence , global stability , and robustness with respect to perturbations of the forcing , of families of waves with arbitrary period / wave number in some range , for every value of the parameters in the system .
_ keywords _ : nonlinear waves , forced feedforward chains , coupled oscillators , type i neural oscillator @xmath0 laboratoire de probabilits et modles alatoires + cnrs - universit paris 7 denis diderot + 75205 paris cedex 13 france + [email protected] + @xmath1 department of mathematics + the cooper union + new york ny 10003 usa + [email protected] + |
the field of molecular electronics@xcite is driven by the quest for functional electronic devices that are smaller than those produced by standard semiconductor technology . the microscopic size and
the reproducibility in the production of molecules provide decisive advantages even if many , identical , molecules should be needed to build fault - tolerant devices a negative differential conductance ( ndc ) , a promising feature for functional devices , has been found recently in organic molecules.@xcite furthermore , many - body effects such as the coulomb blockade and kondo effect , known from semiconductor quantum dots,@xcite have been observed in molecular devices@xcite and carbon nanotubes.@xcite since the typical single - particle level spacing of quantum dots ( or short nanotubes ) is small often only a fraction of a mev low temperatures are required for the resolution of transport through individual levels .
low temperatures are also helpful for the observation of quantum or shot noise . for example , at temperatures above 30k noise transport through molecules between gold break junctions@xcite appears to be dominated by 1/f - like noise , believed to be generated by thermally - induced fluctuations of the gold atoms .
such effects are suppressed at sub - kelvin temperatures , at which the shot noise associated with the discreteness of the charge of the transfered electrons@xcite can be detected . both the current and the shot noise depend on details of the discrete level spectrum and the coupling strengths of these levels to the electrodes.@xcite the combined measurements of current and shot noise , thus , provide a spectroscopic tool to gain information about the level structure .
negative differential conductance through multi - level systems can occur when two adjacent levels have different coupling strengths to the leads .
once the level with weaker coupling is occupied , transport through the other level is suppressed , which reduces the total current . for molecules
, it is well known that the coupling of different molecular orbitals to the electrodes may vary strongly due to differences in the spatial structure of the corresponding wave functions.@xcite in metallic single - walled nanotubes ( swnts ) , two bands cross the fermi surface as the doping level is varied . for short tubes ,
these bands break up in a set of single - particle levels , separated by a level spacing @xmath0 of about 1 mev or less ( see also ref . for multi - walled nanotubes ) . in general , these levels differ in their spatial structure and coupling strength , particularly if they derive from two different bands at different points of the one - dimensional brillouin zone . for semiconducting nanotubes similar considerations hold.@xcite ndc has also been observed in semiconductor quantum dots.@xcite to study transport through systems which display a ndc , we start from an effective model of a few single - particle levels with couplings to the electrodes which vary strongly from level to level , and which also may differ for the source and drain electrodes .
furthermore , we include the possibility of relaxation among the levels : at finite bias voltage , electrons might enter the molecule ( quantum dot or nanotube ) at a high - lying excited state .
provided that the relaxation is fast as compared to the tunneling , the molecule might relax to the ground state or other low - lying state before the electron has the chance to leave the molecule .
the relaxation is accompanied by the emission of a boson , either a photon or a phonon .
such relaxation processes can have a strong impact on the negative differential conductance by destroying the blocking mechanism.@xcite related models ( without coupling to a bosonic bath and relaxation ) were studied in refs . .
the main purpose of this work is to study shot noise for the model described above in the regime where ndc might occur .
we predict that in the absence of relaxation , the ndc is accompanied with super - poissonian noise .
this is formally similar to transport through semi - conducting resonant tunneling devices,@xcite though the origin of ndc in these devices ( chemical potential passes through the semiconductor band edge ) is entirely different from the one discussed here .
relaxation processes enhance the current and reduce the noise in the ndc regime .
the shot noise shows rather rich behavior depending on the coupling and relaxation strength . in particular
, we find that the shot noise is a non - monotonic function of the relaxation rate .
this behavior contrasts with the current , which monotonically increases as the relaxation rate becomes larger .
we are able to present analytic results for the shot noise , which might be useful for the interpretation of future experiments .
we also relate our results to models of transport through several quantum dots .
as a model for electron transport through a molecule or nanotube with @xmath1 molecular orbitals ( levels ) and coulomb interaction we consider a generalized anderson impurity model coupled to a bosonic bath , described by the hamiltonian @xmath2 with : + @xmath3 where @xmath4 and @xmath5 . here ,
@xmath6 and @xmath7 model the non - interacting electrons with density of states @xmath8 in the left and right electrode ( @xmath9 are the fermi operators for the states in the electrodes ) .
the molecule term @xmath10 describes a `` molecule '' with @xmath1 relevant molecular orbitals of energy @xmath11 and coulomb interaction on the molecule ( @xmath12 are fermi operators for the molecular levels , and @xmath13 is the number operator ) .
the charging energy @xmath14 accounts for the classical energy cost to add charge on a confined system with many electrons and ions that are not explicitly considered in the hamiltonian .
in addition , the hubbard - like term with energy @xmath15 punishes double occupancy within the same orbital .
these two kinds of interaction terms are the most important parts of the full two - body interactions present in a real molecule .
other terms could be considered by much more elaborate models , as done in ref . for computation of the . for the ndc / relaxation effects on the shot noise that we wish to study , the above simple molecule model suffices .
tunneling between leads and molecule levels is modeled by @xmath16 and @xmath17 .
the coupling strength is characterized by the intrinsic line width @xmath18 , where @xmath19 are the tunneling matrix elements . in order to allow for relaxation between different molecular levels , we add @xmath20 , which describes a bosonic bath ( where @xmath21 are the corresponding bose operators ) coupled to the molecule by the coupling constants @xmath22 .
this allows relaxation processes where electrons on the molecule can change the orbital by emitting or absorbing a boson .
note that a diagonal coupling , @xmath23 , would not be associated with relaxation but would give rise to `` boson - assisted tunneling '' leading to additional steps in the when the boson bath has a discrete spectrum.@xcite since in the present paper we are not interested in those boson - assisted tunneling processes , we take into account off - diagonal coupling contributions only . to be specific ,
we assume in the following that the bosonic bath consists of photons , although vibrational effects due to phonons could also be described within our model . for simplicity , we assume the constants @xmath24 to be independent of @xmath25 and @xmath26 , and introduce the coupling @xmath27 as @xmath28 , where @xmath29 is the density of states of the bosonic bath for the relaxation due to photons we choose a power law behavior @xmath30 , corresponding to photons with @xmath31 spatial degrees of freedom . for the case of phonon - mediated relaxation , which we are not going to discuss in detail in the present paper ,
the density of states is sharply peaked around the vibration modes of the molecule .
the above model is an extension of the anderson impurity model with one level ( and in the absence of relaxation effects ) , which was described and discussed in ref . .
we are interested in transport through the molecule , in particular in the current @xmath32 and the ( zero - frequency ) current noise @xmath33 .
they are related to the current operator @xmath34 , with @xmath35 being the current operator for electrons tunneling into lead @xmath36 , by @xmath37 and @xmath38 where @xmath39 .
for the calculation of the current @xmath32 and current noise @xmath33 , we use the diagrammatic technique developed in ref . and expanded for the description of the noise in ref . .
in lowest - order perturbation theory in the coupling strengths @xmath40 , the following expressions the current and the noise were found : @xmath41 @xmath42 the bold face indicates matrix notation related to the molecular state labels @xmath43 ( for the @xmath1 level system there are @xmath44 different states ) .
the vector @xmath45 is given by @xmath46 for all @xmath43 .
the zeroth - order stationary probabilities @xmath47 can be expressed in terms of first - order transition rates @xmath48 ( forming a matrix @xmath49 ) between two molecular states @xmath43 and @xmath50 as : @xmath51 the matrix @xmath52 is identical to @xmath49 but with one ( arbitrarily chosen ) row @xmath53 being replaced with @xmath54 .
then the vector @xmath55 is given by @xmath56 .
the total transition rates @xmath48 ( in the absence of relaxation ) are the sum of transition rates associated with electron tunneling through either the left or the right barrier , @xmath57 .
the matrix elements of @xmath58 and @xmath59 are given by @xmath60 and @xmath61 , where @xmath62 is the total number of electrons on the molecule within the state @xmath43 .
the indices @xmath32 or @xmath63 indicate that one or two vertices in the corresponding diagram are due to current operators present in the definition for the current @xmath32 and the noise @xmath33 .
the matrix @xmath64 is associated with the propagation between two blocks @xmath58 containing one current operator each .
to lowest order in @xmath40 , @xmath65 with @xmath66 , i.e. , @xmath64 is of order @xmath67 , thus leading to a non - vanishing contribution of the second part in eq .
( [ s1 ] ) even in lowest ( first ) order perturbation theory in the coupling to the electrodes .
similar expressions for the calculation of current and noise were derived by other means by hershfield _
_ in ref . and korotkov in ref .
+ in order to include relaxation processes , we need to extend the theory of ref . by introducing corresponding transition rates @xmath68 .
assuming weak coupling to the bosonic bath ( in addition to weak tunneling ) , we only keep contributions to either first order in @xmath27 or to first order in @xmath40 .
the total transition rates are , thus , given by @xmath69 , where @xmath68 describe pure relaxation while @xmath70 and @xmath71 models pure tunneling .
the tunneling rates are given by @xmath72 \end{aligned}\ ] ] for @xmath73 , together with @xmath74 , where @xmath75 .
the bosonic rates are @xmath76 for @xmath73 , and @xmath77 , with @xmath78 . here , @xmath79 are the fermi and bose functions and @xmath80 is the energy difference between the many - body states @xmath81 and @xmath50 . while the presence of relaxation leads to a modification of @xmath47 and @xmath64 , the matrices @xmath58 and
@xmath59 are not affected .
the rules for calculating the irreducible blocks @xmath82 describing electron tunneling and relaxation are as follows : \1 ) for a given order @xmath83 draw all topologically different diagrams with @xmath84 vertices connected by @xmath83 tunneling ( electron ) lines or boson ( photon ) lines ( for orders @xmath85 both kinds of lines might be contained in a diagram ) .
assign the energies @xmath86 to the propagators , and energies @xmath87 ( @xmath88 ) to each one of these lines .
\2 ) for each of the ( @xmath89 ) segments enclosed by two adjacent vertices there is a resolvent @xmath90 with @xmath91 , where @xmath92 is the difference of the left - going minus the right - going energies .
\3 ) each vertex containing dot operators @xmath93 ( with @xmath94 different operator structures ) gives rise to a matrix element @xmath95 , where @xmath96 is the dot state entering ( leaving ) the vertex with respect to the keldysh contour ( for our model we have : @xmath97 ) .
\4 ) the contribution of a tunneling line of reservoir @xmath36 is @xmath98 , taking the plus - sign if the line is going backward with respect to the closed time path , and the minus - sign if it is going forward . the same way
the contribution of a bosonic line is given by @xmath99 .
5 ) there is an overall prefactor @xmath100 , where @xmath101 is the total number of vertices on the backward propagator plus the number of crossings of tunneling lines ( no bosonic lines ) plus the number of vertices connecting the state @xmath102 with @xmath103 .
\6 ) integrate over the energies @xmath87 of the tunneling and boson lines and sum over all reservoir and spin indices .
in the following we discuss current and shot noise for the model of eqs .
( 1)-(4 ) with two single - particle levels ( @xmath104 ) in first order perturbation theory in the tunnel couplings @xmath105 and the coupling @xmath27 to the bosonic bath .
we express the different coupling parameters @xmath106 in units of a scale @xmath40 that has the same order of magnitude as the largest of the tunnel couplings @xmath107 .
( in the case of equal tunnel couplings , the natural choice is @xmath108 . )
our perturbation expansion is valid for temperatures larger than the tunnel couplings . throughout this paper ,
we choose @xmath109 .
the molecule can acquire 16 possible states , as each level can be either unoccupied , occupied with spin @xmath103 or @xmath110 , or doubly occupied .
the system described is characterized by level energies @xmath111 and @xmath112 , the hubbard repulsion @xmath15 and the charging energy @xmath14 .
furthermore the electron and photon reservoirs have temperature @xmath113 ( set as @xmath114 ) and are connected to the molecule via the coupling parameters @xmath107 and @xmath27 .
transport is achieved by applying a bias voltage @xmath115 , which is dropped symmetrically at the electrode - molecule tunnel junctions , meaning that the energies of the molecular states are independent of the applied voltage even if the couplings @xmath107 are not symmetric .
the effects of asymmetric voltage drop and a possible gate voltage are straightforward to anticipate , but would only obscure the results presented below
. + an illustration of the transport situation is shown in fig .
[ fig : tt_sketch ] . the fano factor , which is given by the noise to current ratio , @xmath116 , provides additional information about transport properties , not contained in the current - voltage characteristics alone .
therefore , we are interested in studying its dependence on different couplings to the electrodes , the strength of relaxation , the coulomb charging energy , etc . in order to make predictions of the importance of those parameters for a given experiment . and shot noise @xmath33 vs. voltage for @xmath117 , @xmath118 , @xmath119 , @xmath120 , @xmath121 , symmetric bias ( @xmath122 ) and @xmath123 .
the height of the plateaus labeled by @xmath124 are discussed in the text and depend on the choice of the coupling parameters . for suppressed coupling @xmath125 current and shot noise break down leading to negative differential conductance ( ndc ) at a threshold energy .
the curves are normalized to @xmath126 and @xmath127 , respectively .
, width=283 ] we focus on the easiest case which exhibits ndc .
the set of energy parameters @xmath128@xcite describes a molecule that is uncharged at zero bias , as the energy to occupy the first single particle level is @xmath129 ( state @xmath130 ) . without coupling to the boson bath and in contrast to the one level system discussed in ref .
we find a negative differential conductance ( ndc ) regime , see figure [ fig : i - s - asym ] , in dependence on the different coupling strength between the molecular orbitals and the reservoirs , as was previously discussed in refs . .
the shot noise behaves qualitatively similar , the important quantitative details are discussed in the following .
if we chose equal tunnel coupling , @xmath131 ( solid line ) , we find that current and shot noise @xmath33 increase , each time , as a new transport `` channel '' ( controlled by the excitation energies ) opens .
this leads to plateaus , separated by thermally broadened steps .
the first four plateaus are shown and discussed in the following . at a bias voltage of @xmath132 , sequential transport through the state @xmath130 with one electron on the lower lying level
becomes possible . at @xmath133 , additionally transport through the @xmath134
state opens up , with the upper level being occupied with one electron .
the different regions of interest are labeled by @xmath135 with @xmath124 . for a bias voltage above @xmath136 , transport channels with two or more electrons on the molecule open up . in the large - bias regime (
not indicated in the plots ) and for symmetric coupling , the values @xmath126 and @xmath127 are reached .
if now the coupling parameter @xmath125 is suppressed with respect to the other couplings , this leads to suppressed curves for the current and shot noise in region @xmath137 , resulting in ndc at the threshold of @xmath133 , when the state @xmath134 becomes relevant , see in fig .
[ fig : i - s - asym ] for @xmath138 and @xmath139 .
the reason for the ndc is a combination of the pauli principle , coulomb blockade and suppressed coupling , as discussed in ref . .
in our case , an electron , entering the upper molecular orbital from the left electrode , can not leave the molecule , if the coupling of this orbital to the right electrode is entirely suppressed .
transport through the lower molecular orbital is also not possible , since the simultaneous occupation of both orbitals is energetically forbidden in the considered bias regime .
the electron gets stuck in the upper molecular orbital blocking other electrons from tunneling through the molecule .
consequently , the current collapses . since in lowest - order perturbation
theory in @xmath40 the plateau _
heights _ are given by the coupling parameters only , we find , that for @xmath140 ndc can be observed , whereas the shot noise is suppressed below its lower bias plateau only , if @xmath141 .
this difference can already give a rough idea about the coupling strength @xmath125 for a given set of current and noise measurements .
if the shot noise is sufficiently suppressed in the ndc region , a peak in the shot noise appears around the resonance energy of the second level .
this peak is due to temperature induced fluctuations that in certain situations enhance the shot noise over the surrounding plateau values ( where temperature fluctuations are exponentially suppressed ) . as the resonance is approached from lower bias , within the range of temperature broadening the noise
`` detects '' the opening of the second transport channel and increases .
if the bias is beyond the resonance , the redistribution of occupation has taken place and the noise is algebraically suppressed .
the result is the observed peak with width of the temperature .
however , the peak height is only determined by the coupling parameters and is independent of the temperature .
the current never shows such a peak , as it decreases proportional to the loss of occupation of the first level , the transport channel with `` good '' coupling .
the effect of ndc on the fano factor , which is given by @xmath142 , is shown in figure [ fig : fano - asym ] . at small bias , @xmath143 ,
the noise is dominated by thermal noise , described by the well known hyperbolic cotangent behavior which leads to a divergence of the fano factor.@xcite the plateau for bias voltages below @xmath144 ( region @xmath145 ) corresponds to the coulomb blockade regime , where transport is exponentially suppressed . in the regions @xmath146 ( @xmath137 ) transport through the state @xmath130 ( @xmath134 ) is possible . the suppressed coupling strength @xmath125 does not affect the plateau height @xmath147 , whereas @xmath148 reaches values larger then @xmath146 , and up to @xmath31.@xcite this `` super - poissonian '' noise ( @xmath149 ) is predicted for @xmath150 . if the bias is larger then @xmath151 tunneling through states is allowed where both orbitals are occupied simultaneously , i.e. the molecule can be doubly occupied .
the fano factor is sub - poissonian again ( @xmath152 ) in this regime .
comparing the figures [ fig : i - s - asym ] and [ fig : fano - asym ] graphically allows one to determine roughly the strength of the suppression of @xmath125 . in general , however , depending on the underlying energy parameters ( giving the ordering of a sequence of plateaus ) and the coupling parameters ( giving their height ) other values for the fano factor are possible . in the coulomb blockade regime ( region @xmath145 ) , for example , there can also be super - poissonian noise , if both energy levels are below the equilibrium fermi energy .
super - poissonian noise is also possible for a single anderson level , if the spin degeneracy is lifted by a magnetic field ( in the coulomb blockade regime ) or by ferromagnetic leads , see refs . .
the energy and coupling parameters can be fully determined only by considering several transport regimes , e.g. by application of a gate voltage .
it should be noted that the non - monotonic behavior of the fano factor in regions @xmath146 and @xmath137 is entirely due to the second term of the noise expression eq .
( [ s1 ] ) that accounts for `` propagation '' ( and transitions ) of molecule states between the two current vertices at different times.@xcite on the plateaus of regions @xmath146 and @xmath137 the first term of eq .
( [ s1 ] ) is always ( i.e. for any coupling / relaxation parameter values ) identical to the current times the electric charge @xmath153 .
therefore , it contributes a term 1/2 to the fano factor @xmath142 . and shot noise @xmath33 vs. voltage for the same parameters as in fig .
[ fig : i - s - asym ] but fixed coupling @xmath154 .
coupling to a bosonic bath allows for relaxation processes .
the coupling parameter @xmath27 which is varied relative to @xmath40 .
the ndc effect is destroyed by strong relaxation.,width=283 ] let us consider next the effect of relaxation processes on the current and shot noise curves . in figure
[ fig : i - s - ph ] we keep the same set of energy parameters as in figure [ fig : i - s - asym ] and fix the coupling strength @xmath125 at @xmath139 suppressed relatively to the other molecule - electrode couplings .
now a parameter @xmath27 describes the coupling of the molecule with a boson bath .
a value of @xmath155 is below even the relatively weak dipole coupling of photons to molecule states of small aromatic molecules such as benzene.@xcite for this small photon coupling ( solid line ) current and shot noise are still reduced in region @xmath137 relative to the plateau heights @xmath156 and @xmath157 in region @xmath146 .
if now @xmath27 increases , we find that both @xmath32 and @xmath33 also increase in the ndc region , at least initially .
if the value @xmath158 is exceeded , the ndc is gone ( see also figure [ fig : isf - plateaus ] , dashed line ) .
the behavior of the shot noise peak at the resonance energy is now further complicated by the effect of relaxation .
the noise value at the resonance energy is non - monotonic , i.e. it first decreases and then increases again with increasing relaxation .
this is due to redistribution of occupation by the relaxation processes in favor of the first level . for our chosen parameters ,
the value @xmath158 is larger than a reasonable molecule - photon coupling .
however , phonon ( vibrational ) couplings could easily be strong enough to achieve such fast relaxation .
on the other hand , molecule vibrations have a discrete spectrum , much different to the power law assumed in our calculations . relaxation due to phonons can be only effective , if the energies of a phonon and the electronic excitation match within the smearing provided by temperature .
this obviously depends on the details of the molecule and can not be discussed within the model considered here . , [ fig : fano - ph ] .
the values @xmath159 of the first plateau do not depend on the bosonic coupling parameter , whereas on second plateau ( ndc region ) both the shot noise and the fano factor show a non - monotonic dependence.,width=226 ] the destruction of ndc by bosonic transition rates is easily explained .
an electron which formerly was stuck on the upper molecular orbital can now relax onto the lower molecular orbital , from which tunneling to the right electrode is possible via the coupling @xmath160 . for the fano factor in figure
[ fig : fano - ph ] an increase of @xmath27 leads to a decreasing value for the plateau @xmath148 , which passes the poissonian value @xmath161 at @xmath162 .
different to the current , however , the fano factor does not show monotonic behavior with increasing @xmath27 .
the dashed - dotted line corresponding to @xmath163 lies below the dotted one with @xmath164 .
the non - monotonic behavior is even more pronounced for the shot noise .
it has a maximum and a minimum for @xmath165 before increasing again at @xmath166 , see figure [ fig : isf - plateaus ] .
the richness of the noise behavior in the ndc regime might allow a detailed determination of coupling - parameter values .
contrary to the ( monotonic ) @xmath27-dependence of the current , which can be explained by a redistribution of occupation probability form the `` blocking '' upper level to the `` conducting '' lower level , it is difficult to present a simple physical picture for the non - monotonic shot noise behavior in the ndc region . as noted above
, it is the second term of the noise expression eq .
( [ s1 ] ) that is responsible for variation of the noise with coupling parameters . for our set of parameters , the second term of eq .
( [ s1 ] ) has a peak at about @xmath167 and a minimum at @xmath168 , where it almost reaches zero .
the increase of this term at small @xmath27 is again explained by the lifting of the blockade , i.e. the redistribution of occupation probability . the decrease after the maximum is the result of a ( near ) cancellation of contributions from the different states participating in transport .
some of the state contributions are negative and counter the positive contributions that produce the maximum .
such a non - monotonicity with coupling parameters is only possible ( in first order perturbation theory ) for a reducible observable like the shot noise , where aside of the stationary occupation probabilities also the `` molecule propagator '' ( in the form of * p * of eq .
[ def_p ] ) plays an important role .
since in lowest - order perturbation theory temperature only leads to a thermal broadening of the steps , the plateau heights in the different transport regimes are given by the coupling parameters , both the tunnel coupling as well as the relaxation strength .
however , note that the actual relaxation _ rate _ depends also on the position of the energy levels via the boson density of states .
this will complicate matters in the general case with many levels , which are not equidistant from each other .
in our case with two levels , we can extract analytical expressions for the plateau values current , noise and fano factor within the low bias transport regimes as indicated in the figures [ fig : i - s - asym ] to [ fig : fano - ph ] .
we find for the plateau of the ndc - region @xmath137 ( @xmath169 ) @xmath170 for the current and @xmath171 + [ 8\gamma^{\rm l}_1\gamma^{\rm l}_2(\gamma^{\rm r}_1-\gamma^{\rm r}_2)^2 + 4(\gamma^{\rm l}_1\gamma^{\rm r}_2+\gamma^{\rm l}_2\gamma^{\rm r}_1)^2+(\gamma^{\rm r}_1\gamma^{\rm r}_2)^2 ] \over [ 2\gamma^{\rm l}_2 ( \alpha_{\rm ph}+\gamma^{\rm r}_1 ) + ( 2\gamma^{\rm l}_1+\gamma^{\rm r}_1)(\alpha_{\rm ph}+\gamma^{\rm r}_2)]^2}.\ ] ] for the fano factor .
since only bosonic transition between singly occupied levels @xmath146 and @xmath137 are possible in this bias region , the above expressions include only one bosonic rate @xmath172 .
since the temperature is much smaller than @xmath173 , only relaxation processes matter for the plateau values . for completeness , we also give the expressions for the transport regime @xmath146 ( transport through the lower level only ) .
they can be found from the above by setting the couplings @xmath174 and @xmath27 equal to @xmath145 .
then electrons can never enter the upper level at positive bias , leading to an effective one level system with the result@xcite @xmath175 the derivation of analytical expressions in the low bias regime allows us a quick study of current , noise and fano factor for arbitrary coupling situations .
for the special situation where @xmath176 and @xmath177 the fano factor @xmath148 is presented in a contourplot ( see figure [ fig : fano-2d ] ) .
this choice allows the coupling parameters to the left and right reservoir to vary ( independently ) between 0 and @xmath40 , while having the sum of the couplings to each reservoir fixed .
although not all of the possible coupling situations can be visualized this way , the following features which can be extracted from this plot are valid in general : a super - poissonian fano factor @xmath149 can only be found , if @xmath179 and additionally @xmath180 .
furthermore a fano factor @xmath181 is possible only if @xmath182 besides the above conditions . in the absence of relaxation processes ( @xmath177 ) we can also find a point symmetry of @xmath148 .
this symmetry is broken if @xmath183 , as adsorption and emission rates of bosons differ due to the boson occupation factors .
the special case with the settings @xmath184 describes spinless transport through a two level - system in the absence of relaxation processes .
this situation was discussed in ref . where values of the fano factor between @xmath185 and @xmath31 were found in the case of equal couplings @xmath186 . in the case
@xmath187 we can even find fano factors with values much larger than @xmath31 , as the shot noise is strongly enhanced compared to a current that is still sizable itself ( not exponentially suppressed ) , see figure [ fig : fano-2d ] .
this again only happens for a special set of coupling parameters , thus allowing a detailed analysis of the coupling parameters , if such high values for the fano factor were observed in experiment .
besides the super - poissonian noise with fano factors @xmath188 due to positive correlations and values between @xmath189 in the sub - poissonian regime , we can also find values of coupling parameters in which the fano factor drops to values below @xmath185 .
this behavior , however , can only be observed in the presence of relaxations , when the coupling strength @xmath190 and @xmath125 are suppressed relative to the other tunnel couplings .
if the above couplings are vanishing , there is only one path for the electrons to tunnel through the molecule , namely from the left electrode to the upper molecular orbital , then via relaxation onto the lower molecular orbital until finally the electrons leave the molecule by tunneling to the right electrode . by choosing specifically @xmath191 and @xmath192
the value of the fano factor can be minimized and is found to be @xmath193 .
the probabilities to find an unoccupied molecule or an occupied molecule with one electron in the lower level doublet or in upper level doublet are all equal in this case ( @xmath194 ) .
this special situation reminds of a system , where a chain of quantum dots are coupled in series , having interdot tunnel couplings of the same size than the couplings of the chain ends to the leads . for an infinite chain of such dots ( effectively a one - dimensional wire )
the fano factor also reaches @xmath193.@xcite in summary , we have discussed the interplay of coulomb interactions , level - dependent coupling and relaxation in a model suitable for quantum dots , molecules and short nanotubes .
we find super - poissonian shot noise in a bias region where the electronic current is suppressed due to blocking effects .
relaxation due to bosonic excitations has a strong impact on both current and shot noise in this region , for example , the shot noise behaves non - monotonically as a function of the coupling strength to the bosonic bath .
the fano factor ( noise to current ratio ) can become arbitrarily large in the blocking regime . in another special set of couplings the fano factor
can be reduced to 1/3 , resembling that of a one - dimensional wire .
m. h. hettler , h. schoeller , and w. wenzel , europhys . lett . * 57 * , 571 ( 2002 ) ; _ ibid .
_ in _ nano - physics and bio - electronics : a new odyssey _ , eds .
t. chakraborty , f. peeters , and u. sivan ( elsevier 2002 ) .
j. knig , h. schoeller , and g. schn , phys .
* 76 * , 1715 ( 1996 ) ; j. knig , j. schmid , h. schoeller , and g. schn , phys .
b * 54 * , 16820 ( 1996 ) ; h. schoeller , in _ mesoscopic electron transport _ , edited by l.l . sohn , l.p .
kouwenhoven , and g. schn ( kluwer , dordrecht , 1997 ) ; j. knig , _ quantum fluctuations in the single - electron transistor _ ( shaker , aachen , 1999 ) . | we consider charge transport through a nanoscopic object , e.g. single molecules , short nanotubes , or quantum dots , that is weakly coupled to metallic electrodes .
we account for several levels of the molecule / quantum dot with level - dependent coupling strengths , and allow for relaxation of the excited states .
the current - voltage characteristics as well as the current noise are calculated within first - order perturbation expansion in the coupling strengths . for the case of asymmetric coupling to the leads we predict negative - differential - conductance accompanied with super - poissonian noise .
both effects are destroyed by fast relaxation processes .
the non - monotonic behavior of the shot noise as a function of bias and relaxation rate reflects the details of the electronic structure and level - dependent coupling strengths . |
erasure channels , the first example of which was introduced in @xcite , have attracted an increasing attention in the last decades . originally regarded as purely theoretical channels , they turned out to be a very good abstraction model for the transmission of data over the internet , where packets get lost randomly due to
, for example , buffer overflows at intermediate routers .
erasure channels also find applications in wireless and satellite channels where deep fading events can cause the loss of one or multiple packets .
traditionally , mechanisms have been used in order to achieve reliable communication .
a good example is the that is used for data transmission over the internet .
relies on feedback from the receiver and retransmissions and it is known to perform poorly when the delay between transmitter and receiver is high or when multiple receivers are present ( reliable multicasting ) .
an early work on erasure coding is @xcite , where reed - solomon codes and ( dense ) linear random codes are proposed .
however , those techniques become impractical due to their complexity already for small block lengths .
more recently tornado codes were proposed for transmission over erasure channels @xcite .
tornado codes have linear encoding and decoding complexities ( under decoding ) . however , the encoding and decoding complexities are proportional to their block lengths and not their dimension . hence , they are not suitable for low rate applications such as reliable multicasting in which the transmitter needs to adapt its code rate to the user with the worst channel ( highest erasure probability ) .
codes have also been proposed for use over erasure channels @xcite and they have been proved to be practical in several scenarios even under decoding .
fountain codes @xcite are erasure codes potentially able to generate an endless amount of encoded symbols .
they find application in contexts where the channel erasure rate is not known a priori . the first class of practical fountain codes , codes ,
was introduced in @xcite together with an iterative decoding algorithm that is efficient when the number of input symbols @xmath0 is large .
one of the shortcomings of codes is that in order to have a low probability of unsuccessful decoding , the encoding cost per output symbol has to be @xmath1 .
raptor codes were introduced in @xcite @xcite as an evolution of codes .
they were also independently proposed in @xcite , where they are referred to as online codes .
raptor codes consist of a serial concatenation of an outer code @xmath2 ( usually called precode ) with an inner code .
the code design can thus be relaxed requiring only the recovery of a fraction @xmath3 of the input symbols with @xmath4 small .
this can be achieved with linear encoding complexity .
the outer code is responsible for recovering the remaining fraction of input symbols , @xmath4 . if the outer code @xmath2 is linear - time encodable , then the raptor code has a linear encoding complexity , @xmath5 , and therefore the overall encoding cost per output symbol is constant with respect to @xmath0 .
if decoding is used , the decoding complexity is also linear in the dimension @xmath0 and not in the blocklegth @xmath6 , as it is the case for ldpc and tornado codes .
this leads to a constant decoding cost per symbol , regardless of the blocklength ( i.e. , of the rate ) .
furthermore , in @xcite it was shown that raptor codes under decoding are universally capacity - achieving on the binary erasure channel .
most of the works on and raptor codes consider decoding which has a good performance for very large input blocks ( @xmath0 at least in the order of a few tens of thousands symbols ) . often in practice ,
smaller blocks are used .
for example , for the raptor codes standardized in @xcite and @xcite the recommended values of @xmath0 range from @xmath7 to @xmath8 . for these input block lengths ,
the performance under decoding degrades considerably . in this context
, an efficient decoding algorithm in the form of inactivation decoding @xcite may be used in place of .
some recent works have studied the decoding complexity of raptor and codes under inactivation decoding @xcite . in @xcite
lower bounds on the distance and error exponent are derived for a concatenated scheme with random outer code and a fixed inner code . in @xcite
it is shown how the rank profile of the constraint matrix of a raptor code depends on the rank profile of the outer code parity check matrix and the generator matrix of the lt code . in @xcite
upper and lower bounds on the bit error probability of and raptor codes under decoding are derived .
the outer codes there considered in this work are picked from a linear ensemble in which the elements of the parity check matrix are independently set to one with a given probability @xmath9 .
this work is extended in @xcite , where upper and lower bounds on the codeword error probability of codes under decoding are developed .
another extension of this work is @xcite where a pseudo upper bound on the performance of raptor codes under decoding is derived under the assumption that the number of erasures correctable by the outer code is small .
hence , this approximation holds only if the rate of the outer code is sufficiently high .
in @xcite lower and upper bounds on the probability of successful decoding of codes under decoding as a function of the receiver overhead are derived , while corresponding bounds are developed in @xcite for raptor codes . in @xcite finite length protograph - based raptor - like ldpc codes
are proposed for the awgn channel . despite their rateless capability ,
raptor codes represent an excellent solution also for fixed - rate communication schemes requiring powerful erasure correction capabilities with low decoding complexity .
it is not surprising that raptor codes are used in a fixed - rate setting by some existing communication systems ( see , e.g. , @xcite ) . in this context
, the performance under erasure decoding is determined by the distance properties of the fixed - rate raptor code ensemble .
in contrast to @xcite , in this work we consider raptor codes in a fixed - rate setting analyzing their distance properties .
in particular , we focus on the case where the outer code is picked from the linear random code ensemble .
the choice of this ensemble is not arbitrary .
the outer code used by the r10 raptor code , the most widespread version of binary raptor codes ( see @xcite ) , is a concatenation of two systematic codes , the first being a high - rate regular code and the second a pseudo - random code characterized by a dense parity check matrix
. the outer codes of r10 raptor codes were designed to behave as codes drawn from the linear random ensemble in terms of rank properties , but allowing a fast algorithm for matrix - vector multiplication @xcite .
thus , the ensemble we analyze may be seen as a simple model for practical raptor codes with outer codes specifically designed to mimic the behavior of linear random codes .
this model has the advantage to make the analytical investigation tractable .
moreover , although it is simple , the results obtained using this model allow predicting the behavior of binary raptor codes in the standards rather accurately , as illustrated by simulation results in this paper .
for the considered raptor ensemble we develop a necessary and sufficient condition to guarantee a strictly positive normalized typical minimum distance , that involves the degree distribution of the inner fixed - rate code , its rate , and the rate of the outer code .
it identifies a positive normalized typical minimum distance region on the @xmath10 plane , where @xmath11 and @xmath12 are the inner and outer code rates .
this can be used as an instrument for fixed - rate raptor code desing . in particular , for a given overall rate @xmath13 of the fixed - rate raptor ensemble , it allows to identify the smallest fraction of @xmath13 that has to be assigned to the outer code to obtain good distance properties .
a necessary condition is also derived which , beyond the inner / outer code rates , depends on the average output degree only .
finally we show how the analytical results presented in this paper may be used to predict the performance of finite length fixed - rate raptor codes .
this work extends the earlier conference paper @xcite .
the rest of the paper is organized as follows . in section [ sec : ensemble ]
we introduce the main definitions .
section [ sec : dist ] provides the derivation of the average weight distribution of the raptor code ensemble considered and the associated growth rate .
section [ sec : rate_reg ] provides necessary and sufficient conditions for a linear growth of the minimum distance with the block length ( positive normalized typical minimum distance ) .
numerical results are presented in section [ sec : results ] .
the conclusions follow in section [ sec : conclusions ] .
we consider fixed - rate raptor code ensembles based on the encoder structure depicted in figure [ fig : raptor ] .
the encoder is given by a serial concatenation of an @xmath14 outer code with an @xmath15 inner fixed - rate code .
we denote by @xmath16 the outer encoder input , and by @xmath17 the corresponding random vector . similarly , @xmath18 and
@xmath19 denote the input and the output of the fixed - rate encoder , with @xmath20 and @xmath21 being the corresponding random vectors .
the vectors @xmath16 , @xmath18 , and @xmath19 are composed by @xmath0 , @xmath22 , and @xmath6 symbols respectively .
the symbols of @xmath16 are referred to as _ source _ symbols , whereas the symbols of @xmath18 and @xmath19 are referred to as _ intermediate _ and _ output _ symbols , respectively .
we restrict ourselves to symbols belonging to @xmath23 .
we denote by @xmath24 the hamming weight of a binary vector @xmath25 . for a generic output symbol @xmath26
, @xmath27 denotes the output symbol degree , i.e. , the number of intermediate symbols that are added ( in @xmath28 ) to produce @xmath26 .
we will denote by @xmath29 , @xmath30 , and @xmath31 the rates of the outer , inner codes .
we consider the ensemble of raptor codes @xmath32 obtained by a serial concatenation of an outer code in the @xmath33 binary linear random block code ensemble @xmath34 , with all possible realizations of an @xmath35 fixed - rate code with output degree distribution @xmath36 , where @xmath37 is the probability of having an output symbols of degree @xmath38 .
we also denote by @xmath39 the average output degree , @xmath40 . picking randomly one code in the ensemble @xmath32
is performed by randomly drawing the parity - check matrix of the linear random outer code and the low density generator matrix of the fixed - rate encoder .
the parity - check matrix of the outer code is obtained by drawing @xmath41 bernoulli uniform random variables .
the generator matrix of the fixed - rate encoder is generated by independently drawing @xmath6 degrees @xmath38 according to the probability mass function ( p.m.f . ) @xmath42 and , for each such degree @xmath38 , by choosing uniformly at random @xmath38 distinct symbols out of the @xmath22 intermediate ones .
we make use of the notion of exponential equivalence @xcite .
two real - valued positive sequences @xmath43 and @xmath44 are said to be exponentially equivalent , writing @xmath45 , when @xmath46 if @xmath43 and @xmath44 are exponentially equivalent , then @xmath47 given two pairs of reals @xmath48 and @xmath49 , we write @xmath50 if @xmath51 and @xmath52 .
in this section we characterize the expected of a fixed - rate raptor code picked randomly in the ensemble @xmath32 .
an expression for the expected is first obtained .
then , the asymptotic exponent of the is analyzed .
[ theorem : we ] let @xmath53 be the expected multiplicity of codewords of weight @xmath54 for a code picked randomly in the ensemble @xmath32
. for @xmath55 we have @xmath56 where @xmath57 for a serially concatenated code we have @xmath58 where @xmath59 is the average of the outer code , and @xmath60 is the average of the inner fixed - rate code .
the average of an @xmath14 linear random code is known to be @xcite @xmath61 we now focus on the average of the fixed - rate code .
let us denote by @xmath62 the hamming weight of the input word to the encoder and let us denote by @xmath63 the probability that any of the @xmath6 output bits of the encoder takes the value @xmath64 given that the hamming weight of the intermediate word is @xmath65 and the degree of the code output symbol is @xmath66 , i.e. , @xmath67 for any @xmath68 .
this probability may be expressed as @xmath69 removing the conditioning on @xmath66 we obtain @xmath70 , the probability of any of the @xmath6 output bits of the fixed - rate encoder taking value @xmath64 given a hamming weight @xmath62 for the intermediate word , i.e. , @xmath71 for any @xmath68 .
we have @xmath72 since the output bits are generated by the encoder independently of each other , the hamming weight of the codeword conditioned to an intermediate word of weight @xmath62 is a binomially distributed random variable with parameters @xmath6 and @xmath70 .
hence , we may write @xmath73 the average of a code may now be easily calculated multiplying by the number of weight-@xmath65 intermediate words , yielding @xmath74 making use of , and , we obtain .
as opposed to @xmath75 with @xmath76 , whose expression is given by ( 1 ) , the expected number of codewords of weight @xmath77 , @xmath78 , is given by @xmath79 an expected number of weight-@xmath77 codewords larger than one is related to the fact that we have a nonzero probability that the @xmath80 generator matrix of the fixed - rate lt code is not full - rank .
this matrix , in fact , is generated `` online '' in the standard way for lt encoding , i.e. , by drawing @xmath6 i.i.d . discrete random variables with p.m.f .
@xmath42 , representing the weights of the @xmath6 columns . for each such column , the corresponding ` @xmath64 ' entries are placed in random positions .
it will be shown in section [ sec : rate_reg ] , theorem [ theorem : zero_codeword ] , that if the @xmath10 pair belongs to the region there called `` positive normalized typical minimum distance region '' , the expected number @xmath78 of zero weight codewords approaches @xmath64 ( exponentially ) as @xmath6 increases .
next we compute the asymptotic exponent ( growth rate ) of the weight distribution for the ensemble @xmath81 , that is the ensemble @xmath82 in the limit where @xmath6 tends to infinity for constant @xmath11 and @xmath12 .
hereafter , we denote the normalized output weight of the raptor encoder by @xmath83 and the normalized output weight of the outer code ( input weight to the encoder ) by @xmath84 .
the growth rate is defined as @xmath85 [ theorem : growth_rate ] the asymptotic exponent of the weight distribution of the fixed - rate raptor code ensemble @xmath81 is given by @xmath86 where @xmath87 being @xmath88 and @xmath89 defined as follows , @xmath90 @xmath91 & \textrm{otherwise } \
, , \end{array } \right.\end{aligned}\ ] ] with @xmath92 defined as @xmath93 .
\label{eq_npnl}\end{aligned}\ ] ] let us define @xmath94 .
from we have @xmath95 inequality @xmath96 follows from the well - known tight bound @xcite @xmath97 while @xmath98 from @xmath99 and from the fact that the maximum can not be taken for @xmath100 for large enough @xmath101 , hence for large enough @xmath6 ( as shown next ) .
inequality @xmath102 is due again to , to @xmath103 being a monotonically increasing function , and to @xmath104 being a scaling factor not altering the result of the maximization with respect to @xmath62 . that the maximum is not taken for @xmath100 , for large enough @xmath22 ,
may be proved as follows . by direct calculation of for @xmath100 and @xmath105
it is easy to show that we have @xmath106 since @xmath107 for increasing @xmath22 , there exists @xmath108 such that @xmath109 for all @xmath110 .
hence , for all such values of @xmath22 the maximum can not be taken at @xmath100 .
next , by defining @xmath111 the right - hand side of may be recast as @xmath112 the two terms @xmath113 and @xmath114 in the last expression converge to zero as @xmath115 . moreover , also the term @xmath116 converges to zero regardless of the behavior of the sequence @xmath117 .
in fact , it is easy to check that the term @xmath118 converges to zero in the limiting cases @xmath119 @xmath120 and @xmath121 @xmath120 , so it does in all other cases . developing the right hand side of further , for large enough @xmath6 , we have @xmath122 where @xmath123 is the set of rational numbers .
inequality @xmath124 follows from the fact that , as it can be shown , @xmath125 ( uniformly in @xmath126 ) for large enough @xmath6 and from the fact that the supremum over @xmath127 upper bounds the maximum over the finite set @xmath128 .
equality @xmath129 is due to the density of @xmath123 . in equality @xmath129 ,
the function of @xmath130 being maximized is regarded as a function over the real interval @xmath131 ( i.e. , @xmath126 is regarded as a real parameter ) .
the upper bound on @xmath132 is valid for any finite but large enough @xmath6 .
if we now let @xmath6 tend to infinity , all inequalities @xmath96@xmath124 are satisfied with equality . in particular : for @xmath96 this follows from the well - known exponential equivalence @xmath133 ; for @xmath98 from the exponential equivalence @xmath134 ; for @xmath102 from @xmath135 ( due to @xmath136 vanishing for large @xmath6 ) ; for @xmath124 from the fact that , asymptotically in @xmath6 , applying the definition of limit we can show that the maximum over the set @xmath137 upper bounds the supremum over @xmath127 ( while at the same time being upper bounded by it for any @xmath6 ) .
the expression of @xmath92 is obtained by assuming @xmath6 tending to @xmath138 using the expression of @xmath70 .
alternatively , the same expression is obtained by assuming @xmath6 tending to @xmath138 and letting an output symbol of degree @xmath38 choose its @xmath38 neighbors _ with _ replacement . by letting @xmath6 tend to infinity and by cancelling all vanishing terms , we finally obtain the statement . note that we can replace
the supremum by a maximum over @xmath89 as this maximum is always well - defined.$ ] the function @xmath88 diverges to @xmath139 as @xmath140 .
moreover , it diverges to @xmath139 as @xmath141 if @xmath142 for all even @xmath66 and converges as @xmath141 otherwise .
finally , for all @xmath143 $ ] it is continuous for all @xmath144 . ] the next two lemmas , which will be useful in the sequel , characterize the derivative of the growth rate function . for the sake of clarity
, we use the notation @xmath145 instead of @xmath146 .
[ lemma : growth_rate_derivative ] the derivative of the growth rate of the weight distribution of a fixed - rate raptor code ensemble @xmath81 is given by @xmath147 where @xmath148 let us rewrite the expression of @xmath149 in as @xmath150 .
we must have @xmath151 taking the derivative with respect to @xmath152 , after elementary algebraic manipulation we obtain @xmath153 which , applying , yields the statement . [
corollary : der ] for all @xmath154 , the derivative of the growth rate of the weight distribution of a fixed - rate raptor code ensemble @xmath81 fulfills @xmath155 by imposing @xmath156 , from lemma [ lemma : growth_rate_derivative ] we obtain @xmath157 which implies @xmath158 since the function @xmath159 is monotonically decreasing for @xmath160 .
next , due to the definition of @xmath161 in we know that the partial derivative @xmath162 must be zero when calculated for @xmath163 .
the expression of this partial derivative is @xmath164 so we obtain @xmath165 as shown above , for any @xmath152 such that @xmath156 we have @xmath166 . substituting in the latter equation
we obtain @xmath167 which implies @xmath168 .
therefore , the only value of @xmath152 such that @xmath156 is @xmath169 . due to continuity of @xmath170 and to the fact that @xmath171 as @xmath172 ( as shown in subsection b of appendix [ sec : proof_inversion ] ) we conclude that @xmath173 for all @xmath174 . the normalized typical minimum distance of an ensemble @xmath81 is the real number @xmath175 fig .
[ fig : growth ] shows @xmath176 for the ensemble @xmath177 , where @xmath178 is the output degree distribution used in the standards @xcite , @xcite ( see details in table [ table : dist ] ) and @xmath179 for three different @xmath11 values . the growth rate , @xmath176 , of a linear random code ensemble with rate @xmath180 is also shown .
it can be observed how the curve for @xmath181 does not cross the @xmath182-axis , the curve for @xmath183 has @xmath184 and the curve for @xmath185 has @xmath186 . .
the continuous line shows the growth rate of a linear random code with rate @xmath180 .
the dot - dashed , dashed , and dotted lines show the growth rates @xmath176 of the ensemble @xmath187 for @xmath188 , @xmath189 and @xmath190 , respectively . ] fig .
[ fig : gilbert ] shows the overall rate @xmath13 of the raptor code ensemble @xmath191 versus the normalized typical minimum distance @xmath192 .
it can be observed how , for constant overall rate @xmath13 , @xmath192 increases as the outer code rate @xmath12 decreases .
it also can be observed how decreasing @xmath12 allows to get closer to the asymptotic gilbert - varshamov bound . vs. the normalized typical minimum distance @xmath192 .
the continuous line represents the asymptotic gilbert - varshamov bound .
the markers represent raptor codes ensembles @xmath193 with different outer code rates , @xmath12 . ]
in this section we aim at determining under which conditions the ensemble @xmath194 exhibits good normalized typical distance properties .
more specifically , given a distribution @xmath42 and an overall rate @xmath13 , we are interested in the allocation of the rate between the outer code and the fixed - rate lt code to achieve a strictly positive normalized typical minimum distance .
we define the _ positive _ normalized typical minimum distance region of an ensemble @xmath194 as the set @xmath195 of code rate pairs @xmath196 for which the ensemble possesses a positive normalized typical minimum distance .
formally : @xmath197 where we have used the notation @xmath198 to emphasize the dependence on @xmath42 , @xmath11 and @xmath12 .
the positive normalized typical distance region for an lt output degree distribution @xmath42 is developed in the following theorem .
the region @xmath195 is given by [ theorem_inner ] @xmath199 see appendix [ sec : proof_inversion ] .
the next two theorems characterize the distance properties of a fixed - rate raptor code with linear random outer code picked randomly in the ensemble @xmath32 with @xmath200 belonging to @xmath195 .
[ theorem : min_dist ] let the random variable @xmath201 be the minimum nonzero hamming weight in the code book of a fixed - rate raptor code picked randomly in an ensemble @xmath32 . if @xmath202 then @xmath203 exponentially in @xmath6 , for all @xmath204 .
it is well known that this probability can be upper bounded via union bound as @xmath205 we will start by proving that the sequence @xmath206 is non - decreasing for @xmath207 and sufficiently large @xmath6 .
as @xmath115 , the expression @xmath208 converges to @xmath209 , being @xmath210 given in . from lemma [ corollary : der ] we know that @xmath211 for @xmath174 . as @xmath115 , from theorem [
theorem : growth_rate ] we have @xmath212 .
hence , for sufficiently large @xmath6 , @xmath213 , and @xmath206 is non decreasing .
we can now write @xmath214 where we have used @xmath215 , being @xmath210 given in . as
@xmath115 we have @xmath212 .
moreover , @xmath216 for all @xmath204 , provided @xmath217 .
hence , @xmath218 tends to @xmath77 exponentially on @xmath6 . as from theorem [ theorem :
min_dist ] , we have an exponential decay of the probability to find codewords with weight less than @xmath219 when the @xmath200 pair belongs to the region @xmath195 . such an exponential decay shall be attributed to the presence of the linear random outer code characterized by a dense parity - check matrix , which makes the growth rate function monotonically increasing for the values of @xmath152 for which it is negative . as a comparison , for code ensembles characterized by a positive normalized typical minimum distance ,
the growth rate function starts from @xmath220 with negative derivative , reaches a minimum , and then increases to cross the @xmath182-axis . in this case , for @xmath221 the sum in the upper bound is dominated by those terms corresponding to small values of @xmath222 , yielding either a polynomial decay ( as for gallager s codes @xcite ) or even @xmath223 tending to a constant ( as it is for irregular unstructured ensembles @xcite ) . [
theorem : zero_codeword ] let the random variable @xmath224 be the multiplicity of codewords of weight zero in the code book of a fixed - rate raptor code picked randomly in the ensemble @xmath32 . if @xmath202 then @xmath225 in order to prove the statement we have to show that the probability measure of any event @xmath226 with @xmath227 vanishes as @xmath115 .
we start by analyzing the behavior of @xmath228=a_0 $ ] , whose expression is @xmath228 = 1 + 2^{-n { { \mathsf{r}}_{\mathsf i}}(1 - { { \mathsf{r}}_{\mathsf o } } ) } \sum_{l=1}^h \binom { h } { { l } } ( 1 - { { p}_{{l}}})^n$ ] . using an argument analogous to the one adopted in the proof of theorem [ theorem : growth_rate ] , for large enough @xmath6 we have @xmath229 where @xmath230 therefore we can upper bound @xmath228 $ ] as @xmath231 \leq 1 + 2^{n \xi_n}$ ] which , if @xmath217 , implies @xmath228 \rightarrow 1 $ ] exponentially as @xmath115 due to @xmath232 and @xmath233 . is right - continuous at @xmath234 .
this follows from the expression of @xmath176 proved in theorem [ theorem : growth_rate ] and from the fact that @xmath235 is right - continuous at @xmath234 as shown in the proof of theorem [ theorem_inner ] . ]
next , it is easy to show that @xmath228 \geq 1 $ ] and , via linear programming , that the minimum is attained if and only if @xmath236 and @xmath237 for all @xmath238 .
since in the limit as @xmath115 of @xmath228 \rightarrow 1 $ ] , we necessarily have a vanishing probability measure for any event @xmath226 with @xmath227 . from theorem [ theorem : min_dist ] and theorem [ theorem : zero_codeword ] , a fixed - rate raptor code picked randomly in the ensemble @xmath32 is characterized with probability approaching @xmath64 as @xmath115 by a minimum distance at least equal to @xmath219 and by an encoding function whose kernel only includes the all - zero length @xmath0 message ( hence bijective ) . in the following we introduce an outer region to @xmath195 that only depends on the average output degree . [ pro : outer ] the positive normalized typical minimum distance region @xmath195 of a fixed - rate raptor code ensemble @xmath81 fulfills @xmath239 , where @xmath240 with @xmath241 being @xmath242 the only root of @xmath243 in @xmath244 , numerically @xmath245 .
see appendix [ sec : proof_outer ] .
@xmath246 ' '' '' ' '' '' + @xmath247 & 0.0098 & 0.0048 + @xmath248 & 0.4590 & 0.4965 + @xmath249 & 0.2110 & 0.1669 + @xmath250 & 0.1134 & 0.0734 + @xmath251 & & 0.0822 + @xmath252 & & 0.0575 + @xmath253 & & 0.0360 + @xmath254 & 0.1113 & + @xmath255 & 0.0799 & + @xmath256 & & 0.0012 + @xmath257 & & 0.0543 + @xmath258 & 0.0156 & + @xmath259 & & 0.0182 + @xmath260 & & 0.0091 + @xmath39 & 4.6314 & 5.825 ' '' '' ' '' '' + [ table : dist ] [ example_region ] in fig .
[ fig : region ] we show the positive normalized typical minimum distance region , @xmath195 for @xmath178 and @xmath246 ( see table [ table : dist ] ) together with their outer bound @xmath261 .
it can be observed how the outer bound is tight in both cases except for inner codes rates close to @xmath262 .
the figure also shows several isorate curves , along which the rate of the raptor code @xmath263 is constant .
for example , in order to have a positive normalized typical minimum distance and an overall rate @xmath264 , the figure shows that the rate of the ouer code must lay below @xmath265 for both distributions .
let us assume we want to design a fixed - rate raptor code , with degree distribution @xmath178 or @xmath246 , overall rate @xmath264 and for a given length @xmath6 , which we assume to be large .
different choices for @xmath11 and @xmath12 are possible . if @xmath12 is not chosen as @xmath265 the average minimum distance of the ensemble will not grow linearly on @xmath6 . hence , many codes in the ensemble will exhibit high error floors even under erasure decoding .
0.48 and the dashed lines with white markers represents its outer bound @xmath261 .
the gray dashed lines represent isorate curves for different rates @xmath13.,title="fig:",scaledwidth=105.0% ] [ fig : region_a ] 0.48 and the dashed lines with white markers represents its outer bound @xmath261 .
the gray dashed lines represent isorate curves for different rates @xmath13.,title="fig:",scaledwidth=105.0% ] [ fig : region_b ]
in this section experimental results are presented to validate the analytical results obtained in the previous sections . by means of examples
we illustrate how the developed results can be used to make accurate statements about the performance of fixed - rate raptor code ensembles in the finite length regime .
furthermore , we provide some results that show a tradeoff between performance and decoding complexity .
finally we present some simulation results that show that the results obtained for linear random outer codes are a fair approximation for the results obtained with the standard r10 raptor outer code ( see @xcite ) . in this section we will consider raptor code ensembles @xmath266 for different values of @xmath11 , @xmath12 , and @xmath6 but keeping the overall rate of the raptor code constant to @xmath267 . fig . [
fig : region_results ] shows the boundary of @xmath195 and @xmath261 for distribution @xmath178 together with an isorate curve for @xmath267 .
the markers along the isorate curve in the figure represent the two different combinations of @xmath11 and @xmath12 that will be considered in this section .
the first point ( @xmath268 , @xmath269 ) , marked with an asterisk , is inside but very close to the boundary of @xmath195 for @xmath178 .
we will refer to ensembles corresponding to this point as _ bad _ ensembles .
the second point , ( @xmath270 , @xmath271 ) marked with a triangle , is inside and quite far from the boundary of @xmath195 for @xmath178 .
we will refer to ensembles corresponding to this point as _ good _ ensembles .
0.49 its outer bound .
the dashed - dotted line represents the isorate curve for @xmath267 and the markers represent two different points along the isorate curve with the same rate @xmath13 but different values of @xmath11 and @xmath12 .
the lower figure shows the typical minimum distance @xmath272 as a function of the blocklength @xmath6 for ensembles with @xmath273 and @xmath269 and @xmath267 .
the markers represent @xmath272 whereas the lines represent @xmath219.,title="fig : " ] [ fig : region_results ] 0.49 [ fig : d_min_delta_star ] in order to analyze ensembles of finite length raptor codes it is useful to introduce a notion of minimum distance for finite length .
the typical minimum distance , @xmath272 of an ensemble @xmath32 is defined as the integer number @xmath274 this definition will come in handy when we expurgate raptor code ensembles .
in fact , at least half of the codes in the ensemble will have a minimum distance of @xmath272 or larger .
the equivalent of @xmath272 in the asymptotic regime is @xmath192 , the ( _ asymptotic _ ) normalized minimum distance of the ensemble @xmath81 . for sufficiently large @xmath6
one expects that @xmath272 converges to @xmath219 .
[ fig : d_min_delta_star ] shows @xmath272 and @xmath219 as a function of the blocklength @xmath6 .
it can be observed how the good ensemble has a larger typical minimum distance than the bad ensemble .
in fact for all values of @xmath6 shown in fig . [
fig : d_min_delta_star ] @xmath275 for the bad ensemble .
we can also see how already for small values of @xmath6 the @xmath272 and @xmath219 are very similar .
hence , the result of our asymptotic analysis of the minimum distance holds already for small values of @xmath6 .
the expression of the average weight enumerator in theorem [ theorem : we ] can be used in order to upper bound the average over a with erasure probability @xmath276 , @xcite .
however , the upper bound proposed in @xcite needs to be slightly modified to take into account codewords of weight @xmath77 .
we have @xmath277\leq p^{(\mathsf s)}_{b}(n , k,\epsilon ) \nonumber \\ & + \sum_{e=1}^{n - k } { n \choose e } \epsilon^e ( 1-\epsilon)^{n - e } \min \left\{1 , \sum_{w=1}^e { e \choose w } \frac{{a}_w}{{n \choose w}}\right\ } + { a}_0 - 1 \ ] ] where @xmath278 is the singleton bound @xmath279 considering raptor codes in a fixed - rate setting also allows us to expurgate raptor code ensembles as it was done in @xcite for ldpc code ensembles .
let us consider an integer @xmath280 so that @xmath281 we can define the expurgated ensemble @xmath282 as the ensemble of codes in the ensemble @xmath32 whose minimum distance is @xmath283 .
the expurgated ensemble will contain a fraction at least @xmath284 of the codes in the original ensemble .
from @xcite it is known that the average of the expurgated ensemble can be upper bounded by : @xmath285 for each ensemble considered in this section @xmath286 codes were selected randomly from the ensemble .
for each code monte carlo simulations over a were performed until @xmath287 errors were collected or a maximum of @xmath288 codewords were simulated .
we remark that the objective here was not so much characterizing the performance of every single code but rather to characterize the average performance of the ensemble .
[ fig : ub_128 ] shows the vs the erasure probability @xmath276 for two ensembles with @xmath267 and @xmath289 that have different outer code rates , @xmath271 ( good ensemble ) and @xmath269 ( bad ensemble ) .
the good ensemble is characterized by a typical minimum distance @xmath290 whereas the bad ensemble is characterized by @xmath275 ( cf .
[ fig : d_min_delta_star ] ) .
for the two ensembles the upper bound in holds for average cer . however , the performance of the codes in the ensemble shows a high dispersion due to the short blocklength ( @xmath291 ) .
in fact in both ensembles there are codes with minimum distance equal to zero which have @xmath292 ( around @xmath293 for the good ensemble and @xmath294 for the bad ensemble ) . comparing fig .
[ fig : ub_good_128 ] and fig .
[ fig : ub_bad_128 ] one can easily see how the fraction of codes performing close to the random coding bound is larger in the good ensemble than in the bad ensemble . for the good ensemble fig .
[ fig : ub_good_128 ] shows also an upper bound on the average for the expurgated ensemble with @xmath295 , that has a lower error floor . for the bad ensemble
no expurgated ensemble can be defined ( no @xmath280 exists that leads to @xmath296 in ) .
0.49 for two ensembles with @xmath267 and @xmath289 but different values of @xmath12 .
the solid , dashed and dot - dashed lines represent respectively the singleton bound , the berlekamp random coding bound and the upper bound in .
the dotted line represents the upper bound for the expurgated ensemble for @xmath295 .
the markers represent the average of the ensemble and the thin gray curves represent the performance of the different codes in the ensemble , both obtained through monte carlo simulations.,title="fig : " ] [ fig : ub_good_128 ] 0.49 for two ensembles with @xmath267 and @xmath289 but different values of @xmath12 .
the solid , dashed and dot - dashed lines represent respectively the singleton bound , the berlekamp random coding bound and the upper bound in .
the dotted line represents the upper bound for the expurgated ensemble for @xmath295 .
the markers represent the average of the ensemble and the thin gray curves represent the performance of the different codes in the ensemble , both obtained through monte carlo simulations.,title="fig : " ] [ fig : ub_bad_128 ] fig .
[ fig : ub_256 ] shows the vs @xmath276 for two ensembles using the same outer code rates as in fig .
[ fig : ub_128 ] but this time for @xmath297 .
it can be observed how the shows somewhat less dispersion than for @xmath289 . if we compare fig .
[ fig : ub_good_256 ] and fig .
[ fig : ub_good_128 ] we can see how for the good ensemble ( @xmath271 ) the error floor is much lower for @xmath297 than for @xmath289 , due to an increase in the typical minimum distance .
in fact , whereas for @xmath289 there were some codes with minimum distance zero for @xmath298 we did not find any code with minimum distance zero out of the @xmath286 codes which were simulated . for the good ensemble it is possible again to considerably lower the error floor by expurgation
however , comparing fig .
[ fig : ub_bad_256 ] and fig .
[ fig : ub_bad_128 ] we can see how the error floor is approximately the same for @xmath289 and @xmath297 , because in both cases the typical minimum distance is zero .
0.49 for two ensembles with @xmath267 and @xmath297 but different values of @xmath12 .
the solid , dashed and dot - dashed lines represent respectively the singleton bound , the berlekamp random coding bound and the upper bound .
the dotted line represents the upper bound for the expurgated ensemble for @xmath299 .
the markers represent the average of the ensemble and the thin gray curves represent the performance of the different codes in the ensemble , both obtained through monte carlo simulations.,title="fig : " ] [ fig : ub_good_256 ] 0.49 for two ensembles with @xmath267 and @xmath297 but different values of @xmath12 .
the solid , dashed and dot - dashed lines represent respectively the singleton bound , the berlekamp random coding bound and the upper bound .
the dotted line represents the upper bound for the expurgated ensemble for @xmath299 .
the markers represent the average of the ensemble and the thin gray curves represent the performance of the different codes in the ensemble , both obtained through monte carlo simulations.,title="fig : " ] [ fig : ub_bad_256 ] so far we have only considered the performance under decoding . in practical systems
one needs to consider decoding complexity as well .
when inactivation decoding is used the decoding throughput is largely determined by the number of inactivations needed for decoding @xcite , since the decoding complexity is cubic in the number of inactivations .
[ fig : inact ] shows the averaged number of inactivations needed for ensembles of raptor codes with output degree distribution section .
it can be observed how the good ensembles ( @xmath271 ) need more inactivations than bad ensembles ( @xmath269 ) .
hence , the better performance obtained by using an outer code with lower rate comes at the cost of a higher decoding complexity . and the square markers for @xmath269 .
the solid line stands for @xmath289 and the dashed line for @xmath297 . ] in this section we illustrate by means of a numerical example how the results obtained for linear random outer code closely approximate the results with the standard r10 raptor outer code @xcite@xcite .
we consider raptor codes with an degree distribution @xmath300 .
[ fig : region_compare ] shows the positive growth rate region for such a degree distribution ( assuming a linear random outer code ) and the three different rate points , two of which are inside the region @xmath195 while the third one lays outside .
the @xmath10 rate pairs for the three points are specified in the figure caption .
[ fig : cer_compare ] shows the average obtained through monte carlo simulations for the ensembles of raptor codes with @xmath301 , output degree distribution @xmath302 and two different outer codes , the standard r10 outer code and a linear random outer code .
for the three rate points considered the average cer using the standard outer code and a linear random outer code are very close .
as it can be observed , the error floor behavior of the raptor code ensemble with r10 outer code is in agreement with the position of the corresponding point on the @xmath10 plane with respect to the @xmath195 region , although this region is obtained using the simple linear random outer code model . for rate points inside @xmath195
the error floor is much lower , and it tends to become lower the further the point is from the boundary of @xmath195 .
the markers represent three different rate points all of them with @xmath303 but with different inner code rates , @xmath304 , @xmath305 and @xmath306 .
the lower figure shows the average for raptor code ensembles using @xmath302 as output degree distribution and two different outer codes , the standard outer code of r10 raptor codes and a linear random outer code , ( l.r . ) in the legend.,title="fig : " ] [ fig : region_compare ] 0.49 [ fig : cer_compare ]
in this work we have considered ensembles of binary fixed - rate raptor codes which use linear random codes as outer codes .
we have derived the expression of the average of an ensemble and the expression of the growth rate of the as functions of the rate of the outer code and the rate and degree distribution of the inner code . based on these expressions
we are able to determine necessary and sufficient conditions to have raptor code ensembles with a positive typical minimum distance .
a simple necessary condition has been developed too , which only requires ( besides the inner and outer code rates ) the knowledge of the average output degree .
simulation results have been presented that demonstrate the applicability of the theoretical results obtained for finite length raptor codes .
moreover , simulation results have been presented that show that the performance of raptor codes with linear random outer codes is close to that of raptor codes with the standard outer code of r10 raptor codes .
thus , we speculate that the results obtained for raptor codes with linear random outer codes hold as first approximation for standard r10 raptor codes . the work presented in this paper helps to understand the behavior of fixed - rate raptor codes under decoding and can be used to design raptor codes for decoding . despite the fact that only the fixed - rate setting has been considered , we speculate that raptor code ensembles with a good fixed - rate performance will have also a good performance in a rateless setting .
although only binary raptor codes have been considered , the authors believe that the work can be extended to non - binary raptor codes with a limited effort .
we will first prove that for all @xmath10 pairs in @xmath195 we have a positive normalized typical minimum distance .
then we will prove that this is not possible for any other @xmath10 pair .
a sufficient condition for a positive normalized typical minimum distance is @xmath307 which , from theorem [ theorem : growth_rate ] , is equivalent to @xmath308 as done in lemma [ lemma : growth_rate_derivative ] and lemma [ corollary : der ] , let us use the notation @xmath309 to emphasize the dependence on @xmath130 .
we now show that @xmath310\end{aligned}\ ] ] that is we can invert maximization with respect to @xmath126 and limit as @xmath311 , so that the region @xmath195 in is obtained .
this fact is proved by simply showing that @xmath312 that is the function @xmath313 is right - continuous at @xmath314 .
it suffices to show @xmath315 where @xmath316 is an interval independent of @xmath317 and such that the function @xmath318 is bounded over it , i.e. , @xmath319 in fact , under these conditions we have uniform convergence of @xmath320 to @xmath321 in the interval @xmath316 as @xmath172 , namely , @xmath322 the second inequality in implies @xmath323 .
moreover , denoting by @xmath324 the maximizing @xmath130 , we have @xmath325 which implies @xmath326 .
so we have @xmath327 which yields @xmath328 as desired .
next , we prove .
we first observe that in the case @xmath142 for all even @xmath66 ( in which case @xmath145 is strictly increasing ) by direct computation we have @xmath329 for all @xmath330 and for all @xmath331 .
hence in this case we can take @xmath332 . in all of the other cases there
exists @xmath333 such that @xmath334 for all @xmath335 and we can take @xmath336 .
the existence of @xmath337 ( independent of @xmath330 ) such that the maximum is not taken for all @xmath338 is proved as follows . denoting @xmath339 and @xmath340 , we have @xmath341 since @xmath342 for all @xmath343 and since @xmath344 there exists @xmath345 such that @xmath346 this latter inequality implies @xmath347 uniformly with respect to @xmath348 , which is equivalent to @xmath349 for all @xmath350 , independently of @xmath348 .
therefore the maximum can not be taken between @xmath77 and @xmath351 , with @xmath351 independent of @xmath348 .
so far we have proved that the condition on @xmath10 expressed by theorem [ theorem_inner ] is sufficient to have a positive normalized typical minimum distance .
now we need to show that this condition is also necessary .
we need to prove that for the ensemble @xmath81 all rate pairs @xmath10 such that @xmath352 , the derivative of the growth rate at @xmath77 is positive , @xmath353 . according to lemma [ lemma : growth_rate_derivative ]
the expression of @xmath170 corresponds to @xmath354 hence , since @xmath170 is the sum of two terms the first of which diverges to @xmath355 as @xmath172 , a necessary condition for the derivative to be negative is that the second term diverges to @xmath139 , i.e. , @xmath356 .
this case is analyzed in the following lemma .
[ lemma : limit_p ] if @xmath357 then @xmath358 in case the lt distribution @xmath42 is such that @xmath142 for all odd @xmath66 , and @xmath359 for any other lt distribution @xmath42 .
let us recall that @xmath145 is the probability that the lt enconder picks an odd number of nonzero intermediate bits ( with replacement ) given that the intermediate codeword has hamming weight @xmath360 .
if @xmath361 for at least one odd @xmath66 , then the only case in which a zero lt encoded bit is generated with probability @xmath64 is the one in which the intermediate word is the all - zero sequence .
if @xmath142 for all odd @xmath66 , there is also another case in which a nonzero bit is output by the lt encoder with probability @xmath64 , i.e. , the case in which the intermediate word is the all - one word .
consider now a pair @xmath10 such that @xmath352 .
for a fixed - rate raptor code ensemble corresponding to this pair , we have a positive typical minimum distance if and only if @xmath362 . by lemma
[ lemma : limit_p ] this implies @xmath363 when @xmath364 for at least one odd @xmath66 .
it implies either @xmath363 or @xmath365 otherwise . that @xmath366 can not converge to @xmath77 follows from the proof of sufficiency ( as shown
, the maximum for @xmath367 is taken for @xmath368 ) . to complete the proof
we now show that , in the case where @xmath142 for all odd @xmath66 , assuming @xmath365 leads to a contradiction . in case @xmath142 for all odd @xmath66 ,
a taylor series for @xmath145 around @xmath369 is @xmath370 .
assuming @xmath365 , we consider the left - hand side of and calculate its limit as @xmath172 .
we obtain @xmath371 where the last equality follows from the above - stated taylor series . according to ,
the last expression must be equal to zero , a constraint which requires the second limit to diverge to @xmath355 ( as the first limit diverges to @xmath139 ) .
this , however , can not be fulfilled in any case when @xmath152 converge to zero and @xmath372 to one .
in fact , using standard landau notation , when @xmath373 or @xmath374 the second limit converges , while when @xmath375 it diverges to @xmath139 . the proof consists of deriving a lower bound for @xmath176 and evaluating it for @xmath376 . to derive a lower bound for @xmath176
we first derive a lower bound for @xmath377 .
observing we see how @xmath377 is obtained as a summation over all possible intermediate hamming weights .
a lower bound to @xmath377 can be obtained by limiting the summation to the term @xmath378 yielding to @xmath379 where we have introduced @xmath380 representing the probability that the inner encoder outputs a codeword with hamming weight @xmath381 given that the encoder input has weight @xmath360 .
hence , we can write @xmath382 we shall now lower bound @xmath383 .
we denote by @xmath384 note that @xmath385 .
we have that @xmath386 with @xmath387 being the probability that the @xmath66 intermediate symbols selected to encoder @xmath388 are all zero . for large @xmath22
, we have @xmath389 denoting by @xmath390 , we have by jensen s inequality @xmath391 we have thus that @xmath392 replacing in and recalling that @xmath393 we get @xmath394 if we now impose the @xmath395 we obtain : @xmath396 this expression is only valid when the denominator is negative , that is , for @xmath397 , being @xmath242 the only root of the denominator in @xmath244 , whose approximate numerical value is @xmath245 .
the authors would like to to thank prof .
massimo cicognani for the useful discussions about the proof of theorem 3 .
e. paolini , g. liva , b. matuz , and m. chiani , `` maximum likelihood erasure decoding of ldpc codes : pivoting algorithms and code design , '' _ ieee trans .
_ , vol . 60 , no . 11 , pp . 32093220 , nov . 2012 .
b. schotsch , h. schepker , and p. vary , `` the performance of short random linear fountain codes under maximum likelihood decoding , '' in _ proc . of the 2011 ieee int .
conf . on commun .
_ , kyoto , japan , jun .
2011 , pp . 15 .
f. lzaro blasco , e. paolini , g. liva , and g. bauch , `` on the weight distribution of fixed - rate raptor codes , '' in _ proc . of the 2015 ieee int .
symp . on inf .
theory _ , hong kong , china , jun .
2015 , pp . 28802884 . c. di , d. proietti , i. telatar , t. richardson , and r. urbanke , `` finite - length analysis of low - density parity - check codes on the binary erasure channel , '' _ ieee trans .
inf . theory _ ,
48 , no . 6 , pp . 1570 1579 , jun . 2002 | raptor code ensembles with linear random outer codes in a fixed - rate setting are considered .
an expression for the average distance spectrum is derived and this expression is used to obtain the asymptotic exponent of the weight distribution . the asymptotic growth rate analysis is then exploited to develop a necessary and sufficient condition under which the fixed - rate raptor code ensemble exhibits a strictly positive typical minimum distance .
the condition involves the rate of the outer code , the rate of the inner fixed - rate code and the lt code degree distribution .
additionally , it is shown that for ensembles fulfilling this condition , the minimum distance of a code randomly drawn from the ensemble has a linear growth with the block length .
the analytical results can be used to make accurate predictions of the performance of finite length raptor codes .
these results are particularly useful for fixed - rate raptor codes under erasure decoding , whose performance is driven by their weight distribution .
fountain codes , raptor codes , erasure correction , maximum likelihood decoding . |
recent advances in nanotechnology enable successful fabrication and control of systems in length scales where quantum features and fluctuations are dominant .
if we are to use these quantum systems for useful purposes , such as heat to work conversion ( i.e , as a quantum heat engine ) , we have to deal with the limitations imposed by quantum mechanics , such as quantum friction , on the thermodynamical transformations and cycles @xcite .
quantum friction is the hallmark of finite - time thermodynamical transformations .
the infinitely long lasting ones are the reversible processes , which can drive the quantum systems from an equilibrium state to the another one .
reversible processes are , in general , optimal for the work output and the operational efficiency of the quantum heat engines , but they are in the expense of power output . for better powered quantum engines ,
one typically requires faster thermodynamical transformations , which are irreversible and drive the system outside of equilibrium states , leading to ( unwanted ) entropy production . in the present contribution
, we investigate the concept of internal friction , which is the irreversibility in closed quantum systems and arises when a quantum system undergoes non - ideal , finite - time parametric adiabatic transformation @xcite . consider a parametrically driven quantum system with a time - dependent hamiltonian , @xmath0 .
let the system be prepared initially in an equilibrium state with a heat bath .
then , let the system be detached from the heat bath and undergo a parametric unitary change of its hamiltonian from an initial @xmath1 to a final value @xmath2 in a time interval , @xmath3 . if we differ the average work done on the system in a finite - time , @xmath4 , and the one performed in an infinite time , @xmath5 , the difference is non - negative and introduces the non - adiabatic work performed on the system by the driving agent , i.e. , @xmath6 @xcite .
this indeed defines the internal friction in the system .
the system state adiabatically follows the path of equilibrium states in an infinitely long process .
therefore , @xmath7 can be considered as a quantitative measure from the deviation of adiabaticity . if the final density matrix for the reversible adiabatic process @xmath8 has a well - defined temperature , for example @xmath9 , then the irreversible work is directly related to the quantum relative entropy between the relevant states through the relation , @xmath10 @xcite .
the origin of internal friction is entirely quantum mechanical , as it arises when the system hamiltonian at different times does not commute , i.e. , @xmath11 \neq 0 $ ] .
it can be associated with the diabatic transitions between energy eigenbasis of the hamiltonian .
when the transformation lasts in a finite time , the system state is unable to follow the temporary changes in the hamiltonian and therefore develops coherences in the energy frame .
this leads to store additional parasitic internal energy in the system , corresponding to the `` waste energy '' , which should be released to a heat bath if we want to thermalize the system at the temperature @xmath9 by a subsequent process @xcite .
the parametric adiabatic transformations are constituents of the quantum otto and carnot cycles .
therefore , the role of internal friction in quantum thermal devices has been devoted a considerable attention , recently @xcite .
as expected , the quantum friction is found to limit the performance of the quantum heat / refrigerator devices .
the strategies to fight against dissipative effects of the internal friction have also been investigated in the context of quantum thermal engines .
these strategies can be collect under the common titles as `` shortcuts to adiabaticity '' @xcite and `` quantum lubrication '' @xcite . in the present study
, we will not discuss the minimizing strategies of the internal friction in quantum systems .
our main concern is the concept of the internal friction and to discuss how it arises in a driven quantum system .
we follow the ideas and mathematical tools introduced in refs .
@xcite by using a simple , experimentally accessible quantum system via nmr setups @xcite where the quantum system is a two - level system placed in a transverse time - dependent magnetic field .
the hamiltonian of the system is subject to a parametric change from an initial value @xmath1 to a final value @xmath2 in a unitary process ( named as forward protocol ) followed by a reverse unitary transformation of the hamiltonian from @xmath2 back to its initial value @xmath1 ( named as backward protocol ) . due to the misalignment in the magnetic fields of the system ,
the internal friction naturally arises .
we quantitatively study the deviation from the adiabaticity by using the quantum relative entropy between the initial density matrix of the forward protocol and the final density matrix of the backward protocol .
we investigate how the internal friction depends on the transformation time as well as the path that the parametric transformation follows .
our results reveal the non - monotone character of the non - adiabatic work as a function of total protocol time .
we also show that considering different possible scheduling pulses that generates the unitary transformation , almost frictionless protocol can be obtained in finite times .
our results can be relevant to the relation between the macroscopic and microscopic irreversibility ( loschmidt paradox ) @xcite and related time symmetry arguments @xcite .
developments in nano - engineering systems makes the question of irreversibility in nano - scale thermodynamics a practical concern @xcite .
moreover , they can serve for solid examples to clarify and shed some light onto discussions among the novel perspectives of quantum thermodynamics such as unified quantum formulation of mechanics and thermodynamics @xcite .
in addition , extension of our approach to non - equilibrium thermodynamical systems can be an intriguing application @xcite .
we emphasize that dissipative quantum dynamics , entropy production and irreversibility has been widely discussed from different point of views @xcite ; including nonlinear systems @xcite , general description of friction in quantum mechanics with stochastic master equation treatments can be found in the literature , too @xcite .
our particular case here is limited to irreversibility associated with finite time adiabatic transformations which is the source of a profound quantum internal friction effect when the adiabatic control process is incompatible with the free evolution .
we consider a driven quantum system which undergoes unitary evolution developed by a parametric time - dependent hamiltonian of the form , @xmath12 where @xmath13 is a static magnetic field in @xmath14-direction , while @xmath15 is a time - dependent magnetic field along the @xmath16-axes , and @xmath17 ( @xmath18 ) are the components of the spin angular momentum which obey the canonical commutation relation , @xmath19=i\epsilon_{\alpha \beta \gamma } i_{\gamma}$ ] . for simplicity , we consider a spin-@xmath20 particle where @xmath21 ( @xmath22 are the pauli matrices ) . throughout the paper ,
we use a unit system where the planck constant @xmath23 , the gyromagnetic ratio @xmath24 , and the boltzmann constant @xmath25 are all set to unity . the proposed unitary parametric transformation of the hamiltonian eq .
( [ eq : ssh ] ) has two independent parts .
the first unitary transformation is named as the forward protocol and includes the transformation of the hamiltonian from @xmath1 to @xmath2 in a finite - time .
the latter protocol is the backward transformation of the hamiltonian from @xmath2 back to @xmath1 .
the main consequence of such a protocol is to quantitatively measure the internal friction using the initial density matrix of the forward protocol and the final density matrix of the backward protocol , which have the same hamiltonian @xmath1 . the schematic representation of the proposed unitary forward - backward protocol is given in fig .
[ fig:0 ] .
the details and the mathematical formulation are as follow . _ ( a ) forward protocol . _
the system with initial hamiltonian @xmath26 is prepared in an equilibrium with a heat bath at an inverse temperature @xmath27 .
the initial density matrix of the system can be given by the boltzmann - gibbs distribution @xmath28 , where @xmath29 $ ] .
the system is detached from the heat bath and undergoes an adiabatic parameter change of the magnetic field @xmath15 from @xmath30 to @xmath31 in a time interval @xmath32 .
the forward protocol can be defined by a unitary evolution of the density matrix given by liouville - von neumann equation @xmath33 $ ] , where @xmath0 is given in eq .
( [ eq : ssh ] ) and the initial condition is @xmath34 . the hamiltonian @xmath35 at @xmath36 is changed to @xmath37 at the end of the protocol _ ( a ) _ with the final density matrix @xmath38 .
there are many possible choices for the scheduling function @xmath15 to generate this transformation . in the present study , we propose four different driving pulses having explicit forms as @xmath39 or @xmath40 ( here @xmath41 ) . in all cases
, we have @xmath42 and @xmath43 . _
( b ) backward protocol .
_ this unitary process defines the transformation of the hamiltonian @xmath44 back to its initial value @xmath35 .
the time - evolution of the density matrix can be again given by @xmath33 $ ] , where now the initial density matrix is @xmath45 and @xmath0 is given in eq .
( [ eq : ssh ] ) with the time - dependent magnetic fields having explicit forms as @xmath46 or @xmath47 ( @xmath41 ) . here
@xmath48 and @xmath49 . the final density matrix of the backward protocol is denoted by @xmath50 .
( color online . )
the schematic diagram of the forward unitary protocol ( a ) followed by the backward unitary protocol ( b ) .
here @xmath51 ( @xmath52 ) and @xmath53 ( @xmath54 ) are the density matrix and hamiltonian at the terminal points of the protocols .
one should remark that the protocols ( a ) and ( b ) are named according to the change of the hamiltonian . here
, the backward protocol is simply an application a rise - fall protocol for forward time . ] using eq .
( [ eq : ssh ] ) , one can easily show that the commutator @xmath11=-ib_0\left(b(t_1)-b(t_2)\right)i_y$ ] , which is always different than zero , since the above proposed magnetic fields are always non - uniform ( i.e. , @xmath55 ) .
therefore , the system density matrix can not follow the temporary changes in the hamiltonian for a fast transformation ; the density matrix during the protocol deviates from the equilibrium and thus @xmath56 and @xmath50 become different
. the proper quantification between initial equilibrium state @xmath56 and the actual final state of the backward protocol @xmath50 can be used as a quantitative measure for the irreversibility in the system @xcite .
we study the `` closeness '' of @xmath50 to @xmath56 through the entropy - like distance measure , namely the quantum relative entropy , defined as @xcite @xmath57\geq 0.\ ] ] according to the klein s inequality @xcite , it is non - negative and equals to zero provided that @xmath58 . technically , it is not a metric , since it is generally not symmetric , i.e. , @xmath59 .
the quantum relative entropy has many interpretations in quantum information and computation theory .
we refer the readers to recent review @xcite for more details . remark that if we replace @xmath50 by @xmath56 in the second term of the right - hand - site of eq .
( [ eq : rent0 ] ) , then it will correspond to the difference between von neumann entropies ( @xmath60 ) of @xmath50 and @xmath56 which would be zero since von neumann entropy remains invariant in a unitary path .
we would also like to mention about the calculation procedure of the quantum relative entropy .
( [ eq : rent0 ] ) can be directly evaluated using an appropriate computer software which can handle the logarithm of a matrix .
another calculation for eq .
( [ eq : rent0 ] ) can be done by using the orthonormal decomposition of the density matrices . denoting @xmath61 and @xmath62 ,
the `` ln '' of the density matrices become @xmath63 and @xmath64 . after some calculations , eq .
( [ eq : rent0 ] ) can be rewritten in an equivalent form which is more practical for handmade calculations , @xmath65 on the other hand , eq .
( [ eq : rent0 ] ) has a deep thermodynamical interpretation ; it is intimately related to the internal friction @xcite . to make the paper self - consistent ,
let us prove this assertion here .
first , we consider the work performed during the forward - backward protocol in fig .
[ fig:0 ] . for a reversible adiabatic process ( @xmath66 ) , the work performed on the system is zero
, since we return to the initial state @xmath56 and there would be no change in the internal energy of the system . in this case
, the work done on the system by the driving agent in a finite time @xmath3 would be equal to the irreversible work , i.e. , @xmath67-tr[h_1\rho_0]$ ] . if we consider the thermalization of the state @xmath50 to @xmath56 during an additional isochoric stage , the average heat supplied to the system by the heat bath at temperature @xmath9 would be equal to @xmath68 .
now we consider the relation between eq .
( [ eq : rent0 ] ) and the internal friction .
since the von neumann entropy remains invariant during a unitary evolution , it can be written as @xmath69 $ ] . using the above spectral decomposition of @xmath56 with @xmath70 where @xmath71 are the eigenvalues of @xmath35 , then following some calculation steps
, we have @xmath72 $ ] .
if we distribute @xmath73 over the sum and use the trace property , we can simplify it as @xmath74-tr[h_1\rho_0]\right)$ ] . as a result , the following relation directly holds : @xmath75 with respect to klein s inequality @xcite , the irreversible work is non - negative ; it is stored in the system as a parasitic internal energy ( waste energy ) , as it should have to be released to the heat bath if we want to thermalize the system at the temperature @xmath9 with a subsequent thermalization process .
we should stress here that the validity of the eq .
( [ eq : rfric ] ) requires a strong assumption that the developed coherences could be stored in perfect isolation during the entire finite time adiabatic process .
( color online . )
the quantum relative entropy between the density matrix at the end of the backward protocol @xmath50 and the initial thermal state @xmath56 , @xmath76 , as a function of total allocated protocol time @xmath3 for the parameters @xmath77 , @xmath78 and @xmath79 .
the unitary transformations are assumed to be generated by four different driving pulses as discussed in the text .
note that the thermodynamical interpretation of the plotted quantity is given in eq .
( [ eq : rfric ] ) . ]
the parameter @xmath3 defines the total transformation time , as we allot equal time intervals to both protocols .
our main investigation is to analyze how the quantum relative entropy between the state @xmath50 and the initial thermal state @xmath56 depends on the total protocol time @xmath3 . in fig .
[ fig:1 ] , we plot @xmath80 as a function of @xmath3 for the parameters @xmath77 and @xmath78 .
each line in the figure corresponds to the different driving pulses that generates the unitary evolution .
[ fig:1 ] shows that the quantum relative entropy is non - zero , signifying the irreversible nature of the finite - time transformations ; for @xmath81 , the system state can not return back to its initial state after the backward transformation even when the hamiltonian can do .
[ fig:1 ] is a demonstrative example that non - ideal finite - time quantum evolutions are the root of irreversibilities in closed quantum systems . on the other hand , in the two extreme cases of the total protocol time
, we have @xmath58 .
the first limiting case , as shortly discussed above , is the infinitely slow transformation ( @xmath82 ) . in this case
, the system state follows the temporary changes in the instantaneous hamiltonian without changing the initial level populations and remains diagonal in terms of the eigenstate representation .
consequently , at the end of the backward protocol , the system state returns to @xmath56 .
this is the condition for the quantum adiabatic theorem to hold in the system and provides frictionless solutions .
we should stress here that the transformation timescale should be smaller than the internal timescale of the system , otherwise decoherence effects due to the spontaneous emission of the atoms can take place .
the second limiting case is the completely diabatic case ( @xmath83 ) . in this sudden transformation case ,
even if the hamiltonian is transformed , the system density matrix can not find a chance to be transformed , so it remains unchanged .
therefore , at the end of the backward protocol we will have @xmath58 in the limit @xmath83 . in this scenario , a large variance in energy may take place due to the energy - time uncertainty relation .
( color online . )
the trajectory of the bloch vector in the forward - backward protocol for the parameters given in fig .
[ fig:1 ] and @xmath84 .
the red , triangular line corresponds to time evolution in the forward protocol , while the green , circle line shows the trajectory in the backward protocol .
the sub - figures ( a)-(d ) are plotted for the driving pulses labeled by sinusoidal , @xmath85 , respectively .
remark that the bloch vector indicates the points in the bloch sphere corresponding to mixed states . in the forward - backward unitary protocol , the length of the bloch vector remains intact.,title="fig : " ] ( color online . ) the trajectory of the bloch vector in the forward - backward protocol for the parameters given in fig .
[ fig:1 ] and @xmath84 .
the red , triangular line corresponds to time evolution in the forward protocol , while the green , circle line shows the trajectory in the backward protocol .
the sub - figures ( a)-(d ) are plotted for the driving pulses labeled by sinusoidal , @xmath85 , respectively .
remark that the bloch vector indicates the points in the bloch sphere corresponding to mixed states . in the forward - backward unitary protocol , the length of the bloch vector remains intact.,title="fig : " ] ( color online . ) the trajectory of the bloch vector in the forward - backward protocol for the parameters given in fig .
[ fig:1 ] and @xmath84 .
the red , triangular line corresponds to time evolution in the forward protocol , while the green , circle line shows the trajectory in the backward protocol .
the sub - figures ( a)-(d ) are plotted for the driving pulses labeled by sinusoidal , @xmath85 , respectively .
remark that the bloch vector indicates the points in the bloch sphere corresponding to mixed states . in the forward - backward unitary protocol , the length of the bloch vector remains intact.,title="fig : " ] ( color online . ) the trajectory of the bloch vector in the forward - backward protocol for the parameters given in fig .
[ fig:1 ] and @xmath84 .
the red , triangular line corresponds to time evolution in the forward protocol , while the green , circle line shows the trajectory in the backward protocol .
the sub - figures ( a)-(d ) are plotted for the driving pulses labeled by sinusoidal , @xmath85 , respectively . remark that the bloch vector indicates the points in the bloch sphere corresponding to mixed states . in the forward - backward unitary protocol ,
the length of the bloch vector remains intact.,title="fig : " ] fig .
[ fig:2 ] shows the trajectory of the bloch vector , @xmath86 , ( i.e. , the movement of the density matrix in bloch space ) in the forward - backward protocol .
the bloch vector can not return to its initial case due to the quantum friction .
[ fig:3 ] signifies the time evolution of the energy gap in the protocol .
the figure shows that each driving pulse follows different paths for the transformation of the hamiltonian and defines different protocol rates .
this indeed explains the differences in the dynamical behavior of the quantum relative entropy for the different pulses in fig .
[ fig:1 ] and the trajectory in fig .
[ fig:2 ] .
what is more , the oscillatory behavior of the quantum relative entropy suggests that @xmath76 needs not to depend on the total protocol time monotonically .
@xmath76 is maximal around @xmath87 for each driving pulse .
[ fig:1 ] also elucidates that the driving pulses labeled by sinusoidal , @xmath88 and @xmath89 provide `` almost '' frictionless solutions for shorter protocol time , @xmath90 , while @xmath91 labeled driving pulse requires much longer transformation times .
another interesting observation in fig .
[ fig:1 ] is the path independent behavior for the almost frictionless solutions around @xmath92 .
( color online . )
the time evolution of the energy gap @xmath93 as a function of @xmath94 in the forward transformation for the parameters given in fig .
[ fig:1 ] and for the considered driving pulses . the mirror symmetry according to @xmath95 and the rotational symmetry give the time evolution of the energy gap in the backward protocol .
the concept of internal friction , the irreversibility in a closed quantum system driven out of equilibrium in a unitary transformation , is investigated in a simple , experimentally accessible quantum system .
the irreversible production of the excess energy due to the quantum friction is quantitatively analyzed in a forward - backward transform of the hamiltonian by using the quantum relative entropy between the actual density matrix obtained during the parametric transformation and the one taken in the infinitely slow limit .
the role of total transformation time and the different pulse control schemes on the internal friction are investigated in detail .
the transformations taken in finite - times are irreversible processes , where the system density matrix can not return back to its initial state .
this is also verified from the trajectory plots of the density matrix in bloch space .
our proposed pulse control scheme reveals the non - monotone dependence of the internal friction on the total protocol time and the possibility for almost frictionless solutions in finite - time transformations .
the authors thank dieter suter for fruitful discussions .
s. . warmly thanks azmi genten for useful discussions . f.a . thanks department of physics of the ko university for hospitality .
the authors acknowledge support from ko university and lockheed martin corporation research agreement .
plastina f , alecce a , apollaro t j g , falcone g , francica g , galve f , gullo n lo and zambrini r 2014 _ phys .
lett . _ * 113 * 260601 beretta g p 1986 _ frontiers of nonequilibrium statistical physics : proc of the nato advanced study institute , ( santa fe , 1984 , series b : physics , vol .
135 ) _ ed g.t .
moore and m.o .
scully ( new york : plenum ) p 205 | the concept of internal friction , a fully quantum mechanical phenomena , is investigated in a simple , experimentally accessible quantum system in which a spin-1/2 is driven by a transverse magnetic field in a quantum adiabatic process .
the irreversible production of the waste energy due to the quantum friction is quantitatively analyzed in a forward - backward unitary transform of the system hamiltonian by using the quantum relative entropy between the actual density matrix obtained in a parametric transformation and the one in a reversible adiabatic process . analyzing the role of total transformation time and the different pulse control schemes on the internal friction
reveal the non - monotone character of the internal friction as a function of the total protocol time and the possibility for almost frictionless solutions in finite - time transformations . |
bloom et al . ( 1999 ) first identified a second component to the optical afterglow of grb 980326 .
this second component grew brighter than the extrapolation of the early afterglow ( which faded as a power law ) around 1 week after the burst , and peaked around 20 30 days after the burst before fading away .
rough spectra taken 2.4 and 28 days after the burst show that the second component was significantly redder than the early afterglow . bloom et al .
( 1999 ) attributed this behavior to a supernova at a redshift of @xmath2 .
supernovae peak after about @xmath3 days and are significantly redder than afterglows ( blueward of the supernova s spectral peak ) . unfortunately
, neither a redshift nor multiband photometry is available for this afterglow .
reichart ( 1999 ) identified a similar second component to the optical afterglow of grb 970228 , the first burst for which an afterglow was detected .
reichart ( 1999 ) and galama et al .
( 2000 ) showed that the light curve of this second component is consistent with the expectation of a supernova at the spectroscopically - measured redshift of this burst ( z = 0.695 ; bloom , djorgovski , & kulkarni 2001 ) .
more convincingly , vrijk photometry of the second component taken near its peak showed that its spectral flux distribution is also consistent with the expectation of a supernova at this redshift : ( 1 ) the spectral flux distribution is very red blueward of the spectral peak and very blue redward of the spectral peak , and ( 2 ) the spectral peak is at the frequency of the spectral peak of a supernova at @xmath4 ( figure 1 ; reichart 1999 ; galama et al .
2000 ; reichart , castander , & lamb 2000 ; reichart , lamb , & castander 2000 ) .
this is the best evidence to date for the grb / sn connection .
the grb / sn connection has recently been challenged by alternative explanations of these observations , called dust echoes ( waxman & draine 2000 ; esin & blandford 2000 ; see also lamb 1997 , 2000 ) .
if the progenitors of bursts are massive stars ( see lamb & reichart 2000 for a review of the indirect evidence that bursts are associated with star - forming regions ) , due to the progenitor s wind , the density of the surrounding medium should fall off as @xmath5 out to the radius @xmath6 of the wind termination shock ( chevalier & li 1999 ) . at @xmath6 ,
the density should jump by a factor of @xmath7 , and beyond @xmath6 , the density should remain approximately constant out to great distances ( chevalier & li 1999 ) .
dust in the vicinity of the burst should be sublimated by uv light from the optical flash ( the reverse shock ; waxman & draine 2000 ) .
let @xmath1 be the radius out to which dust is sublimated .
if @xmath8 , and if the dust is sublimated isotropically , the dust should occupy a relatively thin shell at @xmath9 due to the rapidly decreasing density of the circumburst medium with distance .
light that is emitted by the afterglow ( the forward shock ) from a distance that is much less than @xmath1 at an angle @xmath10 relative to the line of sight , is scattered ( esin & blandford 2000 ) or absorbed and thermally re - emitted ( waxman & draine 2000 ; esin & blandford 2000 ) by dust at @xmath11 into the line of sight , and arrives at the observer , arrives with an observer - frame time delay of @xmath12 relative to light that was emitted into the line of sight ( @xmath13 ) and not scattered or absorbed . this dust and light geometry
is depicted in the top panel of figure 2 .
waxman & draine ( 2000 ) and esin & blandford ( 2000 ) find that this results in secondary components to the afterglow , called dust echoes , that may peak about 20 30 days after the burst , and that may be significantly redder than the early afterglow , i.e. , that the late - time afterglows of grb 970228 and grb 980326 may in fact be due to dust echoes , and not supernovae .
there are a number of problems with this dust and light geometry .
first of all , @xmath14 pc ( waxman & draine 2000 ) ; however , @xmath6 is probably only a few parsecs ( chavalier & li 2000 ) .
consequently , the dust would probably not occupy a thin shell .
secondly and more importantly , uv light from the optical flash will only sublimate dust within @xmath15 of the line of sight , where @xmath15 is the initial half - opening angle of the jet .
( the lorentz factor @xmath16 of the shocked material in the jet is much greater than @xmath17 when the optical flash occurs . )
if the shock is adiabatic , @xmath18 where @xmath19 is the number density of gas atoms in the circumburst medium , @xmath20 is the isotropic - equivalent energy of the shock , and @xmath21 is observer - frame time ( e.g. , sari , piran , & narayan 1998 ) .
let @xmath22 . since @xmath23 ( e.g. , sari , piran , & narayan 1998 ) , @xmath24^{-1/4 } \left(\frac{e}{10^{52}\rm{\,erg}}\right)^{1/4 } \left(\frac{t}{\rm{1\,dy}}\right)^{-1/4}. \label{gamma}\ ] ] the afterglow emits light into a half - opening angle @xmath25 .
given the rate at which @xmath16 decreases with time , @xmath26 quickly grows significantly greater than @xmath15 : for example , @xmath27 after only a few minutes if @xmath28 : , the jet begins to expand laterally , and the afterglow fades much more rapidly ( e.g. , panaitescu , mszros , & rees 1998 ) .
applying equation ( [ gamma ] ) to the afterglows of grb 990510 ( e.g. , stanek et al .
1999 ; harrison et al .
1999 ) , grb 991216 ( halpern et al .
2000 ) , grb 000301c ( e.g. , rhoads & fruchter 2000 ; berger et al . 2000 ) , and grb 000926 ( price et al .
2001 ) , we find that @xmath29^{1/4}$ ] , @xmath30^{1/4}$ ] , @xmath31^{1/4}$ ] , and @xmath32^{1/4}$ ] , respectively . ]
@xmath33^{-1 } \left(\frac{e\theta_{jet}^2}{10^{52}\rm{\,erg}}\right ) \left(\frac{\theta_{jet}}{0.1}\right)^2{\rm\,min}.\ ] ] light emitted at @xmath34 would be scattered , absorbed , and thermally re - emitted by dust at radii that are substantially smaller than @xmath1 . due to the substantially higher density of dust at these radii , this light would dominate the dust echo .
radiative shocks yield similar results .
this more realistic dust and light geometry is depicted in the bottom panel of figure 2 . in this paper
we consider a dust and light geometry of which the dust and light geometry depicted in the bottom panel of figure 2 is a limiting case .
high - energy light from the progenitor may destroy dust ( e.g. , via electrostatic disruption ; draine & salpeter 1979 ; waxman & draine 2000 ) out to a smaller radius .
similarly , irregularities in the progenitor s wind may destroy dust in shocks . in either case
, the dust would be destroyed isotropically .
let @xmath1 instead be the radius out to which dust is destroyed by the progenitor , and let @xmath35 be the radius out to which dust is destroyed by the burst , the optical flash , and/or the afterglow , by whatever dust destruction mechanisms .
if @xmath35 is even twice as great as @xmath1 , and if the jet is oriented along the line of sight , the dust echoes should brighten substantially at a time given by equation ( [ delay ] ) with @xmath11 and @xmath36 , and dim substantially at a time given by equation ( [ delay ] ) with @xmath11 and @xmath37 , since we assume that the jet is bipolar .
if @xmath38 , then the dust echoes `` turn on '' at @xmath39 and `` turn off '' at @xmath40 though they may fade earlier than @xmath41 due to other effects ( see 2 and 3 ) .
the jet may be oriented as much as @xmath42 from the line of sight ( any more and we would not have seen the burst ) .
consequently , the dust echoes may `` turn on '' half of the way as early as @xmath43 and all of the way as late as @xmath44 .
the `` turn off '' time would be similarly affected . the dust and light geometry depicted in the bottom panel of figure 2
is recovered in the limit that @xmath45 .
another dust and light geometry that we consider in this paper is that light from the optical flash , as well as optical light that may be associated with the burst itself , may result in dust echoes .
consequently , dust echoes may be a way to measure the brightness of the optical flash and/or the optical brightness of the burst . in this paper ,
we model and compute dust echo light curves and spectra for both @xmath46 a few times @xmath1 , and @xmath47 ( in which case dust is destroyed only isotropically ) .
any other case would be fine tuning .
we also model and compute dust echoes resulting from a pulse of light ( i.e. , the optical flash or optical light associated with the burst ) , and an afterglow light curve .
finally , we model and compute dust echo light curves and spectra for both scattered light ( 2 ) , and absorbed and thermally re - emitted light ( 3 ) .
we find that both types of dust echoes have light curves and spectra that can be distinguished from each other , and from the light curves and spectra of supernovae . in 4
, we re - examine the late - time afterglow of grb 970228 , and find that it can not be explained by a dust echo .
we draw conclusions in 5 .
before modeling the general case of the dust echo of collimated light that is emitted as a function of time ( e.g. , an afterglow ) , we first model the simpler case of the dust echo of an isotropic pulse of light ( e.g. , an optical flash or optical light associated with the burst ) that is emitted from a distance that is much less than @xmath1 .
consider the light that is emitted at an angle @xmath10 relative to the line of sight , is scattered by dust at @xmath48 into the line of sight , and arrives at the observer .
the optical depth at observer - frame frequency @xmath49 along this path is given by : @xmath50 where @xmath51 is the number density of dust particles at @xmath48 and @xmath10 , and @xmath52 is the total ( absorption plus scattering ) cross section of these particles . the number density of dust particles at @xmath53 is given by ( 1 ) : @xmath54 where ( 1 ) @xmath55 assuming that @xmath56 a few times @xmath1 , we find that @xmath57 & $ ( f\theta_{jet } < \theta < \frac{\pi}{2}$ or $ r\sin\theta > r)$ \cr \tau_{\nu ( 1+z)}\left[1+\frac{r}{r } \left(\frac{\theta - f\theta_{jet}-2\cos^{-1}\frac{r\sin\theta}{r}}{\sin{\theta } } - 1\right)\right ] & $ ( \frac{\pi}{2 } < \theta < \pi - f\theta_{jet}$ and $ r\sin\theta < r)$ \cr \tau_{\nu ( 1+z)}(3-\frac{2r}{r } ) & $ ( \theta = \pi$ and $ f = 0)$ } , \label{tau}\ ] ] where @xmath58 is the optical depth along any line of sight if @xmath59 , or any line of sight between @xmath60 if @xmath61 .
the probability that light takes this path , i.e. , that light that is emitted within @xmath62 of @xmath10 makes it to within @xmath63 of @xmath48 without being scattered or absorbed , is scattered within @xmath63 of @xmath48 into the line of sight , and makes it to the observer without being scattered or absorbed , is given by : @xmath64 we adopt the same differential dust scattering cross section as do esin & blandford ( 2000 ; white 1979 ) : @xmath65^{3/2 } } , \label{scat}\ ] ] where @xmath66 , @xmath67 is the mean value of @xmath68 at @xmath69 for the dust particles , and @xmath70 is the mean albedo at @xmath69 for these particles .
we adopt @xmath71 ( white 1979 ) . the solid angle across which this light is scattered
is @xmath72 .
consequently , @xmath73 using equation ( [ delay ] ) , we rewrite equation ( [ prob ] ) as a function of @xmath48 and @xmath21 : @xmath74 where @xmath10 is a function of @xmath48 and @xmath21 given by equation ( [ delay ] ) . let @xmath75 be the fluence of the pulse of light at @xmath49 . ignoring light that multiply scatters into the line of sight , the light curve of the dust echo
is then given by : @xmath76}}{(5 - 4\cos{\theta})^{3/2}}d\left(\frac{r ' } { r}\right ) f_\nu dt , \label{pulse}\end{aligned}\ ] ] where @xmath77 $ \cr 1 & $ [ r(\pi - f\theta_{jet } ) < r ' < r(f\theta_{jet})]$ \cr 0 & $ [ r ' > r(f\theta_{jet})]$ } , \label{p}\end{aligned}\ ] ] @xmath78 $ \cr r(\pi - f\theta_{jet } ) & $ [ t >
t(\pi - f\theta_{jet})]$},\ ] ] and @xmath79 $ \cr \pi - f\theta_{jet } & $ [ t >
t(\pi - f\theta_{jet})]$},\ ] ] and where @xmath80 @xmath81,\ ] ] and @xmath82 if @xmath59 , or if @xmath83 and @xmath84 [ equations ( [ p ] ) and ( [ tau ] ) , respectively ] , then equation ( [ pulse ] ) is approximately given by : @xmath85 & $ ( t_{sc } < t < t_{\tau})$ \cr a\left(\frac{r}{r}\right)^{-2 } \left(\frac{t_{\tau}}{t_{sc}}\right)^{-3/2 } \left(\frac{t}{t_{\tau}}\right)^{-7/4 } \left[4 - 3\left(\frac{t_{\tau}}{t_{sc}}\right)^{-1/2 } \left(\frac{t}{t_{\tau}}\right)^{-1/4}\right ] & $ ( t_{sc } < t_{\tau } < t)$ \cr a\left(\frac{r}{r}\right)^{-2}\left(\frac{t}{t_{\tau}}\right)^{-1 } & $ ( t_{\tau } < t$ and $ t_{\tau } \ll t_{sc})$ } , \label{limlc}\ ] ] where @xmath86 @xmath87 is the time when the scattering cross section begins to fall off due to scattering at greater angles , and @xmath88 is the time when the optical depth grows greater by 1 due to the greater path length at greater scattering angles . this interaction of the dust and light geometries is depicted in the top panel of figure 3 .
we plot example light curves for different values of @xmath89 , @xmath90 , and @xmath91 in the top panels of figures 4 7 ; however , consider the following limiting cases : @xmath92 ( figures 4 and 5 , top panels ) : if @xmath93 , then @xmath94 .
consequently , @xmath95 from @xmath96 until @xmath97 , after which the light curve slowly rolls over to @xmath98 .
if @xmath99 , then this continues until @xmath100 , after which @xmath101 . if @xmath102 , then this continues until @xmath103 , after which the light curve slowly rolls over to @xmath104 until after @xmath105
after this time , @xmath101 .
if @xmath106 , then @xmath107 is initially less than @xmath108 .
consequently , @xmath95 from @xmath109 until @xmath103 , after which @xmath110 until after @xmath105 .
after this time , @xmath101 .
@xmath111 ( figures 6 and 7 , top panels ) : the light curve has the same dependence on time as in the @xmath59 case except that it does not `` turn on '' until @xmath112 . also , since the dust is destroyed along the line of sight , the pulse of light is a factor of @xmath113 brighter than in the @xmath59 case .
consequently , the light curve , which is normalized to the fluence of the pulse of light , is a factor of @xmath114 fainter than in the @xmath59 case .
let @xmath115^{\alpha_{\nu(1+z)}}$ ] . the value of @xmath116 is canonically assumed to be 1 ; however , the true value is closer to @xmath117 at nir values of @xmath69 , and levels off to @xmath118 in the uv ( reichart 2000 ) .
the spectral index of the dust echo at @xmath49 as a function of @xmath21 is then given by : @xmath119 & $ ( t < t_{sc}$ and $ t < t_{\tau})$ \cr a_{\nu } + \alpha_{\nu(1+z)}\left[1-\frac{1}{10}\frac{8 - 3\left(\frac{t}{t_{sc}}\right)^{-5/ 2}}{4 - 3\left(\frac{t}{t_{sc}}\right)^{-1/2}}\frac{t}{t_{\tau}}-f\tau_{\nu ( 1+z)}\right ] & $ ( t_{sc } < t < t_{\tau})$ \cr
a_{\nu } + \alpha_{\nu(1+z ) } \left[1-\frac{1}{10}\frac{8 - 3 \left(\frac{t_{\tau}}{t_{sc}}\right)^{-5/2 } \left(\frac{t}{t_{\tau}}\right)^{-5/4}}{4 - 3 \left(\frac{t_{\tau}}{t_{sc}}\right)^{-1/2 } \left(\frac{t}{t_{\tau}}\right)^{-1/4}}-f\tau_{\nu ( 1+z)}\right ] & $ ( t_{sc } < t_{\tau } < t)$ \cr a_{\nu } + \alpha_{\nu(1+z)}\left[\frac{1}{2}-f\tau_{\nu ( 1+z)}\right ] & $ ( t_{\tau } < t$ and $ t_{\tau } \ll t_{sc})$ \cr } , \label{limspec}\end{aligned}\ ] ] where @xmath120 is the spectral index of the pulse of light .
this approximation holds if @xmath59 , or if @xmath61 and @xmath121 .
we plot how the spectral index of the dust echo differs from the spectral index of the pulse of light as a function of time for different values of @xmath89 , @xmath90 , and @xmath122 in the bottom panels of figures 4 7 ; however , consider the following limiting cases : @xmath92 ( figures 4 and 5 , bottom panels ) : since blue light scatters with a higher probability than red light , and since both the dust echo and the pulse of light are extinguished equally at @xmath96 , the dust echo is initially bluer than the pulse of light , independent of the value of @xmath91 : @xmath123 . if @xmath124 , then @xmath94 .
consequently , the dust echo reddens to @xmath125 at @xmath97 .
if @xmath126 , then it grows redder to @xmath127 at @xmath100 , after which it grows bluer and returns to its original color . if @xmath128 , then it grows redder to @xmath129 at @xmath103 , after which it grows bluer to a color of @xmath130 until after @xmath105 . after this time , it grows bluer still and returns to its original color .
if @xmath131 , then @xmath107 is initially less than @xmath108 .
consequently , the dust echo reddens to @xmath132 at @xmath133 , after which its color does not change until after @xmath134 . after this time , its color grows bluer and returns to its original color .
@xmath111 ( figures 6 and 7 , bottom panels ) : the spectral index of the dust echo has the same dependence on time as in the @xmath59 case , but is redder than the pulse of light by an additional @xmath91 : since the dust is destroyed along the line of sight , the pulse of light is not similarly reddened . in any case , if @xmath124 , then the spectral index of the dust echo does not change in time by more than @xmath135 , and if @xmath136 , then it does not change at all .
if @xmath137 , then the dust echo is probably too extinguished to be observed ; consequently , we do not explore this case in detail in this paper .
we now model the more general case of the dust echo of a collimated pulse of light that is emitted from a distance that is much less than @xmath1 .
let the pulse of light be emitted between @xmath138 and @xmath139 ( we assume bipolar emission ) , where @xmath140 .
the only difference from the isotropic case is that @xmath141 if @xmath142 .
consequently , the light curve is again given by equation ( [ pulse ] ) , but with @xmath77 $ \cr 1 & $ [ r(\pi - f\theta_{jet } ) < r ' < r(\pi-\theta_{col})]$ \cr 0 & $ [ r(\pi-\theta_{col } ) < r ' < r(\theta_{col})]$ \cr 1 & $ [ r(\theta_{col } ) < r ' < r(f\theta_{jet})]$ \cr 0 & $ [ r ' > r(f\theta_{jet})]$ } , \label{pr}\end{aligned}\ ] ] @xmath143 $ \cr r(\theta_{col } ) & $ [ t(\theta_{col } ) < t < t(\pi-\theta_{col})]$ \cr r & $ [ t(\pi-\theta_{col } ) < t < t(\pi - f\theta_{jet})]$ \cr r(\pi - f\theta_{jet } ) & $ [ t >
t(\pi - f\theta_{jet})]$ } , \label{r}\ ] ] and @xmath144 $ \cr \theta_{col } & $ [ t(\theta_{col } ) < t < t(\pi-\theta_{col})]$ \cr \theta(r ) & $ [ t(\pi-\theta_{col } ) < t < t(\pi - f\theta_{jet})]$ \cr \pi - f\theta_{jet } & $ [ t > t(\pi - f\theta_{jet})]$}. \label{theta}\ ] ] this interaction of the dust and light geometries is depicted in the bottom panel of figure 3 .
we plot example light curves and color histories for different values of @xmath26 and @xmath91 in figures 8 and 9 , but basically , they are the same as in the isotropic case until @xmath145 . after this time , the light curve falls off as @xmath101 if @xmath146 , as @xmath104 if @xmath147 and @xmath148 [ @xmath149 , and as @xmath150 if @xmath147 and @xmath151 [ @xmath152 , because the observed light no longer includes light that scattered close to the evacuated region where the density ( and the optical depth ) is highest .
a direct consequence of this is that the dust echo grows bluer .
the light curve and color history partially recover their isotropic characteristics after @xmath153 , when back - scattered light from the far side of the evacuated region can first reach the observer ; however , the light curve falls off rapidly after @xmath100 .
we now model the even more general case of the dust echo of collimated light that is emitted as a function of time ( i.e. , an afterglow ) from a distance that is much less than @xmath1 .
since the afterglow emits light continuously , at any time @xmath21 after the burst , we receive scattered light that was emitted by the afterglow at all prior observer - frame times : @xmath154 .
consequently , the dust echo light curve is given by integrating equation ( [ pulse ] ) over these times : @xmath155}}{(5 - 4\cos{\theta})^{3/2 } } d\left(\frac{r'}{r}\right)f_\nu(t')dt',\ ] ] where @xmath156 , @xmath48 , and @xmath10 are given by equations ( [ pr ] ) , ( [ r ] ) , and ( [ theta ] ) , respectively , but with @xmath157 @xmath158,\ ] ] and @xmath159 @xmath160 is the light curve of the afterglow , and @xmath26 is now a function of time that we approximate by : @xmath161 let @xmath162 be the time when @xmath163 .
then from equation ( [ gamma ] ) , @xmath164 and @xmath165^{-1}\left(\frac{e\theta _
{ jet}^2}{10^{52}\rm{\,erg}}\right)\left(\frac{\theta_{jet}}{0.1}\right)^2{\rm\,h r}. \label{tjet}\ ] ] the value of @xmath166 after @xmath162 ( e.g. , sari , piran , & halpern 1999 ) ; this is true in either a @xmath167 or a @xmath168 circumburst medium .
finally , we model the light curve of the afterglow . in this paper , we simply take : @xmath169 which is typical of observed afterglows after @xmath170 , where @xmath170 may be as little as minutes or as much as hours after the burst ( e.g. , sari , piran , & narayan 1998 ) . before @xmath170 , the light curve can depend on time in a variety of ways ( e.g. , sari , piran , & narayan 1998 ) ; however , we do not model it any more carefully than a constant . instead , we show that the value of @xmath170 has little effect on the light curve and color history of the dust echo ( see figure 13 ) .
we plot example light curves and color histories for different values of @xmath15 , @xmath162 , and @xmath170 in figures 10 13 .
the results are similar to the case of a collimated pulse of light with collimation half - angle given by the value of @xmath26 when most of the afterglow light is emitted ; i.e. , by the value of @xmath15 , since most of the afterglow light is emitted between @xmath171 , and @xmath172 during these times . in figure 10
, we take @xmath61 , @xmath173 pc , @xmath174 , @xmath175 hr , @xmath176 , @xmath177 , @xmath178 , and @xmath179 , and @xmath180 , 1.4 , 5.6 , and 22 dy , respectively . these values of @xmath162 are given by equation ( [ tjet ] ) with @xmath181 , @xmath1821 pc@xmath183 @xmath184 , @xmath185 erg , and these values of @xmath15 .
clearly , for these values of the parameters , the dust echo of the afterglow is substantially fainter than the afterglow itself .
changing the value of @xmath91 only makes the dust echo fainter relative to the afterglow : decreasing @xmath91 decreases the amount of light scattered into the line of sight , and increasing @xmath122 increases how much the dust echo is extinguished , but does not increase how much the afterglow is extinguished since @xmath61 .
the dust echo can be made brighter than the afterglow if @xmath59 and @xmath122 is substantially increased , but in this scenario , both the afterglow and the dust echo are substantially extinguished and consequently difficult to detect .
this leaves only three scenarios in which a dust echo can be brighter than an afterglow and detectable : ( 1 ) the value of @xmath90 is substantially greater than 0.1 pc , in which case the dust echo peaks at a substantially later time ( but with a substantially lower peak brightness ) , after the light curve of the afterglow has rolled over and gotten out of the way " of the dust echo ( figure 11 ) ; ( 2 ) the value of @xmath1861 pc@xmath187^{-1}e\theta_{jet}^2 $ ] is substantially less than the value assumed above , in which case the light curve of the afterglow rolls over at a substantially earlier time , again getting out of the way " of the dust echo ( figure 12 ) ; and ( 3 ) the dust echo is dominated by scattered light not from the afterglow , but rather , from the optical flash and/or optical light from the burst itself .
this last scenario can be modeled simply as a collimated pulse of light ( 2.2 ) with @xmath188 .
we show that the choice of @xmath170 has little effect on the light curve and color history of the dust echo of the afterglow in figure 13 .
one aspect of the dust and light geometry that we do not model in this paper is that the afterglow is not emitted from the center of the evacuated region , but rather from the head of the jet , which is moving away from the center ( toward the observer ) at relativistic speeds .
since @xmath189 ( 1 ) , the head of the jet reaches the radius @xmath48 at which most of the scattering is occurring at time @xmath190 our model holds as long as most of the light is emitted before some reasonable fraction of @xmath191 . for afterglow light curves like the one described in equation ( [ afterglow ] ) ,
most of the afterglow light is emitted before @xmath162 , by a factor of a few .
consequently , our model only holds as long as @xmath192 , or @xmath193^{-1}\left(\frac{e\theta_{jet } ^2}{10^{52}\rm{\,erg}}\right){\rm\,pc}.\ ] ] however , if the dust echo is dominated by scattered light from the optical flash and/or optical light from the burst , as opposed to light from the afterglow , then the light is obviously emitted long before @xmath191 , and consequently , the model holds for substantially smaller values of @xmath48 as well .
the dust echo from absorbed and thermally re - emitted light is easier to model than the dust echo from scattered light : unlike equation ( [ scat ] ) , the differential cross section for dust absorption has no dependence on @xmath10 : @xmath194 consequently , the light curve of the dust echo of the afterglow is given by @xmath195}d\left(\frac{r'}{r}\right)f_{uv}(t')dt',\ ] ] where @xmath196 is the bolometric light curve of the dust echo , and @xmath197 is the light curve of the afterglow , integrated over uv values of @xmath69 .
if the re - radiating dust particles had a common temperature @xmath198 , the spectrum of the dust echo would be thermal .
however , the temperature of the dust particles decreases with radius as @xmath199 ( waxman & draine 2000 ) ; consequently , the spectrum should fall off more slowly than a blackbody redward of the spectral peak .
since dust is rapidly sublimated at temperatures @xmath200 k , the spectral peak should occur only at frequencies @xmath201 hz . for a burst at @xmath202 , this is in the k band .
we plot example light curves for different values of @xmath15 in figure 14 .
however , consider the limiting cases for an isotropic pulse of light .
the thermal dust echo has the same limiting cases as the scattered dust echo ( e.g. , equation [ limlc ] ) , except that @xmath203 .
consequently , if @xmath204 , then @xmath205 until @xmath145 , after which @xmath206 until @xmath103 . after this time , @xmath207 . if @xmath208 , then @xmath209 until @xmath103 , after which @xmath210 .
first of all , the late - time afterglow of grb 970228 can not be explained by a thermal dust echo , because its spectral flux distribution ( figure 1 ) peaks between the v and b bands in the source frame , not in or beyond the k band ( 3 ) .
secondly , the late - time afterglow of grb 970228 can not be explained by a dust echo due to scattered afterglow light , because its light curve ( e.g. , reichart 1999 ) does not roll over at sufficiently early times for the dust echo to be able to outshine the afterglow when the second component to the light curve peaks , about 20 30 days after the burst ( 2.3 ) .
this leaves only the possibility of a dust echo due to scattered light from the optical flash and/or optical light from the burst itself . to investigate this possibility ,
we have fitted the spectral flux distribution of figure 1 with two power laws , one to the reddest three points and one to the bluest three points .
we find that the spectral index redward of and including the source - frame v band is @xmath211 , and that the spectral index blueward of and including the source - frame b band is @xmath212 .
the corresponding optical depths as a function of the spectral index of the pre - scattered , unextinguished light are given by equation ( [ limspec ] ) with @xmath89 set to 1 .
taking @xmath213 and @xmath214 ( 2.1 ) , we find that for an intrinsic spectral index of -1/2 , @xmath215 and @xmath216 , and for an intrinsic spectral index of 1/3 , @xmath217 and @xmath218 .
intrinsic spectral indexes of -1/2 and 1/3 are typical of what one expects for the optical flash blueward and redward of the synchrotron peak , respectively , which one expects to be at optical / nir wavelengths during the optical flash ( e.g. , sari & piran 1999 ) . in the case of optical light from the burst itself , one expects a spectral index of 1/3 across the optical and nir bands ( e.g. , sari & piran 1999 ) .
consequently , in either interpretation , the value of @xmath219 is borderline consistent with being physical ( i.e. , non - negative ) , but the rapid transition between @xmath220 or @xmath221 to @xmath222 or @xmath223 or @xmath224 is unphysical .
consequently , we find that the late - time afterglow of grb 970228 can not be explained by a dust echo . unfortunately , the late - time afterglow of grb 980326 was not sufficiently well sampled to distinguish between a supernova and a dust echo explanation
we have modeled and computed dust echo light curves and spectra for dust and light geometries that are more realistic than had previously been considered . in particular , bursts are very likely to be collimated ( on both theory and energetics grounds ) , and dust may be destroyed by the progenitor , as well as the burst , optical flash , and/or afterglow . also , the optical flash and/or optical light from the burst itself may result in dust echoes . consequently
, dust echoes may be a way to measure the brightness of the optical flash and/or the optical brightness of the burst .
we find that dust echo spectra can be used to distinguish different types of dust echoes from each other , and from supernovae .
for example , thermal dust echoes peak in the source - frame k band , while supernovae peak in the source - frame v or b bands . in the absence of extinction ( e.g. , in the source - frame nir )
, dust echoes due to scattered light are bluer than the pre - scattered light by about 1 1.6 in the spectral index , and consequently , can peak in the source - frame v or b bands , or other nearby bands , depending on the amount of extinction .
however , dust echoes due to scattered light do not peak as sharply as do thermal dust echoes or supernovae .
it is for these reasons that we find that the late - time afterglow of grb 970228 can not be explained by a dust echo .
furthermore , dust echo light curves can often be used to distinguish dust echoes from supernovae .
while dust echoes can peak about 20(1+z ) days after a burst , they also can peak substantially earlier and substantially later .
however , supernovae can peak only at this time .
finally , dust echoes can be used to probe the circumburst environment , and the effects of the progenitor , burst , optical flash , and/or afterglow on this environment .
for example , if the dust echo light curve brightens , levels off for a long time , and then fades , then @xmath59 and dust had not ( yet ) been destroyed along the line of sight .
if the dust echo light curve brightens and fades without leveling off , then @xmath61 and dust had been destroyed along the line of sight .
the time at which the dust echo light curve begins to fade gives you ( most likely ) a degenerate combination of @xmath1 and @xmath15 .
the color of the dust echo , particularly near peak , gives you @xmath91 , and consequently a degenerate combination of @xmath1 and @xmath1821 pc@xmath225 .
the time at which the afterglow light curve rolls over due to lateral expansion of the jet gives you a degenerate combination of @xmath15 and @xmath1821 pc@xmath225 .
consequently , @xmath1 , @xmath15 , and @xmath1821 pc@xmath225 can be separated and solved for .
furthermore , the brightness of the dust echo , particularly near peak , gives you a degenerate combination of @xmath1 , @xmath91 , and the fluence of the pre - scattered , unextinguished light ; consequently , this fluence can be solved for as well .
also , by comparing @xmath91 to the same quantity measured from the afterglow , and to absorption measured in x rays , one can further probe dust destruction along the line of sight , and the metals - to - dust ratio off the line of sight where dust is not destroyed , respectively .
support for this work was provided by nasa through the hubble fellowship grant # hst - sf-01133.01-a from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
i am very grateful to reem sari for invaluable discussions about the dust and light geometries .
i am also grateful to don lamb for discussions about dust echoes in general .
reichart , d. e. , castander , f. j. , & lamb , d. q. 2000 , in cosmic explosions : tenth astrophysics conference , aip conference proceedings 522 , eds . s. s. holt and w. m. zhang ( melville , new york : aip ) , 253 reichart , d. e. , lamb , d. q. , & castander , f. j. 2000 , in gamma - ray bursts : 5th huntsville symposium , aip conference proceedings 526 , eds .
kippen , r. s. mallozzi , and g. j. fishman ( melville , new york : aip ) , 414 | the relative brightening and reddening of the optical afterglows of grb 970228 and grb 980326 around 20 30 days after these bursts have been attributed to supernovae , which are red ( blueward of their spectral peak ) , and which peak in time after @xmath0 days .
however , this direct evidence for a grb / sn connection has recently been challenged .
it has been suggested that the late afterglows of these bursts can be explained by dust echoes , of which we consider two cases : ( 1 ) the scattering of light from the afterglow ( the forward shock ) , the optical flash ( the reverse shock ) , and/or optical light from the burst itself by a shell of dust at a radius @xmath1 from the progenitor , and ( 2 ) the absorption and thermal re - emission of this light by this shell of dust . in this paper , we model and compute dust echo light curves and spectra for both cases .
although the late - time afterglow of grb 980326 was not sufficiently well sampled to rule out a dust echo description , we find that the late - time afterglow of grb 970228 can not be explained by a dust echo . |
the concepts and methods of effective field theory ( eft ) @xcite have recently elucidated the successful nuclear phenomenology of relativistic field theories of hadrons , called quantum hadrodynamics ( qhd ) @xcite .
the eft framework shows how qhd models can be consistent with the symmetries of quantum chromodynamics ( qcd ) and can be extended to accurately reproduce its low - energy features .
the eft perspective accounts for the success of relativistic mean - field models and provides an expansion scheme at the mean - field level and for going beyond it @xcite .
one of the most compelling features of qhd mean - field phenomenology is the reproduction of spin - orbit splittings in finite nuclei after fitting only to the equilibrium properties of infinite nuclear matter .
this successful result occurs when the velocity dependence of the equivalent central potential that leads to saturation arises primarily because of a reduced nucleon effective mass .
the spin - orbit interaction in nuclei is then also specified by the structure of the four - component dirac spinors @xcite .
does this connection survive in the extended eft framework ? here
we re - examine the nature of the spin - orbit force in chiral effective field theories of nuclei . the strong correlation in qhd models between the effective mass @xmath1 of the nucleon in equilibrium nuclear matter and
the strength of the spin - orbit force in nuclei has been well documented @xcite .
typical values of @xmath2 in successful models are roughly 0.6 , which yield spin - orbit splittings close to experimental values ( e.g. , 6mev for @xmath3 states in @xmath4o ) . in ref .
@xcite , a wider class of models , which included variations of the linear sigma model as well as all of the standard qhd models used phenomenologically , was studied systematically .
accurate reproductions of observables in finite nuclei , including the spin - orbit splitting in particular , tightly constrained the value of @xmath2 to the range @xmath5 . in ref .
@xcite , an effective hadronic lagrangian consistent with the symmetries of qcd and intended for application to finite density systems was constructed .
this involved a different strategy than in conventional qhd phenomenology .
rather than maximizing predictivity by minimizing the free parameters ( as in renormalizable models ) , the goal was to test a systematic expansion for low - energy observables , which included the effects of hadron compositeness and the constraints of chiral symmetry .
the degrees of freedom are ( valence ) nucleons , pions , and the low - lying non - goldstone bosons . a scalar - isoscalar field with a mass of roughly 500mev was also included to simulate the exchange between nucleons of two correlated pions in this channel .
the lagrangian was expanded in powers of the fields and their derivatives , with the terms organized using georgi s `` naive dimensional analysis '' ( nda ) @xcite .
the coefficient of each term is written as a combination of the pion - decay constant @xmath6mev and a larger scale @xmath7gev times a dimensionless coupling constant .
the effective lagrangian is said to be `` natural '' if these dimensionless coefficients are of order unity .
the result is a faithful representation of low - energy , non - strange qcd , as long as all nonredundant terms consistent with symmetries are included .
in addition , the mean - field framework provides a means of approximately including higher - order many - body and loop effects , since the scalar and vector meson fields play the role of auxiliary kohn
sham potentials in relativistic density functional theory @xcite
. fits to nuclear properties at the mean - field level showed that the effective lagrangian could be truncated at the first few powers of the fields and their derivatives , with natural [ @xmath8 coefficients for each term .
of the two full parameter sets identified , one ( model g2 ) provided an excellent fit , including spin - orbit splittings , with a nuclear matter value of @xmath9 .
more recently , new parameters sets with excellent fits have been found with @xmath10 @xcite .
an analogous study was made of relativistic `` point - coupling '' ( pc ) models in ref .
@xcite . in these models ,
non - goldstone mesons are not included explicitly .
one instead has an expansion of the nucleon scalar and vector potentials in powers and derivatives of local nucleon scalar and vector densities . as in ref .
@xcite , an effective lagrangian consistent with chiral symmetry was constructed and fits to nuclear observables were made to test naturalness .
excellent fits were found with unprecedented values of @xmath2 , as high as @xmath11 at nuclear matter equilibrium .
our purpose here is to examine the nature of the nuclear spin - orbit force in these generalized models .
we find that the most important new feature affecting spin - orbit splittings in the lagrangians of refs .
@xcite and @xcite is the inclusion of a _
tensor _ coupling of the omega field to the nucleon ( or its point - coupling analog ) .
because the effective lagrangian includes a derivative expansion , the tensor term appears at a relatively low order .
none of the other new terms has a comparable impact on the spin - orbit strength .
we find a trade - off between the size of the scalar potential ( or equivalently , the effective mass @xmath12 ) and the size of the tensor coupling .
the existence of such a trade - off has been noted in previous studies of the isoscalar tensor coupling in qhd models , but good fits to nuclei were found only with relatively small values of the coupling and thus small @xmath13 @xcite .
furthermore , this coupling is usually taken to be zero in one - boson - exchange potentials @xcite or limited to small values due to constraints from _ free _ nucleon form factors and the assumption of vector meson dominance @xcite . however , as an effective coupling in nuclei , which might absorb higher - order many - body effects at the mean - field level , the isoscalar tensor coupling could be much larger than these constraints dictate and still be of natural size . in fits made using more complete meson and point - coupling models , larger ( but still natural ) values
are favored , particularly in the point - coupling models @xcite . in this work ,
we show that the increase in @xmath2 in these models is well accounted for by the contribution from the tensor coupling .
first , we consider two - component reductions of the models , which are reviewed in sects . [
sect : two ] and [ sect : three ] .
the spin - orbit potential in the two - component single - particle hamiltonian manifests the influence of both @xmath2 and the tensor coupling , and a simple local - density approximation leads to quantitative predictions for the spin - orbit strength in nuclei .
we find a strong correlation between the size of the isoscalar tensor coupling and the increase of @xmath14 above 0.6 .
this discussion applies to a very general class of qhd models , including the quark - meson - coupling ( qmc ) model @xcite and the zimnyi - moszkowski ( zm ) model @xcite ( both of which are found to be deficient ) .
a second approach is to transform the nucleon fields in models with tensor couplings .
a feature of effective lagrangians is the freedom to perform field redefinitions without changing the physics ( if all nonredundant terms have been included ) .
it is not possible to completely transform away the tensor interaction in a qhd effective lagrangian . to study the consequences for nucleon nucleus scattering , however
, we can perform an energy - dependent transformation to replace the potential in the dirac equation with an equivalent potential ( that is , one that generates the same scattering observables ) with no tensor term @xcite .
the new scalar and vector potentials can then be compared to empirical optical potentials .
it is well known that in relativistic models applied to hypernuclei , the isoscalar tensor coupling of the omega to the hyperon plays a critical role in _ reducing _ the strength of the hyperon spin - orbit potential @xcite .
quark - model arguments imply that the anomalous coupling is equal to ( minus ) the vector coupling , which leads to a strong cancellation between contributions to the spin - orbit force and therefore small splittings , as needed phenomenologically . to our knowledge
, there is no corresponding implication for the tensor coupling of the omega to nucleons .
in this section , we explore the relationship between the spin - orbit force and the nucleon potentials in models without tensor potentials .
the connection between the spin - orbit splittings and the scalar and vector potentials is most easily seen by making a reduction to a two - component spin - orbit hamiltonian .
we can write the isoscalar part of the dirac single - particle hamiltonian ( for spherical nuclei ) as @xmath15 where @xmath16 is the nucleon mass , @xmath17 is the scalar potential , and @xmath18 is the vector potential . a tensor potential is considered in the next section .
note that we need not assume that @xmath17 is simply proportional to a scalar meson field @xmath19 .
in fact , @xmath17 could be proportional to @xmath19 ( as in conventional qhd models ) , or could be expressed as a sum of scalar and vector densities ( as in relativistic point - couplings models ) , or could be a nonlinear function of @xmath19 ( as in the qmc and zm models ) . to leading - order in @xmath20 , the spin - orbit term in a foldy
wouthuysen reduction of @xmath21 takes the form @xcite
@xmath22 \ldots \ \ .
\label{eq : hfw}\ ] ] the expansion is in powers of momenta ( i.e. , derivatives ) and potentials divided by the nucleon mass . the qualitative impact of the potentials on the spin - orbit force is clear from this expression and leads to some basic observations common to all successful models : 1 .
successful models are natural @xcite , which implies that the potentials closely follow the baryon density .
this result follows from the equations for the potentials and the close relationship between scalar and vector densities , @xmath23 where @xmath24 and @xmath25 are @xmath26 .
the precise expressions for the constants @xmath24 and @xmath25 will depend on the model ; for example , in conventional qhd models @xmath27 and @xmath28 . in every case , however , naturalness implies that the corrections are order @xmath29 @xmath30 smaller than the leading term in the interior of nuclei and smaller still on the surface .
furthermore , the scalar density and baryon density are equal to within 5 to 10% at ordinary nuclear densities .
thus a simple local density approximation for @xmath31 and @xmath32 should be quite reasonable : @xmath33 where @xmath34 and @xmath35 are the values of the potentials in nuclear matter at equilibrium density @xmath36 .
the remaining factor from the expectation value of @xmath37 will be universal for different models , as long as they accurately reproduce nuclear density profiles .
3 . the hugenholtz
van hove theorem relates @xmath34 and @xmath35 through the values of the chemical potential @xmath38 and the equilibrium fermi momentum @xmath39 : @xmath40 with @xmath41 .
since @xmath38 and @xmath39 are the same in all successful models to very good approximation @xcite , the dependence on @xmath35 can be eliminated in favor of @xmath34 .
4 . therefore , the strength of the spin - orbit interaction ( [ eq : hfw ] ) and the resulting magnitude of the splitting in nuclei should be , to a good approximation , directly proportional to the size of @xmath34 , the scalar potential in nuclear matter .
if this analysis is quantitatively valid , we would expect to find that plots of spin - orbit splittings in nuclei versus equilibrium @xmath2 for different models that reproduce nuclear density distributions would lie roughly on a straight line .
such plots are shown in figs .
[ fig : oxy][fig : pb ] for splittings in @xmath4o , @xmath42ca , and @xmath43pb using a large number of models .
the open circles are for qhd
i models @xcite ( the original linear walecka model plus cubic and quartic scalar self - interactions ) .
the circles marked with crosses also include one - loop vacuum corrections ; this is essentially equivalent to adding a particular quintic scalar self - interaction term @xcite .
each qhd i model was fit to a standard set of nuclear matter saturation conditions ( equilibrium density , binding energy , and symmetry energy ) and the charge radius of @xmath42ca @xcite .
the diamonds are for qhd models from ref .
@xcite , squares are for point - coupling models from refs .
@xcite and @xcite , and triangles are for qmc and zm models from refs .
@xcite and @xcite .
( the models with filled symbols contain tensor interactions and are discussed in sect .
[ sect : three ] . )
it is evident that there is a universal curve , but it is not a straight line except for @xmath1 very close to @xmath16 ( i.e. , small @xmath17 ) , in contrast to the conclusion reached in point ( 4 ) , above .
for a more _ quantitative _ local - density estimate of the spin - orbit splitting , we must go beyond leading order in the foldy
wouthuysen reduction , which is not an efficient expansion at finite density , even though @xmath44 is quite small .
the problem is that the foldy
wouthuysen expansion also involves @xmath29 and @xmath45 , which are less than unity but not negligible .
reinhard has proposed a more useful expansion , which counts powers of a ( relatively ) small nucleon velocity @xmath46 @xcite .
specifically , @xmath47 % ( \phi - w)/m \qquad & \longrightarrow & \qquad { \rm order\ } v^2 \ , \nonumber\\[3pt ] % \mstar / m = ( m-\phi)/m \qquad & \longrightarrow & \qquad { \rm order\ } 1 \ , \nonumber\\[3pt ] % \displaystyle { { \bbox{\sigma\cdot p}\over m } \ , { m\over m+e_\alpha-\phi - w}\ , { \bbox{\sigma\cdot p}\over m } } \qquad & \longrightarrow & \qquad { \rm order\ } v^2 \ .
\label{eq : orders}\end{aligned}\ ] ] here @xmath48 is the energy eigenvalue for state @xmath49 , including the rest mass . since nuclei are weakly bound , the kinetic energy proportional to @xmath50 balances the net potential energy @xmath51 , which leads to the first two results .
but while the difference of the potentials is counted as a small quantity , the potentials themselves are not . on the other hand , they are still small enough compared to @xmath16 to count @xmath52 as order 1 , which explains the final two counting rules .
reinhard s procedure constructs order - by - order in @xmath50 a single - particle hamiltonian @xmath53 that acts on two - component ( normalized ) wavefunctions @xcite .
the latter are related to the upper components of the dirac four - component wavefunctions by a normalization operator .
he equates the expectation value @xmath54 calculated with the four - component wavefunctions with the corresponding expectation value @xmath55 calculated with the two - component wavefunctions .
this condition specifies @xmath53 at each order in @xmath50 , up to corrections of the next order .
we have considered two possibilities for the spin - orbit part of @xmath53 to order @xmath50 , @xmath56 \ldots = \biggl [ { 1\over 4\mbar^{2}}{1 \over r } \left ( { d \phi \over dr } + { d w \over dr } \right ) + o(v^4)\biggr ] \ldots \ .
\label{eq : hreinhard}\ ] ] where @xmath57 is defined as @xmath58 and @xmath59 \ldots = \biggl [ { 1\over 2\mstar{}^{2}}{1 \over r } { d \phi \over dr } + o(v^4)\biggr ] \ldots \ .
\label{eq : hpreinhard}\ ] ] we have also constructed the @xmath60 corrections to each . by taking explicit matrix elements of @xmath61 and @xmath62 using two - component wavefunctions for spin - orbit pairs and comparing to the energy eigenvalues for the corresponding dirac wavefunctions , we can determine the accuracy of these truncations .
we find that @xmath61 systematically underpredicts the splittings by roughly 5% while @xmath62 overpredicts the splittings by about 15% .
the predictions using the @xmath60 hamiltonians are quite similar in the two cases and are within 3% of the dirac splittings .
we will use @xmath61 to @xmath63 in the sequel .
we can estimate spin - orbit splittings as follows .
we use the local - density expressions from eqs .
( [ eq : philda ] ) and ( [ eq : wlda ] ) in the numerator of eq .
( [ eq : hreinhard ] ) and replace @xmath64 in the denominator by an average value @xmath65 we use an interpolated value of @xmath12 : @xmath66 and eq .
( [ eq : hvh ] ) to define @xmath67 given @xmath68
. one would expect @xmath69 to be given roughly by @xmath70 , where @xmath71 is an average density seen by the nucleons making up the spin - orbit pair . in @xmath42ca ,
this would imply @xmath72 for the @xmath3 states and @xmath73 for the @xmath74 states .
so the energy splitting takes the form @xmath75 where the constant @xmath76 depends on the average of @xmath77 over the nucleus ( which is essentially the same for the members of a spin - orbit pair ) , and @xmath78 is the difference in the expectation values of @xmath79 for the spin - orbit pair .
the nuclear matter value @xmath80 follows from @xmath2 using eqs .
( [ eq : mbardef ] ) and ( [ eq : hvh ] ) .
this local - density estimate of the splitting as a function of nuclear matter @xmath81 is shown in figs .
[ fig : oxy][fig : pb ] as a dotted line .
the curves are normalized to the point with the largest value of @xmath81 to absorb the constant @xmath76 and any overall normalization errors .
the least - bound states are well fit with @xmath82 , and the deeply bound states , with @xmath83 .
the local - density estimates account quantitatively for the spin - orbit trend.ca shifts the dotted curve up ( for a smaller radius ) or down ( for a larger radius ) .
that is why the open diamonds tend to be slightly above the curve , since the @xmath42ca radius used is 3.45fm rather than 3.48fm as in the other models . ]
the shaded regions in figs .
[ fig : oxy ] , [ fig : cad ] , and [ fig : pb ] indicate the empirical spin - orbit splittings .
since we do not calculate rearrangement effects , there are intrinsic uncertainties in comparing to experimental splittings .
we have therefore been conservative in defining these regions .
however , the intersection of the universal curve with each region implies that models with effective masses outside the range @xmath5 proposed in ref .
@xcite will not be able to reproduce empirical spin - orbit splittings ( when there is no tensor coupling ) .
this is true of the qmc @xcite and zm @xcite models , which are indicated by triangles in figs .
[ fig : oxy][fig : pb ] .
one of the appealing features of relativistic models is that one can come to a similar conclusion about @xmath2 without considering spin - orbit splittings .
for example , the study by rufa et al .
@xcite of conventional qhd
i models ( i.e. , the linear walecka model plus cubic and quartic scalar meson self - couplings ) revealed that a good fit to bulk nuclear properties _ excluding spin - orbit splittings _ still required @xmath2 to be close to @xmath84 at equilibrium .
this result can be understood in terms of a key feature of relativistic models : the additional saturation mechanism that arises from a lorentz scalar interaction .
the velocity dependence in the interaction due to the scalar potential appears in the energy of nuclear matter ( to leading order in @xmath85 ) as a modification of the kinetic energy per particle : @xmath86 the extra density - dependent repulsion from the @xmath87 factor leads to saturation @xcite ; in successful mean - field models ( which include qhd
i models with @xmath5 ) it is the dominant saturation mechanism . in these cases ,
a fit to the nuclear matter equilibrium density and binding energy alone constrains @xmath2 and the implied velocity dependence ; the four - component structure of the dirac wave functions then _ automatically _ generates an appropriate spin - orbit force in finite nuclei . in contrast , in qhd
i models with larger values of @xmath2 , the empirical equilibrium point is obtained from repulsive three- and four - body contributions that become more and more important as @xmath2 increases . in summary ,
models without isoscalar tensor couplings exhibit a strong correlation between the effective nucleon mass at nuclear matter equilibrium density and the spin - orbit splittings in nuclei .
comparison to experiment reinforces the constraint @xmath5 for successful models of this type .
next we consider the generalization that includes tensor couplings .
we generalize @xmath21 from eq .
( [ eq : hdirac ] ) by adding a tensor potential @xmath88 : @xmath89 a tensor potential of this form arises in a meson model from the tensor coupling of an isoscalar vector meson ( @xmath0 ) to the nucleon . using the conventions of ref .
@xcite , @xmath90 which implies @xmath91 ( note that @xmath92 as defined here corresponds to the combination @xmath93 typically appearing in one - boson - exchange models @xcite . ) the analog of eq .
( [ eq : lvmdt ] ) in the point - coupling model of ref .
@xcite is @xmath94 which implies @xmath95 thus we can make the rough correspondence @xmath96 .
[ note that a spin-2 meson in the ( 1,1 ) representation is described by a _
symmetric _ tensor @xmath97 .
thus it does not couple to @xmath98 and will not produce a tensor potential at the mean - field level . ]
we repeat the procedure that led to eq .
( [ eq : hreinhard ] ) to find @xmath99 \ldots \ , \label{eq : hlstensor}\ ] ] and the result for point - coupling models is found from the substitution @xmath100 now suppose that we want to know how @xmath12 in a model with @xmath101 compares to the value @xmath102 in an `` equivalent '' model with @xmath103 .
if we define `` equivalent '' to mean that they have the same spin - orbit strength , we can use eq .
( [ eq : hlstensor ] ) to make a rough comparison .
let us assume the local - density relations from eqs .
( [ eq : philda ] ) and ( [ eq : wlda ] ) , so that if we equate @xmath104 in the two models , the @xmath105 dependence cancels ( in an averaged sense ) , and we have the condition @xmath106 the analogous condition for the point - coupling models follows after applying eq .
( [ eq : hlstensort ] ) .
we can use eqs .
( [ eq : equiv ] ) and ( [ eq : hvh ] ) to transform the region @xmath107 , which applies to @xmath108 , to the `` equivalent '' region for @xmath109 or @xmath110 .
values of @xmath111 and @xmath112 are defined by eq .
( [ eq : mbarav ] ) in terms of @xmath1 and @xmath113 using eq .
( [ eq : interp ] ) with @xmath83 , and then @xmath67 and @xmath114 follow from eq .
( [ eq : hvh ] ) .
one then inverts the resulting relation for @xmath1 given an input @xmath113 .
the transformed regions are shown in figs .
[ fig : one ] and [ fig : two ] , along with diamonds and squares marking the generalized ( @xmath115 ) models from refs .
@xcite , @xcite and @xcite .
while the shaded regions are not definitive indicators of the relation between the tensor coupling and @xmath12 in successful models , they should provide a reliable estimate of the basic trend .
evidently , the bands are consistent with the models with nonzero tensor coupling that were fit directly to nuclei .
these results are also consistent with new fits in the point - coupling models from ref .
@xcite , in which @xmath116 is held fixed at zero .
the new sets have @xmath117 between 0.60 and 0.64 . as part of an extensive optimization analysis of conventional qhd models ,
rufa et al . considered including an isoscalar tensor coupling @xcite .
they noted the trade - off between the magnitude of the spin - orbit splitting and the size of the scalar field .
however , their chi - squared optimization yielded only a small improvement from the new term , and @xmath116 from optimized fits in refs .
@xcite also yielded only a small improvement in @xmath118 ( a reduction of about one out of forty ) compared to similar models with these parameters fixed at zero . ] and the best value had a small value of @xmath119 .
( note that their conventions differ from ours by a minus sign . )
reinhard concludes that including the tensor would make a difference only in the details , but not in the qualitative picture @xcite . in ref .
@xcite , bir and zimnyi propose adding a tensor coupling of the @xmath0 meson to the nucleon to the lagrangian proposed by zimnyi and moszkowski @xcite , in order to bring the spin - orbit force up to the strength implied by nuclear spin - orbit splittings .
they perform a similar expansion to the one above . however , the value of @xmath2 in the zm model is roughly 0.85 .
figure [ fig : one ] shows that this would require a very large value of @xmath92 , which does not appear to be compatible with good fits to nuclear observables .
what constraints exist on the value of @xmath92 ?
in free space there are constraints from nucleon electromagnetic form factors under the assumption of vector dominance . depending on the form of vector dominance ( i.e. , tensor or vector representation @xcite )
, we obtain a limit on @xmath92 from the isoscalar anomalous moment of the nucleon or from the isoscalar magnetic radius of the nucleon . in either case
, one finds the small value @xmath120 @xcite . from fig .
[ fig : one ] , we see that such a limit implies an unimportant change in @xmath12 from the results with @xmath103 . in one - boson - exchange potentials ,
the _ isovector _ tensor coupling is important , but the isoscalar tensor coupling is usually taken to be zero @xcite .
on the other hand , without further assumptions , the value of @xmath116 in the point - coupling models is not related to the electromagnetic structure of the nucleon and is therefore unconstrained .
it is important to remember , however , that the mean - field parameters obtained from fitting to finite nuclei can be very different from the `` free - space '' tree - level values .
these mean - field parameters must incorporate the effects of many - body physics beyond the hartree level , which could induce strong renormalizations .
we anticipate that isovector couplings will be less affected ( since the free - space values are already of natural size and nuclei have roughly equal numbers of neutrons and protons ) , and therefore argue that fitting isovector parameters using free - space electromagnetic form factors is a good approximation .
however , the situation with isoscalar parameters is unclear .
we might even expect that in the absence of symmetry constraints in the medium , they take natural values .
this would mean that any values shown in figs .
[ fig : one ] or [ fig : two ] are accessible ( here , as in ref .
@xcite , @xmath121 is the natural combination ) . one might hope to constrain the value of @xmath92 from the energy dependence of the optical potential .
a single - particle hamiltonian of the form @xmath21 in eq .
( [ eq : hdirac ] ) implies an energy - dependent optical potential for nucleon - nucleus scattering of the form @xcite @xmath122 if the potential @xmath123 follows the nuclear density , the slope in energy is given by @xmath124 [ see eq .
( [ eq : wlda ] ) ] , and is constrained by low - energy neutron - nucleus scattering to be about @xmath125@xcite . to be consistent with experiment
, @xmath35 must be as large as the values implied by @xmath5 through eq .
( [ eq : hvh ] ) .
the larger values of @xmath1 , and consequently smaller values of @xmath35 , when @xmath109 ( or @xmath110 ) could set a limit on the magnitude of the tensor coupling .
we can not identify @xmath126 directly from @xmath127 because of the tensor potential .
however , we can apply field redefinitions to the nucleon wavefunctions , so that the transformed version of eq .
( [ eq : hdiracp ] ) does not contain a tensor potential .
such a transformation was applied to dirac wavefunctions in ref .
given a solution @xmath128 that satisfies @xmath129 for scattering states ( @xmath130 ) , we define a transformed wavefunction @xmath131 by @xmath132 where @xmath133 in the meson model and @xmath134 in the point - coupling model .
the function @xmath135 goes to zero rapidly outside the nucleus , so scattering observables are unchanged .
the new wavefunction @xmath19 satisfies @xmath136 \phi_\alpha({\bf r } ) = e e^{2u\gammazero } \phi({\bf r } ) \ .
\label{eq : newphi}\ ] ] by rewriting the exponents we can make the identifications : @xmath137 therefore the transformation introduces energy dependence into the effective scalar and vector potentials .
if evaluated in nuclear matter for @xmath138 , the _ predicted _ values of @xmath139 and @xmath140 for the point - coupling models with @xmath141 are consistent with fig .
[ fig : two ] ; in particular , one obtains @xmath142 for each model shown .
the predicted values for the qhd models in fig .
[ fig : one ] , however , are slightly smaller than expected [ @xmath117 is roughly 0.55 ] . if we use @xmath102 and @xmath143 from eqs .
( [ eq : mstara ] ) and ( [ eq : wa ] ) to evaluate the optical potential in eq .
( [ eq : vopt ] ) , we find @xmath144 that is , the additional energy dependence in eqs .
( [ eq : mstara ] ) and ( [ eq : wa ] ) cancels out when evaluating @xmath126 .
thus the energy dependence of the optical potential is simply proportional to the _ untransformed _ vector potential @xmath18 and is _ independent _ of @xmath92 ( or @xmath116 ) .
if the tensor coupling is too large , figures [ fig : one ] and [ fig : two ] imply that @xmath35 will be too small to account for the experimental energy dependence . therefore , smaller values of @xmath92 ( or @xmath116 ) are favored , with an upper limit around unity .
this analysis is not conclusive , however , since the radial dependence of the potentials may invalidate the simple argument leading to the comparison with experiment @xcite .
this issue will be considered in future work .
we examined the role of an isoscalar tensor coupling to the nucleon in chiral effective lagrangians for nuclei .
effective lagrangians must contain all nonredundant terms consistent with the underlying symmetries .
since tensor couplings of the vector mesons to the nucleon ( or their analogs in point - coupling models ) will appear at low order in an expansion in powers of fields ( or densities ) and derivatives , they should be included . in models that ignore such couplings
, there is a strong correlation between the value of the nucleon effective mass @xmath1 at equilibrium nuclear matter density and spin - orbit splittings in nuclei .
in fact , given a value of @xmath1 , one can predict the _ calculated _ splittings of weakly bound levels to an accuracy of roughly 0.3mev .
moreover , to accurately reproduce the _ empirical _ splittings , one requires an equilibrium @xmath14 between 0.58 and 0.64 .
interestingly , similar values of @xmath1 are obtained when fits are performed without including information from these splittings .
when an isoscalar tensor coupling is included and allowed to vary freely , fits to nuclear properties slightly favor natural - sized values of the coupling , which are significantly larger than values deduced from free - space electromagnetic observables .
this is particularly true in point - coupling models .
these larger values of the coupling modify the simple relationship between @xmath1 and the spin - orbit splittings noted above ; however , it is possible to estimate the contribution of the tensor interaction to the spin - orbit force and to show that the larger values of @xmath1 obtained in nuclear matter are consistent with the observed splittings .
we emphasize that the small size of the isoscalar tensor coupling deduced from free - space electromagnetic data is not incompatible with the larger values obtained in fits to nuclei , since the in - medium parameters must contain many - body effects . in all of the models discussed here ,
the resulting tensor coupling is still natural .
it remains to be seen whether explicit calculation of the many - body contributions can account for the larger values favored by nuclei .
moreover , it may be possible to constrain the tensor coupling by including additional observables in the fits ; the energy dependence of the optical potential for nucleon - nucleus scattering is a promising candidate .
we thank b. c. clark for useful discussions .
this work was supported in part by the department of energy under contracts no .
de fg0287er40365 and de fg0287er40328 , and the national science foundation under grants no .
phy9511923 and phy9258270 .
we thank the department of energy s institute for nuclear theory at the university of washington for its hospitality and the department of energy for partial support during the completion of this work .
s. weinberg , physica * a96 * ( 1979 ) 327 .
g. p. lepage , in _ from actions to answers _
( tasi-89 ) , edited by t. degrand and d. toussaint ( world scientific , singapore , 1989 ) , p. 483 .
j. polchinski , in _ recent directions in particle theory : from superstrings and black holes to the standard model _
( tasi-92 ) , edited by j. harvey and j. polchinski ( world scientific , singapore , 1993 ) , p.235 .
h. georgi , ann .
nuc . part .
* 43 * ( 1993 ) 209 .
s. weinberg , _ the quantum theory of fields , vol .
i : foundations _
( cambridge university press , new york , 1995 ) . b. d. serot and j. d. walecka , adv .
. phys . * 16 * ( 1986 ) 1 .
b. d. serot , rep .
* 55 * ( 1992 ) 1855 .
r. j. furnstahl , b. d. serot , and h .- b .
tang , nucl .
* a615 * ( 1997 ) 441 . b. d. serot and j. d. walecka , int .
e ( 1997 ) , in press .
john j. rusnak and r. j. furnstahl , ohio state preprint , nucl - th/9708040 , ( 1997 ) .
reinhard , rep . prog .
* 52 * ( 1989 ) 439 .
y. k. gambhir , p. ring , and a. thimet , ann . of phys .
( 1990 ) 132 .
a. r. bodmer , nucl .
* a526 * ( 1991 ) 703 .
r. j. furnstahl , h .- b .
tang , and b. d. serot , phys .
c * 52 * ( 1995 ) 1368 .
h. georgi and a. manohar , nucl .
* b234 * ( 1984 ) 189 . j. l. friar , d. g. madland , and b. w. lynn , phys .
c * 53 * ( 1996 ) 3085 .
john j. rusnak , ph.d thesis , ohio state university , 1997 .
m. rufa , p .-
reinhard , j. a. maruhn , w. greiner , and m. r. strayer , phys .
* c38 * ( 1988 ) 390 .
r. machleidt , adv .
* 19 * ( 1989 ) 189 .
g. hhler , e. pietarinen , i. sabba - stefanescu , f. borkowski , g. g. simon , v. h. walther , and r. d. wendling , nucl . phys .
* b114 * ( 1976 ) 505 . g. e. brown , m. rho , and w. weise , nucl . phys .
( 1986 ) 669 .
ulf - g .
meissner , phys .
rep . * 161 * ( 1988 ) 213 .
p. mergell , ulf - g .
meissner , and d. drechsel , nucl .
* a596 * ( 1996 ) 367 .
k. saito , k. tsushima , a. w. thomas , nucl .
* a609 * ( 1996 ) 339 .
t. s. bir and j. zimnyi , phys .
b * 391 * ( 1997 ) 1 .
b. c. clark , s. hama , and s. g. klbermann , e. d. cooper , and r. l. mercer , phys .
c * 31 * ( 1985 ) 694 .
b. jennings , phys .
b * 246 * ( 1990 ) 325 ; nucl .
* a529 * ( 1991 ) 589 .
j. cohen and h. j. weber , phys .
c * 44 * ( 1991 ) 1181 .
j. f. donoghue , e. golowich , and b. r. holstein , _ dynamics of the standard model _
( cambridge university press , new york , 1992 ) .
r. j. furnstahl , c. e. price , and g. e. walker , phys .
c * 36 * ( 1987 ) 2590 .
r. j. furnstahl , b. d. serot , and h .- b .
tang , nucl .
* a618 * ( 1997 ) 446 .
b. a. nikolaus , t. hoch , d. g. madland , phys .
c * 46 * ( 1992 ) 1757 .
m. chiapparini , a. delfino , m. malheiro , a. gattone , z. phys .
a * 357 * ( 1997 ) 47 .
j. zimnyi and s. moszkowski , phys .
c * 42 * ( 1990 ) 1416 .
b. borasoy and ulf - g .
meissner , int .
j. mod .
* a111 * ( 1996 ) 5183 .
p. j. siemens and a. s. jensen , _ elements of nuclei _
( addison - wesley , 19987 ) . | a compelling feature of relativistic mean - field phenomenology has been the reproduction of spin - orbit splittings in finite nuclei after fitting only to equilibrium properties of infinite nuclear matter .
this successful result occurs when the velocity dependence of the equivalent central potential that leads to saturation arises primarily because of a reduced nucleon effective mass .
the spin - orbit interaction is then also specified when one works in a four - component dirac framework . here
the nature of the spin - orbit force in more general chiral effective field theories of nuclei is examined , with an emphasis on the role of the tensor coupling of the isoscalar vector meson ( @xmath0 ) to the nucleon . # |
there is considerable interest to heavy ion reaction at sufficiently high energies , because they provide the possibility for conditions of the transitions to deconfined state of hadronic matter - quark - gluon plasma ( qgp ) . at pb+pb and au+au collisions
there is considerable number of net nucleons ( @xmath0 ) in central interval of rapidity , which it is necessary take into account on plasma and hadron stage .
therefore we have the baryon chemical potential @xmath13 . in the works a number of authors ( p.braun-mnzinger et.al .
@xcite ) was shown , that statistical thermal model gives good description of particle ratios in central collisions of heavy nuclei at high energies ( for sps and rhic ) .
the temperature @xmath14 for all this is close to critical temperature @xmath15 mev obtained from lattice monte - carlo simulations of qcd , and @xmath16 is close to 50 mev for rhic(at @xmath17=130 gev ) this correspond to chemical freeze - out , when inelastic collisions cease to be important and the particle composition is fixed .
the trajectory of chemical freeze - out @xmath18 was obtained also in the work @xcite on the basic of relation for hadronic gas @xmath19 gev . for sps
was found the value @xmath20 mev .
one can to suppose that trajectory @xmath18 for various energies of nuclei is close to boundary of deconfined quark - gluon plasma phase .
for all this the potential @xmath21 correspond to strangeness neutrality ( i.e. overall strangeness = 0 ) .
this correspond in fact to hadronic part of mixed phase , where there are strange mesons and baryons ( i.e. @xmath22 ) . in the present work
we try to define the physical characteristic of initial plasma phase @xmath1 and @xmath2 . for this aim
we use hypothesis of conservation of entropy and of number of net nucleons in initial stage ( where @xmath23 ) and on stage of phase transition at @xmath24 , where we take into account also of constituent quarks phase .
the plasma at finite temperature was considered perturbatively up to order @xmath25 ) @xcite .
however in the experimentally accessible region ( close to temperature @xmath3 of phase transition ) the strong coupling constant is sufficiently large : @xmath26 .
the perturbative expansion in powers of g probably gives bad convergence .
recently in series of papers @xcite for description of quark - gluon plasma have been used quasiparticle model , which allow apparently to sum up partially the perturbative expansion . at high temperature the plasma consist of quasifree quarks and gluons and
probably is described well by perturbative theory .
the lattice model show rapid increase of entropy density and other values near by @xmath3 in su(3 ) gluon plasma .
in quasiparticle model the interacting plasma of quarks and gluons is described as system of massive quasiparticles .
for thermal momenta @xmath27 the quark - particle excitation and transversal gluons will propagate with dispersion relations @xmath28 where @xmath29 are given by the asymptotic values of the hard thermal self energies @xcite @xmath30g^2 , \\
\pi_g&= & \frac{1}{6}[(n_c+\frac{n_f}{2})t^2+\frac{6}{2\pi^2}\mu_q^2]g^2 .
\nonumber \end{aligned}\ ] ] the contributions @xmath29 are generated dynamically by the interaction within the medium , and @xmath31 acts as effective mass .
the quantity @xmath32 is to be considered as effective coupling in quasiparticles model .
the entropy and energy density for example for gluon plasma takes the form : @xmath33 @xmath34 where @xmath35^{-1}$ ] , @xmath36 the value @xmath37 , introduced in expression for energy ( and also for pressure ) is necessary in order to maintain thermodynamic consistency , that is must be fulfilled the relations @xmath38 and @xmath39 . the function @xmath37 compensates the additional @xmath14 derivatives from @xmath40 in pressure and it represent in integral form in the paper @xcite . in our paper we do not use the function @xmath37 .
however the phenomenological parameterization of coupling constant @xmath41 in accordance with perturbative @xmath42 in quasiparticle model @xcite is very likely unjust close to temperature of phase transition @xmath3 , where it should be expected of nonperturbative dynamics . in this model
we have the increase of thermal gluon mass in vicinity of phase transition . in sec .
[ sec2 ] we give the brief review of phenomenological model of confinement @xcite , which agrees with @xmath43 lattice data .
this model is applied also to system with dynamical quarks . in sec .
[ sec3 ] we calculate for isentropic evolution of quasiparticles system with chemical potential @xmath13 the various physical characteristic : the initial temperature @xmath1 , the volume @xmath2 , the initial entropy @xmath44 etc . for nuclear collisions at sps and rhic energies .
we use the values @xmath3 and @xmath16 close to above - mentioned : @xmath45 mev and for rhic and sps . at calculation
we use also additional information - experimental meaning of average number of net nucleons in central region of rapidity for sps and rhic .
in result we have obtained the equations ( [ eq.22 ] ) , ( [ eq.23 ] ) for definition of @xmath1 .
simultaneously must be fulfilled the relations ( [ eq.24]),which follow from conservation of entropy and of number net nucleons . besides of study of characteristic of initial phase we describe in sec .
[ sec3 ] another new aspect - we show , that from conservation of entropy and of number of net nucleons follow , that massive constituent quarks ( @xmath4 and @xmath5 ) appears with decrease of number of degrees of freedom ( i.e. with effective number ) in the presence of octet of pseudogoldstone states .
we show also , that with the same effective number of degrees of freedom appears hadrons and resonances in hadron part of mixed phase ( for sps and rhic ) .
certainly it should be noted , that in nucleus - nucleus collisions at high energies there is possibility of two stages of equilibrium .
the gluons reach of thermal equilibrium to a considerable extent faster , than quarks . according to estimations , equilibrium time of gluons is @xmath46 fm / c and for quarks one is @xmath47 fm / c , i.e. it is possible the production of hot glue at first stage @xcite .
the corresponding estimations of initial conditions for hot glue in effective quasiparticle model there are in the work @xcite in sec .
[ sec3 ] we investigate the initial condition ( @xmath1 , @xmath2 ) and evolution of second plasma stage , where is the equilibrium also for quarks . in this paper
we do not consider the particle ratios .
we have considered the ratios for various baryons and mesons with accounting of resonance decays in the work @xcite .
the particle ratios in the main agrees well with experimental data for sps and rhic ( @xmath9 mev . in sec .
[ sec.4 ] we calculate in effective quasiparticle model the some baryons and mesons ( pions and kaons ) spectra with accounting of resonance decays and transverse flow .
we use for heavy nuclei the thermal freeze - out temperature @xmath48 mev .
the normalization of baryons and mesons spectra here is no free parameter , but is defined by initial condition in plasma - by initial entropy @xmath44 and by condition of entropy conservation @xmath49 .
we show that values of averaged transverse flow velocity @xmath11 increase for rhic in comparison with sps .
the main object of present work is no study the whole of spectra for sps and rhic , but study of dependence spectra from initial state and from stage of phase transition plasma into hadrons .
therefore in sec .
[ sec5 ] we consider the analogous isentropic problem of nuclear collisions by the use of ordinary perturbative theory up to order @xmath50 with qcd parameter @xmath51 gev .
we calculate here also the values @xmath1 , @xmath2 , @xmath44 .
we do not find noticeable difference for spectra of particles in comparison with effective quasiparticle model .
however this model disagrees with su(3 ) lattice data in region of phase transition ( i.e. near of temperature @xmath3 ) . in sec .
[ sec5 ] we consider also the quasiparticle model with phenomenological parametrization of running coupling @xmath12 , which gives good fit of new lattice data .
we calculate here also the initial parameteres @xmath52 .
the coupling strength @xmath53 here increase at @xmath7 from above ( unlike effective quasiparticle model ) .
however baryons and mesons spectra in this model practically do not differ from spectra in effective quasiparticle model .
thus the spectra weakly depend on region of phase transition .
however it can be shown @xcite , that in ordinary peturbative model ( taking into account the hot glue ) we have too great energy loss @xmath54 of gluon jet at rhic energy : @xmath55 ( unlike effective quasiparticle model ) .
thus the jet quenching essentially depend on description of phase transition stage .
we show also , that in second above - mentioned model the situation with energy loss @xmath54 is more dramatic . in sec .
[ sec7 ] - conclusion .
in the papers @xcite for gluon plasma was used the phenomenological parameterization of coupling constant @xmath32 in accordance with perturbative qcd with the two fit parameters @xmath56 and @xmath57 .
the resulting equation of state ( eos ) is in good agreement with lattice data over a wide temperature range between @xmath3 and 5@xmath3 .
however in spite of this the thermal gluon mass have strong increase in the vicinity of the phase transition at decrease of @xmath14 . furthermore , close to a phase transition the coupling constant @xmath58 , that is correctness of perturbative of calculations is questionable .
further it is expected that correlation length @xmath59 , which is proportional @xmath60 , grows when @xmath7 from above ( @xmath61 is gluonic debye mass ) .
the lattice calculations shows that @xmath61 drops by factor of 10 when @xmath14 decrease from 2 @xmath3 to @xmath3 @xcite also for three colors the value @xmath62 remains large , but finite . in hard thermal loop ( @xmath63 )
perturbation theory the thermal mass @xmath64 and gluonic debye mass @xmath61 are connected by relation @xcite @xmath65 however in nonperturbative theory this relation can be broken down @xcite . recently in the paper @xcite was considered interesting phenomenological model of confinement . in region below @xmath3 in a pure @xmath43 gauge theory are color singlet of heavy glueballs . approaching @xmath3
the gluons are liberated , followed by sudden increase of energy and entropy density .
when approaching the phase transition from above , the decrease of thermodynamic values is not caused by increase of masses(unlike @xcite ) , but caused by decrease of number active degrees of freedom .
when @xmath14 comes closer to @xmath3 , more and more of gluons are absorbed by glueballs .
it is assumed that thermal gluons mass @xmath64 follows roughly the behavior of the debye mass , i.e. it decreases .
however the entropy density @xmath66 will overshoot the lattice entropy because light masses near @xmath3 .
this difference may be accounted in quasiparticle model by modifying of number effective degrees of freedom in ( [ eq.3 ] ) , ( [ eq.4 ] ) @xmath67 the explicit value @xmath68 may be estimated as the ratio of the lattice entropy and entropy ( [ eq.3 ] ) with a dropping mass @xmath64 . at @xmath69
we expect @xmath70 and near @xmath71 we have @xmath72 . at @xmath73
it can estimate @xmath74 from lattice data .
the value @xmath68 is a smooth , monotonously increasing function with @xmath14
. it is assumed also that thermal mass ( for example gluons mass ) have form @xmath75 .
the lattice results for debye mass @xmath76 can be parameterized well by formula @xmath77 where @xmath78 and there is small gap at @xmath79 with @xmath80 .
the value @xmath81 is determined here by the asymptotic value of thermal gluon mass , which coincide with lattice mass for instance at @xmath82 .
below of this value the explicit @xmath63 resummation is expected to fail @xcite .
one can to use the asymptotic value @xmath12 with renormalization group inspired parametrization @xcite : @xmath83 the parameteres @xmath56 and @xmath57 are used for fitting new lattice results with @xmath84=2 and @xmath84=4 at @xmath85 ( and also for @xmath84=3 @xcite ) .
the thermal gluon mass at chemical potential @xmath86=0 is analogous with formula ( [ eq.7 ] ) where @xmath87 .
effective coupling @xmath88 have the form @xmath89 .
the gluon and quark mass coincide with lattice data at @xmath90 9.4 for @xmath84=2 and at @xmath90 9.8 for @xmath84=3 .
that gives @xmath91 1.886 for @xmath84=3 .
the like value @xmath88=1.9 we have in formula ( [ eq.7 ] ) .
the such value @xmath92 we use in calculations .
it should be noted , that lattice results at finite chemical potential @xmath86 point to greatly weak dependence of asymptotic thermal masses from @xmath86 @xcite . since ( in accord with assumption )
the values @xmath64 and @xmath76 have similar trends near @xmath93 , the mass @xmath64 is parameterized by analogy with formula ( [ eq.7 ] ) . in order to account for certainties , were investigated @xcite some a range of values @xmath88 , @xmath94 and @xmath95 .
a decreasing effective coupling strength @xmath12 at from above can be understood , since at decreasing of @xmath14 more and more gluons become confined and form heavy glueballs .
the effective glueball exchange interaction between gluons are reduced .
the total interaction can be interpreted as a superposition nonperturbative multigluon and weak glueball exchange .
the good fit @xcite for @xmath68 ( [ eq.6 ] ) have form similar with ( [ eq.7 ] ) @xmath96 where @xmath97 1.25 , @xmath98 0.0026 , @xmath99 0.31 .
the relations ( [ eq.7 ] ) , ( [ eq.9 ] ) give good description of @xmath43 lattice data for @xmath100 , @xmath101 , @xmath102 for @xmath14 close to @xmath3 .
further it is possible to extrapolate effective quasiparticle model to system with dynamical quarks .
unfortunately no lattice data on thermal masses with dynamical quarks are available .
we assume that @xmath103 and @xmath84 dependence of @xmath104 and @xmath4 are given by formulas ( [ eq.2 ] ) , where effective coupling @xmath12 have the form : . for example , the thermal masses of @xmath105 and @xmath106 quarks are : @xmath107 for @xmath108 quarks we replace @xmath109 , @xmath110 .
it is assumed that parameterization of function @xmath111 in the presence of quarks and parameterization ( [ eq.9 ] ) for gluons are similar , with some variation of parameters .
for example , @xmath112 , @xmath113 , @xmath114 for two light quarks @xcite .
the relations of type ( [ eq.9 ] ) for @xmath111 and relations of type ( [ eq.10 ] ) for @xmath4 and @xmath104 are inserted into formulas for entropy density of quasiparticles .
for @xmath115 quarks we have the entropy density : @xmath116 where @xmath117 } \nonumber\end{aligned}\ ] ] the formula ( [ eq.11 ] ) gives correct term already in order @xmath118 at perturbative expansion
. for @xmath108 quarks @xmath119 , and @xmath120 . for gluons
the entropy density is : @xmath121 for the net nucleons density we have : @xmath122 , \nonumber\end{aligned}\ ] ] where @xmath123 we shall show below that in massive constituent quarks phase and hadron part of mixed phase we have also decrease of the number of effective degrees of freedom .
in standard model of relativistic nuclear collisions after a pre - equilibrium period the system becomes thermalised in form of quark - gluon plasma with formation time @xmath124 .
the system then expand , cools off and at the time @xmath125 it begins to hadronize . at temperature @xmath3 appears the mixed phase at first plasma part at @xmath125 , then hadronic part of mixed phase at @xmath126 . at temperature near to @xmath127
mev particles first undergo a chemical freeze out , and then a thermal freeze out at temperature @xmath128 mev .
the particles ratio is defined by temperature close to @xmath3 . for @xmath129 collisions
we have the initial volume @xmath130 , where @xmath131 .
we consider isentropic evolution .
the entropy per unit rapidity is conserved during the expansion @xcite .
the value @xmath132 determine initial entropy density , where @xmath133 is the average multiplicity charged plus neutral particles per unit central rapidity .
in this case part of initial energy goes into collective motion of the expanding system , that is the initial energy density must have been higher than in the free flow scenario . in the work @xcite
was investigated the isentropic longitudinal expansion of ideal gas of gluons and three flavors of massless quarks . in this work
was found the dependence @xmath134 for initial energy density in plasma . in this paper
we have investigated more common case for quasiparticles with @xmath13 .
the quasiparticles density in plasma phase is : @xmath135 where @xmath136 is density of @xmath137 quarks and gluons .
we have here : @xmath138 , \nonumber\end{aligned}\ ] ] @xmath139 @xmath140 by analogy it can be written for summary entropy density : @xmath141 where @xmath142 given by ( [ eq.11 ] ) , ( [ eq.12 ] ) .
we must express the initial temperature @xmath1 across the number of secondary particles per unit central rapidity .
for convenience of calculations we introduce the designation : @xmath143 , @xmath144 , @xmath145 , that is we single out the factors @xmath146 , @xmath147 .
we single out also the constant factors @xmath148 , @xmath149 thus we have @xmath150 , @xmath151 .
hence we have here @xmath152 , where @xmath153 .
thus the initial temperature @xmath1 have been expressed across the number of secondary per unit central rapidity : @xmath154 for @xmath129 collisions the value @xmath133 can be extrapolated by the form @xcite : @xmath155 for @xmath156 collisions we have good approximation per unit central rapidity @xcite : @xmath157 the value @xmath158 describes the amount of rescattering in the interaction of the nuclei . from @xmath159 data it can expect @xmath160 .
the value @xmath158 = 1.1 describes the result for various light ion quite well @xcite .
the data for sps ( @xmath161 gev ) is @xmath162 ( i.e. close to ( @xmath159 ) data ) .
we have ( for @xmath158 = 1.1 ) the data of phenix collaboration show increase of @xmath163 from sps ( na49 ) to rhic ( @xmath9 gev ) on value @xmath164 1.7 , i.e. on value @xmath165 . for sps
we assume @xmath166 . from formula ( [ eq.20 ] )
we have for rhic ( @xmath167 ) : @xmath168 the initial volume @xmath2 in formula ( [ eq.19 ] ) can be expressed across experimental meaning of average number of net nucleons @xmath169 in central region of rapidity .
for sps the number of net protons in central region of rapidity is estimated as 28 - 29 @xcite .
we use the value @xmath170 .
the number of net protons for rhic in central region of rapidity ( at @xmath9 gev ) is estimated as 8 - 10 @xcite , i.e. the number of net nucleons is @xmath171 16 - 20 .
we use the value @xmath172 16.3 ( for some greater agreement with spectra ) .
we have : @xmath173 ( from ( [ eq.13 ] ) . hence the value
@xmath1 is : @xmath174 we single out here the factor @xmath146 in formula ( [ eq.13 ] ) for @xmath175 and take into account the values @xmath176 803 and @xmath177 57 for sps and @xmath176 1374 and@xmath177 16.3 for rhic .
in result we have the equation for definition of initial temperature @xmath1 : @xmath178 where @xmath179 for calculations of values @xmath180 we use formulas ( [ eq.14]-[eq.17 ] ) for sps we have with accounting of constant factors : @xmath181 = 1.676 at @xmath182 and for rhic @xmath181 = 3.045 at @xmath158 = 1.11 .
we have from here the estimations of initial temperature @xmath183 mev for sps and @xmath184 219.6 mev for rhic .
we use here the meaning for temperature of phase transition @xmath185 mev , as indicated by result from lattice gauge theory @xcite , and close to @xmath3 the temperature of chemical freeze - out . at sps energies we have for value @xmath186 mev the good approximation for different antiparticles and particles ratios . for rhic ( @xmath9 gev )
we use also @xmath187 mev and @xmath188 mev .
that give the ratio @xmath189 close to found before experimental value .
later was given @xcite the ratio @xmath190 .
in the work @xcite were given more precise values @xmath191 mev , @xmath192 mev and particle ratios(at @xmath9 gev ) .
however there are for the present noticeable statistic and systematic mistake in data .
one should that not great distinction have weak influence for example on spectra baryons and mesons .
we use also the conservation of entropy and of number net nucleons in initial phase at @xmath193 and in mixed phase at @xmath194 ( the entropy density is @xmath195 ) : @xmath196 , @xmath197 , i.e. @xmath198 , @xmath199 .
the entropy density @xmath200 correspond to sum ( [ eq.18 ] ) , where @xmath201 .
hence we have relations : @xmath202 these relations depend also from constant @xmath88
. the relations ( [ eq.23 ] ) and ( [ eq.24 ] ) must be fulfilled simultaneously . for sps
( @xmath16 = 247 mev , @xmath45 mev ) these relations indeed are fulfilled .
it can be seen from mentioned below values .
for example for @xmath166 at @xmath203 , @xmath204 mev we have : @xmath205 .
it should be noted , that relations ( [ eq.24 ] ) are fulfilled only for coincides values @xmath16 at initial and mixed phases .
thus , in this model the initial temperature @xmath1 at sps energy exceed weakly the temperature @xmath3 of phase transition .
we will express all physical values in units @xmath206 @xmath207 . from formulas ( [ eq.11])-([eq.13 ] )
- we find the values of entropy density for sps at @xmath1 and @xmath3 : @xmath208 , @xmath209 and also net nucleons density @xmath210 , @xmath211 . for the number of net nucleons we find the volume of initial plasma phase : @xmath212 @xmath213 ( i.e. @xmath214 fm ) .
the initial entropy is : @xmath215 .
we have also the volume @xmath216 of plasma part of mixed phase , taking into account the conservation of entropy and of net nucleons : @xmath217 @xmath213 ( i.e. @xmath218 fm )
. therefore with accounting of valent @xmath219 quarks we have the large initial @xmath220 fm .
the suitable calculations by formula ( [ eq.19 ] ) for @xmath221 fm ( for sps ) shows that for various values of @xmath88 the number 57 of net nucleons correspond @xmath222 mev , that contradict to experimental data for particles ratios . for correct values of @xmath223 mev the number @xmath224 is too little(@xmath225 ) .
the analogous situation there is and for rhic . at decrease of interaction the thermal masses of quasiparticles @xmath226 and @xmath227 ( [ eq.10 ] ) are diminished at @xmath7 from above , but they do not correspond to constituent masses of quarks in mixed phase . for example , in the papers @xcite was shown , that by means of help of masses constituent quarks @xmath228 mev , @xmath229 mev can be found the masses of baryons in octet . for @xmath43 group
the spontaneous breaking of chiral symmetry and appearance of masses leads to appearance of octet pseudoscalar light goldstone ( or rather pseudogoldstone ) states @xmath230 , @xmath231 @xcite .
however if in place of relation @xmath232 in ( [ eq.24 ] ) to use relations for massive quarks with accounting of pseudogoldstone states , then these relations do not fulfilled . but these relations will be satisfied , if to use decrease of the number of effective degrees of freedom of massive quarks .
we calculate the entropy and net nucleon density @xmath195 and @xmath233 of massive quarks by formulas for ideal gas .
for example for @xmath234 , @xmath235 mev at @xmath236 mev , @xmath185 mev we have @xmath237 , and @xmath238 .
for entropy density @xmath239 of pseudogoldstone states @xmath240 we find by same way : @xmath241 ( at @xmath242 mev , @xmath243 mev ) .
the factor @xmath244 of decrease of number degrees of freedom can be defined from relation .
that gives @xmath245 .
therefore , it is possible to interpret , that there is appearance of massive constituent quarks with effective number of degrees of freedom in the presence of octet of pseudogoldstone states , while the multigluon state is absorbed into constituent quarks masses .
let us list appropriate calculations for rhic ( gev ) . for example , at @xmath246 ( [ eq.20 ] ) , @xmath247 , we find from ( [ eq.23 ] ) : , that is from here : @xmath248 , @xmath249 , , @xmath250 in units of @xmath251 .
the initial volume @xmath252 @xmath213 , that is @xmath253 fm ( for @xmath254 @xmath255 ) .
the initial entropy @xmath256 .
hence we have for rhic @xmath257 , @xmath258 , that is the relations ( [ eq.24 ] ) do not fulfilled .
however one can assume that at sufficiently high initial temperature it is possible already the appearance ( in nonperturbative state at @xmath259 ) of pseudogoldstone with entropy density @xmath239 .
it gives in fact @xmath260 .
that is we have on this stage of mixed phase for rhic @xmath261 , where @xmath262 , i.e. @xmath263 6.5 fm . at high energy ( rhic ) the multigluon state is also absorbed into massive constituent quarks with lesser number of degrees of freedom
can be found the factor @xmath264 of decrease of effective number of degrees of freedom .
we calculate here by analogy with sps the entropy density @xmath265 and net nucleon density @xmath266 for massive quarks , at @xmath188 mev , @xmath185 mev .
we find in result : @xmath267 and @xmath268 we must have ratio @xmath269 390 .
we find from here @xmath270 0.343 .
we have effective volume @xmath271 on the stage of constituent quarks , or - is the same .
let us consider now the hadronic part of mixed phase .
we have seen , that entropy density of massive constituent quarks is defined by effective number of degrees of freedom and by contribution of octet pseudogoldstone .
it can be shown that with such effective number of degrees of freedom ( @xmath272 and @xmath273 ) in hadron part of mixed phase appears nucleons and hadrons . in order to find strange chemical potential @xmath21 for hadron part of mixed phase
, we use condition of disappearance of strangeness : @xmath274 that is : @xmath275 . here
@xmath278 @xmath279 for strange baryons we have @xmath280 except @xmath281 and @xmath282 . at temperature @xmath3 one should take into account the considerable number of resonances .
we took into account the strange mesons up to @xmath283 mev and the strange baryons up to @xmath284 mev . in result of calculatios we have from ( [ eq.25 ] ) : @xmath285 mev at @xmath286 mev and @xmath287 mev at @xmath288 mev ( for @xmath289 mev ) . at sps
we find the entropy density according to calculation by formulas of ideal gas : @xmath290 for nonstrange mesons and @xmath291 for strange ones ( at @xmath3 ) . for summary entropy density of strange baryons and the whole of nonstrange ones ,
i.e. the whole entropy is : @xmath292 . for net nucleons density with accounting of nucleonic resonances @xmath293 the calculation
gives : @xmath294 .
hence the ratio is : @xmath295 , i.e. @xmath296 .
taking into account only nucleons(without of resonances ) , we have @xmath297 . for @xmath298
we have @xmath299 , for @xmath300 the calculation gives @xmath301 , @xmath302 , i.e. @xmath303 and so on .
thus always @xmath304 , i.e. the relations ( [ eq.24 ] ) do not fulfilled .
however for effective number of nucleon resonances @xmath305 with lesser number of degrees of freedom by factor @xmath306 in the presence of octet pseudogoldstones we have the ratio : @xmath307 , i.e. the relations ( [ eq.24 ] ) are fulfilled .
we have also the volume @xmath308 , or the same @xmath309 , where @xmath310 is effective volume of hadron part of mixed phase .
we give now the result of calculations at rhic energy .
the calculation gives for nucleons only ( without accounting of resonances ) : , for @xmath311 this ratio is : @xmath312 , for the whole of nucleon resonance we have for this ratio : @xmath313 and so on , i.e. @xmath314 .
but for the whole resonances the calculation gives : @xmath315 , @xmath316 in units @xmath251 . with lesser number of degrees of freedom by value @xmath273 = 0.343(like for massive quarks ) we have @xmath317 .
now the effective volume is : @xmath318 , or the same @xmath319 .
hence the relations of type ( [ eq.24 ] ) are fulfilled also for ratio @xmath320 in hadron part of mixed but with effective number of degrees of freedom the same as for constituent quarks .
we investigate the spectra for sps and rhic at thermal freeze - out .
apparently the favor results for heavy nuclei ( for example pb + pb ) correspond to thermal freeze - out temperature @xmath321 120 mev @xcite we consider at first sps energy .
we find the volume @xmath322 at from conservation of number net nucleons .
the value @xmath323 at thermal freezing we find from condition : @xmath324 , where @xmath325 . from here
we have @xmath326 mev . from formula ( [ eq.13 ] )
we have net nucleons density at @xmath48 mev , @xmath327 mev : @xmath328 .
for @xmath329 we have @xmath330 .
the value @xmath331 at 120 mev may be found by no condition ( [ eq.25 ] ) ( as the chemical equilibrium already is absent ) , but from ratio : @xmath332 from here we have found with accounting of various weak decays : @xmath333 mev . from relations @xmath334 ( with accounting of decays )
we find for sps the near value : @xmath331=41 mev @xcite .
one can now find by formulas of ideal gas the entropy density ( at @xmath335 ) of nucleons and strange baryons : @xmath336 and for nonstrange and strange mesons up to @xmath337=1820 mev : @xmath338 .
thus the entropy of baryons is , and the entropy of mesons .
the volume of mesons at thermal freeze - out is @xmath339 .
we can estimate the number of @xmath340 : .for of the value ( calculated without accounting of weak decays ) we have .
the experiment per unit of rapidity gives .
by analogy we find also the number of @xmath341 : .
thus in this model we have different volume of freeze - out for baryons and mesons .
the spectra of protons and strange baryons one should to calculate with @xmath322 , and of mesons with volume @xmath342 . the relations of type ( [ eq.24 ] ) must be fulfilled also for thermal freeze - out .
here we have @xmath343 , and for net nucleons : @xmath344 , i.e. we have instead of ( [ eq.24 ] ) the relation : .
thus the relations of type ( [ eq.24 ] ) are fulfilled ( but now without of decrease of number of degrees of freedom ) and backwards from conservation of entropy @xmath345 one can to find the final volume @xmath322 for baryons .
we calculate the spectra @xmath346 , @xmath347 , @xmath348 , @xmath349 direct and with resonance decays .
we take into account the transverse flow .
we use the hydrodynamic model with linear transverse velocity profile @xmath350 , where @xmath11 is the averaged transverse flow velocity .
the spectra of net protons expressed by formula : @xmath351 here @xmath352 , @xmath353 . for @xmath354
we have factors @xmath355 , where correspondingly , and for @xmath341 - factor .
we take into account also spectra at resonance decays , for example : , , , , , @xmath356 . at resonance
decays we must take into account the resonance with transverse flow , and cascade decays also .
the resonance decays without transverse flow were investigated in the paper @xcite .
we have calculated here the some resonance decays spectra at sps and rhic energies . for example , spectra of @xmath347 at decay @xmath357 is calculated thus : @xmath358 here @xmath359 - branching , @xmath360 - stat.weight , @xmath361[(m_{\sigma } -m_1)^2 - m_2 ^ 2]}}{2 m_{\sigma}},\ ] ] @xmath362 mev , @xmath363 .
we have also : @xmath364 here @xmath365 , @xmath366 .
the value @xmath367 in formula ( [ eq.28 ] ) is : @xmath368 here , . at cascade decays , for example @xmath369 ,
@xmath370 , we calculate at first the spectra of @xmath371 , however no at angle of 90 degrees ( i.e. @xmath372 ) , but for the whole y , and then this spectra we use for calculation of spectra @xmath373 . in spectra
@xmath374 the value @xmath375 we replace now by the value @xmath376 . then the spectra @xmath377 we substitute to formula : @xmath378 we apply now to rhic energy ( @xmath17=130 gev ) . at measurement of antiprotons spectra in collaboration star were measured the spectra of direct @xmath379 and also spectra of @xmath379 from weak decays ( with cascade ) .
the our calculation shows that contribution of weak decays is @xmath380 .
we give here the meanings of values , which were found for rhic at @xmath381 mev ( by analogy with sps ) : @xmath323 = 35.3 mev , @xmath331=7.93 mev , @xmath382 , @xmath383 , @xmath384 . from conservation of number net nucleons
we have the volume @xmath322 at without accounting of weak decays : .
the entropy of baryons in this volume is @xmath385 .
the entropy of mesons is .
the mesons entropy density is . from here
@xmath386 .
[ ht ] - are direct and with resonance decays : @xmath387 , @xmath388 , @xmath389 .
the @xmath390 spectra - are direct + @xmath391 + @xmath392 .
the spectra @xmath393 and @xmath394 - are thermal direct . the averaged transverse flow velocity @xmath395 .
the experimental data are from @xcite ) .
[ fig.1],title="fig : " ] spectra ( sps , pb+pb ) are direct + decays @xmath396 + @xmath397 ( @xmath395 ) .
the dashed line - the direct @xmath398 , the dash - dotted line @xmath398 at decays .
we show also the summary spectra at @xmath399 . the data from @xcite ) .
[ fig.2a ] ] spectra ( sps , pb+pb ) - direct + decay @xmath400 ( @xmath395 ) .
we show also the same spectra at @xmath399 . the data from @xcite ) .
[ fig.2b ] ] we have found the particles density at @xmath335 : @xmath401 , @xmath402 .
this gives the estimation of number of particles : , @xmath403 .
the value @xmath404 we use for calculation , for example , of spectra @xmath405 and @xmath406 for energy rhic .
one can also to show that relation of type ( [ eq.24 ] ) for thermal freeze - out are fulfilled and for rhic energy .
one can calculate the contribution of weak decays into net nucleons density at @xmath407 mev ( for rhic ) @xcite : @xmath408 .
with account of nucleons from decays we find equivalent volume @xmath409 : . from here .
the spectra of direct and of daughter @xmath379 at weak decays we calculate by formulas ( [ eq.26])-([eq.30 ] ) with factors type @xmath410 , @xmath411 .
the contribution of weak decays with cascade decays to @xmath379 spectra gives @xmath380 .
( centrality @xmath412 ) at rhic energy with accounting of weak decays(@xmath413 ) .
we show also the same results with @xmath395 and @xmath414 ( quasiparticle model ) . the data from @xcite ) .
[ fig.3a ] ] spectra ( rhic ) - direct + decays @xmath415 ( @xmath416 ) .
it is shown the same spectra with @xmath399 ) .
[ fig.3b ] ] the theoretical @xmath379 spectra and experimental spectra for central bin @xmath412 ( star coll . ) at mid rapidity are shown in fig .
[ fig.3a ] . in fig .
[ fig.3b ] , fig .
[ fig.3c ] we show also the direct @xmath406 spectra and with decay @xmath417 and the @xmath405 spectra direct + @xmath418 and experimental data . in fig .
[ fig.1 ] we show the baryons spectra : @xmath393 , @xmath347 , @xmath348 , and @xmath349 for sps ( at @xmath335 ) in central interval of rapidity .
the @xmath419 and @xmath420 spectra are shown in fig .
[ fig.2a ] , fig .
[ fig.2b ] together with data .
one can to see that spectra for rhic have greater inverse slope than for sps . with increasing mass of particle the inverse slope ( i.e. the value @xmath11 ) increases .
of course , one should note that obtained spectra with sufficiently large value of @xmath11 are correct only for not great values of @xmath421 .
thus , in the present quasiparticle model with isentropic evolution the normalization of baryons and mesons spectra do not defined by free parameteres , but defined by conditions in initial state ( @xmath1 , @xmath2 ) , i.e. by quark - gluon plasma .
we had investigated the physical characteristic of initial and mixed phases and some baryons and mesons spectra for sps and rhic in effective quasiparticle model with isentropic evolution . in this section
we shall consider at first the analogous problem of nuclear collisions by the use of ordinary perturbative decomposition of thermodynamic values in powers of running coupling @xmath422 in the form @xcite : @xmath423 up to order @xmath424 with qcd parameter @xmath425 gev . here
the value @xmath426 in coupling @xmath427 is considering as the average of @xmath426 in s - canal at temperature t. for example , the entropy density for gluons in plasma is : and so on .
we calculate here also the values @xmath428 . the initial temperature @xmath1 can be find from formula analogous ( [ eq.23 ] ) , where we have now : @xmath429 here the values @xmath430 and @xmath431 are extended up to order @xmath432 .
we find from here for , @xmath286 mev , @xmath433 , @xmath434 mev the such meanings(for sps ) : @xmath435 mev , , , , , in units @xmath251 , @xmath436 , @xmath437 ( where @xmath438 ) . in this model the entropy of nucleons and strange baryons is the same as in quasiparticle model : @xmath439 , the entropy of mesons and .
we have here @xmath440 136 , @xmath441 25 . for rhic energy
we find for @xmath246 , @xmath288 mev , @xmath433 , @xmath434 mev the meanings : @xmath442 mev , @xmath443 , @xmath444 , @xmath445 , , @xmath446 , @xmath447 .
we have here , .
one can show , that in constituent quarks phase the relations ( [ eq.24 ] ) here also are fulfilled only with decrease of number of degrees of freedom in the presence of pseudogoldstone states .
we have here by analogy with quasiparticle model @xmath448 , the entropy of baryons @xmath449 , the entropy of mesons is and . that gives the estimation : @xmath450 , @xmath451 . in this simple model
we have not great decrease of normalization of mesons spectra at sps and rhic ( by 12 - 13 % ) in comparison with quasiparticle model , i.e. apparently it takes place within the limits of precision of measurement .
thus we do not have here noticeable difference for spectra of particles in comparison with quasiparticle model .
however this model disagrees with @xmath43 lattice data in region of phase transition .
the values @xmath452 , @xmath453 , @xmath454 decrease very weakly in this case ( for example , at @xmath455 the value @xmath452 decrease only by value @xmath456 ) the coupling strength @xmath422 increase(although not great ) at @xmath7 from above ( unlike effective quasiparticle model ) .
thus we see , that spectra of particles weakly depend on character of phase transition @xmath457 hadrons .
but one can show @xcite , that in this model we have too great energy loss @xmath54 of gluon jet at rhic energy : @xmath458 . that is caused in the main by value of effective coupling strength .
let us consider now the quasiparticle model with phenomenological parametrization of running coupling @xmath12 ( [ eq.8 ] ) we use here the formulas of type ( [ eq.10])-([eq.13 ] ) and of type ( [ eq.14])-([eq.18 ] ) for calculation of thermodynamic values @xmath4 , @xmath459 and @xmath460 , but with coupling @xmath12 ( [ eq.8 ] ) and with @xmath111=1 .
with help of corresponding parameters @xmath56 , @xmath57 it is possible well to fit the new lattice data even near of @xmath3 . for calculation of initial conditions we use the equation of type ( [ eq.23 ] ) .
for example , for possible parametrization @xcite : @xmath57=5.3 , @xmath461=0.73 ( at @xmath84=3 ) we find for rhic energy ( @xmath17=130 gev ) the following meanings : @xmath184 250 mev,@xmath462 2.533 ( i.e. @xmath463 0.509 ) and also @xmath464 1.948 .
we find corresponding values for entropy and net nucleons density : @xmath465 , @xmath466 , @xmath467 , @xmath468 .
from here we have the estimation for initial state : @xmath469 fm . , and by analogy @xmath470 fm .
we have for complete entropy : @xmath471 , i.e. only on @xmath472 the difference from its value in effective quasiparticle model .
thus we have practically identical baryons and mesons spectra in these two models .
however we have calculated the jet quenching in plasma in this model by analogy with effective quasiparticle model @xcite .
the result is dramatic : the energy loss @xmath54 of gluon jet ( for two parametrization of @xmath57 and @xmath461 ) four - five times as much than energy @xmath473 of itself jet ! .
this contradiction is caused by too great value of running coupling @xmath12([eq.8 ] ) in this model , which besides increase at @xmath474 from above .
in this paper we investigate the isentropic evolution of expanding quark - gluon plasma with phase transition at first into plasma part and then hadron part of mixed phase at temperature @xmath3 , and further isentropic expansion up thermal freeze - out at temperature @xmath335 . for investigation of initial plasma phase
we use the quasiparticle model , which allow apparently to sum up partially the perturbative expansion of thermodynamic function of interacting plasma in powers of coupling constant . in the papers
@xcite was used the phenomenological parameterization of coupling constant in quasiparticles model in accordance with lattice data .
however close to @xmath3 the coupling constant is sufficiently great , therefore in the vicinity of the phase transition the perturbative methods are not expected to be reliability . in the paper @xcite
was shown , that at such parameterization the thermal mass of gluon grows in the vicinity of the phase transition when @xmath474 from above .
this do not agrees with su(3 ) lattice data for debye mass . in the paper @xcite
was considered the phenomenological model of confinement , where the decrease of thermodynamic values at @xmath7 from above is caused no increase of mass , but by decrease of the number of active degrees of freedom ( i.e. by modification @xmath475 ) .
it is supposed , that behaviour of thermal gluon mass @xmath64 is similar with behaviour of debye mass , i.e. it decreases with @xmath474 from above .
this gives the good description of su(3 ) lattice data .
the similar modification is used also for quasiparticle model with quarks .
it is assumed that debye gluon mass have form and it has a small gap at @xmath476 in accordance with su(3 ) lattice data .
it is assumed also that the proportionality of type ( [ eq.5 ] ) between @xmath477 and thermal gluon mass @xmath104 remains true in the vicinity of phase transition .
we use such model for description of isentropic evolution in common case of quasiparticles with @xmath478 for calculation of physical characteristic of initial plasma phase : @xmath1 and @xmath2 . in result
we have obtained the formula ( [ eq.23 ] ) for definition of these values , where we use the approximation ( [ eq.20 ] ) for @xmath479 and experimental meaning of average number of net nucleons @xmath480 in central region of rapidity at sps and rhic energies . from conservation of entropy and of number of net nucleons
follow also the relations ( [ eq.24 ] ) , which must be fulfilled simultaneously with ( [ eq.23 ] ) .
the relations ( [ eq.23 ] ) , ( [ eq.24 ] ) depend on effective coupling constant @xmath88 , which is determined by asymptotic value of thermal gluon mass at @xmath481 @xcite .
for that we use the asymptotic value of running coupling @xmath422 @xcite . in result were obtained the initial meaning @xmath1 , @xmath2 at sps and rhic energy : for sps we have @xmath204 mev , @xmath482 @xmath213 , ( i.e. @xmath483 fm ) and @xmath484 @xmath213 ( i.e.@xmath485 fm ) , and for rhic @xmath486 mev , @xmath487 @xmath213 and @xmath488 @xmath213 , i.e. @xmath489 fm , @xmath490 fm .
we have also @xmath491 and @xmath492 for sps and rhic correspondingly .
when the temperature is lowered ( @xmath7 ) , more and more gluons form heavy glueballs , and the weaker becomes the interaction .
this state can be interpreted as superposition of nonperturbative multigluon and weak glueball exchange @xcite .
the interaction between the quarks also becomes weaker .
it is possible to interpret , that there is appearance of massive constituent quarks , while the multigluon state is absorbed into constituent quarks masses .
further we show , that massive constituent quarks appears with decrease of number of degrees of freedom in the presence of octet of pseudogoldstone states .
that follow from conservation of entropy and of number of net nucleons .
the hadron part of mixed phase appears also with decrease of number of degrees of freedom the same as for constituent quarks .
however at thermal freeze - out ( at @xmath493 ) hadrons appears already with normal number of degrees of freedom .
thus we have at sps the short plasma phase . with above - mentioned parameters we have calculated in effective quasiparticle model the some typical baryons and mesons spectra with account of resonance
we use for heavy nuclei the temperature of thermal freeze - out @xmath407 mev ( as in `` standard '' model @xcite the normalization of baryons and mesons spectra in the present model is no free parameter , but defined by initial condition in quark - gluon plasma . one should note that there are the models for description of particle production with simultaneous chemical and thermal freeze - out @xcite , where 2 - 3 free parameters describe well the spectra at rhic energy . in the present work we use `` standard '' scenario with two freeze - out - chemical and thermal .
the main object of present work is no study the whole spectra at sps and rhic , but the study of dependence of spectra from initial state and from phase transition stage .
for this aim we consider in sec.[sec5 ] the analogous isentropic problem by the use of ordinary perturbative theory up to order @xmath50 , where @xmath422 is running coupling .
we do not find in this approximation of noticeable difference for particles spectra in comparison with effective quasiparticle model , although this model disagrees with su(3 ) lattice data in region of phase transition ( unlike quasiparticle model ) .
however it can be shown @xcite that in this model we have the too great energy loss of gluon and quark jets in plasma ( @xmath494 ) , that contradict to suppression of @xmath419-spectra , reported by phenix @xcite . in this section we consider also the quasiparticle model with phenomenological parametrization of coupling @xmath12 , which gives good fit of new lattice data .
we investigate here also the initial condition in plasma at rhic energy ( on the basic of equation of type([eq.23 ] ) .
we find that baryons and mesons spectra in this model practically do not differ from spectra in effective quasiparticle model .
however the situation with energy loss in plasma here yet more dramatic - the energy loss of gluon jet exceed the energy of itself jet .
thus apparently we have correct quantitative description of energy loss of gluon and quark jets in plasma only in considering here effective quasiparticle model .
but of course the question about nonperturbative solution for thermodynamic values in quark - gluon plasma remains open ( for example , for thermal mass of quasiparticles and running coupling and so on ) .
unfortunately in the vicinity of the phase transition is used the phenomenological parameterization for these values . in this model
we do not consider also the intrinsic volume of particles .
we had investigated in the main the ratios of thermodynamic values .
we use the models @xcite , where effects due to excluded volume correction cancel out in ratios .
it should be noted also the decrease of protons suppression for moderate @xmath421 at rhic energy @xcite it is possible to use for this problem the hydrodynamic spectra in quasiparticle model for @xmath495 and @xmath379 with accounting of resonance decays . 40 p.braun-mnzinger et.al .
, hep - ph/0105229 v. 2 ( 2001 ) , phys .
lett . b**365 * * , ( 1996 ) j.cleymans , k.redlich , phys .
c * 60 * , 054908 ( 1999 ) c. zhai and b. kastering , phys .
d * 52 * , 7232 ( 1995 ) a.peshier et.al .
d * 54 * , 2399 ( 1996 ) m.i.gorenstein and s.n.yang , phys .
d * 52 * , 5206 ( 1995 ) a.peshier , b.kmpfer and g.soff , phys .
c * 61 * , 045203 ( 2000 ) , hep - ph/0106090 j.i.kapusta , finite - temperature field theory ( cambridge university , cambridge , 1989 ) o.kaczmarek et.al . ,
d * 62 * , 034021 ( 2000 ) j .- p.blaizot , e.iancu and a.rebhan , phys .
d * 63 * , 065003 ( 2001 ) r.a.schneider and w.weise , phys .
c * 64 * , 055201 ( 2001 ) e.shuryak , phys .
lett . * 68 * , 3270 ( 1992 ) , + e.shuryak , l.xiong , phys . rev
. lett . * 70 * , 2241 ( 1993 ) y.a .
tarasov , hep - ph/0312252 a.peshier , phys .
d * 63 * , 105004 ( 2001 ) k.k .
szabo , a.t.toth , hep - ph/0302255 d.w.duke , j.owens , phys .
d * 30 * , 49 ( 1984 ) j.d.bjorken , phys .
d * 27 * , 140 ( 1983 ) r.c.hwa , k.kajantie , phys .
d * 32 * , 1109 ( 1985 ) f.karsch , h.satz , cern - th 5900/90 bi - tp 90/42 h.satz , cern - th 6216/91 f.abe et.al .
d * 41 * , 2330 ( 1990 ) h.appelshauser et.al .
letters * 82 * , 2471 ( 1999 ) n.xu , m.kaneta , nucl .
a * 698 * , 306c ( 2002 ) , c.adler et.al .
letters * 87 * , 262302-i ( 2002 ) e.laermann , nucl .
a * 610 * , 1c ( 1996 ) , f.karsch , nucl.phys .
b ( proc.suppl . ) * 83 - 84 * , 14 ( 2000 ) b.b.back et al . ,
a * 698 * , 54c ( 2002 ) .
s.gasiorowicz , j.l.rosner , am .
j. phys .
* 49 * , 954 ( 1981 ) m.lavelle et.al .
report * 279 * , n1 ( 1997 ) a.v.smilga , phys .
report * 291 * , n1 ( 1997 ) g.roland et.al . , na49 coll .
a * 638 * , 91c ( 1998 ) * * e.andersen et.al .
b * 433 * , 209 ( 1998 ) i.g.bearden et.al .
a * 638 * , 419c ( 1998 ) , phys .
lett . * 78 * , 2080 ( 1997 ) b.kmpfer , hep - ph/9612336 j.sollfrank , p.huovinen , p.v.ruuskanen , eur .
j. c * 6 * , 525 ( 1999 ) yu.a.tarasov , hep - ph/0307300 m.aggarwal et.al .
nucl . phys .
a * 610 * , 200c ( 1996 ) , phys .
lett . * 81 * , 4087 ( 1998 ) j.sollfrank , p.koch , u.heinz , phys .
c * 56 * , 3625 ( 1997 ) w.broniowski , w.frolkowski , phys .
lett . * 87 * , 272302 ( 2001 ) j.cleymans et.al . , z.phys .
c * 33 * , 151 ( 1986 ) j.cleymans , h.oeschler and k.redlich nucl - th/9809031 p.g.jones et.al .
na49 coll .
a * 610 * , 188c ( 1996 ) k.adcox et.al .
phenix coll .
* 88 * , 022301 ( 2002 ) t.matsui,b.svetitsky,l.d.mclerran , phys .
d * 34 * , 2047 ( 1986 ) | we use quasiparticle description of deconfined matter in nuclear collisions at finite temperature and chemical potential .
we assume that evolution of expanding system is isentropic . using theoretical formulas and experimental meaning for the average multiplicity of charged and neutral particles and also of net nucleons ( @xmath0 ) per unit central rapidity
, we calculate the initial temperature @xmath1 , and the volume @xmath2 for the collisions of heavy nuclei ( pb+pb and au+au ) at sps and rhic energies . at calculations we use the conservation of entropy and of number net nucleons in initial plasma stage and on the stage of phase transition at temperature @xmath3 , where we take into account the phase of constituent quarks @xmath4 and @xmath5 .
we use at calculation the effective quasiparticle model with decrease of thermal gluon mass @xmath6 at @xmath7 from above .
this model agrees with lattice data .
the particle ratios for sps and rhic are defined by early chemical freeze - out at temperature close to @xmath3 ( at values of chemical potential @xmath8 mev and 50 mev for sps and rhic ( at @xmath9 gev ) correspondingly ) .
the spectra of baryons and mesons are calculated at temperature of thermal freeze - out @xmath10 mev .
the quantitative characteristic of these spectra ( for example , the normalization for baryons and mesons ) are determined in the present model by initial state - quark - gluon plasma . at calculation of spectra
we assume that averaged transverse flow velocity @xmath11 have both direct particles and paternal resonance .
we show that values of @xmath11 increase for energy rhic in comparison with sps .
we had investigated also the analogous problem of nuclear collisions in quasiparticle model with phenomenological parametrization of running coupling @xmath12 , which increase at @xmath7 from above ( in difference from effective quasiparticle model ) .
we have here agreement with new lattice data .
we do not found in this model quantitative difference for spectra in comparison with effective quasiparticle model .
however it can be shown , that in this model we have dramatic situation for energy loss of gluon jets in plasma . |
in the last half century , considerable studies have been made on the boltzman - gibbs - shannon entropy and the fisher information entropy ( matrix ) , both of which play important roles in thermodynamics and statistical mechanics of classical and quantum systems @xcite-@xcite .
the entropy flux and entropy production have been investigated in connection with the space volume contraction @xcite . in the information geometry @xcite , the fisher information matrix provides us with the distance between the neighboring points in the rieman space spanned by probability distributions .
the fisher information matrix gives the lower bound of estimation errors in the cramr - rao theorem . in a usual system consisting of @xmath0 particles ,
the entropy and energy are proportional to @xmath0 ( _ extensive _ ) , and the probability distribution is given by the gaussian distribution belonging to the exponential family . in recent year , however , many efforts have been made for a study on _ nonextensive _ systems in which the physical quantity of @xmath0 particles is not proportional to @xmath0 @xcite
. the nonextensivity has been realized in various systems such as a system with long - range interactions , a small - scale system with large fluctuations in temperature and a multi - fractal system @xcite .
tsallis has proposed the generalized entropy ( called the tsallis entropy hereafter ) defined by @xcite @xmath1 where @xmath2 is the entropic index , @xmath3 denotes the probability distribution of a state @xmath4 at time @xmath5 , the boltzman constant @xmath6 is hereafter unity and @xmath7 expresses the @xmath2-logarithmic function defined by @xmath8 .
the tsallis entropy accounts for the nonextensivity of the entropy in nonextensive systems . in the limit of @xmath9
, @xmath7 reduces to the normal @xmath10 and then @xmath11 agrees with the boltzman - gibbs - shannon entropy expressed by @xmath12 the probability distribution derived by the maximum - entropy method ( mem ) with the use of the tsallis entropy is given by non - gaussian distribution @xcite , which reduces to the gaussian and cauchy distributions for @xmath13 and @xmath14 , respectively .
many authors have discussed the fisher information matrix in nonextensive systems @xcite-@xcite . in order to derive the _
generalized _ fisher information matrix g whose components are given by @xcite-@xcite @xmath15 the generalized kullback - leibler distance of @xmath16 between the two distributions @xmath17 and @xmath18 has been introduced : @xmath19 with @xmath20\ : dx , \nonumber \\ & = & - \:\frac{1}{(q-1 ) } \left [ 1-\int p(x)^q\:p'(x)^{1-q } \:dx \right ] , \label{eq : a6}\end{aligned}\ ] ] where @xmath21 and @xmath22 denotes a set of parameters specifying the distribution . in the limit of @xmath23 , @xmath24 given by eq .
( [ eq : a4 ] ) reduces to the conventional fisher information matrix .
it should be remarked that csiszr @xcite had proposed the generalized divergence measure given by @xmath25 dx , \label{eq : a7}\end{aligned}\ ] ] where @xmath26 is assumed to be a convex function with the condition @xmath27 . for @xmath28 , eq .
( [ eq : a7 ] ) yields the conventional kullback - leibler divergence @xcite given by @xmath29 dx .
\label{eq : a8}\end{aligned}\ ] ]
equation ( [ eq : a7 ] ) for @xmath30 leads to the generalized kullback - leibler distance given by eqs .
( [ eq : a5 ] ) and ( [ eq : a6 ] ) . the generalized divergence given by eq .
( [ eq : a6 ] ) , which is in conformity with the tsallis entropy , is equivalent to the @xmath31-divergence of amari @xcite with @xmath32 @xcite .
the escort probability and the generalized fisher information matrix are discussed in refs .
the fisher information entropy in the cramr - rao inequality has been studied for nonextensive systems @xcite .
extensive studies on the tsallis and fisher entropies have been made for reaction - diffusion systems , by using the mem with exact stationary and dynamical solutions for nonlinear fpe @xcite .
these studies nicely unify the concept of normal , super- and sub - diffusions by a single picture .
the purpose of the present paper is to investigate the stationary and dynamical properties of the information entropies in the coupled langevin model which has been widely adopted for a study of various stochastic systems ( for a recent review , see @xcite ) . the langevin model subjected to multiplicative noise
is known to be one of typical nonextensive systems @xcite .
recently the coupled langevin model subjected to additive and multiplicative noise has been discussed with the use of the augmented moment method @xcite which is the second - moment method for local and global variables @xcite .
we will obtain the probability distribution of the nonextensive , coupled langevin model by using the fokker - planck equation ( fpe ) method with the mean - field approximation .
we have made a detailed study on effects on the stationary information entropies of additive and multiplicative white noise , external force , input signal , couplings and the number of constituent elements in the adopted model . by solving the fpe both by the proposed analytical scheme and by the partial difference equation ( pde ) method , we have investigated the transient responses to an input signal and an external force which are applied to the stationary state .
the outline of the paper is as follows . in sec .
2 , we describe the adopted , @xmath0-unit coupled langevin model .
analytical expressions for the tsallis entropy and generalized fisher information entropy in some limiting cases are presented .
numerical model calculations of stationary and dynamical entropies are reported . in sec . 3 ,
discussions are presented on the entropy flux and entropy production and on a comparison between @xmath2-moment and normal - moment methods in which averages are taken over the escort and normal distributions , respectively .
section 4 is devoted to our conclusion . in the appendix
, we summarize the information entropies calculated with the use of the probability distribution derived by the mem .
the cramr - rao inequality in nonextensive systems is shown to be expressed by the _ extended _ fisher information entropy which is different from the _ generalized _ fisher entropy .
we will discuss effects of additive and multiplicative _ colored _ noise on information entropies , by using the result recently obtained by the functional - integral method @xcite .
we have adopted the @xmath0-unit coupled langevin model subjected to additive and multiplicative white noise given by @xmath33 with @xmath34+i(t ) .
\hspace{1cm}\mbox{($i=1 $ to $ n$ ) } \label{eq : f2}\ ] ] here @xmath35 and @xmath36 denote arbitrary functions of @xmath4 , @xmath37 the coupling strength , @xmath38 an external input , @xmath31 and @xmath39 are the strengths of multiplicative and additive noise , respectively , and @xmath40 and @xmath41 express zero - mean gaussian white noises with correlations given by @xmath42 we have adopted the mean - field approximation for @xmath43 given by @xmath44+i(t ) , \label{eq : f3}\end{aligned}\ ] ] with @xmath45 , \label{eq : f4}\end{aligned}\ ] ] where the @xmath46 $ ] expresses the average over the escort distribution to be shown below [ eqs .
( [ eq : g3])-([eq : g4 ] ) ] .
owing to the adopted mean - field approximation given by eq .
( [ eq : f3 ] ) , each element of the ensemble is ostensibly independent . the total probability distribution of @xmath47 is given by the product of that of each element : @xmath48 where the fpe for @xmath49 in the stratonovich representation is given by @xmath50 p_i(x_i , t ) + \left(\frac{\beta^2}{2}\right ) \frac{\partial^2}{\partial x_i^2}p_i(x_i , t ) \nonumber \\ & + & \left ( \frac{\alpha^2}{2 } \right ) \frac{\partial } { \partial x_i } g(x_i ) \frac{\partial } { \partial x_i } g(x_i ) p_i(x_i , t ) .
\label{eq : g2}\end{aligned}\ ] ] the expectation value of @xmath51 is given by @xmath52,\end{aligned}\ ] ] with @xmath53 & = & \int p_{iq}(x_i , t ) \
: x_i^m \:dx_i , \hspace{0.5cm}\mbox{($m=1,2 $ ) } \label{eq : g3}\end{aligned}\ ] ] where the escort probability distribution @xmath54 is given by @xmath55 it is noted that @xmath51 and @xmath49 are self - consistently determined from eqs .
( [ eq : f3 ] ) and ( [ eq : g2 ] ) . the relevant fluctuation ( variance ) of @xmath56 is given by @xmath57 .
\label{eq : g5}\ ] ] when we adopt @xmath35 and @xmath36 given by @xmath58 where @xmath59 denotes the relaxation rate , the fpe for @xmath3 is expressed by ( the subscript @xmath60 is hereafter neglected ) @xmath61 \ : \frac{\partial}{\partial x } p(x , t)\nonumber \\ & + & \left ( \frac{\alpha^2}{2 } x^2 + \frac{\beta^2}{2 } \right ) \frac{\partial^2}{\partial x^2}p(x , t ) , \label{eq : g8}\end{aligned}\ ] ] with @xmath62 from the fpe given by eq .
( [ eq : g8 ] ) , the stationary distribution is given by @xcite @xmath63 + y(x ) , \label{eq : g10 } \end{aligned}\ ] ] with @xmath64 where the entropic index is given for @xmath65 .
equation ( [ eq : g10 ] ) yields the @xmath2-gaussian distribution given by @xmath66 with @xmath67 where @xmath68 stands for the @xmath2-exponential function defined by @xmath69_+^{1/(1-q)}$ ] where @xmath70_+=y$ ] for @xmath71 and 0 for @xmath72 . some limiting cases of eqs .
( [ eq : g14 ] ) are examined in the following .
\(1 ) for @xmath73 and @xmath74 ( _ i.e. _ additive noise only ) @xmath75 which yield @xmath76 \(2 ) for @xmath77 , @xmath78 , ( _ i.e. _ multiplicative noise only ) @xcite , @xmath79 with @xmath80 where @xmath81 and @xmath82 denotes the gamma and heaviside functions , respectively , and @xmath83 diverges for @xmath84 . for @xmath85 , eqs .
( [ eq : g19 ] ) and ( [ eq : g24 ] ) yield @xmath86 the distribution given by eq .
( [ eq : g19 ] ) has a peak at @xmath87 .
\(3 ) for @xmath77 , @xmath74 , @xmath88 ( _ i.e. _ without coupling and external input ) , @xcite @xmath89 which lead to @xmath90 it is noted that when we adopt normal moments averaged over the @xmath2-gaussian given by @xmath91 & = & \int p(x , t ) \ : x(t)^m \:dx , \label{eq : g25}\end{aligned}\ ] ] in stead of the @xmath2-moments given by eq .
( [ eq : g3 ] ) , its stationary variance is given by @xmath92)^2]=\beta^2/2(\lambda-\alpha^2)$ ] which diverges at @xmath93 @xcite .
with the use of the total distribution of @xmath94 given by eq .
( [ eq : g1 ] ) , the tsallis entropies of single - unit and @xmath0-unit ensembles are given by @xmath95 with @xmath96 eliminating @xmath97 from eqs .
( [ eq : x2 ] ) and ( [ eq : x3 ] ) , we get @xmath98 where @xmath99 . equation ( [ eq : x5 ] ) shows that the tsallis entropy is non - extensive except for @xmath100 , for which @xmath101 reduces to the extensive boltzmann - gibbs - shannon entropy : @xmath102 . substituting the stationary distributions given by eqs .
( [ eq : g16 ] ) , ( [ eq : g19 ] ) and ( [ eq : g22 ] ) to eq .
( [ eq : a1 ] ) , we get the analytic expression for the tsallis entropy of a single unit given by @xmath103 , \hspace{1cm}\mbox{for $ \alpha=0 $ , $ \beta \neq 0 $ } \label{eq : h1}\\ & = & \left ( \frac{1-c_q}{q-1 } \right ) , \hspace{0.5cm}\mbox{for $ \alpha \neq 0 $ } \label{eq : h2}\end{aligned}\ ] ] with @xmath104 where @xmath105 stands for the beta function , and @xmath83 in eqs .
( [ eq : h4 ] ) and ( [ eq : h3 ] ) are given by eqs .
( [ eq : g24 ] ) and ( [ eq : g23 ] ) , respectively .
we consider the generalized fisher information entropy given by @xmath106 from eqs .
( [ eq : g1 ] ) and ( [ eq : h7 ] ) , the generalized fisher entropy for the @xmath0-unit system is given by @xmath107 , \label{eq : x7 } \\ & = & q \int \cdot\cdot \int \left(\sum_i \frac{\partial \ln p_i(x_i)}{\partial \theta } \right)^2 \:\pi_i\ : p_i(x_i)dx_i , \label{eq : x8 } \\ & = & q \sum_i \int \left ( \frac{\partial \ln p_i(x_i)}{\partial \theta}\right)^2 \:p_i(x_i)\:dx_i + \delta g_q , \label{eq : x9 } \\ & = & n g_q^{(1 ) } , \label{eq : x10}\end{aligned}\ ] ] because the cross term @xmath108 of eq .
( [ eq : x9 ] ) vanishes : @xmath109 where @xmath110 stands for the generalized fisher entropy in a single subsystem .
the generalized fisher information entropy is extensive in the nonextensive system as shown in @xcite : @xmath111 the probability distribution @xmath112 obtained by the fpe for our langevin model is determined by the six parameters of @xmath59 , @xmath31 , @xmath39 , @xmath37 , @xmath113 and @xmath0 .
when adopt @xmath114 in eq .
( [ eq : h7 ] ) , for example , we get the generalized fisher entropy given by @xmath115 -e\left [ \left ( \frac{\partial y(x)}{\partial i } \right ) \right]^2 \right),\end{aligned}\ ] ] where @xmath116 $ ] expresses the average over @xmath112 [ eq .
( [ eq : g25 ] ) ] .
alternatively we have adopted the generalized fisher entropy given by @xmath117 which is obtainable for @xmath118 with @xmath119 in eq .
( [ eq : h7 ] ) if @xmath112 is given by the mem [ eqs .
( [ eq : b6 ] ) and ( [ eq : d3 ] ) ] .
although @xmath112 derived by the fpe is not explicitly specified by @xmath120 and @xmath121 , we have employed eq .
( [ eq : h11 ] ) in our following discussion , expecting it is meaningful for both cases of the fpe and mem .
substituting the stationary distributions given by eqs .
( [ eq : g16 ] ) , ( [ eq : g19 ] ) and ( [ eq : g22 ] ) to eq .
( [ eq : h11 ] ) , we get the analytic expression for the generalized fisher entropy for @xmath122 given by @xmath123 where @xmath121 in eqs .
( [ eq : h12 ] ) and ( [ eq : h9 ] ) are given by eqs .
( [ eq : g26 ] ) and ( [ eq : g27 ] ) , respectively .
the adopted langevin model includes six parameters of @xmath59 , @xmath31 , @xmath39 , @xmath37 , @xmath113 and @xmath0 .
the dependence of the tsallis entropy and generalized fisher information entropy on these parameters have been studied by numerical methods .
we have calculated the distribution @xmath112 by the fpe [ eqs .
( [ eq : g14 ] ) and ( [ eq : g15 ] ) ] , and also by direct simulations ( dss ) for the langevin model [ eqs .
( [ eq : f1 ] ) and ( [ eq : f2 ] ) ] with the heun method : ds results are averages of 100 trials .
figures [ figa](a)-[figa](c ) show three examples of the stationary distribution @xmath112 for @xmath124 ( 0.0 , 0.0 ) , ( 0.0 , 0.5 ) and ( 0.5 , 0.5 ) with @xmath125 , @xmath126 , @xmath127 and @xmath128 .
solid curves show the results calculated with the use of the fpe whereas dashed curves those of dss for the langevin equation : both results are in good agreement and indistinguishable . when the coupling strength is increased from @xmath129 to @xmath130 with @xmath131 , the width of @xmath112
is decreased because of a decreased @xmath132 in eq .
( [ eq : g12 ] ) .
when an input of @xmath133 is applied , @xmath112 changes its position by an amount of about 0.5 with a slight variation of its shape : @xmath112 for @xmath134 is not a simple translational shift of @xmath112 for @xmath135 . in the following
, we will discuss model calculations of the dependence on @xmath31 , @xmath39 , @xmath113 , @xmath37 and @xmath0 , whose results are shown in figs . [ figb ] , [ figc ] , [ figd ] , [ fige ] and [ figf ] , respectively ( dotted curves in the frames ( a ) and ( b ) in figs .
[ figb]-[fige ] will be explained in sec .
3.6.1 ) .
* @xmath31 dependence * first we show @xmath120 , @xmath121 , @xmath136 and @xmath118 in figs . [
figb](a ) , [ figb](b ) , [ figb](c ) and [ figb](d ) , respectively , plotted as a function of @xmath137 for @xmath131 ( chain curves ) , @xmath133 ( dashed curves ) and @xmath138 ( solid curves ) with @xmath125 , @xmath127 and @xmath129 .
figure [ figb](a ) shows that the @xmath31 dependence of @xmath120 is very weak .
we note in fig .
[ figb](b ) that for @xmath133 and @xmath138 , @xmath121 is linearly increased with increasing @xmath137 though @xmath121 is independent of @xmath31 for @xmath131 .
figure [ figb](c ) shows that with increasing @xmath137 , @xmath136 is increased with broad maxima at @xmath139 for @xmath138 and at @xmath140 for @xmath133 . with increasing @xmath137 from @xmath141 , in contrast
, @xmath118 is decreased for @xmath133 and @xmath138 with broad minima , whereas @xmath118 is independent of @xmath137 for @xmath131 . for larger
@xmath113 , @xmath136 and @xmath118 have stronger @xmath137 dependence .
* @xmath39 dependence * figures [ figc](a ) , [ figc](b ) , [ figc](c ) and [ figc](d ) show @xmath120 , @xmath121 , @xmath136 and @xmath118 , respectively , plotted as a function of @xmath142 for @xmath131 ( chain curves ) , @xmath133 ( dashed curves ) and @xmath138 ( solid curves ) with @xmath125 , @xmath126 and @xmath129 . with increasing @xmath39
, @xmath120 has no changes although @xmath121 is linearly increased . with increasing @xmath142 from @xmath143 , @xmath136 ( @xmath118 )
is significantly increased ( decreased ) .
this trend is more significant for @xmath131 than for @xmath133 and @xmath138 . * @xmath113 dependence * the @xmath113 dependence of @xmath120 , @xmath121 , @xmath136 and @xmath118 are shown in figs . [
figd](a)-[figd](b ) for @xmath144 ( chain curves ) , @xmath126 ( dashed curves ) and @xmath145 ( solid curves ) with @xmath125 .
the gradient of @xmath120 versus @xmath113 is slightly larger for larger @xmath31 . in the case of @xmath144 , @xmath121 , @xmath136 and @xmath118
are independent of @xmath113 [ see eqs .
( [ eq : h1 ] ) and ( [ eq : h8 ] ) ] . with increasing @xmath113 for finite @xmath31
, @xmath136 is increased while @xmath118 is decreased . * @xmath37 dependence * we show the @xmath37 dependence of @xmath120 , @xmath121 , @xmath136 and @xmath118 in figs .
[ fige](a)-[fige](b ) , for @xmath131 ( chain curves ) , @xmath133 ( dashed curves ) and @xmath138 ( solid curves ) with @xmath125 , @xmath126 , @xmath127 and @xmath128 .
we note that @xmath120 is independent of @xmath37 . with increasing @xmath37 , @xmath121 and @xmath136
are linearly decreased whereas @xmath118 is increased . * @xmath0 dependence * figure [ figf ] shows the tsallis entropy per element , @xmath146 , given by [ eq .
( [ eq : x6 ] ) ] @xmath147 for @xmath144 ( dotted curve ) , @xmath148 ( solid curve ) , @xmath126 ( dashed curve ) and @xmath145 ( chain curve ) with @xmath125 , @xmath127 , @xmath131 and @xmath129 .
note that for @xmath144 ( @xmath100 ) , the system is extensive because @xmath149 . for finite @xmath31 ,
however , it is nonextensive : @xmath146 is more significantly decreased for larger @xmath31 , though the generalized fisher information entropy @xmath118 is extensive [ eq . ( [ eq : h10 ] ) ] . in order to discuss the dynamical properties of the entropies
, we have to calculate the time - dependent probability @xmath3 , solving the fpe given by eq .
( [ eq : g8 ] ) . in the case of @xmath100
, we may obtain the exact solution of the gaussian distribution given by @xmath150 ^ 2/2 \sigma_1(t)^2 } , \label{eq : k1}\ ] ] where @xmath151 and @xmath152 satisfy equations of motion given by @xmath153 in order to obtain an analytical solution of the fpe for @xmath154 , we have adopted the following method : \(1 ) starting from an equation of motion for @xmath155th @xmath2-moment of @xmath156 $ ] given by @xmath157}{d t } & = & \frac{d}{d t } \int p_q(x , t)\:x^n \:dx,\\ & = & \frac{q}{c_q}\int \left ( \frac{\partial p(x , t)}{\partial t}\right)\:p(x , t)^{q-1}\:x^n \:dx -\frac{1}{c_q}\left ( \frac{d c_q}{dt } \right ) e_q[x^n ] , \label{eq : l2 } \\ \frac{d c_q}{dt } & = & q \int \left ( \frac{\partial p(x , t)}{\partial t}\right)\:p(x , t)^{q-1}\:dx , \label{eq : l0}\end{aligned}\ ] ] we have obtained equations of motion for @xmath51 ( @xmath158 $ ] ) and @xmath56 ( @xmath159-e[x]^2 $ ] ) , valid for @xmath160 and @xmath161 , as given by @xcite @xmath162 equations ( [ eq : l3 ] ) and ( [ eq : l4 ] ) lead to the stationary solution given by @xmath163 \(2 ) we rewrite the distribution of @xmath112 given by eqs .
( [ eq : g11])-([eq : g15 ] ) in terms of @xmath120 , @xmath121 and @xmath2 , as @xmath164^{\frac{1}{1-q } } e^{y(x ) } , \label{eq : l7 } \end{aligned}\ ] ] with @xmath165 where @xmath83 expresses the normalization factor [ eq .
( [ eq : g15 ] ) ] . in deriving eqs .
( [ eq : l7])-([eq : l9 ] ) , we have employed relations given by @xmath166,\end{aligned}\ ] ] which are obtained from eqs .
( [ eq : g11 ] ) , ( [ eq : l5 ] ) and ( [ eq : l6 ] ) .
\(3 ) then we have assumed that a solution of @xmath3 of the fpe given by eq .
( [ eq : g8 ] ) is expressed by eqs .
( [ eq : l7])-([eq : l9 ] ) in which stationary @xmath120 and @xmath121 are replaced by time - dependent @xmath51 and @xmath56 with equations of motion given by eqs .
( [ eq : l3 ] ) and ( [ eq : l4 ] ) . dotted curves in the frames ( a ) and ( b ) of figs . 3 - 6
express the results of stationary @xmath120 and @xmath121 calculated by eqs .
( [ eq : l5 ] ) and ( [ eq : l6 ] ) for some typical sets of parameters .
they are in good agreement with those shown by solid curves obtained with the use of the stationary distribution of @xmath112 given by eq .
( [ eq : g14 ] ) . as will be shown shortly , the approximate , analytical method given by eqs .
( [ eq : l3 ] ) , ( [ eq : l4 ] ) , ( [ eq : l7])-([eq : l9 ] ) provides fairly good results for dynamics of @xmath51 , @xmath56 and @xmath11 , and also for that of @xmath167 except for the transient period . in order to examine the validity of the analytical method discussed above ,
we have adopted also the numerical method , using the partial difference equation ( pde ) derived from eq .
( [ eq : g8 ] ) , as given by @xmath168 \left(\frac{b}{2 a}\right)[p(x+a)-p(x - a ) ] \nonumber \\ & + & \left(\frac{\alpha^2}{2 } x^2 + \frac{\beta^2}{2 } \right ) \left(\frac{b}{a^2}\right)[p(x+a , t)+p(x - a , t)-2p(x , t ) ] , \label{eq : k4}\end{aligned}\ ] ] with @xmath169 where @xmath170 and @xmath171 denote incremental steps of @xmath4 and @xmath5 , respectively .
we impose the boundary condition : @xmath172 with @xmath173 , and the initial condition of @xmath174 where @xmath175 is the stationary distribution given by eqs .
( [ eq : g14 ] ) and ( [ eq : g15 ] ) .
we have chosen parameters of @xmath176 and @xmath177 such as to satisfy the condition : @xmath178 , which is required for stable , convergent solutions of the pde .
* response to @xmath38
* we apply the pulse input signal given by @xmath179 where @xmath180 and @xmath181 denotes the heaviside function : @xmath182 for @xmath183 and zero otherwise .
figure [ figg ] shows the time - dependent distribution at various @xmath5 for @xmath125 , @xmath126 , @xmath127 and @xmath129 .
solid and dashed curves express the results of the pde method and the analytical method ( sec .
3.6.1 ) , respectively .
when input of @xmath184 is applied at @xmath185 , the distribution is gradually changed , moving rightward .
the results of the analytical method are in good agreement with those obtained by the pde method , except for @xmath186 and @xmath187 .
this change in @xmath3 induces changes in @xmath51 , @xmath56 , @xmath11 and @xmath167 , whose time dependences are shown in figs . [
figh](a ) and [ figh](b ) , solid and dashed curves expressing the results of the pde method and the analytical method , respectively . by an applied pulse input , @xmath120
, @xmath121 and @xmath136 are increased while @xmath118 is decreased .
the result for @xmath11 of the analytical method is in fairly good agreement with that obtained by the pde method .
the calculated @xmath167 of the analytical method is also in good agreement with that of the pde method besides near the transient periods at @xmath188 and @xmath189 just after the input signal is on and off .
this is expected due to the fact that @xmath167 given by eq .
( [ eq : h11 ] ) is sensitive to a detailed form of @xmath3 because it is expressed by an integration of @xmath190 over @xmath3 , while @xmath11 is obtained by a simple integration of @xmath191 . for a comparison ,
we show by chain curves , the results of the pde method when the step input given by @xmath192 is applied .
the relaxation time of @xmath136 and @xmath118 is about 2.0 .
it is noted that input signal for @xmath73 induces no changes in @xmath11 and @xmath167 , which has been already realized in the stationary state as shown by chain curves in figs .
[ figd](c ) and [ figd](d ) .
* response to @xmath193 * we modify the relaxation rate as given by @xmath194 which expresses an application of an external force of @xmath195 ( @xmath196 ) at @xmath197 with @xmath198 .
figure [ figi](a ) and [ figi](b ) show the time dependence of @xmath121 , @xmath136 and @xmath118 with @xmath126 , @xmath127 , @xmath131 and @xmath129 for which @xmath199 .
solid and dashed curves express the results of the pde method and the analytical method , respectively . when an external force is applied , @xmath121 and @xmath136
are decreased whereas @xmath118 is increased .
the results of the analytical method are in good agreement with those of the pde method .
the relaxation times of @xmath136 and @xmath118 are 0.47 and 0.53 , respectively .
in the preceding sec .
2 , we have discussed the information entropies by using the probability distribution obtained by the fpe for the langevin model .
it is worthwhile to compare it with the probability distribution derived by the mem .
the variational condition for the tsallis entropy given by eq .
( [ eq : a1 ] ) is taken into account with the three constraints : a normalization condition and @xmath2-moments of @xmath4 and @xmath200 , as given by @xcite @xmath201=\int p_q(x ) \:x \ : dx , \label{eq : b2 } \\
\sigma_q^2 & = & e_q[(x-\mu_q)^2 ] = \int p_q(x ) \:(x-\mu_q)^2\:dx , \label{eq : b3}\end{aligned}\ ] ] where @xmath46 $ ] expresses the average over the escort probability of @xmath202 given by @xmath203 the entropic index @xmath2 being assumed to be @xmath204 . after some manipulations , we get the @xmath2-gaussian ( non - gaussian ) distribution given by @xcite @xmath205 with @xmath206 in the limit of @xmath9 , @xmath112 in eq .
( [ eq : b6 ] ) becomes the gaussian distribution : @xmath207 the probability distribution given by eq .
( [ eq : b6 ] ) derived from the mem is different from that of eq .
( [ eq : g14 ] ) obtained by the fpe although both expressions are equivalent for @xmath208 with @xmath209 .
note that the former is defined for @xmath204 while the latter is valid for @xmath65 .
a comparison between the probability distributions obtained by the fpe and mem is made in figs .
[ figr ] and [ figs ] .
figure [ figr](a ) shows the probability distributions calculated by the fpe of the langevin model for @xmath144 , 0.5 , 1.0 , 1.5 and 2.0 , which yield @xmath210 , @xmath211 , @xmath212 , @xmath213 and @xmath214 , respectively , with @xmath215 for @xmath138 , @xmath125 , @xmath127 and @xmath129 [ eqs .
( [ eq : l5 ] ) and ( [ eq : l6 ] ) ] . figure [ figr](b ) shows corresponding distributions calculated by the mem with the respective parameters of @xmath2 , @xmath120 and @xmath216 . for @xmath144 ( @xmath100 ) , both distributions of the fpe and mem are gaussian centered at @xmath217 . for @xmath218 , @xmath112 of the fpe
becomes asymmetric with respect to @xmath219 while that of the mem is still symmetric .
the peak position of @xmath112 of the mem is at @xmath217 independent of @xmath2 while that of the fpe moves leftward with increasing @xmath31 .
it is noted that @xmath112 of the fpe for @xmath77 and @xmath220 given by eq .
( [ eq : g19 ] ) has a peak at @xmath221 .
figure [ figs](a ) shows @xmath112 of the fpe for various inputs of @xmath131 , 0.5 , 1.0 , 1.5 and 2.0 , which yield @xmath222 , @xmath223 , @xmath224 , @xmath225 and @xmath226 , respectively , for @xmath125 , @xmath145 , @xmath127 and @xmath129 [ eqs .
( [ eq : l5 ] ) and ( [ eq : l6 ] ) ] ; corresponding @xmath112 of the mem with the respective parameters of @xmath227 , @xmath120 and @xmath216 are plotted in fig . [
figs](b ) . for @xmath228 , both
the distributions agree .
although centers of both distributions move rightward with increasing @xmath120 , their profiles and peak positions are different between the two distributions .
we note that the magnitude of @xmath112 at @xmath229 of the fpe is smaller than that of the mem for @xmath230 .
the information entropies calculated with the use of the distribution given by eq .
( [ eq : b6 ] ) are summarized in the appendix .
one of the advantages of the mem is that its distribution is explicitly specified by the parameters of @xmath231=@xmath232 while that of the fpe is given in an implicit way [ _ cf .
( [ eq : l7])-([eq : l9 ] ) ] .
we may discuss the upper bound of estimation errors by the cramr - rao inequality , which is shown to be expressed by the _ extended _ fisher entropy [ eq .
( [ eq : e0 ] ) ] but not by the generalized fisher entropy [ eq .
( [ eq : d0 ] ) ] . in order to discuss the dynamics within the mem for @xmath233
, we have once tried to obtain an analytic solution of its distribution @xmath3 , assuming that it is given by eq .
( [ eq : b6 ] ) : @xmath234 , \label{eq : l1}\end{aligned}\ ] ] where the @xmath2-dependent coefficient @xmath235 is determined from eqs .
( [ eq : b8])-([eq : b10 ] ) , and equations of motion for @xmath51 and @xmath56 are derived so as to meet the fpe after refs . @xcite .
unfortunately , we could not uniquely determined them : we got two equations for @xmath236 and three equations for @xmath237 which are mutually not consistent ( except for @xmath100 ) .
this implies that the exact analytic solution of the fpe is not given by eq .
( [ eq : l1 ] ) . indeed , the exact solution for @xmath238 in eq .
( [ eq : g8 ] ) does not have a functional form given by eq .
( [ eq : l1 ] ) @xcite .
it is interesting to discuss the entropy flux and entropy production from the time derivative of the tsallis entropy given by @xmath239 with @xmath240 \ :
\nonumber \label{eq : p5}\\ & & \end{aligned}\ ] ] here @xmath241 denotes the entropy flux , and @xmath242 and @xmath243 stand for entropy productions due to additive and multiplicative noise , respectively . by using the stationary distribution given by eq .
( [ eq : g22 ] ) , we get @xmath241 , @xmath242 and @xmath243 in the stationary state with @xmath244 ( _ i.e. _ without couplings and external input ) : @xmath245 where @xmath83 is given by eq .
( [ eq : g23 ] ) . equations ( [ eq : p6])-([eq : p8 ] )
satisfy the stationary condition : @xmath246 .
it is worthwhile to examine the limit of @xmath247 ( @xmath248 ) , in which eqs .
( [ eq : p1 ] ) , ( [ eq : p6])-([eq : p8 ] ) yield @xmath249 with @xmath250 with noticing the relation : @xmath251=1 $ ] @xcite , we may see that eqs .
( [ eq : p11 ] ) and ( [ eq : p12 ] ) lead to @xmath252 and @xmath253 in the limit of @xmath9 . in the opposite limit of @xmath254 , eqs .
( [ eq : p6])-([eq : p8 ] ) yields that each of @xmath241 .
@xmath242 and @xmath243 is proportional to @xmath255 and then divergent in this limit , though @xmath256 .
it is noted that @xmath257 for @xmath258 and @xmath254 @xcite .
we present some model calculations of @xmath241 , @xmath242 and @xmath243 in the stationary state , which are shown in fig .
[ figl ] as a function of @xmath31 for @xmath259 ( dashed curves ) , @xmath127 ( chain curves ) and @xmath260 ( solid curves ) .
we note that @xmath261 and @xmath262 . with increasing @xmath31 , @xmath241
is decreased in the case of @xmath259 , while it is increased in the cases of @xmath127 and 1.0 .
bag @xcite showed that @xmath241 is always decreased with increasing @xmath31 which disagrees with our result mentioned above : eqs .
( [ eq : p6])-([eq : p8 ] ) are rather different from eqs .
( 36 ) and ( 37 ) in ref .
@xcite where non - gaussian properties of the distribution is not properly taken into account . in refs .
@xcite , we have discussed equations of motion for normal moments of @xmath263 ( @xmath264 $ ] ) and @xmath265 ( @xmath266 $ ] ) [ eq .
( [ eq : g25 ] ) ] in the langevin model with @xmath267 , as given by @xmath268 these equations of motion are rather different from those for the @xmath2-moments of @xmath120 and @xmath121 given by eqs .
( [ eq : l3 ] ) and ( [ eq : l4 ] ) . indeed , eqs .
( [ eq : l11 ] ) and ( [ eq : l12 ] ) yield stationary normal moments given by @xmath269 which are different from the stationary @xmath2-moments of @xmath120 and @xmath121 given by eqs .
( [ eq : l5 ] ) and ( [ eq : l6 ] ) , and which diverge at @xmath270 and @xmath93 , respectively .
the time dependence of @xmath271 and @xmath272 becomes considerably different from those of @xmath51 and @xmath56 for an appreciable value of @xmath31 .
figure [ figp](a ) , ( b ) , ( c ) and ( d ) show some examples of @xmath51 , @xmath56 , @xmath271 and @xmath272 , respectively , when a pulse input given by eq .
( [ eq : k7 ] ) is applied with @xmath273 ( chain curves ) , @xmath126 ( dashed curves ) and @xmath274 ( solid curves ) . although @xmath51 is independent of @xmath31 , @xmath56 , @xmath271 and @xmath272
are much increased at @xmath197 for larger @xmath31 . in particular , @xmath272 is significantly increased because of the @xmath137 term in eq .
( [ eq : l12 ] ) .
we have so far considered additive and multiplicative _ white _ noise .
in our recent paper @xcite , we have taken into account the effect of _ colored _ noise by employing the functional - integral method .
we have assumed the langevin model subjected to additive ( @xmath275 ) and multiplicative ( @xmath276 ) colored noise given by @xmath277 with @xmath278 , \label{eq : r2 } \\
\frac{d \phi(t)}{d t } & = & -\frac{1}{\tau_m } [ \phi(t)- \alpha \;\eta(t ) ] , \label{eq : r3 } \end{aligned}\ ] ] where @xmath279 and @xmath39 ( @xmath280 and @xmath31 ) express the relaxation time and strength of additive ( multiplicative ) noise , respectively , and @xmath281 and @xmath282 stand for independent zero - mean gaussian white noise . by applying the functional - integral method to the langevin model given by eqs .
( [ eq : r1])-([eq : r3 ] ) , we have obtained the effective one - variable fpe , from which the effective langevin model is derived as @xcite @xmath283 with @xmath284}. \label{eq : r6 } \ ] ] here @xmath271 is given by eq .
( [ eq : l11 ] ) with @xmath285 , from which @xmath286 is determined in a self - consistent way . in the stationary state where @xmath287 given by eq .
( [ eq : l13 ] ) with @xmath288 , we get @xmath286 from eq .
( [ eq : r6 ] ) : @xmath289 } , \label{eq : r7 } \\ & = & \frac{1}{\tau_m } \left[(1+\lambda \tau_m)-\sqrt{(1+\lambda \tau_m)^2 - 2 \tau_m \alpha^2}\right ] .
\label{eq : r8 } \end{aligned}\ ] ] we get an approximate expression given by @xcite @xmath290 which is shown to be a good approximation both for @xmath291 and @xmath292 @xcite . equations ( [ eq : r5 ] ) and ( [ eq : r9 ] ) show that effects of additive and multiplicative colored noise are described by @xmath293 and @xmath294 which are reduced by factors of @xmath295 and @xmath296 ) , respectively , from original values of @xmath137 and @xmath142 . the @xmath279 dependence of @xmath136 and @xmath118 is plotted in fig .
[ figq](a ) and [ figq](b ) with @xmath297 for @xmath131 ( chain curves ) , @xmath133 ( dashed curves ) and @xmath138 ( solid curves ) with @xmath125 , @xmath126 and @xmath127 .
we note that with increasing @xmath279 , @xmath118 is much increased for smaller @xmath113 whereas @xmath136 is much decreased for smaller @xmath113 .
the dependence of @xmath136 and @xmath118 on @xmath279 may be understood from their @xmath39 dependence shown in figs . [ figc](c ) and [ figc](d ) .
the @xmath280 dependence of @xmath136 and @xmath118 is plotted in fig . [ figq](a ) and [ figq](b ) with @xmath298 with @xmath125 , @xmath126 and @xmath127 . with increasing @xmath280 , @xmath136 ( @xmath118 )
is decreased ( increased ) for @xmath133 and @xmath138 while no changes for @xmath131 .
these behavior may be explained from figs . [
figb](c ) and [ figb](d ) showing the @xmath31 dependence of @xmath136 and @xmath118 .
we have discussed stationary and dynamical properties of the tsallis and fisher entropies in nonextensive systems .
our calculation for the @xmath0-unit coupled langevin model subjected to additive and multiplicative noise has shown the followings : \(i ) the dependence of @xmath136 and @xmath118 on the parameters of @xmath59 , @xmath31 , @xmath39 , @xmath113 , @xmath37 and @xmath0 in the coupled langevin model are clarified ( figs .
[ figb]-[figf ] ) , and ( ii ) dynamical properties are well described by the analytical method for the fpe proposed in sec .
2.6.1 , which shows that the relaxation times in transient responses of @xmath136 and @xmath118 to a change in @xmath59 are short ( @xmath299 ) while those in @xmath113 are fairly long ( @xmath300 ) .
the difference between the parameter dependence of @xmath136 and @xmath118 in the item ( i ) arises from the fact that @xmath136 provides us with a global measure of ignorance while @xmath118 a local measure of positive amount of information @xcite .
we have calculated the information entropies also by using the probability distribution derived by the mem , from which we get the followings : \(iii ) @xmath112 derived by the mem is rather different from that of the fpe for @xmath301 ( figs .
[ figr ] and [ figs ] ) , for which the information entropies of the mem are independent of @xmath120 while those of the fpe depend on @xmath120 ( _ i.e. _ @xmath113 ) , and \(iv ) the cramr - rao inequality is expressed by the extended fisher entropy [ eq .
( [ eq : e0 ] ) ] which is different from the generalized fisher entropy [ eq .
( [ eq : d0 ] ) ] derived from the generalized kullback - leibler divergence [ eq .
( [ eq : a6 ] ) ] .
the item ( iv ) has not been clarified in previous studies on the fisher entropies in nonextensive systems @xcite-@xcite .
the langevin model has been employed for a study of a wide range of stochastic systems @xcite .
quite recently , the present author has proposed the generalized rate - code model for neuronal ensembles which is described by the coupled langevin - type equation @xcite
. it would be interesting to discuss the dynamics of information entropies in such neural networks , which is left for our future study .
this work is partly supported by a grant - in - aid for scientific research from the japanese ministry of education , culture , sports , science and technology . with the use of eqs .
( [ eq : a1 ] ) and ( [ eq : b6 ] ) , the tsallis entropy is given by @xmath302 , \hspace{1cm}\mbox{for $ q = 1 $ } \label{eq : c1 } \\ & = & \left(\frac{1-c_q}{q-1 } \right ) , \hspace{2cm}\mbox{for $ q \neq1 $ } \label{eq : c2}\end{aligned}\ ] ] with @xmath303 which yield @xmath304 here @xmath83 for @xmath305 and @xmath306 are given by eqs .
( [ eq : b8 ] ) and ( [ eq : b10 ] ) , respectively . the distribution @xmath112 given by eq .
( [ eq : b6 ] ) is characterized by two parameters of @xmath307 . by using eqs .
( [ eq : a4 ] ) and ( [ eq : b6 ] ) , we obtain the component of the generalized fisher information matrix g given by @xcite-@xcite @xmath308 , \label{eq : d0 } \\ & = & q\:e [ ( x_i - e[x_i ] ) ( x_j - e[x_j ] ) ] , \label{eq : d1 } \hspace{1cm}\mbox{for $ i , j=1,2$}\end{aligned}\ ] ] with @xmath309 , \label{eq : d2}\end{aligned}\ ] ] where @xmath310 $ ] denotes the average over the @xmath2-gaussian distribution of @xmath112 whereas @xmath46 $ ] stands for the average over the escort distribution of @xmath202 . substituting the probability given by eq .
( [ eq : b6 ] ) to eq .
( [ eq : d0 ] ) , we get @xmath311 which yield @xmath312 a similar calculation leads to the ( 2,2)-component given by @xmath313 next we discuss the cramr - rao inequality in nonextensive systems . for the escort distribution given by eq .
( [ eq : b4 ] ) which satisfies eqs .
( [ eq : b2 ] ) and ( [ eq : b3 ] ) with @xmath317 = \int p_q(x)\ : dx,\ ] ] we get the cramr - rao inequality @xcite @xmath318 here @xmath319 denotes the covariance error matrix whose explicit expression will be given shortly , and is referred to as the _ extended _ fisher information matrix whose components are expressed by @xmath320 , \hspace{1cm}\mbox{for $ i , j=1,2 $ } \label{eq : e0}\\ & = & e_q \left [ ( \tilde{x}_i - e_q[\tilde{x}_i ] ) ( \tilde{x}_j - e_q[\tilde{x}_j ] ) \right],\end{aligned}\ ] ] with @xmath321 , \\ & = & q ( x_i - e[x_i ] ) , \label{eq;e2}\end{aligned}\ ] ] @xmath322 being given by eq .
( [ eq : d2 ] ) . note that @xmath323 is different from @xmath24 given by eq .
( [ eq : d0 ] ) except for @xmath100 . the ( 1,1 ) component of is given by @xmath324 , \label{eq : e3 } \\ & = & \left ( \frac{q^2}{c_q } \right ) \int p(x)^q \left ( \frac{\partial \ln p(x ) } { \partial x}\right)^2 \:dx , \label{eq : e4 } \\ & = & \left(\frac{2q^2}{\nu \sigma_q^2 ( q-1)}\right ) \frac{b(\frac{3}{2},\frac{q}{(q-1)}+\frac{1}{2 } ) } { b(\frac{1}{2},\frac{q}{(q-1)}-\frac{1}{2 } ) } , \hspace{1cm}\mbox{for $ 1 < q < 3 $ } \label{eq : e5 } \\ & = & \frac{1}{\sigma_q^2 } , \hspace{7cm}\mbox{for $ q=1 $ } \label{eq : e6 } \\ & = & \left ( \frac{2q^2}{\nu \sigma_q^2 ( 1-q ) } \right ) \frac{b(\frac{3}{2},\frac{q}{(1-q)}-1 ) } { b(\frac{1}{2},\frac{q}{(1-q)}+1 ) } , \hspace{1cm}\mbox{for $ 1/2 < q < 1 $ } \label{eq : e7}\end{aligned}\ ] ] which lead to @xmath325 similarly , the ( 2,2 ) component of @xmath326 is given by @xmath327 , \label{eq : e10 } \\ & = & \frac{(q+1)}{4 ( 2 q-1)\sigma_q^4}. \hspace{1cm}\mbox{for $ 1/2 < q < 3 $ } \label{eq : e9}\end{aligned}\ ] ] the extended fisher information matrix @xmath326 is expressed by @xmath328 whose inverse is given by @xmath329 a calculation of the @xmath330 component ( @xmath331 ) of the covariance error matrix @xmath319 leads to @xmath332 in the limit of @xmath13 , the matrices reduce to @xmath333 chain and solid curves in fig .
[ fign](a ) express the @xmath2 dependence of @xmath335 and @xmath336 , respectively .
when @xmath2 is further from unity , @xmath337 is much decreased and it vanishes at @xmath338 and 3 .
the lower bond of @xmath339 is expressed by the cramr - rao relation because it is satisfied by @xmath340 : @xmath341 chain , dashed and solid curves in fig . [
fign](b ) show @xmath342 , @xmath343 and @xmath344 , respectively .
it is noted that @xmath345 diverges at @xmath346 .
the following relations hold : @xmath347 equation ( [ eq : e12 ] ) means that @xmath348 can not provide the lower bound of @xmath345
. equations ( [ eq : e11])-([eq : e13 ] ) clearly show that the lower bound of @xmath319 is expressed by the extended fisher information matrix @xmath326 , but not by the generalized fisher information matrix @xmath349 .
a. palstino and a. r. plastino , physica a * 222 * , 347 ( 1995 ) . c. tsallis and d. j. bukman , phys .
e * 54 * , r2197 ( 1996 ) .
a. palstino , a. r. plastino , and h. g. miller , physica a * 235 * , 577 ( 1997 ) .
f. pennini , a. r. plastino , and a. plastino , physica a * 258 * , 446 ( 1998 ) .
l. borland , f. pennini , a. r. plastino , and a. plastino , eur .
j. b. * 12 * , 285 ( 1999 ) .
a. r. plastino , m. casas , and a. plastino , physica a * 280 * , 289 ( 2000 )
. s. abe , phys .
e * 68 * , 031101 ( 2003 ) . | the tsallis entropy and fisher information entropy ( matrix ) are very important quantities expressing information measures in nonextensive systems
. stationary and dynamical properties of the information entropies have been investigated in the @xmath0-unit coupled langevin model subjected to additive and multiplicative white noise , which is one of typical nonextensive systems .
we have made detailed , analytical and numerical study on the dependence of the stationary - state entropies on additive and multiplicative noise , external inputs , couplings and number of constitutive elements ( @xmath0 ) . by solving the fokker - planck equation ( fpe ) by both the proposed analytical scheme and the partial difference - equation method ,
transient responses of the information entropies to an input signal and an external force have been investigated .
we have calculated the information entropies also with the use of the probability distribution derived by the maximum - entropy method ( mem ) , whose result is compared to that obtained by the fpe . the cramr - rao inequality is shown to be expressed by the _ extended _ fisher entropy , which is different from the _ generalized _ fisher entropy obtained from the generalized kullback - leibler divergence in conformity with the tsallis entropy . the effect of additive and multiplicative _ colored _ noise on information entropies is discussed also .
= 1.333 * stationary and dynamical properties of information entropies + in nonextensive systems * hideo hasegawa _ department of physics , tokyo gakugei university + koganei , tokyo 184 - 8501 , japan _ ( ) _ pacs no .
_ 05.10.gg , 05.45.-a |
studies of globular cluster ( gc ) systems play a critical role in our understanding of galaxy formation .
until two decades ago it was thought that all gcs were old , and this shaped our view of how galaxies , especially elliptical galaxies , formed ( see @xcite ) . imaging with the _ hubble space telescope ( hst ) _ then revealed that young compact star clusters form copiously in galaxy mergers @xcite , a fact that supports theories in which giant elliptical galaxies are formed through mergers of spirals @xcite . however , the formation and evolution of gc systems is still not well understood .
we should be able to observe how gc systems evolve from the very young systems with power - law luminosity functions into old systems with log - normal luminosity functions , such as are observed in typical elliptical galaxies .
these observations would greatly clarify exactly which processes ( shocking , evaporation , dynamical friction ; e.g. , @xcite ) are important in the evolution of star cluster systems .
finding intermediate - age gc populations in elliptical galaxies has been difficult .
spectroscopy is often thought to be a relatively accurate method for measuring cluster ages and metallicities .
however , in practice spectroscopy is very time consuming , even for 8m - class telescopes , leaving only the brightest gcs open to such analyses .
@xcite spent 9.3 hours observing nine gc candidates in the e5 galaxy ngc 3610 with keck , and found only one young gc .
the majority of candidate gcs in galaxies identified via morphology and color to be intermediate - age merger remnants are too faint for spectroscopic study .
photometric studies have also had limited success , partly because most have been done only at optical wavelengths ( specifically , in the @xmath1 and @xmath2 passbands ) .
the problem is that the colors of metal - rich gcs with ages of 13 gyr overlap with the colors of @xmath013 gyr old metal - poor gcs , resulting in an age - metallicity degeneracy ( * ? ? ?
fortunately , the combination of optical and near - infrared ( nir ) photometry can break this age
metallicity degeneracy .
this is because optical - to - nir indices are sensitive to metallicity , but only weakly dependent on age ( e.g. , * ? ? ?
* ; * ? ? ?
physically this works because in older stellar populations the @xmath1-band predominantly samples the light of stars near the turn - off , while the -band is more sensitive to cooler stars on the giant branch .
the color of the giant branch is metallicity - dependent , so the index is primarily a metallicity indicator .
thus , we can lift the age metallicity degeneracy by plotting the locations of simple stellar populations ( ssp ) in a color color diagram , specifically vs. .
the four early - type galaxies ngc 1700 , ngc 2865 , ngc 4382 , and ngc 7727 were selected based on their tidal fine - structure content , blue @xmath3 colors , and enhanced balmer absorption lines @xcite . the sample spans the morphology and age gap between on - going mergers and normal elliptical galaxies .
ngc 4382 is similar to the merger remnants ngc 7252 , ngc 3921 , and ngc 7727 , but lacks tidal tails .
its rich fine structure and distortions ( see fig . 1 in @xcite and fig
. 9 in @xcite ) suggest that it experienced a merger 12 gyr ago , a hypothesis strengthened by its still bluish center @xcite , double - peaked nucleus @xcite , and counter - rotating core @xcite .
it was selected specifically to bridge the age gap between ongoing mergers like ngc 4038/39 or ngc 3256 and normal old ellipticals .
the elliptical galaxies ngc 1700 and ngc 2865 both show evidence for recent interactions or mergers and are therefore candidate `` young ellipticals . ''
ngc 1700 features boxy isophotes , a strong isophotal twist , and two broad arms or tails that have long suggested it may be the remnant of a major merger of two disk galaxies ( @xcite , esp .
1b ; @xcite ) .
its kinematics supports this hypothesis : it rotates fast @xcite , yet features a counterrotating core with a distinctly younger stellar population @xcite .
@xcite found that its gcs are consistent with an age of @xmath44 gyr , but do not necessarily form a population of obviously intermediate age .
ngc 2865 features bright irregular shells , a luminous plume of material , and a faint outer loop , all suggesting that it is the remnant of a relatively recent merger ( see fig . 5 of @xcite and fig . 1 of @xcite )
this hypothesis is supported by gas fragments associated with the optical fine structure @xcite , a rapidly rotating small central core with a younger stellar population and gas @xcite , and a body with unusually blue colors .
@xcite note that in a color histogram the galaxy s gcs seem to contain a subpopulation bluer than normal metal - poor gcs and , hence , likely of much younger age ( @xmath40.51 gyr ) . finally , ngc 7727 (= vv67 = arp 222 ) is so obviously tidally disturbed that it was included by both @xcite and @xcite in their catalogs of interacting galaxies . because of its spiral - shaped tidal tails it was originally classified as a peculiar sa galaxy @xcite , but was then reclassified as a recent merger remnant of age @xmath41.3 gyr by @xcite .
the presence of bluish young star clusters in it was noted by @xcite and described in some more detail by @xcite .
we obtained -band images of the above four early - type galaxies with the near - infrared imager and spectrometer ( niri ) on gemini north @xcite and with flamingos -
i on gemini south @xcite , with the results shown in figure [ fig : ngc ] .
the program ids were gn-2006b - q-97 and gs-2002a - dd-4 , respectively .
we used niri in f/6 imaging mode , which yields an image scale of @xmath5/pixel and a field of view of @xmath6 , and flamingos - i also in imaging mode , which yields a scale of @xmath7/pixel and a field of view of @xmath8 . both instrument
have similar fields of view , comparable to those of the wfpc2 and acs cameras on _
all four galaxies were also observed with _ hst _ in @xmath1 and @xmath2 ( go-7468 , pi : schweizer ) , and we used unpublished photometry from that project . in the case of ngc 4382 , we used the most recent photometric data published by @xcite .
table [ table0a ] summarizes our observations and the data from other authors used in this study .
lcccc ngc 1700 & 52 & @xmath9 & _ hst_/wfpc2 & @xcite + & & & gemini / niri & this paper + ngc 2865 & 35 & @xmath9 & _ hst_/wfpc2 & schweizer ( cycle 7 ) + & & & gemini / niri & this paper + ngc 4382 & 16 & @xmath9 & _ hst_/acs & @xcite + & & & gemini / niri & this paper + ngc 7727 & 26 & @xmath9 & hst / wfpc2 & @xcite + & & & gemini / flamingos - i & this paper [ table0a ] the data were reduced with the gemini / niri package of tasks within iraf .
normalized flats were constructed from images taken with the gemini calibration unit .
flat - field images with the ir lamp on and off allowed separation of the instrumental thermal signature from the sensitivity response .
the near - ir sky level and structure varies on time scales of a few minutes . to account for such variations we used a dither pattern in our imaging sequences , which provided blank fields close enough in time to derive the sky levels reliably .
some of the niri frames contain an electronic pattern visible as vertical striping with a period of eight columns and not always present in all quadrants . before including the affected frames in our image processing list we used the stand - alone python routine _
nirinoise.py _ which almost perfectly removed the striping .
the same image reduction steps were applied to the standard star observations as well . all reduced science and standard star exposures of the corresponding nights were registered to a common coordinate system with the task _ imcoadd _ , based on geometric solutions obtained with the task _ geomap _ for niri .
the final science image for each galaxy was derived by averaging individual images , scaled to the galaxy image from the most photometric night , with the iraf task _
to detect as many gcs as possible from our -band images and match them with existing _ hst
_ photometry , we first created a model image of each galaxy with the tasks _ ellipse _ and _ bmodel _ of the _ isophote _ package in iraf .
this model image was then subtracted from the original combined galaxy image .
this procedure facilitated the uniform detection of point sources at a detection threshold of 3@xmath10 above the background , using the task _ daofind _ of the _ daophot _ package .
figure [ fig : technique ] shows an example of this technique .
the photometry of the candidate gcs in our sample of merger remnants was calibrated to on the standard system via 14-aperture photometry matching that used in the 2mass survey .
this method was chosen because the standard stars observed with gemini indicated significant variability in the zero point during our observations .
the conditions that these galaxies were observed under were iq70 and , therefore , not guaranteed to be photometric .
the flamingos - i observations , on the other hand , showed a high level of agreement with the calibration via the 2mass stars ; coincidently , this night had very good photometric conditions .
we matched the optical _ hst _ observations ( already reduced and calibrated ) with our infrared gc candidates using the iraf tasks _ geomap _ and _ geotran_. the brightest common objects in both images served as points of reference .
the initial selection of gc candidates was largely based on applying color limits to the objects identified in the @xmath1 , @xmath2 , and images .
however , it was expected that this method alone would prove insufficient to remove contaminants such as background galaxies and foreground stars . therefore , the photometric completeness of the data was determined , and a visual inspection of each gc candidate was performed . since this study utilized space - based @xmath1 and @xmath2 images , but ground - based images , it was expected that the images would set the completeness limit for the photometric data .
hence , a set of completeness curves were generated from the images obtained with niri and flamingos - i for each galaxy .
these curves were created through the addition of artificial cluster images to each galaxy image , using the daophot task _
addstar_. for each of the galaxies ngc 1700 , ngc 2865 , and ngc 7727 , 100,000 artificial cluster images ranging from 15 to 25 magnitudes were added , with no more than 100 clusters added to any one image to prevent overcrowding . in the case of ngc 4382 , which exhibits an already relatively crowded field of point sources ,
only 50 artificial clusters were added to each image , effectively halving the total number of artificial clusters studied .
the daophot tasks _ daofind _ and _ phot _ were then used to attempt to recover these artificial clusters , and to perform photometry on them .
this process was repeated for background values of 25 , 65 , 85 , 125 , and 250 counts per pixel .
this process allowed us to generate two kinds of completeness curves for the gcs of each galaxy .
curves of the first kind represented the fraction of artificial clusters that were successfully recovered by the task _
daofind_. completeness curves of the second kind represented the fraction of artificial clusters that were not only successfully recovered by _ daofind _ , but that also returned a photometric magnitude within acceptable limits via the task _
phot_. there were two of these limits : the returned photometric magnitude was required to have an error of @xmath110.3 mag and was also required to be within 0.5 mag of the magnitude of the input artificial - cluster image .
a smoothed curve based on the completeness function given in @xcite was fit to each set of completeness curves generated .
the two parameters returned by this fitting procedure were the 50% completeness limit , ( expressed in magnitudes ) , and a steepness parameter @xmath12 .
table [ tablecplte ] presents for each galaxy and background value the two fit parameters derived , both for the first kind of completeness curves ( subscript `` fnd '' for `` found '' ) and for the second kind ( subscript `` acc '' for `` accepted '' ) .
as mentioned above , several completeness curves were generated for each galaxy in order to determine the stability of the photometric completeness limit over a range of background values . given that the background varied considerably over any one image , particularly across the center of the galaxy , such stability was essential .
indeed , the variance in background over each image considerably affected the point - spread functions ( psfs ) that we attempted to construct . in order to negate this effect , the final psfs that were used in the artificial cluster addition
were generated from background - subtracted gc images , as illustrated in figure [ fig : psf ] .
these background - subtracted gc images were created using a @xmath13-pixel median subtraction .
for each panel.,scaledwidth=40.0% ] > from the results presented in table [ tablecplte ] it is evident that although the steepness of the completeness curves did vary
the 50% completeness limit was exceptionally stable over the range of background values investigated .
therefore , we were able to use a single completeness limit for each image , despite the fact that the background values varied dramatically across the cores of the galaxies
. figure [ fig : ngc1700c ] shows the completeness curve derived for gcs in ngc 1700 ( shown for accepted artificial clusters only ) . as table [ tablecplte ]
details , the 50% completeness limits were @xmath1421.91 for both the recovered and the accepted clusters at all background levels . hence , it is evident that photometric errors do not play a significant role in setting the depth of completeness in this case .
rather , the completeness is limited primarily by our ability to detect the clusters .
figure [ fig : ngc2865c ] shows the completeness curve derived for gcs in ngc 2865 . in this galaxy , the 50% completeness limits were @xmath1521.52 for both the recovered and the accepted artificial clusters . as in the case of ngc 1700 , the very small range of these values indicates that the completeness in was limited by our ability to detect the clusters rather than by photometric errors .
figure [ fig : ngc4382c ] shows the completeness curve derived for gcs in ngc 4382 .
again , the 50% completeness limits are similar for the recovered and accepted curves , though with wider scatter than in ngc 1700 and ngc 2865 and with a 0.06 mag difference between the two mean values ( see table [ tablecplte ] ) .
figure [ fig : ngc7727c ] shows the completeness curve derived for gcs in ngc 7727 . as table [ tablecplte ]
details , the various 50% completeness limits scatter more and show a significant systematic difference of 0.09 mag between the recovered artificial clusters ( mean @xmath16 ) and the accepted clusters ( mean @xmath17 ) . notice that these limits for the ngc 7727 gcs are more than 0.5 mag fainter than the limits for gcs in the other three galaxies , which is not surprising given that the images of ngc 7727 were obtained with flamingos -
i on a night of particularly good seeing .
it is because of these fainter limits that the photometric errors are more influential in ngc 7727 than in the other three galaxies .
nevertheless , we adopted the single value of @xmath17 as our acceptance criterion for candidate gcs in ngc 7727 .
lccccc 1700 & 25 & 21.91 & 22.05 & 21.90 & 19.27 + 1700 & 55 & 21.91 & 15.74 & 21.90 & 12.84 + 1700 & 85 & 21.91 & 21.19 & 21.90 & 9.17 + 1700 & 125 & 21.91 & 18.41 & 21.90 & 18.97 + 1700 & 250 & 21.91 & 17.62 & 21.90 & 16.33 + 2865 & 25 & 21.52 & 16.00 & 21.51 & 15.10 + 2865 & 55 & 21.52 & 11.81 & 21.51 & 12.81 + 2865 & 85 & 21.51 & 16.67 & 21.51 & 13.33 + 2865 & 125 & 21.52 & 15.39 & 21.51 & 7.09 + 2865 & 250 & 21.52 & 14.63 & 21.51 & 8.37 + 4382 & 25 & 21.55 & 11.57 & 21.51 & 10.55 + 4382 & 55 & 21.56 & 11.18 & 21.46 & 3.68 + 4382 & 85 & 21.55 & 11.87 & 21.51 & 14.11 + 4382 & 125 & 21.57 & 10.49 & 21.51 & 10.03 + 4382 & 250 & 21.54 & 13.03 & 21.47 & 4.01 + 7727 & 25 & 22.46 & 3.93 & 22.38 & 3.99 + 7727 & 55 & 22.45 & 3.76 & 22.37 & 4.44 + 7727 & 85 & 22.46 & 4.26 & 22.38 & 3.81 + 7727 & 125 & 22.46 & 3.94 & 22.37 & 3.55 + 7727 & 250 & 22.45 & 3.96 & 22.35 & 3.55 [ tablecplte ] in addition to the completeness corrections we performed a visual examination of each gc candidate , using the iraf task _ imexam _ to further remove any contaminants such as background galaxies .
candidates were dropped if their shape was particularly elliptical and/or if the counts per pixel within the object were well below those of the other candidates .
table [ table : candidates ] gives a summary of the gc candidate corrections applied for each gc system .
figures [ fig : ngc2865vik ] and [ fig : ngc7727vik ] show the resulting color color and color magnitude diagrams for the gc systems of ngc 2865 and ngc 7727 , respectively .
lcccc 1700 & 45 & 6 & 13 & 26 + 2865 & 65 & 2 & 20 & 43 + 4382 & 65 & 3 & 2 & 60 + 7727 & 31 & 0 & 9 & 22 [ table : candidates ]
in order to constrain the age of the merger event for each of the four galaxies , the ages of their gc subpopulations were determined .
specifically , we assumed that the parameters of the youngest subpopulation may be associated with the most recent major star - formation episode the galaxy - merger event itself .
conventional studies of gc ages and metallicities are often based on optical - to - nir color
magnitude diagrams such as , e.g. , the @xmath1 vs diagram .
however , such diagrams suffer from an inherent age
metallicity degeneracy ( e.g. , * ? ? ?
* ) , and it is often difficult to disentangle age and metallicity without additional detailed spectra . in an effort to break or at least partly lift this degeneracy , we use color color diagrams ( specifically vs ) and interpret the measured gc colors via the ssp models generated by charlot & bruzual ( 2007 , private communication , hereafter cb07 ; see @xcite and @xcite for published models and details ) . the fitting algorithm used to match the models to our data is briefly described below and closely follows the method and algorithm used by @xcite .
we also present the results of our efforts to verify the technique .
our principal goal was to identify the possible presence of multiple subpopulations among the gcs of our sample galaxies and to estimate their ages , based on simple assumptions about their metallicities .
we assume that there are three predominant cluster subpopulations of interest : one consisting of @xmath013 gyr old metal - poor gcs , a second one of also old ( @xmath1810 gyr ) metal - rich gcs , and a third one of younger , intermediate - age metal - rich gcs .
the universal old metal - poor subpopulation is present in all major galaxies and is represented in our model simulations by clusters of mean logarithmic metallicity @xmath19 = -1.7 $ ] , with colors computed by cb07 ; the size and age of this subpopulation is being held constant during the simulations for each galaxy .
the two metal - rich subpopulations are assumed to both have logarithmic metallicities in the range of @xmath20 < 0.4 $ ] .
the number ratios of metal - rich to metal - poor clusters , and of intermediate - age to old clusters within the metal - rich subpopulation , are then varied systematically .
with these assumptions made , the goal is to simulate the appearance of the mixed gc populations in vs diagrams with bc07 model clusters and to then compare these model diagrams with the observed diagrams for the gcs of each galaxy .
the measure used for the comparison is provided by the cumulative age distributions of the artificial clusters and the observed gcs , both derived from the positions of the clusters in their respective vs diagrams and compared via a chi - squared ( hereafter ) test described below .
to _ simulate _ the observed mixed gc population of each galaxy , a set of model cluster colors was generated using a monte carlo technique .
for each of the many simulations per galaxy , the total number of model clusters was made to equal the number of gc candidates in the galaxy ( table [ table : candidates ] , last column ) .
corresponding model cluster colors were then computed based on the ssp models of cb07 . to simulate observational errors ,
each model cluster population was furthermore convolved with photometric errors based on the indicative errors of our data .
finally , cumulative age distributions were calculated for each model cluster population , based on the same ssp models by cb07 . for each of our four galaxies
, we generated a grid of 121 model cluster populations as follows .
the age of the old metal - rich subpopulation was held fixed at 13 gyr , while the age of the younger metal - rich subpopulation was set to 0.2 , 0.5 , 0.75 , 1 , 1.5 , 2 , 3 , 5 , 7 , 10 , and 13 gyr . for each of these 11 ages , the ratio of young to old clusters
was varied from 0% to 100% in 10% steps , thus yielding a total of 121 model populations for comparison with the observed gc population of each galaxy .
for each of the 121 age and number - ratio combinations , 1000 model samples were generated via the monte carlo technique , and the average cumulative age distribution ( based on the ssp models ) of the simulated clusters was calculated .
the cumulative age distribution of the observed gcs was also calculated based on the same ssp models , via a transformation from the vs plane to the abundance vs log - age plane based on these models ( for details , see @xcite ) .
we then performed fitting of the model cumulative age distributions to the observed cumulative age distribution to determine which of the age and number - ratio combinations best fit the data . for each galaxy ,
a contour plot of the values over the age vs.number-ratio configuration plane was drawn and used to infer the age and number ratio of the young metal - rich subpopulation as compared to the old subpopulation .
these contour plots are shown and discussed in [ sec43 ] below .
in order to validate the use of this technique , we performed a series of verification tests independent from those performed by @xcite .
our verification method involved replacing the observed gc population with a simulated population of model clusters .
the simulation of this population followed the same procedure as the simulations of our model populations described above .
first , we tested the ability of our adopted method to return an accurate value for the age of the younger metal - rich input population . as figure
[ fig : chi2_trial ] shows , while the age returned by the -fitting did not converge exactly to the simulated input age , it was largely consistent ( to within @xmath010% ) .
hence , we can confirm that , over the range of ages ( @xmath015 gyr ) that we are investigating , the method returns a reasonably accurate age for a given input age .
second , we also tested our method s stability with respect to the photometric errors of the input data . for a given age and ratio configuration
, we varied the photometric errors from 0.1 mag to 0.3 mag .
note that ( i ) the indicative photometric errors in our observed data were below 0.3 mag , and ( ii ) while there were some clusters with higher errors , they were cut before the -fitting was performed .
figure [ fig : chi2_trial2 ] highlights that , as expected , increasing the errors decreased the convergence of the fit .
even so , however , the age returned by the fitting contours was still largely consistent with the simulated input age .
thus , we confirm the validity of our method for both the ages and the indicative photometric errors of the input data .
> from the spread in the contours and the deviation of the converged age from the input age we infer an upper limit for the age uncertainty of about @xmath21 gyr .
similarly , for the young - to - old ratio we find an upper limit to the uncertainty of about @xmath22 .
these uncertainties will be used for the gc subpopulations of each galaxy in the following sections .
figure [ fig : chi2_contour ] shows contour plots of values over the modeled configuration space for each gc system of our four sample galaxies . in each contour plot , the convergence of to a minimum value marks the most likely combination of age and number - ratio for the younger gc subpopulation of that galaxy , based on the ssp models by cb07 .
table [ table4 ] summarizes these resulting combinations , which we now briefly discuss for each galaxy
. for this galaxy , figure [ fig : chi2_contour ] shows that the values converge to a minimum at a gc population mix consisting of about 5% of an old subpopulation and 95% of a younger subpopulation with an age of @xmath23 gyr .
note that this age is consistent with the previously published age estimate of @xmath24 gyr @xcite to within the combined errors .
note also that the quoted percentages apply only to the _ observed _ gc candidates , and not to the entire gc population , whose total number and subpopulation percentages are unknown . for ngc 2865 ,
the contours again show a high level of convergence , as can be seen in the top - right panel of figure [ fig : chi2_contour ] . from the contour minimum
we infer a younger metal - rich gc subpopulation of age @xmath25 gyr , comprising about 90% of the observed gc candidates . again
, this newly determined age for the younger gcs compares well with the previously published age of @xmath26 gyr for the young _ stellar _ subpopulation in this galaxy @xcite .
lcccl 1700 & [email protected] & 95% & @xmath28 & @xcite + 2865 & [email protected] & 90% & @xmath26 & @xcite + 4382 & [email protected] & 85% & @xmath29 & @xcite + 7727 & [email protected] & 100% & 12 & @xcite [ table4 ] for ngc 4382 , the contours shown in figure [ fig : chi2_contour ] indicate a younger metal - rich gc subpopulation of age @xmath30 gyr , comprising between about 75% and 95% of the observed gc candidates .
note that the values for the gcs in ngc 4382 are much smaller than those for the gcs in the other three galaxies , suggesting a better agreement between the observed gc data and the simulated cluster populations .
as can be seen , there is also a significantly wider spread for the ratio of young - to - old gcs .
recall , however , that our primary interest is in the age of the younger subpopulation , and note that the convergence in age is very high .
this newly determined age for the younger metal - rich gc subpopulation agrees well with the previously published age of @xmath31 gyr for the young _ stellar _ subpopulation of this galaxy @xcite .
finally , for ngc 7727 the contours of figure [ fig : chi2_contour ] indicate a purely young metal - rich gc population of age @xmath32 gyr .
the observed gc candidates of this galaxy are noteworthy both for their particularly young ages and for the apparent absence of an older subpopulation .
both of these characteristics were previously reported by @xcite , who obtained a gc population age of 12 gyrs .
note , however , that the population of old metal - poor gcs universally present in all major galaxies may be too faint to be detected on our -band images .
hence , its apparent absence may simply reflect a selection effect and does not imply that this old metal - poor gc population does nt exist .
we have presented an age analysis of gc systems in four early - type galaxies that are considered candidate remnants of recent mergers .
new -band photometry of the gcs in these galaxies has been obtained with two near - infrared imagers on gemini and combined with existing _ hst _ photometry in @xmath1 and @xmath2 to produce color color and color magnitude diagrams of the gc systems in .
these data have been fitted to simple toy models of mixed gc populations containing three subpopulations that differ in age and metallicity .
the mixed gc populations have been simulated with ssp models by charlot & bruzual ( 2007 , private communication ) and compared to the observations .
various fits to the cumulative age distributions derived from our data and from the toy models yield clear evidence for the presence of intermediate - age , metal - rich subpopulations of gcs in each of the four early - type galaxies . at the observed , relatively bright magnitudes these intermediate - age subpopulations exceed the old metal - rich and metal - poor subpopulations in cluster numbers , and their estimated ages of @xmath012 gyr are consistent with published ages for the same galaxies based on the integrated light .
thus , photometry of gc populations in early - type galaxies can be used to date epochs of major star and cluster formation presumably due to relatively recent merger events .
we thank stphane charlot and gustavo bruzual for sending us an early version of their new cluster - population models ( `` cb07 '' ) , presently still under development , and maren hempel and markus kissler - patig for helpful information and a copy of their simulation code .
we also thank the anonymous referee , whose constructive comments helped us improve the presentation of our results .
this research is based on observations obtained at the gemini observatory , which is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the particle physics and astronomy research council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , cnpq ( brazil ) and conicet ( argentina ) .
trancho , g. , miller , b. w. , schweizer , f. , & whitmore , b. 2004 , in asp conf . ser . 322 , the formation and evolution of massive young star clusters , ed . h. j. g. l. m. lamers , l. j. smith , & a. nota
( san francisco : asp ) , 219 | we present the results of combining _ hubble space telescope _ optical photometry with ground - based -band photometry from the gemini imagers niri and flamingos - i to study the globular - cluster populations in four early - type galaxies that are candidate remnants of recent mergers ( ngc 1700 , ngc 2865 , ngc 4382 , and ngc 7727 ) .
these galaxies were chosen based on their blue colors and fine structure , such as shells and ripples that are indicative of past interactions .
we fit the combined globular - cluster data with simple toy models of mixed cluster populations that contain three subpopulations of different age and metallicity .
the fits , done via chi - square mapping of the parameter space , yield clear evidence for the presence of intermediate - age clusters in each galaxy .
we find that the ages of @xmath012 gyr for these globular - cluster subpopulations are consistent with the previously estimated merger ages for the host galaxies . |
in the hierarchical structure formation scenario based on a cdm universe , a galaxy forms by the assembly of many proto - galactic fragments .
these small systems collapse first in the high density peaks in the initial density fluctuation fields , and grow by successive mergers between galaxies .
this overall picture is supported by various pieces of observational evidence , such as the existence of dynamically decoupled subgalactic components within a galaxy ( eg . , de zeeuw et al .
2002 ) or on - going mergers with tidal remains .
le fvre et al .
( 2000 ) count the statistical associations of close companions around galaxies in the redshift survey fields ( cfrs and ldss ) based on the high spatial resolution images by wfpc2 on the _ hubble space telescope _ ( hst ) .
they find a rapid increase in the fraction of close pairs with redshift , suggesting that a typical @xmath15 galaxy today has undergone 1 - 2 significant merger events since @xmath1 .
dickinson et al .
( 2002 ) have shown that the most luminous lyman break galaxies at @xmath16 in the _ hubble deep field north _ are an order of magnitude less massive than @xmath17 today .
clusters of galaxies are specially selected areas that grow from high density regions in the large scale structure of the early universe . in these regions ,
the galaxy formation process is likely to be accerelated .
this qualitatively explains why the majority of star formation in galaxies in cluster cores take place much earlier ( @xmath13 ) than that in the general field ( as shown by the extensive analyses based on the fundamental planes ( fp ) and the colour - magnitude relations ( cmr ) both locally and at higher redshifts eg .
, bower et al . 1992 ; 1998 ; ellis et al . 1997 ; van dokkum et al . 1998 ; stanford et al . 1998 ; kodama et al . 1998 , kelson et al.2000 ) . compared to star formation , hierarchical galaxy formation models ( eg . , kauffmann 1996 ) predict that the mass assembly process is more widely spread over time . indeed , van dokkum et al .
( 1999 ) find many mergers or close companions in a @xmath18 cluster , suggesting that mass assembly is still in an active phase in this high redshift cluster , even though the star formation in most of these mergers have been completed ( as suggested from their red colours ) .
the key issue now is to identify the epoch by which the cluster galaxies have accumulated most of their mass .
this is still an open question .
however , measuring total galaxy masses at high redshifts is difficult , since dynamical measurements of internal velocity dispersion for a large statistical sample becomes extremely time consuming even with 8 - 10 m telescopes ( van dokkum & stanford 2002 ) .
in contrast , stellar mass can be more easily estimated based on the near - infrared luminosity .
the near - infrared luminosity of a galaxy is relatively insensitive to its on - going / recent star formation history even at high redshifts ( @xmath1 ) , and hence gives a good measure of the underlying total stellar mass of the galaxy .
kauffmann & charlot ( 1998 ) showed that the evolution of near - infrared luminosity function ( lf ) was a sensitive test of galaxy formation models , where we should expect to see strong evolution with redshift if the mass of the galaxies are growing due to the assembly processes .
kauffmann & charlot argued that this evolution will be most prominent in the number density of massive galaxies , since it takes longer to form these massive systems .
for the brightest cluster galaxies , aragn - salamanca , baugh , & kauffmann ( 1998 ) have shown that there is significant mass evolution between @xmath12 and present - day , typically by a factor 2 - 3 .
the observational result that the central galaxy of a @xmath19 cluster is separated to multiple systems supports this view ( yamada et al .
2002 ) . this can be easily understood , since the brightst cluster galaxy preferentially sits in the dynamical centre of a cluster , where the satellite galaxies tend to sink and merge towards the central galaxy due to the dynamical friction ( binney & tremaine 1987 ) . pioneering work using the @xmath20-band lf to quantify the mass evolution of the cluster galaxy population has been presented by barger et al .
( 1996 ; 1998 ) and de propris ( 1999 ) for many distant clusters out to @xmath21 .
they both find that the change of characteristic luminosities of the @xmath20-band lfs with redshift is entirely consistent with the passive evolution of galaxies formed at high redshifts ( @xmath22 ) , suggesting mass evolution is negligable between @xmath21 and the present .
these results have pushed the mass assembly epoch of the massive cluster galaxies to higher redshift ( @xmath23 ) .
it is worth noting , however , that their analyses are limited to the very central part of the clusters ( typically clustercentric distance of @[email protected] mpc ) , and may not be representative of the whole cluster population due to dynamical mass segregation . in this paper , we present the combined @xmath0-band luminosity function of the two @xmath1 clusters drawn from a wide field ( @xmath24@xmath61 mpc ) in order to study the stellar mass evolution of the representative cluster population as a whole .
we use @xmath25 colours to correct for the effect of star formation on the mass - to - luminosity ( @xmath26 ) ratio in order to derive the stellar mass .
the crucial issue is whether there is a deficit of massive galaxies ( @xmath27 ) at @xmath1 compared to the less massive galaxies .
a detection of this deficit would confirm the importance of galaxy mergers , as suggested by hierarchical galaxy formation models .
an outline of the paper follows . after describing the cluster selection , observation and the data reduction in 2 , and the data processing including the corrections for incompleteness and foreground and background contamination in 3 , we present the @xmath0-band number counts , luminosity function , and the stellar mass function of the @xmath1 clusters in 4 .
discussion and conclusions are given in 5 and
6 , respectively .
the cosmological parameters adopted throughout this paper are ( @xmath28 , @xmath29 , @xmath30)=(0.7 , 0.3 , 0.7 ) , unless otherwise stated . here we define @xmath28 and @xmath31 as @xmath32/(100 km s@xmath33 mpc@xmath33 ) and @xmath32/(70 km s@xmath33mpc@xmath33 ) , respectively .
[ cols="^,^,^,^,^,^,^",options="header " , ] [ tab : ratio ] to quanify the comparison between the models and data in a way that is independent of the normalisation , we define the giant - to - dwarf ratio , @xmath34(giant / dwarf ) , as @xmath35 the ratios are calculated from data shown in figs .
[ fig : mf ] , and are summarised in table 2 .
this shows that the fraction of giant galaxies at @xmath12 in the semi - analytic predictions is factor @xmath363 smaller than the observed value .
the scatter of the ratio in the tokyo model measured from the 100 generated clusters ( 0.05 ) is too small to account for the over - abundance of the massive galaxies in the real high redshift clusters .
how much mass evolution can be allowed between @xmath12 and the present - day within the errors of our analysis ? to answer this question , we shift the local mass function of 2mass / lcrs clusters ( balogh et al .
2001 ) towards lower masses and calculate the statistical deviation from the observed mass function at @xmath1 by a boot - strap resampling method ( ie . , randomly replacing the observed points assuming the gaussian errors on each point ) . from this analysis
, we find that we can reject a mass decrease of a factor of 2 at 99% level ( 2.5@xmath2 ) no matter whether the bright blue galaxies are included or excluded . since the definition of bright blue cut that we introduced in 4.2 is somewhat arbitrary , we consider the effect of further reducing the numbers of galaxies in the brightest bin , moving it two sigma down from its original place . even in this extreme case , the giant - to - dwarf ratio is 0.27 and is still 2.3 @xmath2 off from the factor 2 evolution in mass , corresponding to 98% rejection level .
in contrast to the bright cluster galaxies , faint galaxies seem to have a much greater diversity of formation histories . the faint galaxies may be in general formed at later cosmic times than the massive galaxies , although this appears to be opposite to the hierarchical picture .
if so , then distant clusters should have a lf that has a declining faint - end slope in contrast to the rising faint - end slope of local clusters ( balogh et al .
2001 ; de propris et al . 1998 ) . the progression of formation activity to lower - mass galaxies as the universe ages is often referred to as the `` down - sizing '' effect ( cowie et al . 1996 ; kodama & bower 2001 ) .
kajisawa et al .
( 2000 ) suggested just such a deficit of faint galaxies ( @xmath37 , or @xmath38 + 1.5 ) in the 3c324 cluster at @xmath39 . in our two @xmath1 clusters
, however , such a strong deficit is not confirmed down to @xmath40 ( @xmath4 + 2.7 ) ( fig . [
fig : klf ] ) . a marginal discrepancy ( @xmath32@xmath2 ) between @xmath41 and
1 is seen in the least massive bin of the stellar mass functions ( the solid lines and our data points in figs .
[ fig : mf ] ) . however , given the fact that the faint - end slope of the local mass function of cluster galaxies ( 2mass / lcrs ; balogh et al .
2001 ) is only poorly determined ( @xmath42 ( 1@xmath2 ) ) , we can not discuss further whether there is an evolution between @xmath41 and 1 in the faint end of the mass function from the current data set . to test the hypothesis of different formation epoch depending on galaxy mass , our data would need to go deeper than @xmath43 .
however , we certainly see significant numbers of less massive galaxies below 1.5@xmath710@xmath44m@xmath9 , which are nearly absent in kajisawa et al .
we speculate that the reason why they saw such a deficit of less massive galaxies is their limited field coverage ( cluster - centric distance @xmath24@xmath4540@xmath46 or 0.33 mpc ) .
one might be concerned that our high giant - to - dwarf ratio at @xmath12 clusters and hence the relatively strong constraint on the mass growth rate , is partly driven by the deficiency of the less massive galaxies at @xmath12 as well as the abundance of massive galaxies .
we verify this effect on our conclusions by artificially increasing the numbers of galaxies in the least massive bin at @xmath12 by 2@xmath2 , and by flattening the faint end slope of the @xmath41 curve ( 2mass / lcrs ) by @xmath47 .
however , these effects on the constraint on the mass growth factor is found to be small .
for example , in the first case , the giant - to - dwarf ratio goes down to @xmath34=0.26 and 0.24 with and without the bright blue galaxies , respectively , but the mass evolution of factor 2 between @xmath12 and 0 ( which gives @xmath34=0.07 ) is still rejected at 98% level ( 2.3@xmath2 ) .
it is now generally accepted that the stellar populations of galaxies in clusters are old with little on - going star formation based on the colours and the fundamental planes studies ( eg .
, bower et al . , 1992 ; 1998 ; ellis et al . 1997 ; van dokkum et al . 1998 ; stanford et al .
1998 ; kodama et al .
1998 ; kelson et al .
2000 ) . however , the stellar populations of galaxies may be old , but at the same time the galaxies may not have been completely assembled . between @xmath12 and the present - day , galaxies might grow in mass through collisions .
this key issue can only be tested by comparing the luminosity functions of galaxies in clusters at different redshifts .
early assembly of massive galaxies is a challenge for hierarchical models of galaxy formation since these models tend to grow galaxies through both on - going merging and gas accretion up to the present - day .
the first generation of such models predicted very rapid galaxy growth ( kauffmann & charlot 1998 ) .
more recent models within a @xmath48-cdm universe and revised merging prescriptions ( eg . , cole et al . , 2000
) have predicted evolution that is slower but still significant see cimatti et al .
( 2002 ) for a recent discussion and comparision with field data . in clusters of galaxies
, we might anticipate that the rate of evolution of galaxy mass is less than in the field since rich clusters are over dense regions of the universe in which gravitational collapse has been accelerated .
thus an important issue that has concerned us is that clusters which are observed at @xmath1 are not directly equivalent to the progenitors of clusters in the local universe ( ie . , while the galaxies in the distant clusters form part of the local cluster population , there are many galaxies in local clusters which were located in lower density environments at @xmath12 ) .
care is therefore need to select galaxies from the semi - analytic models , and we have tried to select clusters in the same way in both the semi - analytic models and the data .
we applied two different ways of selecting the systems in the models by mass and by space density but both definitions give similar results and predict a significant reduction in the characteristic stellar mass of cluster galaxies with redshift .
in contrast to the model , however , our observations of two distant clusters show that the evolution of the luminosity function is well modelled by passive evolution of the galaxy population .
the @xmath20-band luminosity function is slightly brighter at @xmath12 than at @xmath41 .
however , when we correct the observed galaxy luminosities for the passive evolution of the stellar population , and cosmological dimming , we find that the galaxy mass function has evolved little over this time .
a factor of 2 evolution is only marginally consistent with the data ( @xmath32.5@xmath2 ) .
however , this level of uncertainty means that our data can not yet completely rule out the heirarchical model . if we could fix the mass function normalisation using the total ( ie .
, dark plus stellar mass ) of the clusters the evolution in galaxy mass predicted by the theoretical models would be much clearer . unfortunately the clusters we have studied have no reliable mass estimates that we can use to independently secure the lf normalisation ( the only measure available is a velocity dispersion of 397 km s@xmath33 for the 3c336 cluster , measured from eight cluster members : steidel et al .
this is an important area for future work .
there is no significant difference in the stellar mass function between the 3c336 and q1335 clusters ( @xmath49 ) .
the cluster to cluster scatter in the theoretical models is also small ( @xmath50 @xmath34(giant / dwarf)=0.05 ) . in an individual cluster ,
the measured luminosity function is dependent on the background subtraction .
however , even if we artificially change the background level by a factor 2 , the resulting change in the giant / dwarf ratio is found to be less than 0.1 .
it therefore seems that our conclusion is not significantly affected by the field - to - field variation of the background . finally , there are uncertainties in the theoretical model predictions . for this reason it has been useful to check that we obtain similar results for both the durham and mitaka models . in addition ,
the simulations allow us to check that cluster to cluster variation is not significant .
a concern when comparing theoretical and observational results is the radius over which the comparison is made .
the wide field of our observations ( @xmath24@xmath61 mpc ) means that they include a large fraction of the cluster virial radius , and are thus directly comparable to the observational results .
it is interesting to note that the discrepancy between our results and the deficit of faint ( @xmath51 + 1.5 ) galaxies by kajisawa et al .
( 2000 ) , might result from this type of mass segregation .
we have presented the near - infrared lf of galaxies in the two high confidence @xmath1 cluster candidates based on deep wide - field imaging .
a third cluster was observed , but has then been excluded because of contamination by a foreground system .
we find that the @xmath0-band lf at @xmath1 is consistent with pure luminosity evolution with constant stellar mass according to the passive evolution with old age ( @xmath52 ) , if compared to the lower redshift counterparts , such as 2mass / lcrs lf for local clusters .
we have transformed our lf at @xmath1 to a stellar mass function by correcting for the star formation contribution to the @xmath0-band light as estimated from the @xmath25 colours .
the derived stellar mass function is found to be very similar to the one for the 2mass / lcrs local clusters . in particular ,
massive galaxies ( @xmath53@xmath54m@xmath9 ) make up a significant fraction of the galaxy population .
this result presents a challenge for current hierarchical galaxy formation models which tend to under - predict the fraction of massive galaxies compared to our observed data .
the next steps are threefold .
one is to increase the sample of clusters at similar redshift ( @xmath1 ) and to have better statistics on the derived stellar mass function at this epoch .
the second is to extend this analysis to higher redshift ( @xmath13 ) to identify the epoch at which the massive galaxies are assembled .
the third is to extend this analysis along an environmental axis .
galaxy - galaxy mergers are likely to be most effective in the group environment rather than the cluster cores where velocity dispersion is too high ( binney & tremaine 1987 ) .
therefore , it may be the case that the mass growth can be clearly seen at lower redshifts in these lower density regions .
we are grateful to dr .
masa nagashima , carlton baugh , and sarah reed for providing us with the model predictions for stellar mass functions of galaxies from their semi - analytic models of galaxy formation , and also for useful discussion about the comparison with the observed data .
we acknowledge to the referee , dr . aragn - salamanca , for his useful comments which have improved the clarity of the paper .
we thank ian smail for providing us with scripts for efficient observation with ingrid and data reduction in iraf .
tk thanks the japan society for the promotion of science for support through its research fellowships for young scientists between 1999 - 2001 during which these ingrid data were taken .
rgb thanks pparc and the leverhulme trust for their support .
we also acknowledge the daiwa - adrian prize 2001 given by the daiwa anglo - japanese foundation .
andreon , s. , 2002 , a&a , 382 , 821 aragn - salamanca , a. , baugh , c. m. , kauffmann , g. , 1998 , mnras , 297 , 427 balogh , m. l. , christlein , d. , zabludoff , a. i. , zarisky , d. , 2001 , apj , 557 , 117 barger , a. j. , et al .
, 1998 , apj , 501 , 522 barger , a. j. , aragn - salamanca , a. , ellis , r. s. , couch , w. j. , smail , i. , sharples , r. m. , 1996 , mnras , 279 , 1 baugh , c. m. , benson , a. j. , cole , s. , frenk , c. s. , lacey , c. , astro - ph/0203051 bertin , e. , arnouts , s. , 1996 , a&as , 117 , 393 bessell , m. s. , brett , j. m. , 1988 , aj , 100 , 1134 best , p. n. , 2000 , mnras , 317 , 720 binney , j. , tremaine , s. , 1987 , in _ galactic dynamics _ , princeton university press bower , r. g. , kodama , t. , terlevich , a. , 1998 , mnras , 299 , 1193 bower , r , g. , smail , i , 1997 , mnras , 290 , 292 brinchmann , j. , ellis , r. s. , 2000 , apj , 536 , l77 calzetti , d. , armus , l. , bohlin , r. c. , kinney , a. l. , koornneef , j. , storchi - bergmann , t. , 2000 , apj , 533 , 682 casali , m. m. , hawarden , t. g. , 1992 , jcmt - ukirt newsl .
no.4 , 33 cimatti , a. et al . , 2002 , a&a , 392 , 395 cole , s. , lacey , c. g. , baugh , c. m. , frenk , c. s. , 2000 , mnras , 319 , 168 cowie , l. l. , songaila , a. , hu , e. m. , cohen , j. g. , 1996 , aj , 112 , 839 de propris , r. , eisenhardt , p. r. , stanford , s. a. , dickinson , 1998 , apj , 503 , l45 de propris , r. , stanford , s. a. , eisenhardt , p. r. , dickinson , m. , elston , r. , 1999 , apj , 118 , 719 de zeeuw , p. t. et al , 2002 , mnras , 329 , 513 dressler , a. , smail , i. , poggianti , b. m. , butcher , h. , couch , w. j. , ellis , r. s. , oemler , a. j , 1999 , apjs , 122 , 51 drory , n. , et al .
, 2001 , apj , 562 , l111 hawarden , t. g. , leggett , s. k. , letawsky , m. b. , ballantyne , d. r. , casali , m. m. , 2001 , mnras , 325 , 563 huchings , j. b. , crampton , d. , johnson , a. , 1995 , aj , 109 , 73 kajisawa , m. , et al .
, 2000 , pasj , 52 , 53 kauffmann , g. , 1996 , mnras , 281 , 487 kauffmann , g. , charlot , c. , 1998 , mnras , 297 , 23 kelson , d. d. , illingworth , g. d. , van dokkum , p. g. , franx , m. , 2000 , apj , 531 , 184 kennicutt , r. c. , jr . , 1983 , apj , 272 , 54 kodama , t. , arimoto , n. , 1997 , a&a , 320 , 41 kodama , t. , arimoto , n. , barger , a. j. , aragn - salamanca , a. , 1998 , a&a , 334 , 99 kodama , t. , bower , r. g. , 2001 , mnras , 321 , 18 le fvre , et al
. , 2000 , mnras , 311 , 565 nakata f. , kajisawa , m. , yamada , t. , kodama , t. , shimasaku , k. , tanaka , i. , 2001 , pasj , 53 , 1139 nagashima , m. , yoshii , y. , totani , t. , gouda , n. , 2002 , apj , in press saracco , p. , dodorico , s. , moorwood , a. , buzzoni , a. , cuby , j .-
, lidman , c. , 1999 , a&a , 349 , 751 saracco , p. , giallongo , e. , cristiani , s. , dodorico , s. , fontana , a. , iovino , a. , poli , f. , vanzella , e. , 2001 , a&a , 375 , 1 steidel , c. c. , dickinson , m. , meyer , d. m. , adelberger , k. l. , sembach , k. r. , 1997 , apj , 480 , 568 tanaka , i. , yamada , t. , aragn - salamanca , a. , kodama . t. , miyaji , t. , ohta , k. , arimoto , n. 2000 , apj , 528 , 123 van dokkum , p. g. , franx , m. , fabricant , d. , kelson , d. d. , illingworth , g. d. , 1999 , apj , 520 , l95 van dokkum , p. g. , stanford , s. a. , 2002 , apj , accepted , astro - ph/0210643 yamada , t. , tanaka , i. , aragn - salamanca , a. , kodama , t. , ohta , k. , arimoto , n. , 1997 , apj , 487 , l125 yamada , t. , et al . , 2002 , apj , in press yee , h. k. c. , ellingson , e. , carlberg , r. g. , 1996 , apjs , 102 , 269 | we present the near - infrared ( @xmath0-band ) luminosity function of galaxies in two @xmath1 cluster candidates , 3c336 and q1335 + 28 . a third cluster , 3c289 , was observed but found to be contaminated by a foreground system .
our wide field imaging data reach to @xmath0=20.5 ( 5@xmath2 ) , corresponding to @xmath3@xmath4 + 2.7 with respect to the passive evolution .
the near - infrared luminosity traces the stellar mass of a galaxy due to its small sensitivity to the recent star formation history .
thus the luminosity function can be transformed to the stellar mass function of galaxies using the @xmath5 colours with only a small correction ( factor@xmath62 ) for the effects of on - going star formation .
the derived stellar mass function spans a wide range in mass from @xmath33@xmath710@xmath8 m@xmath9 down to @xmath36@xmath710@xmath10 m@xmath9 ( set by the magnitude limit ) .
the form of the mass function is very similar to lower redshift counterparts such as that from 2mass / lcrs clusters ( balogh et al .
2001 ) and the @xmath11 clusters ( barger et al .
1998 ) .
this indicates little evolution of galaxy masses from @xmath12 to the present - day .
combined with colour data that suggest star formation is completed early ( @xmath13 ) in the cluster core , it seems that the galaxy formation processes ( both star formation and mass assembly ) are strongly accerelated in dense environments and has been largely completed by @xmath12 .
we investigate whether the epoch of mass assembly of massive cluster galaxies is earlier than that predicted by the hierarchical galaxy formation models .
these models predict the increase of characteristic mass by more than factor @xmath14 between @xmath12 and the present day .
this seems incompatible with our data .
galaxies : clusters galaxies : formation galaxies : evolution galaxies : stellar content |
the notion of slope-(semi)stability for vector bundles on a smooth projective varieties over an algebraically closed field @xmath1 , as introduced by d. mumford in the case of curves and generalized by f. takemoto to higher dimensional varieties , is a very important tool in algebraic geometry .
unfortunately , for a concretely given vector bundle it is often very difficult to decide whether it is semistable or even stable . in this paper
we develop an algorithm to determine computationally the semistability of certain vector bundles on the projective space @xmath0 . throughout this paper
we assume that @xmath3 , since for @xmath4 by the theorem of a. grothendieck every vector bundle splits as a direct sum of line bundles .
we restrict ourselves to vector bundles which are given as a kernel of a surjective morphism between splitting bundles , i.e. , vector bundles @xmath5 which sit in a short exact sequence @xmath6 we call such bundles _ kernel bundles_. for instance , by the theorem of horrocks every non - split vector bundle on @xmath7 admits such a presentation . the morphism @xmath8 which defines @xmath5 is given by an @xmath9 matrix @xmath10 , where the entries @xmath11 $ ] are homogeneous polynomials of degrees @xmath12 .
special instances ( @xmath13 and @xmath14 ) of kernel bundles are the so - called _ syzygy bundles _ @xmath15 for @xmath16-primary homogeneous polynomials @xmath17 ( i.e. , @xmath18 ) , i.e. , a syzygy bundle has a presenting sequence @xmath19 where @xmath20 . due to their explicit nature ,
kernel bundles and syzygy bundles are suitable for direct computations , in particular grbner basis methods and combinatorics .
but still in general , not much is known about ( semi)stability of kernel bundles or even syzygy bundles .
one of the most important results in this direction , due to h. brenner , is a combinatorial criterion for ( semi)stability of syzygy bundles given by monomial families : [ monomialcase ] let @xmath1 be a field , @xmath21 $ ] and let @xmath22 denote @xmath16-primary monomials of degrees @xmath23 in @xmath24 $ ] , @xmath25 .
suppose that for every subset @xmath26 , @xmath27 , the inequality @xmath28 holds , where @xmath29 is the degree of the highest common factor of @xmath30 , @xmath31 .
then the syzygy bundle @xmath32 is semistable @xmath33and stable if @xmath34 holds@xmath35 .
see ( * ? ? ?
* corollary 6.4 ) .
another important theorem , due to g. bohnhorst and h. spindler , is a numerical ( semi)stability criterion for kernel bundles of rank @xmath36 on @xmath0 in characteristic @xmath2 : [ bohnhorstspindlerhdoneintro ] let @xmath5 be a vector bundle of rank @xmath3 on the projective space @xmath0 over an algebraically closed field @xmath1 of characteristic @xmath2 .
suppose there is a short exact sequence @xmath37 such that @xmath38 , @xmath39 and @xmath40 for @xmath41 .
then @xmath5 is semistable @xmath33stable@xmath35 if and only if @xmath42 .
this is ( * ? ?
* theorem 2.7 ) applied to the dual bundle @xmath43 .
a general algorithm using grbner bases methods ( computation of syzygy modules ) that detects semistability of syzygy bundles and its imlementation by the first author was already announced by h. brenner in ( * ? ? ?
* remark 5.3 ) . in this article
we describe this algorithm more generally for kernel bundles and describe in detail how to implement it in a computer algebra system ( this has been done concretely by the first author in cocoa @xcite ) .
this semistability algorithm can be used as a tool to examine further problems regarding semi(stability ) of vector bundles by providing interesting examples .
we explain these applications in more detail in the sequel .
the paper is organized as follows . in section [ hoppesemistabilitycriterion ]
we recall a criterion due to h. j. hoppe ( see lemma [ exteriorpowercriterion ] ) which relates ( semi)stability to global sections of exterior powers of a given vector bundle .
in particular , we show that this result , originally only formulated in characteristic @xmath2 , holds in arbitrary characteristic .
hoppe s criterion is the key - result for our algorithm . in section [ syzygyandkernelbundles ]
we discuss some properties of kernel bundles and syzygy bundles on projective spaces .
in particular , for these bundles we discuss necessary bohnhorst / spindler like numerical conditions ( compare theorem [ bohnhorstspindlerhdoneintro ] ) for semistability . the actual semistability algorithm for kernel bundles and its implementation
is explained in section [ algoimplementation ] . besides exterior powers
, we also describe explicitly how to compute global sections of tensor products and symmetric powers of kernel bundles .
these algorithms play an important role in our first application : the computation of tannaka dual groups of polystable vector bundles @xmath5 of degree @xmath2 and rank @xmath44 on @xmath0 in characteristic @xmath2 .
section [ tannaka ] starts with a brief introduction to tannaka duality .
roughly spoken , for a polystable vector bundle @xmath5 of degree @xmath2 one can find a semisimple algebraic group @xmath45 and an equivalence of categories between the abelian tensor category generated by @xmath5 and the category of finite - dimensional representations of @xmath45 .
the algebraic group @xmath45 is called the _ tannaka dual group _ of @xmath5 .
it was shown by the second author in ( * ? ? ?
* lemma 4.4 and proposition 5.3 ) that for stable vector bundles of degree @xmath2 as in theorem [ bohnhorstspindlerhdoneintro ] the almost simple components of the tannaka dual group are of type @xmath46 .
we explain how to compute the tannaka dual group for an arbitrary stable kernel bundle of degree @xmath2 on @xmath0 and construct examples for low - rank syzygy bundles on @xmath7 having the symplectic group @xmath47 as tannaka dual group .
furthermore , we are interested in the behaviour of the tannaka dual group after restricting the bundles to smooth curves .
section [ sectionrestrictiontheorems ] contains a short overview of restriction theorems for sheaves . in section [ tannakaoncurves ] , we describe a method to construct for certain kernel bundles @xmath5 on @xmath0 a finite morphism @xmath48 such that the restriction of the pull - back @xmath49 to certain complete intersection curves of sufficiently large degree has the same tannaka dual group as the vector bundle @xmath49 on @xmath0 .
we show that this works for example for the syzygy bundles constructed in section 5 .
we would like to draw the reader s attention to the paper @xcite by v. balaji .
he shows the existence of a rank @xmath50 bundle @xmath5 with @xmath51 on a smooth surface @xmath52 , such that the restriction to a curve of genus @xmath53 has tannaka dual group @xmath54 , see also ( * ? ? ?
* proposition 3 ) .
his method is completely different from ours .
he uses this result to show that the moduli space of stable principal @xmath55-bundles on @xmath52 with large characteristic classes is non - empty , where @xmath55 is any semisimple algebraic group ( ( * ? ? ?
* chapter 7 ) ) . in section [ stabilityofgenericquadrics ] we close an open case left in the paper @xcite , where l. costa , r. m. mir - roig and p. macias marques show the stability of the generic syzygy bundle on @xmath7 except for the bundle generated by five generic quadrics .
we use the results obtained in section 5 and construct an example for a stable syzygy bundle in this case , which gives the generic result via the openness of stability . in the final section we provide another application of the semistability algorithm concerning the computation of tight / solid closure of homogeneous ideals in the coordinate ring of a smooth projective curve .
this is possible due to the geometric approach to this topic developed by h. brenner .
we would like to thank holger brenner for many useful discussions .
in particular the first author is grateful for the supervision of his phd - thesis @xcite at the university of sheffield where the semistability algorithm is part of one chapter .
we recall that a torsion - free sheaf @xmath5 on a smooth projective variety @xmath52 over an algebraically closed field @xmath1 is _ semistable _ if for every coherent subsheaf @xmath56 the inequality @xmath57 holds .
the sheaf @xmath5 is _ stable _ if the inequality is always strict .
the degree of a sheaf @xmath58 is defined using intersection theory and a fixed very ample invertible sheaf @xmath59 ( which is also called a _ polarization _ of @xmath52 ) as @xmath60 . for every coherent torsion - free sheaf @xmath5 there exists a unique filtration @xmath61 , called the _ harder - narasimhan filtration _ , such that @xmath62 is semistable and @xmath63 .
the slopes @xmath64 and @xmath65 ) are also denoted by @xmath66 and @xmath67 respectively .
if @xmath1 is not algebraically closed , then we define the terms degree , semistable , etc . via the algebraic closure of @xmath1 .
if the characteristic of the base field @xmath1 is @xmath2 , it is well - known that the tensor product @xmath68 of two semistable vector bundles @xmath5 and @xmath58 on a smooth projective polarized variety @xmath69 is again semistable , and this also holds for exterior powers and symmetric powers ( cf .
* theorem 3.1.4 and corollary 3.2.10 ) ) .
this does not longer hold in characteristic @xmath70 .
this is due to the fact that the ( absolute ) _ frobenius morphism _ @xmath71 may destroy semistability , i.e. , the frobenius pull - back @xmath72 of a semistable vector bundle @xmath5 is in general not semistable ; see for instance the example of serre in ( * ? ? ?
* example 3.2 ) .
but for vector bundles on a projective space ( in which we are mainly interested in ) semistability behaves nicely with respect to tensor operations .
[ semistabilitytensorposchar ] let @xmath73 be a smooth projective polarized variety defined over an algebraically closed field @xmath1 of positive characteristic such that @xmath74 .
if @xmath5 and @xmath58 are semistable vector bundles , then @xmath68 , all exterior powers @xmath75 and all symmetric powers @xmath76 are semistable .
in particular , this holds if @xmath52 is an abelian variety , a toric variety or a homogeneous space .
since @xmath74 a semistable vector bundle @xmath5 on @xmath52 is _ strongly semistable _ by ( * ? ? ?
* theorem 2.1 ) ( we recall that this means that the frobenius pull - backs @xmath77 are semistable for all @xmath78 ) .
hence it follows from ( * ? ? ?
* theorem 3.23 ) that @xmath68 , @xmath75 and @xmath76 are also semistable .
the cotangent bundle of an abelian variety is trivial and the cotangent bundles of toric varieties and homogeneous spaces can be embedded into a trivial bundle .
so these varieties fulfill the condition @xmath74 which gives the supplement .
the following result is well - known in characteristic @xmath2 ( see for instance ( * ? ? ?
* proposition 2.1.5 ) or ( * ? ? ?
* proposition 1.1 ) ) .
it gives an algorithmic criterion to check semistability of a vector bundle on @xmath0 in terms of global sections of its exterior powers .
it uses the trivial but useful fact ( in particular for a computational approach to semistability ) that a semistable vector bundle of negative degree ( or slope ) does not have any nontrivial global sections .
since the key - idea goes already back to h. j. hoppe ( see ( * ? ? ?
* lemma 2.6 ) ) , this result is attributed to him .
lemma [ semistabilitytensorposchar ] shows that hoppe s result is also true in positive characteristic .
we mention that the proof which we present below is essentially the same as the one given in @xcite .
[ exteriorpowercriterion ] let @xmath5 be a vector bundle on @xmath0 over an algebraically closed field @xmath1
. then the following holds . 1 .
the bundle @xmath5 is semistable if and only if for every @xmath79 and every @xmath80 there does not exist a non - trivial global section of @xmath81 .
2 . if @xmath82 for every @xmath79 and every @xmath83 , then @xmath5 is stable .
we prove @xmath84 .
if @xmath5 is semistable , then all exterior powers @xmath75 are also semistable by lemma [ semistabilitytensorposchar ] .
we have @xmath85 , which can be easily verified using the splitting principle ( see ( * ? ? ?
* section i.1.2 ) ) .
we have @xmath82 because @xmath86 and @xmath5 is semistable . for the proof of the other direction
, we do not need lemma [ semistabilitytensorposchar ] .
assume that for every @xmath79 and every @xmath80 there does not exist a non - trivial global section of @xmath81 .
let @xmath87 be a coherent subsheaf of rank @xmath79 .
then we also have an inclusion @xmath88 .
the bidual @xmath89 is isomorphic to @xmath90 with @xmath91 .
hence , we have @xmath92 outside a closed subset of codimension @xmath93 ( see ( * ? ? ?
* lemma 1.1.10 ) ) . because @xmath5 is locally free
, there is a non - trivial sheaf morphism @xmath94 , i.e. , @xmath95 . by assumption
we have @xmath96 and therefore @xmath97 .
hence the vector bundle @xmath5 is semistable .
part @xmath98 follows in the same way if we replace @xmath34 by @xmath99 appropriately .
we recall that the concepts of semistability and stability coincide if @xmath100 and @xmath101 are coprime .
as mentioned in @xcite the converse of the stability statement in lemma [ exteriorpowercriterion ] does not hold in general .
the easiest examples are the so - called _ nullcorrelation bundles _ on projective spaces @xmath0 for @xmath36 odd . these bundles are given by a short exact sequence @xmath102 where @xmath103 denotes the tangent bundle .
in particular , we have @xmath104 and @xmath105 . moreover , the nullcorrelation bundles are stable and have the property that if @xmath106 , then @xmath107 ( see [ ibid . , remark below example 1.2 ] ) .
so these bundles do not fulfill the exterior power condition from lemma [ exteriorpowercriterion ] .
nevertheless , it is easy to see that a rank-@xmath50 vector bundle @xmath5 on @xmath0 ( @xmath36 arbitrary ) is stable if and only if @xmath108 for @xmath109 . since @xmath110 ( see ( * ? ? ?
* proof of theorem ii.1.3.1(i ) ) ) , it is clear that a nullcorrelation bundle @xmath111 on @xmath112 is stable and fulfills the exterior power condition .
[ applicationtoothervarieties ] let @xmath69 be a polarized smooth projective variety of dimension @xmath113 defined over an algebraically closed field of characteristic @xmath2 . for a semistable vector bundle @xmath5 on @xmath52
the numerical condition on the exterior powers @xmath75 in lemma [ exteriorpowercriterion ] is still fulfilled , if we replace the degree bound for the global sections by @xmath114 for every @xmath115 . hence , the numerical condition is , up to the factor @xmath116 , always necessary for semistability .
if additionally @xmath117 , then the numerical criterion is again equivalent to the semistability of @xmath5 .
important examples of varieties with this property are general surfaces of degree @xmath118 in @xmath119 ( noether s theorem ) and ( in arbitrary characteristic ) complete intersections of dimension @xmath120 in @xmath0 ( see ( * ? ? ?
* corollary iv.3.2 and iv.4(i ) ) ) . in positive characteristic , ( under the assumption @xmath117 )
the numerical condition on the exterior powers still implies semistability , but the equivalence in lemma [ exteriorpowercriterion ] only holds if every semistable vector bundle on @xmath52 is strongly semistable ( see lemma [ semistabilitytensorposchar ] ) .
thus it is clear that lemma [ exteriorpowercriterion ] does not provide an algorithmic tool to detect semistability of vector bundles on curves . for algorithmic methods to determine semistability and strong semistability of vector bundles over an algebraic curve in positive characteristic
see ( * ? ?
* chapter 3 ) .
[ calabiyauexample ] let @xmath121 $ ] , @xmath122 , be a homogeneous polynomial of degree @xmath123 such that the hypersurface @xmath124/(f))$ ] is smooth . by remark [ applicationtoothervarieties
] we have @xmath125 and thus hoppe s criterion [ exteriorpowercriterion ] is applicable to determine semistability of vector bundles on @xmath52 .
now we assume that @xmath126 .
then the canonical bundle @xmath127 is trivial which implies the semistability of the cotangent bundle @xmath128 ( see ( * ? ? ?
* theorem 3.1 ) ) .
in particular , @xmath52 is a calabi - yau variety .
we consider @xmath52 as the generic fiber @xmath129 of the generically smooth projective morphism @xmath130/(f ) ) \lto \spec \zz\ ] ] of relative dimension @xmath131 . up
to finitely many exceptions , the special fiber @xmath132 over a prime number @xmath133 is a smooth projective variety over the finite field @xmath134 with @xmath135 . by the openness of semistability , the cotangent bundle @xmath136 of the special fiber @xmath132 is semistable too for almost all prime numbers @xmath133 . since @xmath137 ,
every semistable vector bundle is strongly semistable on @xmath132 by ( * ? ? ?
* theorem 2.1 ) .
thus , for @xmath138 we can also use lemma [ exteriorpowercriterion ] to detect semistability of vector bundles on @xmath132 ( in positive characteristic ) .
in the sequel of this article , we restrict ourselves to vector bundles on @xmath0 , @xmath3 , which are kernels of surjective morphisms between splitting bundles , i.e. , bundles sitting inside a short exact sequence of the form @xmath139 where @xmath140 .
the morphism @xmath8 is given by an @xmath9 matrix @xmath141 , where the entries @xmath11 $ ] are homogeneous polynomials of degrees @xmath142 . in this paper , we call such a vector bundle a _ kernel bundle_. special instances of kernel bundles are _ syzygy bundles _ which correspond to the case @xmath13 and @xmath143 , i.e. , a syzygy bundle @xmath32 is given by a short exact sequence @xmath144 where @xmath145 $ ] are homogeneous polynomials of degrees @xmath146 , @xmath25 .
if one of the polynomials is constant , the syzygy bundle @xmath32 is obviously split . to exclude this case ,
one often demands that the ideal @xmath147 is @xmath16-primary , i.e. , @xmath148 .
the most prominent example of a syzygy bundle is the _ cotangent bundle _
@xmath149 of @xmath0 .
we can compute the topological invariants of a kernel bundle @xmath5 from the presenting short exact sequence above .
we have @xmath150 and thus @xmath151 since the chern polynomial is multiplicative on short exact sequences , it is also easy to compute higher chern classes of kernel bundles ( see also section [ sectionrestrictiontheorems ] ) .
if @xmath5 does not split as a direct sum of line bundles , then the dual bundle @xmath43 of a kernel bundle has homological dimension one , and therefore we obtain the inequality @xmath152 ( see ( * ? ? ?
* corollary 1.7 ) ) .
every vector bundle @xmath5 on the projective plane @xmath7 which does not split as a direct sum of line bundles has homological dimension @xmath153 .
this is easy to see , since there exists a surjective sheaf morphism @xmath154 for some splitting bundle @xmath155 , which is also surjective on global sections ( for this standard argument see for instance ( * ? ? ?
* lemma 1.5 ) ) .
thus , there exists a short exact sequence @xmath156 where the first cohomology @xmath157 of the kernel @xmath158 vanishes .
but by the theorem of horrocks ( see ( * ? ? ?
* theorem i.2.3.1 ) ) this means that @xmath158 splits as a direct sum of line bundles .
since @xmath5 is semistable if and only if @xmath43 is semistable , we can dualize the short exact sequence and apply algorithm [ algorithmpn ] to the kernel bundle @xmath43 . hence , our semistability algorithm is applicable to every ( non - split ) vector bundle on @xmath7 and to vector bundles of homological dimension @xmath153 on @xmath0 in general . in the sequel
, we show that the twists @xmath159 and @xmath160 , which occur in the presenting sequence of a kernel bundle , have to fulfill a certain numerical condition which is necessary for semistability ( stability ) .
we remark that this condition is also necessary for the semistability ( stability ) of kernel bundles ( and in particular of syzygy bundles ) on arbitrary smooth projective varieties .
if @xmath5 is a vector bundle on @xmath0 of rank @xmath44 and @xmath87 a subsheaf of rank @xmath161 , then the quotient @xmath162 is outside codimension @xmath50 isomorphic to @xmath163 for some @xmath164 .
this is equivalent to a section @xmath165 . for kernel bundles we are able to control such sections by an easy numerical condition .
in particular , we can replace the condition on the global sections of the @xmath166th exterior power in hoppe s criterion [ exteriorpowercriterion ] by this condition . before we state the result ,
we recall that a resolution @xmath167 of a vector bundle @xmath5 on @xmath0 with splitting bundles @xmath168 , @xmath169 , is _ minimal _ if the global evaluation @xmath170 is exact too for every @xmath171 and no line bundle can be omitted for two consecutive splitting bundles @xmath168 and @xmath172 , which is equivalent to say that there are no constant entries ( @xmath173 ) in the matrices representing the differentials in the resolution ( see ( * ? ? ?
* section 7.2 ) ) . [ linebundlequotients ]
let @xmath5 be a vector bundle on @xmath0 , @xmath3 , sitting in a short exact sequence @xmath174 where @xmath175 . if @xmath176 , then there are no mappings from @xmath5 to line bundles which contradict the semistability @xmath33stability@xmath35 of @xmath5 . moreover ,
if the dualized sequence is a minimal resolution for @xmath43 , then this numerical condition is necessary for semistability @xmath33stability@xmath35 .
we twist the short exact sequence presenting @xmath5 with @xmath163 and look at the dual sequence @xmath177 since @xmath178 for all @xmath164 , every global section of @xmath179 comes from @xmath180 .
consequently , for @xmath181 there exists no non - trivial morphism @xmath182 . by assumption
, we have @xmath183 .
so we have @xmath184 for @xmath185 .
hence there are no mappings to line bundles which contradict the semistability .
analogously , one obtains the corresponding statement for stability .
next , we prove the supplement .
assume that @xmath186 .
it follows from the assumption on the resolution of @xmath43 that the mapping @xmath187 is non - zero .
but such a morphism does not exist for @xmath5 semistable or stable .
[ rankthreecase ] if @xmath5 is a rank-@xmath188 bundle ( not necessarily a kernel bundle ) , then @xmath5 is semistable if and only if @xmath189 for @xmath190 and @xmath191 for @xmath192 ( see ( * ? ? ?
* remark ii.1.2.6 ) ) .
so for kernel bundles of rank @xmath188 on @xmath7 ( for @xmath112 see theorem [ bohnhorstspindlerhdoneintro ] ) we only have to check global sections and the numerical condition of proposition [ linebundlequotients ] .
it is easy to check that the subsheaf @xmath193 destabilizes the kernel bundle @xmath5 if @xmath194 and the resolution of @xmath43 is minimal .
[ minimalitybospi ] in ( * ? ? ?
* proposition 2.3 ) , g. bohnhorst and h. spindler show that for a kernel bundle @xmath5 of rank @xmath36 on @xmath0 , the corresponding resolution of @xmath43 is minimal if and only if @xmath38 , @xmath39 and @xmath40 for @xmath41 holds .
hence , in characteristic @xmath2 , theorem [ bohnhorstspindlerhdoneintro ] shows that for kernel bundles of this type the numerical condition of lemma [ linebundlequotients ] is even sufficient for semistability ( stability ) .
for syzygy bundles , we give an easy characteristic free proof of theorem [ bohnhorstspindlerhdoneintro ] .
[ parametersyzygybundle ] let @xmath1 be a field and let @xmath195 $ ] be homogeneous parameters of degrees @xmath196 .
if @xmath197 , then the syzygy bundle @xmath198 is semistable on @xmath0 .
if the inequality is strict , then @xmath198 is a stable bundle .
we use lemma [ exteriorpowercriterion ] to check the semistability of @xmath198 . for a subset
@xmath199 we use the notation @xmath200 .
we consider the koszul complex @xmath201 on @xmath0 associated to the parameters @xmath202 , where @xmath203 for @xmath204 .
we have to show for every @xmath205 and every @xmath206 that @xmath207 . for every @xmath208
we have a surjection @xmath209 since the koszul complex of a regular sequence is also globally exact , every global section of @xmath210 comes from the bundle @xmath211 .
we have @xmath212 for @xmath213 .
the assumption @xmath214 implies @xmath215 for @xmath216 ( for this easy computation see ( * ? ? ?
* corollary 2.4 ) ) .
but this is equivalent to @xmath217 , @xmath216 , and we obtain the assertion .
the supplement follows analogously .
the aim of this section is to use hoppe s semistability criterion [ exteriorpowercriterion ] to obtain a semistability algorithm for kernel bundles , which can be implemented in a computer algebra system .
all implementations described in this section can be found on http://www2.math.uni-paderborn.de/people/ralf-kasprowitz/cocoa.html .
the major advantage of kernel bundles compared to arbitrary vector bundles is that we can compute the global sections @xmath218 of their exterior powers in a way which is suitable for a computer algebra system .
we do not claim that such a computational approach is impossible for other vector bundles , but at least it requires more technical effort . for a kernel bundle @xmath5 we give (
probably well - known ) presentations of the tensor operations ( tensor powers , exterior powers and symmetric powers ) as kernels of mappings between splitting bundles ( but these mappings are in general not surjective ) .
this enables us to compute the global sections of these vector bundles described by applying the left exact functor @xmath219 . for our semistability algorithm
we require such a presentation only for the exterior powers , but we will need the other tensor operations in section [ tannaka ] .
[ exteriorpowerpresentations ] let @xmath5 be a vector bundle on @xmath0 , which sits in a short exact sequence @xmath220 set @xmath221 , @xmath222 and @xmath223 for a fixed @xmath224 .
the @xmath225th tensor product of @xmath5 sits in the exact sequence @xmath226 where the map @xmath227 is given by @xmath228 here @xmath229 means the @xmath230-tuple @xmath231 without the @xmath133th element .
the @xmath225th exterior power of @xmath5 , @xmath232 , sits in the exact sequence @xmath233 where the map @xmath227 is given by @xmath234 @xmath33the subset @xmath235 is supposed to have the induced order and @xmath236 let @xmath237 .
the @xmath225th symmetric power of @xmath5 sits in the exact sequence @xmath238 where the map @xmath227 is given by @xmath239here @xmath240 means that this element is omitted .
we start with formula ( [ formel1 ] ) .
it is obviously true for @xmath241 , and assume that we proved it for some @xmath224 .
we consider the locally free sheaves @xmath242 and obtain a complex @xmath243 : @xmath244 .
the complex @xmath245 is by definition the complex @xmath246 see for example @xcite .
since the kernels and images of all morphisms occuring in the complex @xmath247 are locally free , we may apply the formula of knneth ( ( * ? ? ?
* theorem vi.3.1 ) ) and get a short exact sequence @xmath248 where the tor sheaves on the right are trivial due to the fact that the sheaves @xmath249 are locally free for all @xmath250 .
observe that the kernel of the morphism @xmath227 is @xmath251 , hence we get with the knneth formula @xmath252 furthermore , it is not difficult to obtain the identity @xmath253 under the obvious identifications @xmath254 and @xmath255 . for formula ( [ formel2 ] ) ,
let @xmath256 be a short exact sequence of locally free sheaves .
consider the koszul complex of the @xmath257-algebra @xmath258 , locally defined by the sequence of elements in @xmath258 consisting of a basis of @xmath58 , i.e. , the complex @xmath259 with @xmath260 and @xmath261 ( @xmath262 means that we omit the @xmath263-th factor ) .
one easily checks that for every @xmath115 this yields an exact complex @xmath264 with @xmath265 . applying this argument for the short exact sequence defining the kernel bundle @xmath5 together with the isomorphism @xmath266 yields the claim .
the case ( [ formel3 ] ) works analogously to the second one , but now we take the koszul complex of @xmath267 locally with respect to the regular sequence consisting of a basis of @xmath5 .
this gives the exact complex @xmath268 applying this to the dual exact sequence that defines the kernel bundle @xmath5 and dualizing again yields the exact complex @xmath269 furthermore , there are isomorphisms @xmath270 and @xmath271 if the characteristic of the field @xmath1 does not divide @xmath225 , see ( * ? ? ?
* satz 86.12 ) .
now , we describe the implementation of the semistability algorithm for kernel bundles .
[ globalsectionconecomputation ] let @xmath5 be a vector bundle on @xmath0 .
the twisted dual bundle @xmath272 is generated by global sections for @xmath273 ( see ( * ? ? ? * theorem 5.17 ) ) , i.e. , we have a surjection @xmath274 for some @xmath250 .
let @xmath158 be the kernel of this morphism . taking duals yields a short exact sequence @xmath275 of vector bundles .
as in the proof of proposition [ exteriorpowerpresentations ] , we derive for every @xmath276 an exact sequence @xmath277 since we also have a surjection @xmath278 for @xmath279 and some @xmath280 , we obtain a diagram where the maps @xmath227 and @xmath281 have the same kernel . in this sense , the algorithmic methods to determine semistability which we develop in this chapter , are applicable to every vector bundle on @xmath0 .
[ cokernelpresentation ] let @xmath282 be a finitely generated graded @xmath283-module
( @xmath284 $ ] ) .
if we fix homogeneous generators @xmath285 , we obtain a surjection @xmath286 from this map we can derive , for every positive integer @xmath287 , the well - known exact sequence @xmath288 where the map @xmath231 is given by @xmath289 ( see ( * ? ? ? * , aufgabe 26 ) ) .
fixing homogeneous generators of @xmath290 gives a diagram in this way we obtain a presentation of @xmath291 as a cokernel of a map between free modules . since all these mappings are graded , we have a corresponding sequence @xmath292 of the associated coherent sheaves on @xmath0 ( with suitable twists ) .
but the map @xmath293 is in general not surjective .
hence this sequence can not be the basis of an algorithmic approach .
if the depth of @xmath294 is at least @xmath50 , then this map is surjective if and only if @xmath295 is surjective ( see ( * ? ? ?
* theorem a4.1 and theorem a4.3 ) ) . an illustrating example is the following . [ monomialexample ]
we consider the syzygy bundle @xmath296 on @xmath297 $ ] ( see also ( * ? ? ?
* example 7.4 ) ) and use brenner s criterion [ monomialcase ] .
the slope of this bundle equals @xmath298 .
for the subsheaves of rank @xmath153 coming from two monomials we have ( we only list the combinations having a common factor ) @xmath299 hence we see that the global sections of @xmath300 do not contradict the semistability .
but the monomial subfamily @xmath301 yields the rank-@xmath50 subbundle @xmath302 of slope @xmath303 .
thus , @xmath300 is not semistable with harder - narasimhan filtration @xmath304 moreover , the mapping @xmath305 is not surjective . since we assume that a vector bundle and its exterior powers are given as kernels of morphisms between splitting bundles , we have to know how to compute the kernel of an @xmath283-linear map @xmath306 between finitely generated free modules over the polynomial ring @xmath283 .
the answer is given by the following well - known lemma which shows that we can compute a minimal nontrivial global section with grbner bases .
[ kernelcomputation ] let @xmath307 $ ] be the polynomial ring over a field @xmath1 and let @xmath308 be an @xmath283-linear map .
denote by @xmath309 the standard basis vectors of @xmath310 .
with the notation @xmath311 , @xmath312 , we have @xmath313 in other words , the kernel of @xmath8 is the @xmath314-@xmath35syzygy module of the columns of the matrix which describes @xmath8 .
see ( * ? ? ? *
proposition 3.3.1(a ) ) .
so we have accumulated the necessary technical tools to formulate an algorithm , based on hoppe s criterion [ exteriorpowercriterion ] , to determine semistability of a given kernel bundle on @xmath0 .
note that every instruction can be performed with any computer algebra system which is able to handle grbner bases calculations . in our case
, we implemented it in cocoa @xcite .
[ semistability ] [ algorithmpn ] two lists @xmath315 $ ] , @xmath316 $ ] and a homogeneous @xmath9 matrix @xmath10 with no constant polynomial entries @xmath317 of degrees @xmath318 defining a kernel bundle @xmath319 with @xmath175 @xmath33as usual @xmath320 denotes the sheaf associated to the graded @xmath283-module @xmath321 . the decision whether @xmath5 is semistable in terms of a boolean value true or false respectively . 1 .
compute the invariants @xmath322 , @xmath323 and @xmath324 .
2 . if the slope condition @xmath325 of proposition [ linebundlequotients ] is fulfilled , then continue .
else return false and terminate .
3 . set @xmath326 .
4 . construct the matrix @xmath327 which describes the map @xmath227 in proposition [ exteriorpowerpresentations ] .
compute the syzygy module @xmath328 of the columns of @xmath327 .
compute the initial degree @xmath329 of the graded @xmath283-module @xmath328 @xmath33i.e .
, @xmath330 is the minimal twist @xmath331 such that @xmath332 .
if @xmath333 , then return false and terminate .
else set @xmath334 and continue .
if @xmath335 , then go back to step @xmath336 .
else return true and terminate .
we require the lists @xmath315 $ ] and @xmath337 $ ] as input data , since they define the degree of every entry @xmath338 in the matrix @xmath339 .
this becomes clear if the matrix @xmath339 contains zeros ( which is often the case if @xmath340 ) .
an illustrating example is @xmath341 on @xmath7 for @xmath16-primary elements @xmath342 . since @xmath343 the semistability heavily depends on the degree @xmath123 which we have allocated to @xmath2 .
one easily verifies that the map given by the matrix @xmath339 is surjective if and only if the ideal generated by all @xmath344 minors of @xmath339 is @xmath16-primary .
the latter property can be checked computationally .
we give a first example of an application of our semistability algorithm .
let @xmath1 be an arbitrary field .
we consider the monomials @xmath345 $ ] and the corresponding sheaf of syzygies @xmath346 .
is @xmath300 a semistable sheaf ?
since the ideal generated by these monomials is not @xmath16-primary , we can neither apply theorem [ monomialcase ] nor ( at first sight ) algorithm [ algorithmpn ] .
we compute a resolution of @xmath300 ( for instance with cocoa ) , namely , @xmath347 where @xmath348 since @xmath300 is a reflexive sheaf , it is locally free on @xmath7 ( cf .
* lemma 1.1.10 ) ) .
so if we dualize the resolution , we obtain a short exact sequence @xmath349 i.e. , @xmath350 is a kernel bundle .
hence , we apply algorithm [ algorithmpn ] to @xmath350 in order to obtain the answer to our question ( we recall that @xmath300 is semistable if and only if @xmath350 is semistable ) .
a cocoa computation gives : 1 .
@xmath351 for @xmath352 and @xmath353 .
2 . @xmath354 for @xmath355 and @xmath356 .
in particular , we obtain no information about stability .
3 . the numerical condition of proposition [ linebundlequotients ] is fulfilled .
so there are no mappings @xmath357 into line bundles which contradict the semistability .
eventually , we conclude via lemma [ exteriorpowercriterion ] that @xmath350 is semistable and so is @xmath300 .
as a first application of the algorithms described in the previous sections we will compute the tannaka dual groups of certain stable syzygy bundles of degree @xmath2 on the projective plane .
we start with describing the setting . from now on
, @xmath1 denotes an algebraically closed field of characteristic @xmath2 and @xmath52 a smooth , irreducible and projective variety over @xmath1
. furthermore , denote by @xmath358 the category of polystable vector bundles of degree @xmath2 on @xmath52 .
it is well - known that @xmath358 is an abelian tensor category that possesses the faithful fiber functor @xmath359 , where @xmath360 is the category of finite - dimensional @xmath1-vector spaces and @xmath361 maps a bundle @xmath362 to its fiber @xmath363 for a point @xmath364 . in other words , it is a neutral tannaka category and hence there exists an affine group scheme @xmath365 over @xmath1 and an equivalence of categories @xmath366 for the theory of tannaka categories , see e.g. @xcite .
we denote by @xmath367 the smallest tannaka subcategory of @xmath358 containing the vector bundle @xmath5 and by @xmath45 its tannaka dual group . the group scheme @xmath365 is pro - reductive and @xmath45 is a reductive linear algebraic group ( not necessarily connected ) .
it is in a natural way an algebraic subgroup of @xmath368 .
furthermore , there is a faithfully flat morphism @xmath369 . since global sections of vector bundles in @xmath358
correspond to @xmath45-invariant elements of the fiber , it follows from ( * ? ? ?
* proposition 3.1 ) that the algebraic group @xmath45 is uniquely determined by the global sections of @xmath370 for @xmath371 .
it even suffices to know the global sections of @xmath372 for @xmath373 , since the dual of a stable bundle @xmath5 occurs as a direct summand in some tensor power of @xmath5 .
[ nottannakian]the restriction to fields of characteristic @xmath2 is essential , as the following example shows .
it was communicated to us by h. brenner .
let @xmath1 be a field of positive characteristic @xmath133 , @xmath374 .
consider the plane curve @xmath375 .
this curve is smooth by the jacobian criterion .
now we look at the syzygy bundle @xmath376 on @xmath377 of degree @xmath2 .
since @xmath5 is stable on @xmath7 by proposition [ parametersyzygybundle ] and @xmath374 , it remains stable on @xmath377 by langer s restriction theorem [ langerrestrictiontheorem ] .
the frobenius pull - back @xmath378 has the non - trivial section @xmath379 because we have the equation @xmath380 on the curve .
it is easy to see that @xmath250 has no zeros on @xmath377 and that there is no further non - trivial section of @xmath72 .
hence @xmath72 is a non - trivial extension of the structure sheaf by itself and therefore not polystable . since @xmath381
, it follows that @xmath382 is not polystable either .
the same holds for @xmath383 since @xmath382 is a quotient of the @xmath133-fold tensor product .
so we see that @xmath384 is not a tensor - category and in particular not tannakian .
let us now consider stable bundles @xmath5 of degree @xmath2 on the projective space @xmath0 .
[ connectedlemma ] the tannaka dual group @xmath45 of a stable vector bundle @xmath5 of degree @xmath2 on @xmath0 is a connected semisimple group .
suppose that the algebraic group @xmath45 is not connected .
then the representations of the finite quotient @xmath385 would correspond to a subcategory of @xmath367 containing nontrivial finite vector bundles , see ( * ? ? ?
* lemma 3.1 ) .
but the latter form , together with the obvious fiber functor , a neutral tannaka category ( see ( * ? ? ?
* proposition 3.7 ) ) , and the @xmath1-valued points of the tannaka dual group are well known to coincide with the tale fundamental group if the characteristic of the ground field is @xmath2 .
hence there are no nontrivial finite vector bundles on the projective space .
furthermore , the reductive group @xmath45 does not have any nontrivial characters due to the fact that @xmath386 , hence it has to be semisimple .
it was shown in ( * ? ? ?
* lemma 4.4 and proposition 5.3 ) that for stable vector bundles of degree @xmath2 and rank @xmath44 as in theorem [ bohnhorstspindlerhdoneintro ] the almost simple components of the tannaka dual group have to be of type @xmath46 .
we conjecture that it is always the standard @xmath44-dimensional representation of @xmath387 or its dual .
one motivation for this paper was to construct examples of syzygy bundles on the projective space having a tannaka dual group of type different from @xmath46 .
note that one can not simply try to guess an example for a syzygy bundle with group different from @xmath388 , see the remark [ moduli ] below .
our idea is to exclude this case by constructing self - dual bundles . with the algorithmic methods described in [ semistability ] and
proposition [ exteriorpowerpresentations ] , it is in principle possible to compute the tannaka dual group and its representation for an arbitrary stable kernel bundle of degree @xmath2 .
however , the necessary computations grow very fast with the rank of the bundle , so we were only able to handle syzygy bundles up to rank @xmath389 on @xmath7 .
furthermore , we only found syzygy bundles having the almost simple tannaka dual group @xmath390 , where @xmath391 .
there are no stable self - dual syzygy bundles of odd rank on @xmath7 , see corollary [ nooddrank ] . for stable rank @xmath50 bundles of degree @xmath2 on @xmath7 ,
there is only one possible tannaka dual group , namely the @xmath50-dimensional irreducible representation of @xmath392 .
an example for this is the syzygy bundle @xmath393 .
it is stable due to theorem [ bohnhorstspindlerhdoneintro ] . to find higher rank syzygy bundles whose tannaka dual group is not the group @xmath387 with an @xmath44-dimensional representation
, we will use the following simple lemma .
[ selfdual ] let @xmath394 $ ] be homogeneous polynomials such that the ideal @xmath395 is @xmath16-primary and minimally generated by @xmath396 .
then @xmath397 is self - dual @xmath33up to a twist with a line bundle@xmath35 if and only if @xmath398 is gorenstein .
the minimal free resolution of @xmath398 has length @xmath188 and is self - dual up to twist since @xmath398 is gorenstein : @xmath399 and with @xmath400 we have @xmath401 .
in particular , there is an isomorphism @xmath402 with @xmath403 .
conversely , one easily sees that if @xmath5 is self - dual up to twist , then the minimal free resolution of @xmath398 ends with a free module of rank @xmath153 , that is @xmath398 is gorenstein .
[ nooddrank ] there are no self - dual @xmath33up to a twist with a line bundle@xmath35 non - split syzygy bundles of odd rank on @xmath7 .
this is @xcite , corollary 2.2 , which says that the minimal number of generators of a gorenstein ideal of grade @xmath188 is odd .
all stable syzygy bundles of degree @xmath2 and odd rank @xmath404 on the projective plane have a semisimple tannaka dual group whose simple components are of type @xmath46 .
a table of representations of lie algebras ( e.g. @xcite ) shows that the smallest non - self - dual irreducible representation of a semisimple algebraic group with some simple component not of type @xmath46 is the @xmath33up to duality@xmath35 @xmath405-dimensional representation of the semisimple group @xmath406 , which is the tensor product of the @xmath188-dimensional irreducible representation of @xmath407 with the @xmath408-dimensional irreducible representation of @xmath409 .
now looking at rank @xmath408 , what possibilities are there for the lie algebra of the tannaka dual group ?
there are three different self - dual and irreducible representations , where we use the notations of the tables of simple lie algebras and their representations in @xcite : @xmath410 with highest weight @xmath188 , @xmath411 with highest weight @xmath412 , and @xmath413 with highest weight @xmath414 . using the computer algebra package for lie group computations lie @xcite
, one finds that @xmath415 in the cases of type @xmath46 and @xmath416 in the case of @xmath411 . to find a bundle with tannaka dual group of type @xmath411
, we have to construct a gorenstein ideal @xmath417 with @xmath418 minimal generators , such that the associated syzygy bundle @xmath5 has degree @xmath2
. then we have to check its stability and the global sections of @xmath419 . to find a suitable gorenstein ideal @xmath417
, we consider the polynomial ring @xmath420 $ ] as a module over itself by interpreting a polynomial in @xmath283 as differential operator acting on itself , e.g. , @xmath421 .
choose a homogeneous polynomial @xmath422 of degree @xmath44 and define @xmath423 .
this ideal is artinian and gorenstein , see for example @xcite , lemma 2.12 .
[ exampleofstablebundle ] the gorenstein ideal of the homogeneous form @xmath424 is the ideal @xmath425 .
if we pull back the associated syzygy bundle on @xmath7 via the finite morphism @xmath426 , we obtain the twisted syzygy bundle @xmath427 it has degree @xmath2 , rank @xmath408 and is self - dual by lemma [ selfdual ] .
we apply algorithm [ algorithmpn ] to @xmath5 and obtain with the help of cocoa : 1 .
@xmath428 for @xmath429 ; 2 . @xmath430 for @xmath431 , @xmath432 for @xmath433 ; 3 .
@xmath434 for @xmath435 ( this computation is , by proposition [ linebundlequotients ] , actually not necessary since the degrees of the polynomials are constant ) .
hence , we see that @xmath5 @xmath33and all its twists@xmath35 are semistable , but we get no information about stability because of @xmath332@xmath35 . observe that hoppe s criterion can never reveal stability of a self - dual bundle of rank @xmath408 , since in this case there must be nontrivial global sections of @xmath436 , see again @xcite .
fortunately , the self - duality of the bundle allows us to prove its stability . by the computation
above , we only have to consider subsheaves of rank @xmath50 which may destroy the stability of @xmath437 .
assume that @xmath438 is a stable subsheaf of rank @xmath50 and degree @xmath2 . since we can pass over to the reflexive hull ,
and since we are working on @xmath7 , we may assume that @xmath58 is locally free ( see ( * ? ? ?
* lemma 1.1.10 ) ) .
in particular , we have @xmath439 by ( * ? ? ?
* proposition 1.10 ) .
that is , the subsheaf @xmath58 is self - dual too .
so the composition of the morphisms @xmath440 yields an endomorphism of @xmath437 , which is not a multiple of the identity . but a computation of global sections using the implementation of proposition [ exteriorpowerpresentations ] yields that @xmath441 that is , the bundle @xmath437 is simple and a morphism as above does not exist .
hence , the bundle @xmath437 is stable .
finally , another computation shows @xmath442 , hence the tannaka dual group is in fact almost simple of type @xmath411 and the representation of its lie algebra has highest weight @xmath412 .
it is well - known that this corresponds to the irreducible and faithful representation @xmath443 .
[ secondexample ] the gorenstein ideal associated to the homogeneous form @xmath444 via the correspondence described above is the ideal @xmath445 we consider the same pull - back as in the previous example to be able to twist the associated syzygy bundle to degree @xmath2 .
the same computations as above show that the corresponding syzygy bundle @xmath446 , defined as @xmath447 is semistable of degree @xmath2 , where a subsheaf @xmath58 that destroys stability has to be of rank @xmath448 or @xmath449 .
but since @xmath446 is self - dual , we can always assume @xmath58 to be of rank @xmath50 and obtain stability for the same reason as above , since the bundle again turns out to be simple . again using lie @xcite
, we find that it is possible to determine the lie algebra of the tannaka dual group by computing @xmath450 , which shows that it has to be simple of type @xmath451 , with highest weight @xmath452 .
this corresponds to the faithful irreducible representation @xmath453 .
[ moduli ] there is a good reason why one should expect to find the group @xmath47 in these cases .
it is well - known that the moduli space of stable bundles of fixed rank and chern classes exists as a quasi - projective variety , see for example
* theorem 4.3.4 ) .
let us denote by @xmath294 the moduli space containing the syzygy bundle of example [ exampleofstablebundle ] respectively of example [ secondexample ] .
let @xmath454 be the quasi - universal bundle on @xmath455 , see ( * ? ? ?
* chapter 4.6 ) .
this means that the restriction of @xmath454 to @xmath456 is a finite product of copies of the stable bundle on @xmath7 corresponding to the point @xmath133 .
applying the semicontinuity theorem @xmath33 e.g. ( * ? ? ?
* theorem 12.8 ) @xmath35 for @xmath457 one finds that the locus of the vector bundles having tannaka dual group @xmath387 is open in @xmath294 since the dimension of @xmath458 is minimal in this case and strictly higher in all other cases .
hence for a generic choice of a stable syzygy bundle on @xmath7 one expects to find the group @xmath387 as tannaka dual group .
the locus of self - dual bundles is closed in @xmath294 ( apply the semicontinuity theorem for @xmath459 ) , containing the bundles with tannaka dual group @xmath460 as an open locus ( since the dimension of @xmath461 is minimal for a self - dual stable bundle with this tannaka dual group , see the discussion of the examples above ) .
it follows that for a generic choice of a self - dual syzygy bundle one expects the group @xmath460 as tannaka dual group
. it would be interesting to find a method for constructing stable syzygy bundles having a tannaka dual group different from @xmath387 or @xmath460 .
furthermore , one could try to determine the geometry of the tannaka - strata of the moduli spaces discussed above .
algorithm [ algorithmpn ] enables us to determine semistability ( and in some cases stability ) of kernel bundles on projective spaces . if our algorithm gives a positive answer , then what can be said about the semistability ( respectively stability ) of these bundles when we restrict them to a hypersurface @xmath462 ?
the answer is given by so - called _ restriction theorems _ which ensure the semistability ( stability ) of the restriction of a bundle to hypersurfaces of sufficiently large degree .
these theorems hold in more general situations , we only give formulations for vector bundles on projective spaces . by applying restriction theorems
, we can use algorithm [ algorithmpn ] to produce examples of semistable vector bundles on more complicated projective varieties .
we omit the famous restriction theorem of mehta and ramanathan ( see ( * ? ? ?
* theorem 6.1 ) or ( * ? ? ?
* theorem 7.2.8 ) ) , which works over an arbitrary algebraically closed field @xmath1 and provides the existence of an integer @xmath463 such that for @xmath464 general elements @xmath465 the restriction of a semistable torsion - free sheaf @xmath5 on @xmath0 to the smooth complete intersection @xmath466 is semistable for all @xmath467 .
but their theorem provides no bound for @xmath463 . unlike the theorem of mehta and ramanathan , the following restriction theorem of h. flenner gives an explicit bound for @xmath463 , but only works in characteristic @xmath2 .
[ flennerrestrictiontheorem ] let @xmath1 be an algebraically closed field of characteristic @xmath2 and let @xmath5 be a semistable coherent torsion - free sheaf of rank @xmath44 on @xmath0 .
then for @xmath263 and @xmath468 fulfilling @xmath469 and for @xmath464 general elements @xmath470 the restriction @xmath471 is semistable on the smooth complete intersection @xmath472 .
see ( * ? ? ?
* theorem 1.2 ) or ( * ? ? ?
* theorem 7.1.1 ) .
the strongest restriction theorem is due to a. langer .
it works in arbitrary characteristic and gives a degree bound for arbitrary smooth hypersurfaces in projective space .
it involves the _ discriminant _ @xmath473 of a locally free sheaf @xmath5 of rank @xmath44 , where @xmath474 and @xmath475 denote the first and second chern class respectively .
[ langerrestrictiontheorem ] let @xmath1 be an algebraically closed field and let @xmath5 be a stable coherent torsion - free sheaf of rank @xmath476 on @xmath0 and let @xmath477 be a smooth divisor such that @xmath478 is torsion - free .
if @xmath479 then the restriction @xmath478 is stable . see ( * ? ? ?
* theorem 2.19 ) . for a kernel bundle @xmath5 given by a short exact sequence @xmath480
we have already mentioned in section [ syzygyandkernelbundles ] that @xmath481 .
since the chern polynomial is multiplicative on short exact sequences , it is also easy to see that we have @xmath482 and hence @xmath483 in particular , we have for a syzygy bundle @xmath484 given by homogeneous @xmath16-primary polynomials @xmath485 $ ] of degrees @xmath486 the formula @xmath487 . in @xcite h. brenner has shown that there is no restriction theorem for strong semistability in the sense of theorem [ langerrestrictiontheorem ] . for general hypersurfaces
@xmath462 there is the following flenner - type restriction theorem which is also due to a. langer .
[ langerstronglysemistablerestriction ] let @xmath5 be a semistable locally free sheaf of rank @xmath476 on @xmath0 over an algebraically closed field @xmath1 of positive characteristic .
let @xmath263 be an integer such that @xmath488 and @xmath489 then @xmath478 is strongly semistable on the general hypersurface in @xmath490
. see ( * ? ? ?
* theorem 3.1 ) .
we return to the situation of section [ tannaka ] . recall that the ground field @xmath1 is algebraically closed and of characteristic @xmath2 . here
we investigate the problem of the behavior of tannaka dual groups after restricting a stable bundle of degree @xmath2 on @xmath0 to smooth connected curves @xmath52 such that the restricted bundle is still stable .
the main problem is that the connected tannaka dual group of a stable vector bundle on @xmath0 may become disconnected after restriction to a curve : let @xmath491 be a connected smooth curve of genus @xmath53 .
then there exists a stable bundle of degree @xmath2 on @xmath7 such that its restriction to @xmath52 is again stable with nonconnected tannaka dual group .
recall that a finite vector bundle is a vector bundle which is trivialized by a finite tale morphism .
there is a one - to - one correspondence between finite vector bundles on @xmath52 and representations of the tale fundamental group @xmath492 having finite image .
now choose such an irreducible representation of dimension @xmath44 with trivial determinant , which certainly exists if the genus of the curve is @xmath53 .
it is well - known that the associated vector bundle @xmath362 is stable of degree @xmath2 , with tannaka dual group equal to the image of the representation .
further , its determinant is the trivial bundle . since @xmath493 is generated by @xmath494 global sections ( ( * ? ?
* lemma 2.3 ) ) , it is easy to see that one obtains a short exact sequence @xmath495 the dual of the morphism @xmath8 lifts to an exact sequence @xmath496 on @xmath7 . in particular , @xmath497 and the singularities of @xmath5
are of codimension @xmath498 , hence @xmath5 is locally free and of course stable of degree @xmath2 .
the tannaka dual group of @xmath5 is connected by lemma [ connectedlemma ] , but the restriction of the bundle @xmath5 to @xmath52 has a finite tannaka dual group . hence we need a criterion for the tannaka dual group to be connected after restricting the bundle @xmath5 to the smooth and connected curve @xmath52 .
for the rest of this section we denote by @xmath133 a prime number and by @xmath499 an algebraic closure of the p - adic numbers with ring of integers @xmath500 and residue field @xmath501 .
we call a finitely presented , flat and proper scheme @xmath502 over @xmath500 together with an isomorphism @xmath503 a _ model _ of @xmath52 . note that any scheme @xmath502 over @xmath500 is the disjoint union of the generic fiber @xmath504 , which is open in @xmath502 , and the special fiber @xmath505 , which is closed . for the rest of this section set @xmath506 .
[ connectedtheorem ] let @xmath362 be a vector bundle on the smooth , connected and projective curve @xmath52 over @xmath499 .
if there exists a model @xmath502 of @xmath52 together with a vector bundle @xmath5 on @xmath502 such that @xmath507 and @xmath508 is a trivial bundle on the scheme @xmath509 , then @xmath362 is semistable of degree @xmath2 with connected tannaka dual group @xmath510 .
the proof uses results from non - abelian @xmath133-adic hodge theory .
the semistability of the bundle is shown in ( * ? ? ?
* theorem 13 ) , the assertion that @xmath510 is connected follows from ( * ? ? ? * theorem 3.12 ) .
[ exampleofrestriction ] let @xmath5 be a vector bundle on @xmath0 sitting in the short exact sequence @xmath511 where the morphism @xmath8 is defined by homogeneous polynomials @xmath512 with @xmath513 $ ] .
the vector bundle @xmath5 is stable of degree @xmath2 due to theorem [ bohnhorstspindlerhdoneintro ] . if we consider @xmath5 as a sheaf on @xmath514 , it is easy to see that it is a locally - free sheaf outside the closed subset @xmath515\}\subset \pp^n_{\kappa } \subset \pp^n_{\mathfrak{o}}$ ] .
then for every smooth connected curve @xmath462 which has a model @xmath516 such that the special fiber does not intersect the subspace @xmath515\}$ ] , the vector bundle @xmath517 has @xmath36 linearly independent global sections modulo @xmath133 and hence is trivial .
if the curve @xmath52 is the intersection of @xmath518 smooth divisors of degree @xmath519 , theorem [ langerrestrictiontheorem ] yields that the restriction @xmath362 of @xmath5 to @xmath52 is a stable bundle , and it follows from theorem [ connectedtheorem ] above that @xmath510 is connected . see also ( * ? ? ?
* example 4.6 and remark 4.8 ) . a similar argument was used by h. brenner to provide examples for stable vector bundles on @xmath133-adic curves with semistable reduction , see ( * ? ? ?
* remark p. 571 ) .
in general , we can not apply theorem [ connectedtheorem ] to an arbitrary kernel bundle @xmath5 on @xmath0 , but in many cases one can apply it to the pull - back @xmath520 with respect to a suitable chosen finite morphism @xmath521 such that @xmath520 is defined over @xmath500 and such that it is trivial modulo @xmath133 outside a closed subset of codimension @xmath93 .
then for all curves @xmath52 which have a model @xmath516 whose special fiber does not intersect this closed subset , we have @xmath522 if the curve @xmath52 is a complete intersection of smooth divisors of sufficiently high degree
. we will illustrate this method in the following examples .
consider the syzygy bundles @xmath523 @xmath524 on @xmath7 from example [ exampleofstablebundle ] and example [ secondexample ] . to find a suitable finite morphism @xmath525 as explained above , we try to construct a nontrivial morphism @xmath526 such that the pull - back @xmath527 is the trivial bundle
. then we can choose a rational map @xmath528 that is defined outside the point @xmath529\ } \in \pp^2_{\ff_p } \subset \pp^2_{\zz_p}$ ] such that modulo @xmath133 there is the commutative diagram @xmath530\ } \ar[d ] \ar[r]^<<<<<{\pi_{i,\ff_p } } & \pp^2_{\ff_p } \\
\pp^1_{\ff_p } \ar[ur]_{g_i } & } \ ] ] where the vertical morphism is defined as @xmath531 \mapsto [ x;y]$ ] .
it is then clear that @xmath532 is modulo @xmath133 the trivial bundle on the open subset @xmath533\}$ ] . a computation with cocoa shows that the morphism @xmath534 can for example be chosen as @xmath535 \mapsto [ x;y;2x+y]$ ] if the prime @xmath133 is odd .
then the rational map @xmath536 can be chosen as @xmath531 \mapsto [ x;y;2x+y+pz]$ ] .
the restriction to the generic fiber gives the finite morphism @xmath525 we were looking for .
on the generic fiber , the bundles @xmath537 are polystable , and one computes @xmath538 using proposition [ exteriorpowerpresentations ] .
it follows that they have to be stable .
let @xmath539 be a model of a smooth connected curve @xmath540 such that the special fiber @xmath541 does not contain the point @xmath542 $ ] .
if the degree of the plane curve @xmath52 is large enough , we may use theorem [ langerrestrictiontheorem ] and assume that the restriction of the bundle @xmath537 is still a stable bundle .
it follows from theorem [ connectedtheorem ] that its tannaka dual group is a connected semisimple group .
furthermore , we have @xmath543 for curves of sufficiently large degree .
hence the tannaka dual groups satisfy @xmath544 , @xmath545 .
the syzygy bundle @xmath546 on @xmath7 is stable of degree @xmath2 due to theorem [ monomialcase ] , with tannaka dual group @xmath547 because of @xmath548 .
the morphism @xmath549 can be chosen as @xmath535 \mapsto [ x^2+y^2;x^2;x^2+xy]$ ] and hence @xmath550 for example as @xmath531 \mapsto [ x^2+y^2;x^2;x^2+xy + pz^2]$ ] . the same computations and arguments as above then show that @xmath551 for a plane curve @xmath52 of sufficiently large degree as in the above example .
it is natural to ask if this method works for all semistable vector bundles of degree @xmath2 on @xmath0 : let @xmath5 be a semistable vector bundle of degree @xmath2 on @xmath0 .
is there always a finite morphism @xmath521 and a model @xmath552 of @xmath0 such that @xmath520 lifts to a sheaf on @xmath552 which is modulo @xmath133 a free sheaf outside a closed subset of codimension @xmath93 ?
it is an open question of whether for generic forms @xmath17 of degrees @xmath486 in the polynomial ring @xmath307 $ ] over an algebraically closed field @xmath1 the corresponding syzygy bundle is semistable or even stable .
there is no chance if the @xmath146 s do not satisfy the necessary degree condition of proposition [ linebundlequotients ] .
hence , the question only makes sense if the necessary condition on the degrees is fulfilled , e.g. , if we consider forms of constant degree .
since semistability is an open property , it is enough to find a single @xmath16-primary family @xmath285 having the same degree configuration such that @xmath553 is semistable . via @xmath16-primary monomial families
@xmath22 , @xmath23 , one can use brenner s result [ monomialcase ] to establish generic semistability in a combinatorial way by producing examples of monomial families with semistable syzygy bundle .
this has been done recently in ( * ? ? ?
* theorem 4.6 ) , where p. macias marques and r. m. mir - roig have proved the stability of the syzygy bundle @xmath32 on @xmath0 for generic forms of degree @xmath123 with @xmath554 , @xmath555 .
this extends ( * ? ? ?
* theorem 3.5 ) of l. costa , r. m. mir - roig and p. macias marques , where only the case @xmath556 has been proven .
the general result of @xcite has also been obtained simultaneously by i. coand in @xcite . for the case
@xmath556 , @xmath557 and @xmath558 , where only semistability has been shown , macias marques asks the following question ( see ( * ? ? ?
* problem 2.9 ) ) .
[ questionmarques ] is there a family of five quadratic homogeneous polynomials in @xmath559 $ ] such that their syzygy bundle is stable ?
note that one can not establish generic semistability via a monomial example since for the only candidate we have @xmath560 we answer macias marques s question in the following proposition .
[ answertomarquesproblem ] the syzygy bundle @xmath561 is stable on @xmath562 $ ] .
moreover , the syzygy bundle for five generic quadrics in @xmath563 $ ] is stable on the projective plane .
the syzygy bundle @xmath564 , which we have considered in example [ exampleofstablebundle ] , is the pull - back of @xmath565 under the finite morphism @xmath566 since @xmath300 is a stable bundle , so is @xmath5 .
the supplement follows from the openness of stability . in the recent preprint @xcite i. coand
has independently proved in [ ibid .
, example 1.3 ] the stability of the generic syzygy bundle for @xmath567 .
but his proof is more complicated and does not provide an explicit example of a family of five homogeneous quadrics in three variables .
let @xmath568 be the moduli space of stable vector bundles of rank @xmath569 and chern classes @xmath570 on the projective space @xmath0 .
denote by @xmath571 the stratum of stable syzygy bundles @xmath5 defined by the short exact sequence @xmath572 in his thesis @xcite , m. marques computes the dimension of the syzygy stratum and the codimension of its closure in the irreducible component of the moduli space .
however , he could not give an answer for the case @xmath573 due to the lack of a stable syzygy bundle of five homogeneous quadrics .
our example also closes this gap .
we recall that the moduli spaces of stable bundles on @xmath7 with fixed invariants are irreducible and their dimensions are known , see for example @xcite . the syzygy stratum @xmath574 has dimension @xmath418 .
in particular , @xmath575
. see ( * ? ? ?
* proposition 4.2 and theorem 4.3 ) , where the proof presented there works analogously for the case @xmath556 , @xmath557 , @xmath558 due to proposition [ answertomarquesproblem ] .
our semistability algorithm also has impact on certain ideal closure operations in commutative algebra due to a geometric interpretation by h. brenner .
we recall briefly the notions of _ tight closure _ and _ solid closure _ , where we restrict ourselves to the case where the ring @xmath283 under consideration is a noetherian integral domain . for a detailed exposition of these closure operations and their backround
see @xcite .
let @xmath576 be an ideal and @xmath577 .
the @xmath283-algebra @xmath578/(f_1 t_1 + \ldots + f_n t_n+f)$ ] is called the _ forcing algebra _ for the elements @xmath579 . the element @xmath422 belongs to the _ solid closure _ , which we denote by @xmath580 , if and only if the following holds : for every maximal ideal @xmath581 of @xmath283 the top - dimensional local cohomology module @xmath582 does not vanish , where @xmath583 is the forcing algebra for the given data over the local complete domain @xmath584 and @xmath585 .
now assume that @xmath283 is of positive characteristic @xmath133 ( i.e. , @xmath283 contains a field of positive characteristic ) .
then the _ tight closure _ of @xmath417 is defined as the ideal @xmath586 } \mbox { for all } q = p^e\},\ ] ] where @xmath587 } = ( f_1^q \komdots f_n^q)$ ] denotes the extended ideal under the @xmath588-th iteration of the frobenius @xmath589 , @xmath590 .
an important fact due to m. hochster is that @xmath591 holds in positive characteristic for a normal @xmath1-algebra @xmath283 of finite type ( in fact this is true under weaker assumptions ) ; see ( * ? ? ?
* theorem 8.6 ) .
it follows already from the definitions of these closure operations that they are hard to compute . for a normal standard graded integral @xmath50-dimensional algebra @xmath283 over an algebraically closed field @xmath1
there is a well - developed theory by h. brenner for solid closure and tight closure which connects these notions with ( strong ) semistability of the corresponding syzygy bundle @xmath15 on the smooth projective curve @xmath592 ; see @xcite for an excellent survey .
this geometric approach combined with the restriction theorems presented in section [ sectionrestrictiontheorems ] enables us to use our semistability algorithm [ algorithmpn ] to compute inclusion bounds for solid closure and tight closure in homogeneous coordinate rings of smooth projective curves , particularly plane curves , of sufficiently large degree .
[ inclusionsolidclosure ] let @xmath1 be an algebraically closed field of characteristic zero and @xmath283 be a normal standard graded @xmath1-domain of dimension two . further let @xmath593 be an @xmath16-primary homogeneous ideal
. if @xmath15 is semistable on @xmath592 then @xmath594 where @xmath20 for @xmath595 .
so theorem [ inclusionsolidclosure ] gives for an element @xmath596 an inclusion @xmath597 for @xmath598 and for @xmath599 the question of whether @xmath422 belongs to @xmath580 reduces to an ideal membership test which is a well - known procedure in computational algebra ( cf . for instance ( * ? ? ?
* proposition 2.4.10 ) ) .
the following theorem works in positive characteristic under the assumption that the syzygy bundle is strongly semistable .
we recall that a vector bundle @xmath5 is _ strongly semistable _ if for every @xmath78 the frobenius pull - backs @xmath600 are semistable .
so our algorithmic methods for tight closure can either be applied to homogeneous coordinate rings of the general curve of large degree ( via theorem [ langerstronglysemistablerestriction ] ) or to coordinate rings of elliptic curves ( there semistable bundles are strongly semistable by ( * ? ? ?
* theorem 2.1 ) ) .
[ inclusiontightclosure ] let @xmath1 be an algebraically closed field of characteristic @xmath70 and @xmath283 be a normal standard graded @xmath1-domain of dimension two . further let @xmath593 be an @xmath16-primary homogeneous ideal such that @xmath15 is strongly semistable on @xmath592
. denote the genus of @xmath377 by @xmath601 . then the following hold . 1 .
if @xmath602 then @xmath603 .
2 . if @xmath599 and @xmath596 then @xmath604 if and only if 1 .
@xmath605}=(f_1^p \komdots f_n^p)$ ] if @xmath606 2 . or
@xmath607}=(f_1^q \komdots f_n^q)$ ] for @xmath608 if @xmath609 .
let @xmath610 be a generically smooth projective relative curve and @xmath611 be an @xmath16-primary ideal .
in this situation one can deduce tight closure information of the reductions @xmath612 in the fiber rings @xmath613 from semistability in characteristic @xmath2 .
let @xmath614 denote the syzygy bundle on the total space @xmath615 .
if @xmath616 is semistable on the generic fiber @xmath617 and @xmath618 then @xmath619 has positive degree and is therefore ample ( see ( * ? ? ?
* theorem 2.4 ) ) .
since ampleness is an open property , the reductions to positive characteristic @xmath620 and @xmath621 are also ample on the special fibers @xmath622 for almost all prime numbers @xmath623 .
in this situation brenner s geometric approach also yields results on the tight closure of @xmath624 for @xmath138 ; see ( * ? ? ?
* section 4 ) for a detailed treatment of ampleness criteria for tight closure .
in particular , by [ ibid . , proposition 4.17 ] we have @xmath625 ( in fact @xmath626 already belongs to the _ frobenius closure _ @xmath627 } \mbox { for some } q = p^e\ } \subseteq i_p^*$ ] ) .
what can be said in higher dimensions ? as usual , we consider @xmath16-primary homogeneous polynomials @xmath17 in @xmath628 $ ] .
we can compute a minimal graded free resolution @xmath629 of the ideal @xmath147 ; see ( * ? ? ?
* section 4.8.b ) for the computational backround .
since the quotient @xmath630 is artinian , the length of @xmath629 equals @xmath631 by the auslander - buchsbaum formula .
consequently , the corresponding resolution of the associated sheaves on @xmath0 gives a resolution @xmath632 of the structure sheaf with splitting bundles @xmath168 , @xmath633 .
instead of looking at @xmath634 , we consider the bundle @xmath635 to check whether @xmath636 is semistable , we can apply algorithm [ algorithmpn ] to its dual @xmath637 which is a kernel bundle .
if the answer is positive , then we obtain an inclusion bound for the tight closure @xmath638 in the homogeneous coordinate ring @xmath283 of a generic hyperplane @xmath462 of sufficiently large degree .
this works as follows .
if we restrict the resolution @xmath629 to @xmath52 , we obtain an exact complex of sheaves on @xmath52 and the sheaf @xmath639 is strongly semistable for @xmath640 by theorem [ langerstronglysemistablerestriction ]
. then brenner s result ( * ? ? ?
* theorem 2.4 ) gives the inclusion bound @xmath641 , where @xmath642 .
note that we can compute all necessary invariants ( rank , degree , discriminant ) of @xmath636 from the resolution @xmath629 .
we obtain the same inclusion bound if we restrict @xmath636 to smooth hypersurfaces @xmath462 for which every semistable bundle on @xmath52 is strongly semistable ; compare for instance lemma [ semistabilitytensorposchar ] and example [ calabiyauexample ] . here
the degree bound for @xmath643 , which ensures the semistability of @xmath639 , is given by theorem [ langerrestrictiontheorem ] .
to3em , _ tight closure and vector bundles _ , three lectures on commutative algebra ( j. elias s. zarzuela g. colom - nin , t. cortodellas benitez , ed . ) , university lecture series , vol .
42 , ams , 2008 , pp . 171 .
h. cartan and s. eilenberg , _ homological algebra _ ,
princeton landmarks in mathematics , princeton university press , princeton , nj , 1999 , with an appendix by david a. buchsbaum , reprint of the 1956 original .
a. iarrobino and v. kanev , _ power sums , gorenstein algebras , and determinantal loci _ , lecture notes in mathematics , vol .
1721 , springer - verlag , berlin , 1999 , appendix c by iarrobino and steven l. kleiman .
w. g. mckay , j. patera , and d. w. rand , _ tables of representations of simple lie algebras .
i _ , universit de montral centre de recherches mathmatiques , montreal , qc , 1990 , exceptional simple lie algebras , with a foreword by r. v. moody .
t. peternell , _ subsheaves in the tangent bundle : integrability , stability and positivity _ , school on vanishing theorems and effective results in algebraic geometry ( j. p.
demailly et al . , ed . ) , ictp lect .
, vol . 6 , the abdus salam international centre for theoretical physics , 2001 , pp . 285334 . | we develop a semistability algorithm for vector bundles which are given as a kernel of a surjective morphism between splitting bundles on the projective space @xmath0 over an algebraically closed field @xmath1 .
this class of bundles is a generalization of syzygy bundles .
we show how to implement this algorithm in a computer algebra system .
further we give applications , mainly concerning the computation of tannaka dual groups of stable vector bundles of degree @xmath2 on @xmath0 and on certain smooth complete intersection curves .
we also use our algorithm to close an open case left in a recent work of l. costa , p. macias marques and r. m. mir - roig regarding the stability of the syzygy bundle of general forms .
finally , we apply our algorithm to provide a computational approach to tight closure .
all algorithms are implemented in the computer algebra system cocoa .
mathematical subject classification ( 2010 ) : primary : 14j60 , 14q15 ; + secondary : 13p10 keywords : semistable vector bundle , syzygy bundle , tannaka duality , monodromy group , tight closure |
the study of hadronic properties at finite temperature @xmath2 is one of the theoretical ingredients needed to understand the behaviour of matter created in relativistic heavy ion collision experiments , such as those in rhic and lhc ( alice ) . in particular , the qcd transition involving chiral symmetry restoration and deconfinement plays a crucial role , as it is clear from the many recent advances of lattice groups in the study of the phase diagram and other thermodynamical properties @xcite . for vanishing baryon chemical potential ,
the qcd transition is a crossover for 2 + 1 flavours with physical quark masses , the transition temperature being about @xmath5 150 - 160 mev , determined by the vanishing behaviour of the quark condensate and the peak of the scalar susceptibility . in the chiral limit
it becomes a second - order phase transition consistent with the @xmath6-model universality class @xcite , which is supported in lattice simulations by the mass and temperature scaling of thermodynamical quantities as well as chiral partner degeneration @xcite .
the expected reduction in the transition temperature from the physical mass case to the chiral limit one based on those analysis is about 15 - 20@xmath7 @xcite . from the theoretical side , it is therefore important to provide solid analysis of this chiral restoration pattern based on effective theories , given the limitations of perturbative qcd at those temperature scales .
such effective description should start from a proper understanding of the lightest component , i.e. the pion gas .
pions are actually the most abundant particles after a heavy ion collision and most of their properties from hadronization to thermal freeze - out can be reasonably described within the temperature range where effective theories are applicable .
in fact , approaches based on effective theories for the lightest mesons provide a good description of the physics involved , especially in what concerns the effect of the lightest resonances , as we discuss below .
a more accurate treatment of thermodynamic quantities near @xmath8 would require including heavier degrees of freedom , which can be efficiently achieved through the hadron resonance gas framework @xcite . a systematic and model - independent framework that takes into account the relevant light meson degrees of freedom and their interactions is chiral perturbation theory ( chpt ) @xcite .
the effective chpt lagrangian is constructed as a derivative and mass expansion @xmath9 , where @xmath10 denotes generically a meson energy scale compared to the chiral scale @xmath11 1 gev .
the lowest order lagrangian @xmath12 is the non - linear sigma model ( nlsm ) .
thus , chiral restoring behaviour is qualitatively obtained within chpt through the vanishing quark condensate for different orders @xcite , although a critical description is not obtained in the chiral limit . despite the model - independent character of the chpt predictions for chiral restoration , the low-@xmath2 nature of this theory implies a continuous behaviour for order parameters and susceptibilities , even in the chiral limit .
thus , the quark condensate is a dropping continuous function around the critical point and the scalar susceptibility grows also continuously .
in addition , chpt is unable to describe resonant states , which play a crucial role in the description of the hadronic medium .
some of these limitations are improved within the unitarized framework at finite temperature @xcite , which provides an accurate description of several effects of interest in a heavy - ion environment , such as thermal resonances and transport coefficients @xcite .
it provides also a novel understanding of the role of the @xmath13 @xmath14 thermal pole ( a broad resonant state ) in chiral symmetry restoration .
thus , the scalar susceptibility saturated with this @xmath15-like state within the so - called inverse amplitude method ( iam ) unitarization , develops a maximum near @xmath8 @xcite compatible with lattice data and chiral partners in the scalar - pseudoscalar sector are understood through degeneration of correlators and susceptibilities .
the role of the @xmath4 state for chiral restoration could become more complicated if its possible tetraquark component is also considered at finite temperature @xcite .
a complementary approach is the large-@xmath1 one , where @xmath1 is the number of light nambu - goldstone bosons ( ngb ) . within this framework , the lowest order chiral effective lagrangian for low - energy qcd will be the @xmath0 nlsm , whose corresponding symmetry breaking pattern is @xmath16 .
as we have just commented , the latter is believed to take place in chiral symmetry restoration for @xmath17 , since @xmath6 and @xmath18 are respectively isomorphic to the isospin groups @xmath19 and @xmath20 . in this limit ,
many of the features discussed above arise naturally . for instance , the pion scattering amplitude generates the @xmath4 resonance in accordance with scattering and pole data .
when extended at finite temperature , thermal unitarity holds exactly and the thermal pole gives rise to a saturated scalar susceptibility diverging in the chiral limit at the critical temperature as a second - order phase transition @xcite . within a similar context , it is worth mentioning also recent large-@xmath1 analysis in the vector @xmath21 model regarding the @xmath15 spectral properties at finite @xmath2 @xcite .
the large-@xmath1 framework for the nlsm has been analyzed in earlier works under various approximations . at @xmath22 , functional methods were developed in @xcite ; the scattering amplitude , including its renormalizability , was studied in the chiral limit in @xcite and to leading order in mass corrections in @xcite . at finite temperature , apart from the previously mentioned work @xcite , the free energy and the quark condensate within a saddle - point approximation for the auxiliary field were analyzed in @xcite .
other studies in the chiral limit that also utilize functional methods @xcite provide a chiral restoring analysis and the @xmath2-dependence of the pion decay constant @xcite .
a later work @xcite studied the nlsm as the infinite coupling limit of the @xmath21 vector model , their results for the nlo pressure in the chiral limit not being fully consistent with those in @xcite .
various renormalization - group studies of the critical properties of this model were compiled in @xcite . in this work
we will analyze the leading large-@xmath1 contributions to the quark condensate and the scalar susceptibility , derived diagrammatically from the partition function or the free energy .
since we are mostly interested in the study of the critical behaviour , we will restrict to the leading order in the chiral limit , which has the additional advantage of yielding results that are not to be renormalized . the scalar susceptibility and the analysis of the critical behaviour are new from this work .
in addition , another motivation for the present study and a prominent difference with respect to previous analysis is that we will work directly within the diagrammatic approach to the partition function , identifying the dominant contributions to the free energy .
thus , that approach will not introduce additional assumptions within the auxiliary field method , such as saddle point approximations , but will require a careful evaluation of the diagrams and effective vertices involved .
in fact , for the susceptibility analysis we will deal with a resummation of an infinite set of closed ring diagrams , which will confirm qualitatively the analysis performed previously in terms of pion scattering and the thermal @xmath4 saturation .
that comparison is also another motivation for the present work . we will need to calculate the large-@xmath1 free energy up to order @xmath23 in order to extract properly the leading order susceptibility near the chiral limit , which actually means going beyond previous analysis of the nlsm for large @xmath1 . in this sense , our present study aims to set up the correct diagrammatic framework for future analysis beyond the chiral limit .
the paper is organized as follows : we will present our main formalism in section [ sec : form ] , the detailed diagrammatic analysis will be performed in section [ sec : diag ] and the results will be presented in section [ sec : res ] , where we compare the obtained critical behaviour with previous theoretical approaches , as well as with lattice results for critical exponents .
we start from the lagrangian of the nonlinear @xmath24 model with a explicit symmetry breaking term which generates the pion mass @xcite @xmath25\partial_{\mu}{\bm{\pi}}^{a}\partial^{\mu}{\bm{\pi}}^{b } + nf^2m^2\sqrt{1-\frac{{\bm{\pi}}^2}{nf^2 } } \nonumber\\ & = & nf^2m^2+\frac{1}{2}\partial_{\mu}{\bm{\pi}}^{a}\partial^{\mu}{\bm{\pi}}^{a}-\frac{1}{2}m^2{\bm{\pi}}^2+\frac{1}{nf^2}{\bm{\pi}}^a{\bm{\pi}}^b \partial_\mu{\bm{\pi}}_a\partial^\mu{\bm{\pi}}_bf\left(\frac{{\bm{\pi}}^2}{nf^2}\right)-\frac{m^2}{8nf^2}({\bm{\pi}}^2)^2g\left(\frac{{\bm{\pi}}^2}{nf^2}\right ) , \label{nlsm}\end{aligned}\ ] ] with @xmath26 and @xmath27 and @xmath28 are respectively the pion mass and the pion decay constant in the chiral limit . here
we have explicitly separated the kinetic free lagrangian and write the interaction part , with and without derivatives , in terms of the functions @xmath29=-8\sum_{k=0}^{\infty}(-1)^k \begin{pmatrix } 1/2\cr k+2 \end{pmatrix } x^k=1+\frac{x}{2}+\frac{5}{16}x^2+\cdots,\end{aligned}\ ] ] where @xmath30 .
both @xmath31 and @xmath32 are normalized so that the leading order in the @xmath33 chpt expression ( @xmath34 ) corresponds to @xmath35 .
similar functions are used in the analysis of pion scattering in the massive case at @xmath22 within the auxiliar field method @xcite .
the free energy from which all the thermodynamic variables can be extracted is given by @xmath36 , \label{freeenergy}\ ] ] where @xmath37 is the qcd partition function in the pionic sector , hence expected to be dominant at low and moderate temperatures , and @xmath38 , @xmath39 , @xmath40=\mathcal{l}_{nlsm}[{\bm{\pi}}]+\dots$ ] , the dots indicating higher order lagrangians in derivatives and masses , which would eventually have to be included to renormalize the theory , within the same approach followed in previous works @xcite .
the light quark condensate behaves as an order parameter for chiral symmetry restoration in the chiral limit : @xmath41 where @xmath42 is the light quark mass , @xmath43 with @xmath44 the number of light flavours and we have used the standard relation between the ngb mass and the quark mass @xmath45 , related also to the @xmath22 quark condensate in the chiral limit @xmath46 .
the quark condensate correlator defines the scalar susceptibility , namely : @xmath47}. \label{chil4q}\end{aligned}\ ] ] an important comment is that throughout this work , we will be interested only in the chiral limit @xmath48 , which , as explained above , should capture the essential features of the chiral phase transition , both for the quark condensate and for the scalar susceptibility .
however , from their previous definitions and , we see that it is necessary to keep @xmath27 finite and send it to zero only after differentiation .
thus , we will consider the large-@xmath1 leading contribution for finite mass and then we will keep only the relevant terms in the @xmath27 expansion near the chiral limit .
this is a distinctive feature with respect to previous large-@xmath1 nlsm analysis at finite @xmath2 , which analyze @xmath49 within the limit of massless pions , but not its @xmath27 corrections @xcite , particularly relevant for the case of @xmath50 , since , from its divergent nature near the transition it is indeed expected to behave as an @xmath51 quantity @xcite , which we will also obtain in this approach by keeping the relevant @xmath23 terms . in the following analysis
, the free energy density will be expressed in terms of different thermal functions .
following the same notation as in @xcite , we define @xmath52 satisfying @xmath53 and whose expansion in @xmath54 reads ( we keep the terms relevant for this work ) @xmath55\label{g0},\\ g_1(m , t)&=&\frac{t^2}{12}\left[1 - 3\frac{m}{\pi t}+{{\cal o}}\left(\left(\frac{m}{2\pi t}\right)^2\log\left(\frac{m}{2\pi t}\right)\right)\right]\label{g1},\\ g_2(m , t)&=&\frac{t}{8\pi m}+{{\cal o}}\left(\log\left(\frac{m}{2\pi t}\right)\right).\label{g2}\end{aligned}\ ] ] the above @xmath56 functions arise naturally from thermal parts of loop functions .
thus , defining @xmath57 as the tadpole function with @xmath58 the free pion propagator , we have : @xmath59 , \label{g1}\end{aligned}\ ] ] with @xmath60 the matsubara frequencies , @xmath61 the renormalization chpt scale and @xmath62 the euler gamma function .
we follow the same notation as in @xcite .
the @xmath63 function corresponds to the thermal part of the scattering loop with zero external momenta : @xmath64 similarly , we define the loop function @xmath65 \qquad \mbox{for } \
k\geq 3 , \label{gk}\end{aligned}\ ] ] where we have extracted the leading order in the @xmath66 expansion , from the asymptotic expansion .
in order to identify the leading order contribution in the large-@xmath1 limit for fixed mass @xmath27 , it is useful to examine the different possible terms for a given number of vertices , i.e. , to a certain order in the expansion of the interaction part of the lagrangian ( [ nlsm ] ) in the free energy .
thus , we first extract the constant term @xmath67 which is @xmath68 and we will denote by @xmath69 the leading contribution for @xmath70 vertices at large @xmath1 for fixed @xmath27 . the contribution with zero vertices
is that coming from the free kinetic pion part of the nlsm lagrangian ; this corresponds to the free partition function of a @xmath1-component massive boson gas : @xmath71 , \label{z0}\ ] ] which is also @xmath68 .
the contributions @xmath72 and @xmath73 are given by diagrams ( a ) and ( b ) in figure [ fig : parfun ] , respectively .
we also keep the space - time dimension @xmath74 in the dimensional regularization scheme .
the free energy is divergent and needs renormalization , as it is discussed in detail within the chpt expansion in @xcite .
renormalization within the large-@xmath1 framework can also be carried out by including suitable higher order lagrangians in derivatives and masses .
this is reviewed for the scattering amplitude at @xmath22 in @xcite and for @xmath75 in @xcite . at the order considered in this work , as we will see , the dominant contribution is finite
, so we will not need to implement explicitly such renormalization procedure .
and up to @xmath3 .
the black squares denote the effective vertex depicted in figure [ fig : massvertex ] .
the dashed lines in ( e ) indicate multiple insertions of this vertex along the central loop . ]
the one - vertex contribution corresponds to the first order in the expansion of the interaction lagrangian in ( [ nlsm ] ) . in the mass term ,
i.e. , the one with the @xmath76 function , we get the maximal contribution in @xmath1 by contracting all pion pairs with the same isospin indices , so that @xmath77 with @xmath78 in .
therefore , we have to sum all tadpole insertions in the vertex as given by the function @xmath76 in the lagrangian , which we attain by defining an effective vertex as indicated in figure [ fig : massvertex ] . on the other hand
, one realizes that contractions involving the interaction derivative term in the lagrangian ( the term proportional to the function @xmath79 in ) give always subdominant contributions in @xmath1 since @xmath80 by parity , so those contributions are @xmath81 in the large-@xmath1 expansion of the free energy .
we point out that the combinatoric factors shown in figure [ fig : massvertex ] remain the same as those obtained from the @xmath82 expansion of the mass vertex in ( [ nlsm ] ) .
this is because our analysis , albeit diagrammatical , is not describing any specific scattering process as in previous analysis @xcite but closed diagrams for the free energy , so we do not have to consider extra factors in the resummation given by the function @xmath83 .
another significative difference with the scattering case is that here derivative vertices do not show up to leading order .
thus , we obtain for the large-@xmath1 leading one - vertex contribution to the free energy @xmath84+{{\cal o}}(n^0 ) \nonumber \\ & = & -nm^2f^2 \left\ { h\left[\frac{t^2}{12f^2}\right]-\frac{mt}{4\pi f^2}h ' \left[\frac{t^2}{12f^2}\right ] \right\ } + { { \cal o}}\left[m^4\log m , n^0\right ] , \label{z1 } \end{aligned}\ ] ] according to and , with @xmath85
. corresponds to @xmath86 in @xcite . ] the above contribution @xmath87 corresponds to the connected diagram ( c ) in figure [ fig : parfun ] .
the topology of this diagram is the same as in chpt @xcite but here effective mass vertices enter and derivative vertices are absent . for two vertices , following the above considerations , the dominant contribution in @xmath1 , which is again @xmath68 , is obtained by taking the two vertices as the two mass effective vertices of figure [ fig : massvertex ] .
the corresponding connected diagram is shown in figure [ fig : parfun ] ( d ) .
thus , those effective vertices count as @xmath82 each , the external bubbles connected to the vertex also count as @xmath1 , and the internal bubble connecting the two vertices counts an additional @xmath88 coming from @xmath89 .
any other combination is subleading , including the derivative vertices in ( [ nlsm ] ) . hence , even if the @xmath90 contractions in @xmath31 are taken maximally in @xmath1 , since @xmath91 can not be contracted at the same point one ends for diagram ( d ) in fig .
[ fig : parfun ] with the structure @xmath92 for one mass vertex and one derivative vertex and @xmath93 for two derivative vertex , instead of the @xmath94 structure of the two mass vertices .
following the same arguments , other topologies of diagrams considered in the chpt expansion @xcite are subdominant for large @xmath1 at finite mass , like diagram ( a ) in figure [ fig : otherdiagrams ] .
expansion for fixed @xmath95 , while ( b ) and ( c ) are of leading @xmath68 order but subleading in the @xmath27 expansion . ] to calculate the dominant @xmath96 contribution from figure [ fig : parfun ] ( d ) , we have to include a combinatoric factor accounting for all the possible ways to choose a @xmath90 in each vertex that is to be connected with the other vertex .
thus , at each vertex there will be an additional @xmath97 factor , where @xmath98 is the integer labelling the power @xmath99 at vertex @xmath100 .
hence , we end up with a modified effective vertex function for that diagram , namely : @xmath101=\frac{1}{\sqrt{1-x}}-1 . \label{modeffvert}\end{aligned}\ ] ] in addition , we have to multiply by 2 from the two ways to contract @xmath102 and by @xmath103 from the lagrangian expansion , so that the leading @xmath68 term for @xmath96 is : @xmath104 ^ 2 g_2(m , t ) + { { \cal o}}(n^0)= -\frac{nm^3t}{32\pi}\left[\tilde g\left(\frac{t^2}{12f^2}\right)\right]^2 + { { \cal o}}(m^4\log m , n^0 ) \label{z2}\end{aligned}\ ] ] with @xmath105 in and using the expansions and around the massless limit .
now , for three or more vertices @xmath70 , following the previous arguments , the dominant connected diagrams are those with only effective mass vertices and @xmath106 bubbles .
those bubbles can be either closed around one of the vertices or connecting a subset of them , so that two given vertices are connected at most with two lines .
examples of these general @xmath68 foam or super - daisy type diagrams @xcite are diagram ( e ) in figure [ fig : parfun ] and diagrams ( b ) and ( c ) in figure [ fig : otherdiagrams ] .
however , a crucial point of our present approach is that only a subclass of those foam diagrams is dominant to leading order in the @xmath27 expansion , namely the ring or daisy diagrams of the type depicted in figure [ fig : parfun ] . to understand this , note first that any of those generic diagrams contains a product of the type @xmath107 , multiplied by the corresponding effective vertices ( which are functions of @xmath78 only ) , with @xmath108 the number of loops and @xmath109 given in .
hence , we see that the minimum power of @xmath95 in a foam diagram is obtained when setting @xmath110 to its maximum allowed value , which is @xmath111 and hence @xmath112 , which corresponds to the loop function @xmath113 connecting those @xmath70 single bubbles , and the other @xmath114 .
including also the effective vertices described above , that combination corresponds to the ring diagrams depicted in figure [ fig : parfun](e ) , whose @xmath27 counting is then @xmath115 times a function of @xmath116 , near the chiral limit .
other foam diagrams are subleading , like diagram ( b ) in figure [ fig : otherdiagrams ] which is proportional to @xmath117 and so on for other diagrams like diagram ( c ) in figure [ fig : otherdiagrams ] .
we recall that the dominance of ring diagrams near the infrared region is a known feature of thermal field theory @xcite and allows us precisely to extract the @xmath3 term needed to calculate the leading order of the scalar susceptibility .
in addition , and as announced above , to this leading order all the results are finite , since the @xmath22 corrections are subleading . therefore , at this level of approximation we will not need to discuss the renormalization details of the calculation in terms of the infinite set of coupling constants arising from higher order lagrangians @xcite . within that standard large-@xmath1 approach , we can consider then that those low - energy constants of higher order are subleading .
we will just keep in mind that they may introduce subleading corrections to our expressions , which for instance could modify numerically the transition temperature , as we discuss below .
the energy density up to @xmath3 in the large-@xmath1 limit is obtained then by summing all ring diagrams , taking into account that we have to multiply by : i ) @xmath118^n$ ] from the modified effective vertices , which count all the ways to take in each effective vertex the two pion lines shared with the adjacent vertices , ii ) by @xmath119 for the two possible ways to take those shared lines in every link , iii ) by @xmath120 from the series expansion of the interaction lagrangian , iv ) by @xmath121 , which are the topologically different ways of sorting @xmath70 points in a circle , and v ) by @xmath122 from the loops .
finally we obtain : @xmath123-\frac{mt}{4\pi f^2}h ' \left[\frac{t^2}{12f^2}\right]\right\ } \nonumber\\ & - & \frac{nm^3t}{8\pi}h\left[-\frac{1}{2 } \tilde g\left(\frac{t^2}{12f^2}\right)\ \right ] + { { \cal o}}\left[m^4\log m , n^0\right ] , \label{zorderm3}\end{aligned}\ ] ] where @xmath124 . the combinatoric factor given here resembles that obtained for a finite - temperature effective higgs potential in @xcite .
we remark that the result is meaningful only as a expansion around the chiral limit @xmath48 so we can extract the leading order for the quark condensate and the scalar susceptibility in that limit .
although the mass expansion performed here could yield mass corrections to the free energy , quark condensate and scalar susceptibility , we must be careful at this point .
the @xmath95 expansion that we have carried out here comes from two different dimensionless parameters , namely @xmath54 and @xmath125 .
thus , while we expect the behaviour of thermal functions to be dominated by the chiral limit contribution in the high temperature regime @xmath126 , @xmath125 is a pure @xmath22 parameter coming from the mass vertex in the lagrangian . therefore , mass corrections might not be reliable for physical masses .
a related complication introduced by the large-@xmath1 resummation is that , as showed above , from @xmath96 onwards the effective vertex is actually @xmath127 , with @xmath128 given in , which may introduce spurious divergences near @xmath129 when keeping a finite mass @xmath95 . for these reasons
, we will stick here to the strict chiral limit @xmath48 since it ensures that those two dimensionless scales can be treated as perturbatively equivalent .
thus , we will keep only the @xmath130 term in for the condensate and the @xmath131 term for the scalar susceptibility . in this work
we are interested in the critical behaviour of the large-@xmath1 expansion and , therefore , it makes sense to restrict to the chiral limit as a first approximation to the problem . a proper treatment of finite mass effects within the large-@xmath1 expansion would require in principle to sum all types of foam diagrams like that in figure [ fig : otherdiagrams ] ( c ) .
from the analysis of the previous section , we can easily extract the quark condensate in the chiral and large-@xmath1 limits , from and : @xmath132 where we denote by @xmath133 the temperature at which the quark condensate vanishes , and @xmath134 .
the result is certainly expected from the chiral limit analysis in @xcite , where the same temperature dependence is found for the scaling of the effective sigma field @xmath135 .
although the identification @xmath136 is very natural from the viewpoint of the quark model assignment , note that within our present approach we do not need to introduce any such @xmath15 field , since we have the exact leading mass dependence of the free energy .
the above result is also numerically compatible with the condensate obtained from the large-@xmath1 framework in @xcite , which relies on a saddle point approximation developed via the auxiliary field method and depends on a cutoff parameter .
therefore , the quark condensate at this level of approximation vanishes at @xmath129 and , more importantly , is not defined above @xmath129 , unlike the chpt expression which is just a polynomial whose leading order is recovered just by expanding in powers of @xmath137 , namely @xmath138 ; this corresponds to the leading order chpt result in the chiral limit for @xmath17 @xcite .
note also that , consistently with previous studies @xcite @xmath129 is independent of @xmath1 to leading order . as for the particular value of the transition temperature @xmath129 , taking the standard value for the pion decay constant in the chiral limit @xmath139 87.1 mev yields @xmath140 174.24 mev , which is high compared to the restoration temperature in the chiral limit expected from lattice data .
however , the present approach is not meant to provide a @xmath129 close to such lattice values , mostly because is based only on ngb degrees of freedom .
heavier states in the partition function would contribute significantly to reduce the condensate and @xmath129 @xcite .
in addition , higher order lagrangians could modify the @xmath22 value of the condensate and hence the critical temperature .
what is more meaningful is to compare the result with the standard chpt analysis to different orders in the chiral limit @xcite , as we show in figure [ fig : cond ] .
the value of @xmath129 is reduced with respect to the nnlo chpt while kept within the expected uncertainty given by a @xmath82 expansion near @xmath17 .
actually , from the chiral expansion of we get already a factor of @xmath141 reduction in @xmath129 with respect to leading order chpt . in the chiral limit compared to the chpt results for different orders as obtained in @xcite ]
as discussed in the introduction , the main improvement of the large-@xmath1 analysis with respect to the chpt one is that the system undergoes a second - order phase transition , as corresponds to qcd in the chiral limit , which is clear from the analytic expression and from figure [ fig : cond ] .
the system does not undergo a first - order phase transition since there are not jump discontinuities in the order parameter ; the latter behaviour is seen for instance when considering auxiliar field methods to analyze finite - temperature effects in an @xmath6 nonlinear model @xcite .
this second - order critical behaviour will be confirmed by our analysis of the scalar susceptibility below . despite the fact that we can not access from this approach the region @xmath142 , let us analyze further the analytic behaviour of the condensate below @xmath129 . at this point
it is important to recall the critical behaviour observed in lattice qcd . in @xcite , the scaling with the quark mass and the temperature of lattice data
is fitted reasonably well to the three - dimensional @xmath143 and @xmath6 universality classes ( see @xcite for a review ) .
thus , the quark condensate is expected to scale in the chiral limit @xmath144 as the second - order behaviour @xmath145 for @xmath146 , where the value of the critical exponent @xmath147 is given in table [ tab : critexp ] for the cases considered in @xcite .
our value is then @xmath148 from , which is not far from the lattice observations , still within the @xmath82 expected error , although giving a stronger critical behaviour .
note that , as commented above , we are including only goldstone bosons and not heavier states and hence we must consider our framework as a qualitative description of the transition , which actually captures its main critical features and certainly improves over standard low-@xmath2 expansions such as chpt . in fact , the critical behaviour is meant to be controlled by goldstone bosons below the transition @xcite .
3d @xmath143 & 0.35 & 0.49 + 3d @xmath6 & 0.38 & 0.54 + this work & 0.5 & 0.75 + saturated thermal pole large @xmath1 @xcite & & 0.975 + saturated thermal pole iam @xcite & & 1.005 + in order to gain more insight into the above mentioned critical features , let us now calculate the leading order contribution of the scalar susceptibility @xmath50 from the free energy result , keeping only the leading term for @xmath48 .
after taking into account the mass expansions of all the functions involved we find the following remarkably simple expression : @xmath149 expression diverges below @xmath129 as a second - order phase transition , as qcd in the chiral limit and confirming our previous analysis of the condensate .
actually , the critical behaviour with @xmath2 and @xmath95 for @xmath143 and @xmath6 3d models is given by @xmath150 for @xmath151 and @xmath48 , where the values of the critical exponent @xmath152 are given in table [ tab : critexp ] .
therefore , we reproduce the expected mass behaviour for @xmath50 and we obtain a critical exponent slightly above but not far from those models , which reproduce fairly well the critical behaviour of lattice qcd @xcite . as explained above , this is more than reasonable for a description based only on the lightest ngb which again lies within the expected numerical uncertainty while capturing the main features of the chiral transition .
note also that , consistently , from we recover the leading order chpt expression for the thermal part of the susceptibility in the chiral limit @xcite as @xmath153 .
the @xmath22 vanishes at this order and is included in the @xmath154 neglected corrections in .
finally , we also compare our results with the analysis performed in @xcite and @xcite , from which one can define a scalar susceptibility saturated by the thermal @xmath4 state , with a mass corresponding to that scalar resonance defined as @xmath155 , where @xmath156 is the position of the pole in the second riemann sheet of the scattering partial wave with isospin and angular momentum @xmath14 calculated also at finite temperature within the large-@xmath1 limit in @xcite and with the iam in @xcite around the chiral limit . in order
that such saturated susceptibility complies properly with the expected low-@xmath2 behaviour given by the lo chpt in the chiral limit , we define it as @xmath157 neglecting the @xmath22 logarithmic contribution near @xmath158 , and we plot @xmath159 for different approaches , which is then @xmath1-independent .
the results are showed in figure [ fig : suscep ] , where we also include the chpt chiral limit one for comparison , which to leading order is just a constant with the normalization chosen .
higher order chpt corrections yield smoothly growing functions of @xmath2 . in the saturated large-@xmath1 case
, we have used the parameters of the so called grayer fit in @xcite .
saturated one defined through , considered in @xcite at large @xmath1 , where the results with grayer fit have been considered .
we also include the saturated susceptibility with the iam @xcite in the chiral limit and the chpt lo chiral limit result for comparison .
temperatures are rescaled to the critical temperature for each case , namely @xmath160 mev for the present approach , @xmath161 mev for the saturated large-@xmath1 one , @xmath162 mev for the saturated iam and @xmath163 mev for the chpt chiral limit one . ]
first of all , we observe that the values of the critical temperature differ considerably between the free - energy analysis , like the one we present here or chpt , and those based on thermal @xmath4 saturation ( either iam or large-@xmath1 ) .
as we have commented above , the numerical value of @xmath129 obtained from the present approach is presumably affected by higher order terms and does not account for any physical state other than the ngb ones .
in contrast , the saturation approach incorporates successfully the thermal @xmath4 , relying on a good description of the physical @xmath22 pole consistent with scattering data and the quoted pdg values for that state .
in fact , the critical temperature of that approach is closer to the expected values from lattice analysis , which highlights the importance of that thermal state .
nevertheless , the most important result here has to do with the critical behaviour , so that in order to provide a clearer comparison of the two approaches in that sense , we have represented the susceptibility in terms of @xmath164 , where @xmath129 is different for each method .
as it can be seen from figure [ fig : suscep ] , although the critical exponent for the saturated @xmath165 are larger ( the numerical results are given also in table [ tab : critexp ] in table [ tab : critexp ] differ from those quoted in @xcite by small numerical corrections which come from taking more points close to the transition temperature in the present work . ] ) , they depart later from the low-@xmath2 chpt value , so in the end the two approaches remain very close near the critical region .
this is an important check of consistency between those two different ways to determine @xmath50 , concerning its critical behaviour .
in this work we have analyzed the quark condensate and the scalar susceptibility of a gas of @xmath1 goldstone bosons to leading order in @xmath1 in the chiral limit . to obtain the leading behavior of the susceptibility
requires to compute up to @xmath23 corrections in the free energy ( which come from an infinite set of closed ring diagrams ) . to this order ,
the results are directly finite and then , it is no necessary to add higher order lagrangian counterterms within the usual large-@xmath1 framework .
this diagrammatic treatment in terms of thermal effective vertices and dominant diagrams would allow to extend this analysis beyond the chiral limit or for higher orders in the @xmath82 expansion .
our results show a critical behaviour in reasonable agreement with the universality classes expected for lattice simulations of the chiral transition , both for the quark condensate and for the scalar susceptibility , within the numerical uncertainties expected for a large-@xmath1 approach .
this is particularly realized in the critical exponents for those quantities .
the quark condensate improves over nnlo chpt in the sense that it reduces the value of its vanishing point and , more importantly , it behaves as the order parameter of a second - order transition , the critical @xmath129 being @xmath1-independent to leading order .
the differences in the critical temperature with respect to chpt remain within the expected @xmath82 uncertainty .
likewise , the scalar susceptibility that we have calculated here diverges at the same @xmath129 as the condensate vanishing , as it should .
in fact , another motivation of the present work was to test the consistency of a recent approach based on saturating the scalar susceptibility with the thermal @xmath4 resonance pole .
our present analysis shows a reasonable agreement between the two methods regarding the critical behaviour in terms of @xmath164 .
however , the value of @xmath129 is much lower in the saturated approaches that in those based on the partition function , like chpt or the one we present here .
this is most likely due to having incorporated properly the physical @xmath4 state and its thermal dependence , which somehow mimics the influence of higher order contributions in the perturbative chiral approach . to obtain higher order corrections in
the mass expansion would require to sum additional infinite sets of diagrams , the so - called foam diagrams .
this would be especially interesting for the case of the scalar susceptibility , since in principle it should lead to a change from a divergent second - order transition to a smooth peak characteristic of a crossover , as observed in lattice simulations .
that analysis is beyond the scope of this work and will be reported elsewhere .
nevertheless , we believe that the present study can be useful as a first approach to this problem , setting up the diagrammatic framework for future studies while extending previous analysis in a nontrivial way and capturing the main features of chiral restoration in qcd .
work partially supported by the spanish research contract fpa2014 - 53375-c2 - 2-p .
we also acknowledge the support of the eu fp7 hadronphysics3 project and the spanish hadron excelence network ( fis2014 - 57026-redt ) .
santiago corts thanks prof .
jos rolando roldn and the high energy physics group of universidad de los andes and colciencias for financial support .
a. bazavov , t. bhattacharya , m. cheng , c. detar , h. t. ding , s. gottlieb , r. gupta and p. hegde _ et al .
_ , phys . rev .
d * 85 * , 054503 ( 2012 ) . t. bhattacharya , m. i. buchoff , n. h. christ , h .-
ding , r. gupta , c. jung , f. karsch and z. lin _ et al .
_ , phys .
lett . * 113 * , 082001 ( 2014 ) .
t. appelquist and c. w. bernard , phys .
d * 23 * , 425 ( 1981 ) ; l. tataru , phys . rev .
d * 12 * , 3351 ( 1975 ) .
a. dobado and j. r. pelaez , phys .
b * 286 * , 136 ( 1992 ) .
a. dobado and j. morales , phys .
d * 52 * , 2878 ( 1995 ) . | we analyze chiral restoration within the @xmath0 non - linear sigma model for large @xmath1 as an effective theory for low - energy qcd at finite temperature @xmath2 .
the free energy is constructed diagramatically to @xmath3 in the pion mass , which allows to derive the quark condensate and the scalar susceptibility in the chiral limit . at this order
, we do not have to deal with renormalization , neither from divergences from mass tadpoles nor from those of higher order loop contributions .
our results for the critical behaviour are consistent with expectations from lattice analysis and with previous works where the susceptibility is saturated by the thermal @xmath4 pole . |
the increasing observational evidence associating supernovae ( sne ) and long gamma - ray bursts ( lgrbs ) ( see reviews by @xcite ) , supports scenarios like the magnetar @xcite and the collapsar @xcite , in which the death of a massive , rapidly rotating star is responsible for the formation of a lgrb . in the collapsar scenario ,
the energy for the production of a lgrb is provided either by neutrino cooling @xcite , by the blandford - znajek ( bz ) mechanism @xcite , or a combination of both .
the formation and evolution of the accretion disk has been studied in many hydrodynamical and mhd simulations of collapsars @xcite .
it has been found that in order to have an effective neutrino cooling mechanism , the infalling material s angular momentum distribution @xmath5 , must be larger than the critical value @xmath6 needed to form an accretion disk around a schwarzschild bh @xcite .
further work has focused in studying the importance of the progenitor s angular momentum distribution in the energy production mechanism @xcite . nevertheless ,
only 2d hydrodynamical simulations by @xcite have studied in detail the effect of transitions in the angular momentum distribution between subcritical ( @xmath3 ) and supercritical ( @xmath2 ) values .
such transitions in the angular momentum distribution have been obtained in sn and grb progenitor models , and are caused by convection at burning shells transporting angular momentum outwards from the inner part of the convection zone @xcite . however , if an effective mixing caused by magnetic torques and convection is present during the evolution of the long grb progenitor , it is possible that such transitions in the angular momentum are suppressed @xcite .
the results from @xcite , show that in the scenario where the transition in the angular momentum distribution persists , the accretion disk formed from the collapse of supercritical material can in principle be completely destroyed by the infall of a massive enough subcritical shell ( @xmath7 times more massive than the supercritical shell ) .
this produces a quiescent time where there is an inefficient conversion of potential to thermal energy in the quasi - radial accretion flow .
however , the collapsar scenario is intrinsically a 3d problem , where instabilities can easily appear and break the axisymmetry assumed in 2d simulations . in a previous paper ( @xcite , hereafter bl14 )
, we studied this process and in particular considered the formation of structure and gravitational instabilities within the infalling envelope when it is entirely supercritical in terms of its angular momentum distribution . here
, we now show the results of a new set of calculations performed in 3d using smoothed particle hydrodynamics ( sph ) with the code gadget-2 @xcite , of the collapse of rotating polytropic envelopes onto a bh .
these simulations explore the importance of qualitative changes in the radial angular momentum distribution @xmath0 , as has been explored in 2d collapsar models by @xcite .
as we shall see , strong asymmetries develop in the accretion flow , leading to important changes in the disk s response to the collapse of subcritical angular momentum material , and in the release of gravitational energy , when compared to the 2d hydrodynamical simulations .
this paper is structured as follows . in section 2
, we describe the initial conditions and outline the input physics included in our simulations , the results are presented and briefly discussed in section 3 , and section 4 includes the analysis and conclusions from this work .
we constructed polytropic envelopes as in bl14 , of mass @xmath8 , extending up to a radius @xmath9 , and with a dynamical time scale @xmath10 s , where @xmath11 accounts for the total mass of the system , which includes the @xmath12 bh fixed at the center of the distribution ( please refer to bl14 for further details ) . the angular momentum given to the envelopes
is initially inspired by the grb progenitors obtained by @xcite , in particular model 16 tb , in which @xmath0 shows drastic falls at different radii ( see top panel from fig .
[ fig : jdist ] ) . such angular momentum distribution ( blue line ) has a supercritical innermost shell with mass @xmath13 extending up to a radius @xmath14 ( leftmost gray region at the top panel from fig .
[ fig : jdist ] ) , where the angular momentum distribution of the envelope decreases drastically below @xmath15 ( red line ) , resulting in a shell extending from @xmath14 to @xmath16 , with mass @xmath17 and subcritical angular momentum ( red shaded region ) . moving outwards from @xmath14 to @xmath16 , the angular momentum of the shell increases , up to the point where it becomes supercritical at @xmath16 .
this supercritical outer shell of mass @xmath18 extends from @xmath16 to @xmath19 ( rightmost gray shaded region in the top panel from fig .
[ fig : jdist ] ) . in order to reproduce and parametrize the features
observed in the angular momentum of model 16 tb , we assigned to all our envelopes a rigid body angular momentum distribution @xmath20 , determined by a piecewise constant angular velocity @xmath21 and the position on the envelope @xmath22 , such that : @xmath23 where , @xmath24 is the distance to the origin , @xmath25 the azimuth angle ( playing no role on the rigid body rotation ) , @xmath26 is the polar angle ( where @xmath27 and @xmath28 correspond to the north pole , equator and south pole respectively ) and @xmath29 is a parameter that will be used to produce a drop in the angular momentum distribution at a radius @xmath14 .
we chose an angular velocity of @xmath30 rad / s with specific angular momentum @xmath31 , to guarantee the formation of an accretion disk from the collapse of the innermost material at @xmath32 , where the critical angular momentum is : @xmath33 assuming that the inner mass @xmath34 will eventually be accreted by the bh .
the angular momentum distribution obtained for such angular velocity @xmath21 can be seen in the leftmost bottom panel from figure [ fig : jdist ] , showing the angular momentum @xmath35 given to a sample of the sph particles in the envelope .
the red line showed in such panel corresponds to the critical angular momentum ( [ eqjc ] ) , and each colored point corresponds to an sph particle located at @xmath22 . at each radius @xmath24 , as the location of the sph particle departs from the equator at @xmath36 , its angular momentum @xmath37 decreases from its maximum value . according to their ratio @xmath38
, each sph particle is colored from red to blue as @xmath39 increases from 0 to 1 .
all particles with @xmath40 ( supercritical ) are shown as dark blue points on top of the red line defined by @xmath41 .
( blue line ) and critical angular momentum @xmath42 ( red line ) in the equatorial plane as a function of the inner mass @xmath34 , for model 16 tb from @xcite .
the angular momentum distribution is scaled by a factor @xmath43 .
gray shaded regions have @xmath2 and masses @xmath13 and @xmath18 respectively , and the red shaded region starting at @xmath14 has @xmath3 and mass @xmath17 .
bottom : angular momentum distribution from eq .
( [ eq : jr ] ) assigned to the sph particles for three different scaling factors @xmath44 and subcritical radius @xmath45 ( from left to right respectively ) .
the accretion factor is multiplied to eq .
( [ eq : jr ] ) for @xmath46
. the thick red line corresponds to the critical angular momentum @xmath15 and the sph particles are colored according to its ratio @xmath39.,scaledwidth=49.0% ] once @xmath21 was set , we defined the subcritical radius @xmath14 , at which the original angular momentum distribution @xmath37 is scaled by a factor @xmath29 , driving the angular momentum of all material within @xmath47 to subcritical values .
this is shown on the middle and rightmost bottom panels from figure [ fig : jdist ] . for a given scaling factor @xmath29 ,
the position of @xmath14 determines the mass of both the supercritical ( at @xmath32 ) and the subcritical ( at @xmath48 ) angular momentum shells with masses @xmath13 and @xmath17 respectively , resulting in increasing mass ratios @xmath49 for smaller @xmath14 . on the other hand
, the scale factor @xmath29 determines the depth of the fall in the angular momentum distribution at @xmath14 , and more importantly , it determines the position of the radius @xmath16 at which @xmath20 becomes supercritical again and the amount of supercritical material in the outermost shell @xmath18 .
the introduction of the supercritical radius @xmath14 and the scaling factor @xmath29 , thus allows for a parametrization in terms of a varying mass , and angular momentum content of the subcritical infalling shell .
table [ table2 ] shows the values used for the radius @xmath14 and the scaling factor @xmath29 in our simulations .
as can be seen in the table , the mass ratio @xmath49 depends on both @xmath14 and @xmath29 , as do the fractional masses of the supercritical shells ( @xmath50 and @xmath51 respectively ) scaled by the total mass on the supercritical shells @xmath52 .
simulations with a scale factor @xmath53 have an outer supercritical shell that roughly accounts for @xmath54 of the total supercritical mass , meanwhile in simulations with a scale factor @xmath55 , it only accounts for @xmath56 of the supercritical mass , providing little supercritical material to replenish the accretion disk .
.subcritical radius @xmath14 and scaling factor @xmath29 used in the simulations .
the subcritical and supercritical envelope s masses ( @xmath17 , @xmath13 and @xmath18 respectively ) are determined by @xmath14 and @xmath29 . [ cols="^,^,^,^,^",options="header " , ] due to the low angular momentum regime explored in this work , most of the material in our simulations is accreted quasi - radially in a few dynamical time scales , decreasing the resolution given by the number of sph particles .
we addressed this issue by increasing the number of sph particles to @xmath57 in order to ensure that after the collapse of the entire envelope , the accretion disk had a similar resolution to the one used in bl14 ( @xmath58 ) .
as in bl14 , we made use of a paczynski - wiita ( pw ) potential for the bh @xcite and defined the accretion radius @xmath59 as the position of the innermost stable circular orbit ( isco ) for a schwarzschild bh @xmath60 .
the bh mass increases as it accretes sph particles , and thus @xmath61 and the accretion radius increase with time , modifying the pw potential and increasing the critical angular momentum @xmath15 needed to form an accretion disk . in the scenario we are exploring ,
the accretion rates are dominated by the accretion of subcritical material .
this would produce only a small increase in the bh s spin parameter @xmath62 , that changes the location of @xmath63 from @xmath64 to @xmath65 as @xmath66 increases from 0 to 1 .
therefore , as long as @xmath67 , as is our case , a non rotating bh and a rotating kerr bh would behave very similarly , and the dynamics of infalling material should be the same . for simplicity , we made use of the original eos from the code gadget-2 which makes use of an ideal gas eos , in terms of an entropic function @xmath68 , where @xmath69 is the adiabatic index .
we adopted the same cooling prescription as in bl14 , defined by the internal energy @xmath70 of the material and a characteristic cooling time scale @xmath71 , considering this to originate from neutrino emission .
we used a single cooling time scale of @xmath72 s for our simulations , which guarantees the production of enough energy to power a grb and is also able to induce the production of spiral structure at the accretion disk ( bl14 ) .
since our cooling implementation has a soft dependence on the internal energy of the envelope ( @xmath73 ) , variations in the cooling rate @xmath74 will not be as drastic as the ones that should be obtained for a more realistic neutrino cooling .
nevertheless , this cooling gives the right order of magnitude for the neutrino luminosity and will track intense variations in the accretion flow as shown in bl14 .
as the innermost supercritical shell reaches its centrifugal barrier around the bh , an accretion disk forms after @xmath75 s. this dense and hot material accumulates near the bh forming an outgoing shock that will slow down the infall of material at larger radii . as this shock evolves , intense asymmetries are formed ( as shown in figure [ fig : rhovr_xy ] ) , which contain high internal energy , low density material with important amounts of subcritical material .
this indicates that they have formed , at least partly , from shocked low angular momentum material that got close to the bh but was not accreted .
figure [ fig : rhovr_xy ] shows density cross sections at the xy plane ( left panels ) , and the angular momentum ratio @xmath39 ( right panels ) for sph particles located in a disk of scale height @xmath76 . the snapshots on this figure correspond to the simulation with the largest mass ratio ( @xmath77 ) .
sph particles with @xmath78 and @xmath79 are colored yellow and red respectively , while particles with @xmath80 and @xmath81 are colored purple and blue respectively . since the mass of the bh and the infalling envelope are comparable , we calculated the critical angular momentum @xmath41 according to equation ( [ eqjc ] ) , considering that all interior mass @xmath34 ( bh , and sph particles ) contributes to determine if material at a radius @xmath24 has enough angular momentum @xmath20 to stay in a stable orbit around a bh with mass @xmath34 .
the snapshots shown on figure [ fig : rhovr_xy ] correspond to three different times : the beginning of the collapse of the subcritical shell , the end of the collapse of the subcritical shell , and during the collapse of the outermost supercritical shell ( top , middle and bottom panels with @xmath82 and @xmath83 s , respectively ) . as figure [ fig : rhovr_xy ] shows
, the dense accretion disk with @xmath84 ( blue to black colored region on the density plots ) is composed mostly of red and purple colored particles , which correspond to barely subcritical material ( @xmath85 ) and supercritical material ( @xmath40 ) respectively .
both of these components are distributed anisotropically throughout the accretion disk , and the initial distribution is determined by a spiral pattern formed from the outgoing shock .
the supercritical material distributed initially along this spiral pattern , gradually mixes with the subcritical material as shown in the middle right panel of figure [ fig : rhovr_xy ] .
besides the mixing between subcritical and supercritical material observed in the spiral structures at the accretion disk s plane , there is also an important mixing of supercritical and subcritical material going on at different polar angles .
this can be seen in figure [ fig : rhovr_xz ] where we show the density cross sections and the sph particles ratio @xmath39 along the xz plane .
the densest part of the accretion disk ( @xmath86 ) is composed mostly by a combination of barely subcritical and supercritical material with @xmath87 ( red and purple colored respectively ) , with an important fraction of the barely subcritical material located above and below the accretion disk .
this barely subcritical material surrounding the dense accretion disk will be able to fall onto the bh , without directly interacting with the supercritical material at smaller latitudes .
the degree of mixing observed between subcritical and supercritical material at the accretion disk is only possible in 3d , since previous 2d calculations do not account for the possible mixing at the accretion disk s plane .
this general morphological behavior was observed in all of our simulations , even in the one with no subcritical shell .
however , by the end of the collapse of the polytropic envelope , the simulations with scaling factor @xmath55 had lost about @xmath88 of the supercritical material contained in the innermost shell with mass @xmath13 , this will be addressed in detail in the following sections . of sph particles located in a disk of scale height @xmath76 , for the simulation with mass ratio @xmath89 .
the snapshots correspond to the end of the collapse of the inner supercritical shell at @xmath90 s , the end of the collapse of the subcritical shell at @xmath91 s , and after the collapse of the outer supercritical shell at @xmath83 s ( from top to bottom respectively ) .
the densest regions of the accretion disk ( blue region with @xmath92 in the left panels ) are composed of an anisotropic mixture of subcritical ( yellow and red ) and supercritical material ( purple and blue ) .
the density is scaled to @xmath93 and the boxes cover from @xmath94 $ ] .
left panel figures made with splash @xcite.,scaledwidth=99.5% ] of sph particles located in a disk of scale height @xmath76 , for the simulation with mass ratio @xmath89 .
the snapshots correspond to the end of the collapse of the inner supercritical shell at @xmath90 s , the end of the collapse of the subcritical shell at @xmath91 s , and after the collapse of the outer supercritical shell at @xmath83 s ( from top to bottom respectively ) .
the densest regions of the accretion disk ( blue region with @xmath92 in the left panels ) are composed of an anisotropic mixture of subcritical ( yellow and red ) and supercritical material ( purple and blue ) .
the density is scaled to @xmath93 and the boxes cover from @xmath94 $ ] .
left panel figures made with splash @xcite.,scaledwidth=99.5% ] but at the xz plane.,scaledwidth=99.5% ] but at the xz plane.,scaledwidth=99.5% ] in order to study more qualitatively how the accretion disk is affected by the infall of low angular momentum material , we followed the evolution of the mass contained in the accretion disk , which has to be properly defined from the properties of the sph particles .
material in a rotationally supported accretion disk , should complete several orbits around the bh , before being accreted .
however , if low angular momentum material falls on top of such an accretion disk at a radius @xmath24 , it will reach the bh in a time @xmath95 , determined by its radial velocity @xmath96 , and the distribution of material at the accretion disk , preventing its direct collapse .
thus , as long as this low angular momentum material is able to spend enough time at the accretion disk before reaching the bh , it will form part of the accretion disk and contribute significantly to the neutrino cooling process and production of energy .
material at the accretion disk will be accreted in a viscous time scale @xmath97 , determined by the disk s radius @xmath98 and the viscosity @xmath99 .
the viscosity is defined by the @xmath100 parameter @xcite , the local sound speed @xmath101 and the disk s height @xmath102 . however , instead of implementing an @xmath100 viscosity in our simulations , we relied on the intrinsic sph artificial viscosity implemented in gadget-2 to mimic naturally occurring dissipative processes , which can be accounted as a rough approximate to an @xmath100 prescription @xcite .
however , the viscosity wo nt be determinant in the evolution of the accretion disk , since the viscous time scale @xmath103 of our simulations ( ranging between @xmath104 s for a characteristic @xmath105 ) , is at least an order of magnitude larger than the other relevant time scales ( @xmath106 s , @xmath107 s and the total duration of the simulations @xmath108 ) . to determine if material forms part of the accretion disk we will compare the time it will take it to complete a fraction of an orbit around the bh ( @xmath109 , where @xmath110 is the angular velocity around the rotation axis ) , with the time it would take it to reach the bh with its current radial velocity ( @xmath111 ) . if @xmath112 then material will spend a significant amount of time at the accretion disk before reaching the bh .
with this definition of accretion disk , material collapsing from large latitudes ( close to the poles ) will mostly have @xmath113 and @xmath114 , and wo nt be considered as part of the accretion disk , meanwhile material close to the equator , will have @xmath115 and @xmath112 and will form part of the accretion disk . in order to discard uncollapsed material with @xmath116 and @xmath112 as part of the accretion disk , we reduced the area of interest where the accretion disk will reside , to a sphere of radius @xmath117 which covers up to @xmath118 for a @xmath119 schwarzschild bh .
this sphere of radius @xmath120 completely covers the region observed in figures [ fig : rhovr_xy ] and [ fig : rhovr_xz ] and guarantees that the outermost uncollapsed material will not be accounted as part of the accretion disk .
figure [ mdisk ] shows the disk s mass @xmath121 ( in solar masses ) obtained for two simulations with @xmath53 and 0.5 and the same @xmath122 , using two different values of the parameter @xmath123 and @xmath124 ( top and bottom panel respectively ) which determines the fraction of the orbit around the bh that needs to be completed to form part of the accretion disk .
as can be seen , by increasing @xmath125 there will be fewer material that will have large enough orbital velocity to complete a fraction @xmath125 of the orbit , however , the qualitative features of the evolution of the disk s mass remains the same .
the lines in the figure are colored according to the average cylindrical radius @xmath126/r_{\rm acc}$ ] of the @xmath127 subcritical particles contained at the disk .
this radius will indicate how far from the bh the subcritical mass is concentrated .
the light grey shaded area roughly covers the time during which the subcritical shell is collapsing on top of the sphere with radius @xmath120 ( from @xmath128 s ) , and the dark shaded gray region covers the extended collapse observed in simulations with @xmath55 and a larger subcritical shell . since the value adopted for @xmath125 does not seem to alter the qualitative evolution of @xmath121 , we will use @xmath129 through the rest of our analysis , which considers that a fraction of subcritical material will form part of the accretion disk , as long as it does not violently collapses towards the bh . and @xmath124 ( top and bottom panels respectively ) for two simulations with the same subcritical radius @xmath130 and different scaling factor @xmath29 .
the lines are colored according to the average radius @xmath131 of the subcritical component , and all material at @xmath132 is assigned a blue color .
the gray shaded region indicates the times during which the subcritical shell collapses , with the darkest gray region corresponding to the extended collapse of the @xmath55 simulation.,scaledwidth=49.0% ] the information provided by figure [ mdisk ] shows that for a given subcritical radius @xmath14 , the scale factor @xmath29 changes dramatically the evolution of @xmath121 . in both simulations , material becomes part of the accretion disk after @xmath133 s , and before the collapse of the subcritical shell ( grey shaded region ) , an important fraction of the mass of the accretion disk is located near the bh ( indicated by the red and orange color of the lines ) , a region occupied mostly by subcritical material ( yellow and red colored particles shown in right panels from figures [ fig : rhovr_xy ] and [ fig : rhovr_xz ] ) . during the collapse of the subcritical shell , the average radius of subcritical material @xmath131 moves away from the bh , indicating that a significant amount of supercritical material is reaching the outer parts of the disk .
however , during the collapse of the subcritical shell , @xmath131 gradually decreases , indicating that subcritical material is moving towards the bh .
both of these simulations have the same innermost supercritical shell with mass @xmath13 extending to the subcritical radius @xmath14 , followed by the subcritical shell with mass @xmath17 whose mass and extension is determined by the scaling factor @xmath29 .
the simulation with @xmath55 has a more massive subcritical shell with a mass ratio of @xmath134 compared to the ratio @xmath135 for the simulation with @xmath53 .
the simulation with @xmath55 also has a less massive outermost supercritical shell that will replenish the accretion disk with supercritical material , which can be observed in the smaller @xmath121 after the collapse of the subcritical shell .
the evolution of @xmath121 for all simulations with a subcritical shell shown in figure [ mdiskall ] , gives important information about the importance of the parameters @xmath14 and @xmath29 in determining the evolution of the accretion disk .
all simulations show a continuous increase in @xmath121 until the collapse of the subcritical shell begins . during the collapse of the subcritical shells ( grey shaded area ) ,
all simulations either maintain or increase @xmath121 , and it is towards the end of the collapse of the subcritical material where the more substantial changes in @xmath121 occur .
simulations with @xmath53 end the collapse of the subcritical shell before reaching the darkest grey shaded region , at that point , the outermost supercritical shell with mass @xmath18 will start replenishing the accretion disk .
this is shown as an steep increase in @xmath121 starting before @xmath136 s for all simulations with @xmath53 .
on the other hand , simulations with @xmath55 show an intense decrease in @xmath121 once the subcritical shell has completely collapsed , which should translate in an intense decrease in the disk s energy production . of the subcritical component .
the gray shaded region indicates the times during which the subcritical shell collapses , with the darkest gray region corresponding to the extended collapse of the @xmath55 simulations.,scaledwidth=49.0% ] figure [ mdiskall ] shows that increasing the mass @xmath13 contained in the innermost supercritical shell will translate into a more massive accretion disk .
however , this does not mean that the subcritical material is not significantly contributing to @xmath121 .
this contribution can be seen in figure [ mdisk_comp ] , which shows the disk s mass @xmath121 separated by its subcritical and supercritical components ( top and bottom panels respectively ) for four different simulations . as can be seen , the amount of subcritical and supercritical material at the accretion disk is comparable throughout the collapse of the subcritical shell . in figure [ mdisk_comp ]
the lines are colored according to the average cylindrical radius of the subcritical component @xmath131 and the supercritical component @xmath137 respectively .
clearly the scaling factor @xmath29 is determining the most important features in the evolution of @xmath121 and the accretion flow , since it regulates not only the amount of subcritical mass that will fall onto the disk but also how far below is the material s angular momentum from the critical value @xmath15 . decreasing
the angular momentum of the subcritical material will translate into a more violent collapse , which could significantly change the amount of time that subcritical material spends at the accretion disk .
what is interesting to notice is that all simulations , retained a large fraction of the supercritical material during the collapse of the subcritical shell , which means that the previously formed accretion disk was not dragged towards the bh by the subcritical shell , even in the most extreme mass ratio @xmath138 in the model with @xmath53 and @xmath139 . , 2.302 , 3.755 and 5.247 ( labeled in the figure by its corresponding @xmath29 and @xmath14 ) .
the color of the lines indicates the average cylindrical radius of the subcritical component @xmath131 and the supercritical component @xmath137 respectively.,scaledwidth=49.0% ] the bh accretion rate can be written in terms of the derivatives of the supercritical and subcritical angular momentum mass components ( @xmath140 and @xmath141 respectively ) as : @xmath142 to compare the bh accretion rate and the supercritical and subcritical accretion rates , we will treat the latter as if they had opposite sign .
thus , a positive value for any of them will imply that such component is transferring mass to the bh and possibly to the other component . if there is mass transfer between the low and high angular momentum components , the one gaining mass will have a negative accretion rate , while the other will have an accretion rate @xmath143 .
figure [ mdotlum ] shows the accretion and the energy loss rates ( top and bottom panel respectively ) for three simulations with scale factors @xmath144 and @xmath124 ( black , red and blue lines respectively ) .
the solid lines represent the bh accretion rate @xmath145 and the energy loss rate @xmath146 .
the dashed and dotted lines stand for the contribution from the subcritical and the supercritical components respectively .
all simulations with the same scaling factor @xmath29 showed a similar evolution for @xmath147 and @xmath74 and since the simulations shown have similar mass ratio @xmath148 , we can conclude that the mass ratio between the supercritical and subcritical shells is not the most important parameter determining the evolution of the accretion flow . as we saw from figures [ mdiskall ] , [ mdisk_comp ] and [ mdotlum ] ,
the scaling factor is actually the main driver of the properties of the accretion flow , thus we will analyze simulations with different @xmath29 separately .
( top panel ) and energy loss rate @xmath149 ( bottom panel ) for simulations with @xmath144 and 0.5 ( black , red and blue lines respectively ) .
the dashed and dotted lines stand for the contribution from the subcritical and the supercritical components respectively , and the solid line corresponds to the sum of both components . both @xmath147 and @xmath74 are scaled to system units , which can be obtained from the system properties.,scaledwidth=48.0% ] these simulations are characterized by having a moderate fall in the angular momentum distribution and a massive outermost shell with supercritical angular momentum , which will replenish the initially formed accretion disk at @xmath150 s. they show episodes of negative @xmath140 during the collapse of the subcritical shell , which could be due to angular momentum transport induced by the spiral structures observed in the first two panels from figure [ fig : rhovr_xy ] ( @xmath151 ) .
this episodes of negative @xmath140 translate into small increases in the supercritical mass @xmath152 shown on figure [ mdisk_comp ] , and they only appear during the collapse of the subcritical shell , where the disk s azimuthal asymmetry is more noticeable . in order to understand the evolution of the accretion and energy loss rates shown in figure [ mdotlum ] we must return to figure [ mdisk_comp ] . in such figure , it is important to notice that subcritical material is not immediately accreted by the bh as it becomes part of the disk .
subcritical material starts to build up at large radii , increasing the disk s subcritical mass component @xmath17 .
this material slowly moves towards the bh , reducing its average radius @xmath153 , and wo nt be accreted until reaching @xmath154 .
only when subcritical material reaches an average radius @xmath155 , a considerable fraction of it will be close enough to the accretion radius to be rapidly accreted in a quasi - radial fashion . during the initial collapse of the spherical envelope
, subcritical material reaches the innermost part of the accretion disk before the supercritical material does . in consequence ,
most of the supercritical material does not reach its centrifugal barrier , and is located along with subcritical material , safely away from the bh at the outer parts of the accretion disk ( as seen in figures [ fig : rhovr_xy ] and [ fig : rhovr_xz ] ) .
figure [ mdisk_comp ] shows that the amount of subcritical material at the accretion disk continues to increase far beyond the end of the collapse of the subcritical shell , since the outermost supercritical shell is still feeding the accretion disk with subcritical material from above the equator . before @xmath156 s ,
the subcritical material becomes the dominant component of the accretion disk , and its average radius @xmath153 approaches the accretion radius .
it is when @xmath155 that the amount of subcritical material starts decreasing due to its rapid accretion onto the bh .
this rapid decrease in @xmath17 reduces the amount of hot material contributing to the energy loss rate and is responsible for the steep decrease in the energy loss rate @xmath157 shown on figure [ mdotlum ] ( red line ) around @xmath158 s. these simulations have a deeper drop in the angular momentum distribution at @xmath14 , compared to simulations with @xmath53 , which further reduces the angular momentum of the subcritical material .
they also have a small outermost supercritical layer , which contains @xmath159 of the supercritical material of the infalling envelope .
the later means that there wo nt be much supercritical material to replenish the accretion disk after the collapse of the subcritical shell .
as subcritical material reaches the outer parts of the accretion disk , it slowly moves towards the bh until reaching an average radius @xmath160 at the end of the collapse of the subcritical shell ( dark shaded grey area on figure [ mdisk_comp ] ) .
since the amount of angular momentum of this subcritical material is considerably lower than the angular momentum of subcritical material from simulations with @xmath53 , from this point on , subcritical material gets rapidly accreted , producing a violent decrease in the subcritical mass @xmath17 and the total mass of the accretion disk ( see figure [ mdiskall ] and top panel of figure [ mdisk_comp ] at @xmath161 s ) .
the intense decrease in the accretion disk s mass produced by the rapid accretion of subcritical material , can be seen as the steep increase in @xmath147 shown in figure [ mdotlum ] around @xmath162 s , and it is also visible as the intense decrease in the energy loss rate @xmath157 reaching @xmath163 of its original value in @xmath164 s. since a significant amount of subcritical material was rapidly accreted , the value of the critical angular momentum @xmath165 , also increased very rapidly , reducing the amount of supercritical material .
after this intense accretion event , there is only very little material to replenish the accretion disk , explaining the significantly reduced accretion and energy loss rates at later times .
the 2d hydrodynamical simulations done by @xcite , found that the collapse of subcritical shells with mass ratios @xmath166 were able to completely disrupt the accretion disk transforming the accretion flow into a quasi radial inflow with low neutrino cooling efficiency .
this result was obtained on both their simulations with subcritical shells with @xmath167 and @xmath168 .
in contrast , our 3d hydrodynamical simulations showed that increasing the mass ratio did not necessarily translated in the destruction of the accretion disk by the subcritical shell , since the simulation with the largest mass ratio @xmath169 managed to endure the collapse of such a massive subcritical shell .
instead , simulations with subcritical shells with less angular momentum support ( smaller scaling factor @xmath29 ) were the ones that showed an overall decrease in the accretion disk s total and supercritical mass ( @xmath121 and @xmath152 respectively ) .
the amount of angular momentum contained in the subcritical shell , determined by @xmath55 and @xmath170 , regulates the evolution of @xmath152 during the collapse of the subcritical shell . in simulations with smaller scaling factor @xmath29 ,
the amount of supercritical material at the accretion disk @xmath152 decreased , contrary to the slight increase observed for simulations with @xmath53 .
this increase in @xmath152 must come from barely subcritical material ( @xmath167 ) at the outer parts of the accretion disk , obtaining angular momentum from innermost material , by means of angular momentum transport mediated by the non - axisymmetic instabilities observed in figure [ fig : rhovr_xy ] , or also from fresh supercritical material being fed to the accretion disk region ( @xmath171 ) where @xmath152 was calculated . however , to explain the decrease observed in @xmath152 for simulations with @xmath55 we have to consider the following factors : * there is very few barely subcritical material ( @xmath167 ) able to gain angular momentum and increase @xmath152 .
simulations with @xmath53 had mostly barely subcritical material at the equator , which seems to have helped to increase and maintain the amount of supercritical material during the collapse of the subcritical shell .
* since the bh is growing at a fast rate from accreting extremely subcritical material , some of the initially supercritical material will become subcritical due to the increase in @xmath15 .
the growing bh mass increases the accretion radius , and the gravitational potential from the bh , reducing the supercritical material s circularization radius and thus , its mean radius @xmath172 .
* finally , it is possible that part of the decrease in @xmath152 is due to the accretion of supercritical material by the bh , but it is only likely to be important when the average radius @xmath172 approaches @xmath173 , which only happens by the end of the collapse of the subcritical shell , thus , the fraction of supercritical material that approached the accretion radius @xmath59 should be small .
results from figures [ mdiskall ] and [ mdisk_comp ] show that a large fraction of the supercritical material survived during most of the collapse of the subcritical shell with mass @xmath17 . however , if there is no `` fresh '' supercritical material being supplied to the accretion disk ,
the formerly supercritical material will become subcritical as the bh s mass and @xmath15 increase from the accretion of subcritical material .
this effect is visible on figure [ mdisk_comp ] for the simulations with @xmath55 , where @xmath152 with average radius @xmath174 reduces drastically after the rapid accretion of subcritical material observed at @xmath175 s. we should also notice that the amount of subcritical material at the accretion disk and the time it takes for it to be accreted , increases with the amount of supercritical material contained in the innermost supercritical shell .
this supercritical material prevents the direct collapse of subcritical material , allowing it to contribute to the energy loss rate before being accreted .
this can be seen in figure [ mdisk_comp ] where an increasing mass ratio @xmath49 ( determined by a decrease in @xmath14 ) , turns into a small maximum mass @xmath17 at the accretion disk , and into an earlier decrease in @xmath17 after the collapse of the subcritical shell . as shown in figure [ mdotlum ] ,
the evolution of the accretion flow @xmath147 and the energy cooling rate @xmath146 , was mostly determined by the scale factor @xmath29 used in the simulation .
this parameter is also critical in determining the cooling efficiency of the system @xmath176 $ ] , which tells us how efficiently are we extracting energy from the accretion flow . figure [ cooleff ] shows the cooling efficiency @xmath177 for all of our simulations . as expected , the simulation with no subcritical shell is the one with the highest cooling efficiency ( solid line ) , retaining a large amount of material close to the bh and contributing to the energy production for a long time .
it is clear that the cooling efficiency is more sensitive to the scale factor @xmath29 than to the mass ratio @xmath49 , decreasing for smaller scale factors and increasing for smaller mass ratios @xmath49 .
evolution as a function of time .
same color lines correspond to simulations with the same radius @xmath14 and each line corresponds to a different initial mass ratio @xmath49 from table [ table2 ] as indicated on the figure.,width=326 ] simulations with @xmath55 , i.e. with the smaller amount of angular momentum , were the ones with the lowest cooling efficiencies .
the extremely low cooling efficiency of simulations with @xmath55 arises due to a rapid increase in the accretion rate followed by a decrease in the energy loss rate @xmath157 ( see figure [ mdotlum ] ) . both caused by the rapid accretion of subcritical material at @xmath175 s shown in the top panel of figure [ mdisk_comp ] .
as figure [ cooleff ] shows , by @xmath178 s all simulations with @xmath179 have very similar cooling efficiencies .
this means that they have all reached a very similar ratio between the accretion and the energy loss rates .
nevertheless , in simulations with @xmath55 , the energy loss rate is considerably reduced by the end of the simulation , due to the small amount of supercritical material that was supplied throughout the collapse .
figure [ rhodens ] shows the evolution of the mean density @xmath180 of the @xmath181 sph particles contained at the accretion disk within @xmath182 , for the simulations shown in figure [ mdotlum ] .
simulations with @xmath55 show a steep decrease in the mean density @xmath183 which coincides with the rapid accretion of subcritical material around @xmath162 s. since our energy loss rate @xmath157 is proportional to the internal energy @xmath70 , and the internal energy is also proportional to the density , the overall energy loss rate decreases accordingly with the density , reaching the lowest values for simulations with @xmath55 .
this intense decrease in @xmath183 would also decrease the total power extractable from the rotation of the bh and the accretion disk , since both the blandford - znajek mechanism , and the release of energy through magnetic torques at the rotating accretion disk , depend on the strength of the large scale magnetic field @xmath184 trapped in the vicinity of the bh , determined by the density of the accretion disk @xcite .
therefore , the question one should ask in order to know if the infall of a subcritical shell will be able to shutdown the inner engine is the following : is the mass of the subcritical shell large enough to increase @xmath15 beyond the angular momentum of the innermost supercritical material ?
if it is nt , the accretion disk is very likely to survive the collapse of the subcritical shell and still maintain a hot and dense disk structure .
however , if the subcritical shell is massive enough to increase @xmath15 beyond the innermost supercritical shell s angular momentum , the survival of a hot and dense accretion disk depends on whether the outermost supercritical shell replenishes the accretion disk before all subcritical material gets accreted .
thus , in order to increase the chances of the survival of a hot and dense accretion disk around the bh after the collapse of a really massive subcritical shell , the infalling envelope should meet the following conditions : * the innermost supercritical shell should form a massive accretion disk .
increasing the mass will help to delay the accretion of subcritical material , allowing outer supercritical material to reach the accretion disk before the bulk of the subcritical material gets accreted . *
have a subcritical shell with barely subcritical material ( @xmath185 ) feeding the accretion disk .
simulations with such envelopes ( @xmath53 ) showed a longer lasting accretion disk structure which allowed for the arrival of the outermost supercritical shell before the subcritical material at the accretion disk was completely accreted .
* a massive outer supercritical shell must feed the disk before the bulk of the subcritical material contained at the accretion disk is accreted by the bh .
otherwise the accretion of subcritical material will increase @xmath15 beyond the angular momentum of material at the accretion disk .
this relatively simple criteria could be used to assess whether a lgrb progenitor with an angular momentum distribution like the ones explored in this work , will be able to show an inner engine shutdown due to the collapse of a subcritical shell on top of the accretion disk . at radii @xmath186 , for the simulations shown on figure [ mdotlum ] .
the huge drop in @xmath183 for the simulation with @xmath55 ( square points ) shown around @xmath187 s , corresponds to the drop in @xmath157 shown on figure [ mdotlum ] .
the color indicates the total mass ( in solar masses ) contained in the accretion disk defined in the results section.,scaledwidth=49.0% ] our 3d hydrodynamical simulations with angular momentum distributions parametrized by the mass ratio @xmath49 and the scale factor @xmath29 , allowed us to study how the energy production is affected by having subcritical angular momentum material falling onto a collapsar accretion disk .
this is a plausible scenario observed in lgrb progenitors ( @xcite ) , however due to the narrow range of parameters explored , one should take care when extrapolating our results to different progenitors . here
we summarize the most important conclusions drawn from our results : * the intrinsic 3d nature of the collapsar scenario and the anisotropies developed during the collapse , prevents the accretion disk from being dragged into the bh by the infall of subcritical angular momentum material .
the added degree of freedom along azimuth in the orbital plane combines with vertical motions ( already allowed in 2d calculations ) to allow for greater mixing of high and low - angular momentum material and distinct flows in the equatorial plane which greatly diminish the impact of infalling material onto a pre - existing disk , and allow for its survival . * the collapse of progenitors with transitions between supercritical and subcritical values in the angular momentum distribution ( as in fig [ fig : jdist ] ) , can lead to the production of large quiescent times originating from the absence of a hot accretion structure near the bh .
however , the formation of such quiescent times is more difficult than previously thought , since it does not only depend on the mass ratio between the infalling subcritical shell and the innermost supercritical shell @xmath49 , but more importantly on the amount of angular momentum contained in the subcritical material , which determines if it will be accreted before fresh supercritical material is supplied to the disk . *
the collapse of extremely subcritical material onto the accretion disk will result in a shutdown of the inner engine if the critical angular momentum @xmath15 increases beyond the angular momentum of material at the accretion disk , before it is resupplied with an important amount of supercritical material .
we thank enrico ramirez - ruiz for useful discussion and feedback concerning this work . the author thanks the kind hospitality of the harvard smithsonian center for astrophysics , where this work was initiated .
these simulations where computed using the cluster atocatl at ia - unam , the cluster diable at icn - unam and the supercomputer hyades at ucsc .
we gratefully acknowledge financial support from dgapa - unam ( ig100414 ) and conacyt ( 101958 ) .
ab is supported by a uc - mexus posdoctoral fellowship . | for the collapsar scenario to be effective in the production of gamma ray bursts , the infalling star s angular momentum @xmath0 must be larger than the critical angular momentum needed to form an accretion disk around a blackhole ( bh ) , namely @xmath1 for a schwarzschild bh . by means of 3d sph simulations , here we study the collapse and accretion onto black holes of spherical rotating envelopes , whose angular momentum distribution has transitions between supercritical ( @xmath2 ) and subcritical ( @xmath3 ) values .
contrary to results obtained in previous 2d hydrodynamical simulations , we find that a substantial amount of subcritical material fed to the accretion disk , lingers around long enough to contribute significantly to the energy loss rate . increasing the amount of angular momentum in the subcritical material increases the time spent at the accretion disk , and only when the bulk of this subcritical material is accreted before it is replenished by a massive outermost supercritical shell , the inner engine experiences a shutdown .
once the muffled accretion disk is provided again with enough supercritical material , the shutdown will be over and a quiescent time in the long grb produced afterwards could be observed .
[ firstpage ] accretion , accretion disks @xmath4 hydrodynamics @xmath4 gamma - ray burst @xmath4 supernovae : general |
in this paper we aim to develop a general , model free , method for analyzing current status data using machine learning techniques .
in particular , we propose a support vector machine ( svm ) learning method for estimation of the failure time expectation for current status data .
svm was originally introduced by vapnik in the 1990 s and is firmly related to statistical learning theory @xcite .
the choice of svms for current status data is motivated by the fact that svms can be implemented easily , have fast training speed , produce decision functions that have a strong generalization ability and can guarantee convergence to the optimal solution , under some weak assumptions @xcite .
current status data is a data format where the failure time @xmath0 is restricted to knowledge of whether or not @xmath0 exceeds a random monitoring time @xmath1 .
this data format is quite common and includes examples from various fields .
@xcite mention a few examples including : studying the distribution of the age of a child at weaning given observation points ; when conducting a partner study of hiv infection over a number of clinic visits ; and when a tumor under investigation is occult and an animal is sacrificed at a certain time point in order to determine presence or absence of the tumor . for instance , in the last example of carcinogenicity testing , @xmath0 is the time from exposure to a carcinogen and until the presence of a tumor , and @xmath1 is the time point at which the animal is sacrificed in order to determine presence or absence of the tumor .
clearly , it is difficult to estimate the failure time distribution since we can not observe the failure time @xmath0 .
these examples illustrate the importance of this topic and the need to find advanced tools for analyzing such data .
we present a support vector machine framework for current status data .
we propose a learning method , denoted by csd - svm , for estimation of the failure time expectation .
we investigate the theoretical properties of the csd - svm , and in particular , prove consistency for a large family of probability measures . in order to estimate the conditional expectation we use a modified version of the quadratic loss . using the methodology of @xcite
, we construct a data dependent version of the quadratic loss .
since the failure time @xmath0 is not observed , our data dependent loss function is based on the censoring time @xmath1 and on the current status indicator .
finally , in order to obtain a csd - svm decision function for current status data , we minimize a regularized version of the empirical risk with respect to this data - dependent loss .
there are several approaches for analyzing current status data .
traditional methods include parametric models where the underlying distribution of the survival time is assumed to be known ( such as weibull , gamma , and other distributions with non - negative support ) .
other approaches include semiparametric models , such as the cox proportional hazard model , and the accelerated failure time ( aft ) model ( see , for example , * ? ? ?
in the cox model , the hazard function is assumed to be proportional to the exponent of a linear combination of the covariates . in the aft model
, the log of the failure time is assumed to be a linear function of the covariates .
several works including @xcite , @xcite and others have suggested the cox proportinal hazard model for current status data , where the cox model can be represented as a generalized linear model with a log - log link function .
other works including @xcite discussed the use of the aft model for current status data and suggested different algorithms for estimating the model parameters . needless to say that both parametric and semiparametric models demand stringent assumptions on the distribution of interest which can be restrictive .
for this reason , additional estimation methods are needed . over the past two decades ,
some learning algorithms for censored data have been proposed ( such as neural networks and splitting trees ) , but mostly with no theoretical justification .
additionally , most of these algorithms can not be applied to current status data but only to other , more common , censored data formats . recently , several works suggested the use of svms for survival data .
@xcite suggested the use of svms for survival analysis , and formulated the task as a ranking problem .
shortly after , @xcite suggested the use of svms for regression problems with censored data ; this was done by asymmetrically modifying the @xmath2-insensitive loss function .
both examples were empirically tested but lacked theoretical justification .
@xcite proposed an empirical quantile risk estimator , which can also be applied to right censoring , and studied the estimator s performance .
@xcite studied an svm framework for right censored data and proved that the algorithm converges to the optimal solution .
@xcite suggested two svm - based formulations for classification problems with survival data .
these examples illustrate that initial steps in this direction have already been taken . however , as far as we know , the only svm - based work that can also be applied to current status data is by @xcite which has a more computational and less theoretic nature .
the authors studied the use of svm for regression problems with interval censoring and , using simulations , showed that the method is comparable to other missing data tools and performs significantly well when the majority of the training data is censored .
the contribution of this work includes the development of an svm framework for current status data , the study of the theoretical properties of the csd - svm , and the development of new oracle inequalities for censored data .
these inequalities , together with finite sample bounds , allow us to prove consistency and to compute learning rates .
the paper is organized as follows . in section [ sec : preliminaries ] we describe the formal setting of current status data and discuss the choice of the quadratic loss for estimating the conditional expectation . in section [ sec : support - vector - machines ] we present the proposed csd - svm and its corresponding data - dependent loss function . section [ sec : theoretical - results ] contains the main theoretical results , including finite sample bounds , consistency proofs and learning rates . in section [ sec :
estimation - of - the ] we illustrate the estimation procedure and show that the solution has a closed form .
section [ sec : simulation - study ] contains the simulations .
concluding remarks are presented in section [ sec : concluding - remarks ] .
the lengthier proofs appear in appendix [ sec : appendix ] .
the matlab code for both the algorithm and for the simulations can be found in the [ sub : supplementary - material ] .
in this section we present the notations used throughout the paper .
first we describe the data setting and then we discuss briefly loss functions and risks .
assume that the data consists of @xmath3 i.i.d.random triplets @xmath4 .
the random vector @xmath5 is a vector of covariates that takes its values in a compact set @xmath6 .
the failure - time @xmath0 is non - negative , the random variable @xmath1 is the censoring time , the indicator @xmath7 is the current status indicator at time @xmath1 , and is contained in the interval @xmath8\equiv\mathcal{y}$ ] for some constant @xmath9 . for example , in carcinogenicity testing , an animal is sacrificed at a certain time point in order to determine presence or absence of the tumor . in this example
, @xmath0 is the time from exposure to a carcinogen and until the presence of a tumor , @xmath1 is the time point at which the animal is sacrificed , and @xmath10 is the current status indicator at time @xmath1 ( indicating whether the tumor has developed before the censoring time , or not ) .
we now move to discuss a few definitions of loss functions and risks , following @xcite .
let@xmath11 be a measurable space and @xmath12 be a closed subset .
then a loss function is any measurable function @xmath13 from @xmath14 to @xmath15 .
let @xmath16 be a loss function and @xmath17 be a probability measure on @xmath18 .
for a measurable function @xmath19 , the @xmath13-risk of @xmath20 is defined by @xmath21=\int_{z\times y}l\left(z , y , f(z)\right)dp(z , y)$ ] .
a function @xmath20 that achieves the minimum @xmath13-risk is called a _
bayes decision function _ and is denoted by @xmath22 , and the minimal @xmath13-risk is called the _ bayes risk _ and is denoted by @xmath23 .
finally , the empirical @xmath13-risk is defined by @xmath24 .
for example , it is well known ( see , for example , * ? ? ? * ) that the conditional expectation is the bayes decision function with respect to the quadratic loss .
let @xmath25 be a reproducing kernel hilbert space ( rkhs ) of functions from @xmath26 to @xmath27 , where an rkhs is a function space that can be characterized by some kernel function @xmath28 . by definition ,
if @xmath29 is a universal kernel , then @xmath25 is dense in the space of continuous functions on @xmath26 , @xmath30 ( see , for example , ( * ? ? ?
* definition 4.52 ) ) .
let us fix such an rkhs @xmath25 and denote its norm by @xmath31 and let @xmath32 be some sequence of regularization constants .
an svm decision function for uncensored data is defined by : @xmath33 we recall that current status data consists of @xmath3 independent and identically - distributed random triplets @xmath4 .
let @xmath34 and @xmath35 be the cumulative distribution functions of the failure time and censoring , respectively , given the covariates @xmath36 .
let @xmath37 be the density of @xmath35 . for current status data ,
we introduce the following identity between risks , following @xcite .
we extend this notion and incorporate loss functions and covariates in the following identity .
let @xmath38 be a loss function differentiable in the first variable .
let @xmath39 be the derivative of @xmath13 with respect to the first variable .
we would like to find the minimizer of @xmath40 over a set @xmath25 of functions @xmath20 .
note that @xmath41\\ = & e_{z}\left[\int_{0}^{\tau}\ell(t , f(z))(1-f(t|z))dt-\left.l(t , f(z))(1-f(t|z))\right|_{0}^{\tau}\right]\\ = & e_{z}\left[\int_{0}^{\tau}\ell(t , f(z))(1-f(t|z))dt\right]+e[l(0,f(z))]\,,\end{aligned}\ ] ] where the equality before last follows from integration by parts .
note also that @xmath42 and thus @xmath43= & e_{z , t}\left[e_{c}\left[\left.\frac{\mathbf{{1}}\{t > c\}\ell(c , f(z))}{g(c|z)}\right|z = z , t = t\right]\right]\\ = & e_{z , t}\left[\int_{0}^{\tau}\frac{\mathbf{{1}}\{t > c\}\ell(c , f(z))g(c|z)}{g(c|z)}dc\right]\\ = & e_{z , t}\left[\int_{0}^{\tau}\mathbf{{1}}\{t > c\}\ell(c
, f(z))dc\right]\\ = & e_{z}\left[\int_{0}^{\tau}\ell(c , f(z))\int_{0}^{\tau}\mathbf{{1}}\{t > c\}df(t|z)dc\right]\\ = & e_{z}\left[\int_{0}^{\tau}\ell(c , f(z))(1-f(c|z))dc\right]\,.\end{aligned}\ ] ] this shows that the risk can be represented as the sum of two terms @xmath44+e[l(0,f(z))]$ ] .
hence , in order to estimate the minimizer of @xmath40 , one can minimize a regularized version of the empirical risk with respect to the data - dependent loss function @xmath45 note that this function need not be convex nor a loss function . for the quadratic loss function ,
our data - dependent loss function becomes @xmath46 note that this function is convex but not necessarily a loss function since it can obtain negative values . in order to ensure positivity we add a constant term that does not depend on @xmath20 , and
so our loss becomes @xmath47 , where for a fixed dataset of length @xmath48 @xmath49 .
note that this additional term will not effect the optimization ( since @xmath50 is just a shift by a constant of @xmath51 ) and thus will be neglected here after . in order to implement this result into the svm framework
, we propose to define the csd - svm decision function for current status data by @xmath52.\label{eq : csd - svm}\ ] ] note that if the censoring mechanism is not known , we can replace the density @xmath53 with its estimate @xmath54 ; in this case our loss function becomes @xmath55 and the svm decision function is @xmath56 $ ] ( note the use of @xmath54 instead of @xmath53 in the denominator ) .
we note that for current status data , the assumption of some knowledge of the censoring distribution is reasonable , for example , when it is chosen by the researcher @xcite . in other cases ,
the density can be estimated using either parametric or nonparametric density estimation techniques such as kernel estimates .
it should be noted that the censoring variable itself is not censored and thus simple density estimation techniques can be used in order to estimate the density @xmath53 .
in this section we prove consistency of the csd - svm learning method for a large family of probability measures and construct learning rates .
we first assume that the censoring mechanism is known , and thus the true density of the censoring variable @xmath53 is known . using this assumption , and some additional conditions , we bound the difference between the risk of the csd - svm decision function and the bayes risk in order to form finite sample bounds .
we use this result , together with oracle inequalities , to show that the csd - svm converges in probability to the bayes risk .
that is , we demonstrate that for a very large family of probability measures , the csd - svm learning method is consistent .
we then consider the case in which the censoring mechanism is not known and thus the density @xmath53 needs to be estimated .
we estimate the density @xmath53 using nonparametric kernel density estimation and develop a novel finite sample bound .
we use this bound to prove that the csd - svm is consistent even when the censoring distribution is not known .
finally we construct learning rates for the csd - svm learning method for both @xmath53 known and unknown .
[ def : normalized loss]let @xmath57 be the normalized quadratic loss , let @xmath58 be its derivative with respect to the first variable , and let @xmath59 be the data - dependent version of this loss . for simplicity
, we use the normalized version of the quadratic loss .
since both @xmath13 and @xmath60 are convex functions with respect to @xmath61 , then for any compact set @xmath62\subset\mathbb{r}$ ] , both @xmath13 and @xmath60 are bounded and lipschitz continuous with constants @xmath63 and @xmath64 that depend on @xmath65 .
[ rem : loss bounds]@xmath66 for all @xmath67 and @xmath68 for all @xmath69 and for some constant @xmath70 .
we need the following assumptions : 1 .
the censoring time @xmath1 is independent of the failure time @xmath0 given @xmath5.[as : a1 ] 2 .
@xmath1 takes its values in the interval @xmath8 $ ] and @xmath71 , for some @xmath72 [ as : a2 ] .
@xmath6 is compact [ as : a3 ] .
@xmath25 is an rkhs of a continuous kernel @xmath29 with @xmath73 [ as : a4 ] .
define the approximation error by @xmath74 define @xmath75 and @xmath76 , where @xmath77 is defined in remark [ rem : loss bounds ] .
in this section we develop finite sample bounds assuming that the censoring density @xmath53 is known .
[ theorem 1 - g known]assume that ( a[as : a1])-(a[as : a4 ] ) hold . then for fixed @xmath78 , and @xmath79 , with probability not less than @xmath80 @xmath81 where @xmath82 is the covering number of the @xmath83 of @xmath84 with respect to supremum norm and where @xmath85 is the unit ball of @xmath25 ( for further details see @xcite ) .
the proof of this theorem appears in appendix [ sub : proof - of - theorem 1 ] .
we now move to discuss consistency of the csd - svm learning method . by definition
, @xmath17-universal consistency means that for any @xmath86 , @xmath87 where @xmath88 is the bayes risk .
universal consistency means that ( [ eq:3 ] ) holds for all probability measures @xmath17 on @xmath18 .
however , in survival analysis we have the problem of identifiability and thus we will limit our discussion to probability measures that satisfy some identification conditions . let @xmath89 be the set of all probablity measures that satisfy assumptions ( a[as : a1])-(a[as : a2 ] ) .
we say that a learning method is @xmath89-universal consistent when ( [ eq:3 ] ) holds for all probability measures @xmath90 . in order to show @xmath89-universal consistency
, we utilize the finite sample bounds of theorem [ theorem 1 - g known ] .
the following assumption is also needed for proving @xmath89-universal consistency : 1 . [ as : a5 ] @xmath91 , for all probability measures @xmath17 on @xmath18 assumptio ( a[as : a5 ] ) means that our rkhs @xmath25 is rich enough to include the bayes decision function .
[ cor - consistency ] assume the setting of theorem [ theorem 1 - g known ] and that assumptio ( a[as : a5 ] ) holds .
let @xmath92 be a sequence such that @xmath93 and @xmath94 choose @xmath95 for some @xmath96 then the csd - svm learning method is @xmath89-universal consistent . in theorem [ theorem 1 - g known ] we showed that @xmath97 with probability not greater than @xmath98 .
choose @xmath99 from assumption ( a[as : a5 ] ) together with lemma 5.15 of @xcite , @xmath100 converges to zero as @xmath3 converges to infinity .
clearly @xmath101 finally , from the choice of @xmath102 , it follows that both @xmath103 and @xmath104 converge to 0 as @xmath105 .
hence for every fixed @xmath106 @xmath107 with probability not less than 1-@xmath98 .
the right hand side of this converges to 0 as @xmath105 , which implies consistency ( * ? ? ?
* lemma 6.5 ) .
since this holds for all probability measures @xmath90 , we obtain @xmath89-universal consistency .
in this section we form finite sample bounds for the case in which the censoring density is not known and needs to be estimated .
we assume that the density of the censoring variable is estimated using nonparametric kernel density estimation . in lemma [ lemma on density ]
we construct finite sample bounds on the differnce between the estimated density @xmath54 and the true density @xmath53 . in theorem
[ theorem g unknown ] we utilize this bound to form finite sample bounds for the csd - svm learning method .
[ kernel of order m]we say that @xmath108 ( not to be confused with the kernel function @xmath29 of the rkhs @xmath109 is a kernel of order @xmath110 , if the functions @xmath111 are integrable and satisfy @xmath112 and @xmath113 .
[ holder class ] the hlder class @xmath114 of functions @xmath115 is the set of @xmath116 times differentiable functions whose derivative @xmath117 satisfies @xmath118 for some constant @xmath119 .
[ lemma on density]let @xmath108 be a kernel function of order m satisfying @xmath120 and define @xmath121where @xmath122 is the bandwidth .
suppose that the true density @xmath53 satisfies @xmath123 .
let us also assume that @xmath124 belongs to the @xmath125 class @xmath114 .
finally , assume that @xmath126 .
then for any @xmath86 , @xmath127 where @xmath128 and @xmath129 are constants , and for some @xmath130 $ ] . the proof of the lemma is based on ( * ? ? ?
* propositions 1.1 and 1.2 ) together with basic concentration inequalities ; the proof can be found in appendix [ sub : proof - of - lemma 1 ] .
we would like to choose @xmath122 that minimizes the sum of @xmath131 and @xmath132 .
define @xmath133 . taking the deivative of @xmath134 with respect to @xmath122 and setting to zero yields : @xmath135 @xmath136 where @xmath137 .
it can be shown that the second derivative of @xmath134 is positive which guarentees that the zero of the derivative above is the minimizer . after substituting @xmath138 in @xmath134
, we obtain that @xmath139 choosing @xmath86 such that @xmath140 and substituting @xmath138 , we obtain by lemma [ lemma on density ] that @xmath141 we now move to construct finite sample bounds for the csd - svm learning method when @xmath53 is unknown using the above lemma .
we assume that @xmath54 is the kernel density estimate of @xmath142 such that the conditions of lemma [ lemma on density ] hold .
[ theorem g unknown]assume that ( a[as : a1])-(a[as : a4 ] ) hold .
assume the setting of lemma [ lemma on density ] and that @xmath143 for some @xmath72 .
choose @xmath144 such that @xmath145 then for fixed @xmath146 , we have with probability not less than @xmath147 that @xmath148 where @xmath149 .
the proof of the theorem appears in appendix [ sub : proof - of - theorem 2 ] . using the above theorem
we show that under some conditions , the csd - svm decision function converges in probability to the conditional expectation .
[ consistency 2]let @xmath92 be a sequence such that @xmath93 and that @xmath94 choose @xmath95 for some @xmath96 assume the setting of theorem 3 , then the csd - svm learning method is consistent .
the proof of the corollary is derived similarly to the proof of corollary [ cor - consistency ] ( consistency - case i ) . in this section
we derive learning rates for cases i and ii .
[ def learning rate]a learning method is said to learn with rate @xmath150 $ ] that converges to zero if for all @xmath151 and all @xmath152 $ ] , @xmath153 , where @xmath154 and @xmath155 are constants such that @xmath156 and @xmath157 . [ theorem learning rate]assume that ( a[as : a1])-(a[as : a4 ] ) hold .
choose @xmath158 and assume that there exist constants @xmath159 such that @xmath160 .
additionally , assume that there exist constants @xmath161 $ ] such that @xmath162 .
choose @xmath92 = @xmath163 .
then \(i ) if @xmath53 is known , the learning rate is given by @xmath164 .
\(ii ) if @xmath53 is not known and the setup of theorem [ theorem g unknown ] holds , then the leraning rate is given by @xmath165 .
the proof of the theorem appears in appendix [ sub : proof - of - theorem 3 ] .
in this section we demonstrate how to compute the csd - svm decision function with respect to the quadratic loss .
in addition we show that the solution has a closed form .
since @xmath166 is convex , then for any rkhs @xmath25 over @xmath26 and for all @xmath167 , it follows that there exists a unique svm solution @xmath168 .
in addition , by the representer theorem @xcite , there exists constants @xmath169 such that @xmath170 .
hence the optimization problem reduces to estimation of the vector @xmath144 .
a more general approach will also include an intercept term @xmath171 such that @xmath172 .
let @xmath173 be the feature map that maps the input data into an rkhs @xmath25 such that @xmath174 .
our goal is to find a function @xmath175 that is the solution of ( [ eq : csd - svm ] ) . from the representer theorem
, there exists a unique svm decision function of the form @xmath176 .
define for each @xmath177 the function @xmath178 by @xmath179 .
then for @xmath180 , the optimization problem reduces to : @xmath181+\frac{1}{2}\|w\|^{2}\ ] ] @xmath182 @xmath183 this is an optimization problem under equality constraints and hence we will use the method of lagrange multipliers .
the lagrangian is given by @xmath184+\frac{1}{2}\|w\|^{2}+\sum_{i=1}^n \alpha_{i}\left(c_{i}-<w,\phi(z_{i})>-b - r_{i}\right)\ ] ] minimizing the original problem @xmath185 yields the following conditions for optimality : @xmath186 @xmath187 @xmath188 since these are equality constraints in the dual formulation , we can substitute them into @xmath185 to obtain the dual problem @xmath189 . by the strong duality theorem ( * ? ? ? * theorem 6.2.4 ) , the solution of the dual problem is equivalent to the solution of the primal problem . @xmath190+\frac{1}{2}\sum_{i=1}^n
\sum_{j=1}^n \alpha_{i}\alpha_{j}k(z_{i},z_{j})\\ + & \sum_{i=1}^n \alpha_{i}\left(c_{i}-\sum_{j=1}^n \alpha_{j}k(z_{i},z_{j})-b-\left(\frac{\alpha_{i}}{c_{\lambda}}+c_{i}-\frac{(1-\vardelta_{i})}{2\hat{g}(c_{i}|z_{i})}\right)\right).\end{aligned}\ ] ] some calculations yield : @xmath191 subject to the constraint @xmath192 and where @xmath193 .
this is a quadratic programming problem subject to equality constraints .
its solution satisfies : @xmath194 note that if we do not require an intercept term , the solution is @xmath195 .
it is interesting to note that this solution is equivalent to the solution attained by the representer theorem for differentiable loss functions : @xmath196 ( * ? ? ?
* section 5.2 ) . in our case ,
@xmath197 ; hence @xmath198 and since @xmath199 we see that @xmath200 i.e. , @xmath195 .
in this section we test the csd - svm learning method on simulated data and compare its performance to current state of the art . we construct four different data - generating mechanisms , including one - dimensional and multi - dimentional settings . for each data type ,
we compute the difference between the csd - svm decision function and the true expectation .
we compare this result to results obtained by the cox model and by the aft model . as a reference , we compare all these methods to the bayes risk . for each data setting , we considered two cases ; : ( i ) the censoring density @xmath53 is known ; and ( ii ) the censoring density is not known . for the second
setting , the distribution of the censoring variable was estimated using nonparametric kernel density estimation with a normal kernel .
the code was written in matlab , using the spider library . in order to fit the cox model to current status data
, we downloaded the ` icsurv ' r package @xcite . in this package ,
monotone splines are used to estimate the cumulative baseline hazard function , and the model parameters are then chosen via the em algorithm .
we chose the most commonly used cubic splines . to choose the number and locations of the knots
, we followed @xcite and @xcite who both suggested using a fixed small number of knots and thus we placed the knots evenly at the quantiles .
for the aft model , we used the ` surv reg ' function in the ` survival ' r package @xcite . in order to call r through matlab
, we installed the r package rscproxy @xcite , installed the statconndcom server , and download the matlab r - link toolbox @xcite . for the kernel of the rkhs @xmath25
, we used both a linear kernel and a gaussian rbf kernel @xmath201 , where @xmath202 and @xmath203 were chosen using 5-fold cross - validation .
the code for the algorithm and for the simulations is available for download ; a link to the code can be found in the [ sub : supplementary - material ] .
we consider the following four failure time distributions , corresponding to the four different data - generating mechanisms : ( 1 ) weibull , ( 2 ) multi - weibull , ( 3 ) multi - log - normal , and ( 4 ) an additional example where the failure time expectation is triangle shaped .
we present below the csd - svm risks for each case and compare them to risks obtained by other methods .
the risks are based on 100 iterations per sample size .
the bayes risk is also plotted as a reference . in setting 1 ( weibull failure - time ) , the covariates @xmath5 are generated uniformly on @xmath204,$ ] the censoring variables @xmath1 is generated uniformly on @xmath8,$ ] and the failure time @xmath0 is generated from a weibull distribution with parameters @xmath205 .
the failure time was then truncated at @xmath206 . figure [ fig : weibull - failure - time ] compares the results obtained by the csd - svm to results achieved by the cox model and by the aft model , for different sample sizes .
it should be noted that both the ph and the aft assumption hold for the weibull failure time distribution . in particular , when the ph assumption holds , estimation based on the cox regression is consistent and efficient ; hence , when the ph assumption holds , we will use the cox regression as a benchmark .
figure [ fig : weibull - failure - time ] shows that when @xmath53 is known , even though the csd - svm does not use the ph assumption or the aft assumption , the results are comparable to those of the cox regression , and are better than the aft estimates , especially for larger sample sizes . however , when @xmath53 is not known , the cox model produces the smallest risks , but its superiority reduces as the sample size grows .
this coincides with the fact that when @xmath53 is not known , the learning rate of the csd - svm is slower .
.[fig : weibull - failure - time ] ] in setting 2 ( multi - weibull failure - time ) , the covariates @xmath5 are generated uniformly on @xmath204^{10},$ ] and the censoring variable @xmath1 is generated uniformly on @xmath8 $ ] , as in setting 1 .
the failure time @xmath0 is generated from a weibull distribution with parameters @xmath207 .
the failure time was then truncated at @xmath208 . note that this model depends only on the first three variables . in figure
[ fig : multi - weibull - failure - time ] , boxplots of risks are presented .
figure [ fig : multi - weibull - failure - time ] illustrates that the csd - svm with a linear kernel is superior to the other methods , for all sample sizes and for both the cases @xmath53 known and uknown .
however , since the data may be sparse in the feature space , the csd - svm with an rbf kernel might require a larger sample size to converge . .
[ fig : multi - weibull - failure - time ] ] in setting 3 ( multi - log - normal ) , the covariates @xmath5 are generated uniformly on @xmath204^{10},$ ] @xmath1 was generated as before and the failure time @xmath0 was generated from a log - normal distribution with parameters @xmath209 .
the failure time was then truncated at @xmath210 . figure [ fig : log - normal - failure - time ] presents the risks of the compared methods .
this example illustrates that for small sample sizes , the csd - svm risks are significantly superior and converge quickly to the bayes risk .
as the sample size grows , the aft also converges to the bayes risk whereas the cox estimates does not , as can be seen by the very high risks they produce .
note that for the log - normal distribution , even though the aft assumption is correct , the csd - svm manages to produce better or equivalent results . .
[ fig : log - normal - failure - time ] ] in setting 4 , we considered a non - smooth conditional expectation function in the shape of a triangle .
the covariates @xmath5 are generated uniformly on @xmath204,$ ] @xmath1 is generated uniformly on @xmath8 $ ] , and @xmath0 was generated according to the following @xmath211 the failure time was then truncated at at @xmath212 . in figure
[ fig : triangle box i ] , the boxplots of risks are presented . as can be seen
, the csd - svm with an rbf kernel is superior in both cases , for sufficently large sample sizes . .
[ fig : triangle box i ] ] to illustrate the flexibility of the csd - svm , we also present a graphical representation of the true conditional expectation and its estimates , as a function of the covariates .
figure [ fig : triangle plot i ] compares the true expectation to the computed estimates for the case that @xmath53 is known ; these estimates are based on the first iteration . as can be seen , the csd - svm with an rbf kernel produces the most superior results .
is known ) .
the true expectation is the blue line .
the following estimates are compared : the csd - svm with an rbf kernel , the csd - svm with a linear kernel , cox and aft for sample sizes @xmath213.[fig : triangle plot i ] ] to summarize , figures [ fig : weibull - failure - time]-[fig : triangle plot i ] showed that the csd - svm is comparable to other known methods for estimating the failure time distribution with current status data , and in certain cases is even better .
specifically , we found that the csd - svm with an appropriate kernel was superior in three out of the four examples , especially when the true density @xmath53 is known .
it should be noted that even when the assumptions of the other models were true the csd - svm estimates were comparable .
additionally , when these assumptions fail to hold , the csd - svm estimates were generally better .
the main advantage of the proposed svm approach is that it does not assume any parametric form and thus may be superior , especially when the assumptions of other models fail to hold .
additionally , it seems that the csd - svm can perform well in higher dimensions .
we proposed an svm approach for estimation of the failure time expectation , studied its theoretical properties and presented a simulation study .
we believe this work demonstrates an important approach in applying machine learning techniques to current status data
. however , many open questions remain and many possible generalizations exist .
first , note that we only studied the problem of estimating the failure time expectation and not other distribution related quantities .
further work needs to be done in order to extend the svm approach to other estimation problems with current status data .
second , we assumed that the censoring is independent of the failure time given the covariates and that the censoring density is positive given the covariates over the entire observed time range .
it would be worthwhile to study the consequences of violation of some of these assumptions .
third , it could be interesting to extend this work to other censored data formats such as interval censoring .
we believe that further development and generalization of svm learning methods for different types of censored data is of great interest .
since @xmath59 is convex , it implies that there exists a unique svm solution ( see * ? ? ? * section 5.1 ) .
for all distributions @xmath214 on @xmath18 , we define the svm decision function by @xmath215 we note that for an rkhs @xmath25 of a continuous kernel @xmath29 with @xmath73 , recall that the unit ball of @xmath25 is denoted by @xmath85 and its closure by @xmath222 ; since@xmath223 we can write @xmath224 . since @xmath6 is compact , it implies that the @xmath225 @xmath222 of the unit ball @xmath85 is compact in @xmath226 ( see * ? ? ?
* corollary 4.31 ) .
@xmath243+\frac{1}{n}\sum_{i=1}^n [ l(0,h(z_{i}))]-\frac{1}{n}\sum_{i=1}^n \left[\frac{(1-\delta_{i})\ell(c_{i},f(z_{i}))}{g(c_{i}|z_{i})}\right]-\frac{1}{n}\sum_{i=1}^n [ l(0,f(z_{i}))]\right|\\ \leq & \left|\frac{1}{n}\sum_{i=1}^n \left[\frac{(1-\delta_{i})\ell(c_{i},h(z_{i}))}{g(c_{i}|z_{i})}\right]-\frac{1}{n}\sum_{i=1}^n
\left[\frac{(1-\delta_{i})\ell(c_{i},f(z_{i}))}{g(c_{i}|z_{i})}\right]\right|+\left|\frac{1}{n}\sum_{i=1}^n [ l(0,h(z_{i}))]-\frac{1}{n}\sum_{i=1}^n [ l(0,f(z_{i}))]\right|\\ \equiv & c_{n,1}+c_{n,2},\end{aligned}\ ] ] where @xmath244\right|\\ = & \left|\frac{1}{n}\sum_{i=1}^n \left[\frac{(1-\delta_{i})}{g(c_{i}|z_{i})}\left(\ell(c_{i},h(z_{i}))-\ell(c_{i},f(z_{i}))\right)\right]\right|\\ \leq & \left|\frac{1}{n}\sum_{i=1}^n \left[\frac{1}{g(c_{i}|z_{i})}\left(\ell(c_{i},h(z_{i}))-\ell(c_{i},f(z_{i}))\right)\right]\right|\\ \leq & \frac{1}{2k}\left|\frac{1}{n}\sum_{i=1}^n \left[\ell(c_{i},h(z_{i}))-\ell(c_{i},f(z_{i}))\right]\right|\leq\frac{1}{2nk}\sum_{i=1}^n |\ell(c_{i},h(z_{i}))-\ell(c_{i},f(z_{i}))|\\ \leq & \frac{1}{2nk}\sum_{i=1}^n c_{l}|h(z_{i})-f(z_{i})|\leq\frac{1}{2nk}\sum_{i=1}^n c_{l}\varepsilon=\frac{c_{l}\varepsilon}{2k},\end{aligned}\ ] ] and where @xmath245\right|\\ \leq & \frac{1}{n}\sum_{i=1}^n \left|l(0,h(z_{i}))-l(0,f(z_{i}))\right|\\ \leq & \frac{1}{n}\sum_{i=1}^n c_{l}|h(z_{i})-f(z_{i})|\leq\frac{1}{n}\sum_{i=1}^n \left[c_{l}\varepsilon\right]=c_{l}\varepsilon\end{aligned}\ ] ] as in ( * ? ? ?
* proposition 1.1 ) , define @xmath259 $ ]
. then @xmath260 , for @xmath261 are i.i.d .
random variables with zero mean and with variance : @xmath262&=e_{g}\left[\left(\eta_{i}(c)\right)^{2}\right]=e_{g}\left[\left(k\left(\frac{c_{i}-c}{h}\right)-e_{g}\left[k\left(\frac{c_{i}-c}{h}\right)\right]\right)^{2}\right]\leq e_{g}\left[k^{2}\left(\frac{c_{i}-c}{h}\right)\right ] \\ & = \int_{u}{k^{2}\left(\frac{u - c}{h}\right)g(u)du}\leq g_{max}\int_{u}k^{2}\left(\frac{u - c}{h}\right)du\stackrel{}{=g_{max}\int_{v}k^{2}\left(v\right)dv = c_{1}h}\end{aligned}\ ] ] where the equality before last follows from change of variables and where @xmath263 . thus @xmath264=\frac{1}{nh^{2}}e_{g}\left[\eta_{1}^{2}(c)\right]\leq\frac{c_{1}h}{nh^{2}}=\frac{c_{1}}{nh}.$ ] by the cauchyschwarz inequality we have that @xmath265\right|\right]\leq\sqrt{e\left[\left|\hat{g}(c)-e\left[\hat{g}(c)\right]\right|^{2}\right]}=\sqrt{v(\hat{g}(c))}.$ ] hence @xmath265\right|\right]\leq\sqrt{\frac{c_{1}}{nh}}.$]therfore , by markov s inequality , @xmath269\right|+\frac{1}{n}\sum_{i=1}^n \left|e\left[\hat{g}(c_{i})\right]-g(c_{i})\right|>\epsilon+c_{2}\cdot h^{\beta}\right)\\ \leq & pr\left(\frac{1}{n}\sum_{i=1}^n \left|\hat{g}(c_{i})-e\left[\hat{g}(c_{i})\right]\right|+c_{2}\cdot h^{\beta}>\epsilon+c_{2}\cdot h^{\beta}\right)\\ = & pr\left(\frac{1}{n}\sum_{i=1}^n \left|\hat{g}(c_{i})-e\left[\hat{g}(c_{i})\right]\right|>\epsilon\right)\leq\sqrt{\frac{c_{1}}{nh\epsilon^{2 } } } \end{alignedat}\ ] ] where @xmath122 is the bandwidth .
note that the proof of this theorem is similar to the proof of of theorem [ theorem 1 - g known ] and thus we will only discuss the parts of the proof where they differ . as in theorem [
theorem 1 - g known ] , equation [ eq : abc ] , @xmath270 @xmath274+\frac{1}{n}\sum_{i=1}^n [ l(0,v(z_{i}))]-\frac{1}{n}\sum_{i=1}^n \left[\frac{(1-\delta_{i})\ell(c_{i},f(z_{i}))}{\hat{g}(c_{i}|z_{i})}\right]-\frac{1}{n}\sum_{i=1}^n [ l(0,f(z_{i}))]\right|\\ \leq & \left|\frac{1}{n}\sum_{i=1}^n \left[\frac{(1-\delta_{i})\ell(c_{i},v(z_{i}))}{\hat{g}(c_{i}|z_{i})}\right]-\frac{1}{n}\sum_{i=1}^n \left[\frac{(1-\delta_{i})\ell(c_{i},f(z_{i}))}{\hat{g}(c_{i}|z_{i})}\right]\right|+\left|\frac{1}{n}\sum_{i=1}^n [ l(0,v(z_{i}))]-\frac{1}{n}\sum_{i=1}^n [ l(0,f(z_{i}))]\right|\\ \equiv & c_{n,1}+c_{n,2}\end{aligned}\ ] ] recall that the loss @xmath235 is bounded by @xmath250 .
define @xmath279 by @xmath280+\frac{1}{n}\sum_{i=1}^n [ l(0,v(z_{i}))].\ ] ] in other words , @xmath279 is the empirical risk with the true censoring density function @xmath53 .
@xmath284-\frac{1}{n}\sum_{i=1}^n \left[\frac{(1-\delta_{i})\ell(c_{i},v(z_{i}))}{\hat{g}(c_{i}|z_{i})}\right]\right|\\ = & \left|\frac{1}{n}\sum_{i=1}^n \left[(1-\delta_{i})\ell(c_{i},v(z_{i}))\left(\frac{1}{g(c_{i}|z_{i})}-\frac{1}{\hat{g}(c_{i}|z_{i})}\right)\right]\right|\\ \leq & \frac{1}{n}\sum_{i=1}^n \left[\left|\ell(c_{i},v(z_{i}))\left(\frac{1}{g(c_{i}|z_{i})}-\frac{1}{\hat{g}(c_{i}|z_{i})}\right)\right|\right]\\ = & \frac{b_{1}}{n}\sum_{i=1}^n \left[\left|\frac{\hat{g}(c_{i}|z_{i})-g(c_{i}|z_{i})}{g(c_{i}|z_{i})\hat{g}(c_{i}|z_{i})}\right|\right]\leq\frac{b_{1}}{2k^{2}n}\sum_{i=1}^n \left[\left|\hat{g}(c_{i}|z_{i})-g(c_{i}|z_{i})\right|\right].\end{aligned}\ ] ] note that these inequalities hold for all functions @xmath285 .
we would like to bound the last expression using lemma [ lemma on density ] . by equation [ eq : h ] ,
let @xmath138 , choose @xmath144 such that @xmath286 and let @xmath287 , then by lemma [ lemma on density ] @xmath288>\eta\right)\\ = & pr\left(\frac{b_{1}}{2k^{2}n}\sum_{i=1}^n \left[\left|\hat{g}(c_{i}|z_{i})-g(c_{i}|z_{i})\right|\right]>\frac{b_{1}\left(\alpha+c_{2}\cdot h^{\beta}\right)}{2k^{2}}\right)\\ = & pr(\frac{1}{n}\sum_{i=1}^n \left[\left|\hat{g}(c_{i}|z_{i})-g(c_{i}|z_{i})\right|\right]>\alpha+c_{2}\cdot h^{\beta})\\ \leq & \sqrt{\frac{c_{1}}{nh\alpha^{2}}}=e^{-\theta}. \end{alignedat}\ ] ] @xmath297 with probability not less than @xmath80 . for any compact set @xmath62\subset\mathbb{r}$ ] , both @xmath13 and @xmath60 are bounded and lipschitz continuous with lipschitz constants @xmath298 and @xmath299 .
hence , recall that @xmath75 and @xmath76 , where @xmath77 is some bound on the derivative of the loss . since @xmath158 , then @xmath306 , and therefor @xmath307 .
earlier we defined @xmath308 such that @xmath309 .
thus , @xmath310 @xmath311 where we define @xmath312 .
@xmath313+\frac{n}{\sqrt{\lambda}}\sqrt{\frac{2\theta}{n}}\\ \leq & \left(\frac{p}{2}\right)^{\frac{-p}{1+p}}\frac{n}{\sqrt{\lambda}}\left[\sqrt{2}\left(\frac{2a}{n}\right)^{\frac{1}{2 + 2p}}+\frac{mp}{2n}\left(\frac{2a}{n}\right)^{\frac{1}{2 + 2p}}\right]+\frac{n}{\sqrt{\lambda}}\sqrt{\frac{2\theta}{n}}\\ \leq & \left(\frac{p}{2}\right)^{\frac{-p}{1+p}}\frac{n}{\sqrt{\lambda}}\left[2\left(\frac{2a}{n}\right)^{\frac{1}{2 + 2p}}+\frac{mp}{n}\left(\frac{2a}{n}\right)^{\frac{1}{2 + 2p}}\right]+\frac{n}{\sqrt{\lambda}}\sqrt{\frac{2\theta}{n } } \end{alignedat}\ ] ] we would like to choose a sequence @xmath92 that will minimize the bound in ( [ eq : learning rate bound ] ) .
define @xmath323.$ ] differentiating @xmath324 with respect to @xmath325 and setting to zero yields : @xmath326=0\\ \leftrightarrow\\ c\gamma\lambda^{\gamma-1}= & \frac{1}{2}n\lambda^{-\frac{3}{2}}\left[6\left(\frac{2a}{n}\right)^{\frac{1}{2 + 2p}}+\sqrt{\frac{2\theta}{n}}\right]\\ \leftrightarrow\lambda= & \left(\frac{1}{2c\gamma}n\left[6\left(\frac{2a}{n}\right)^{\frac{1}{2 + 2p}}+\sqrt{\frac{2\theta}{n}}\right]\right)^{\frac{1}{\gamma+\frac{1}{2}}}\propto\left(\frac{1}{n}^{\frac{1}{2 + 2p}}+\left(\frac{1}{n}\right)^{\frac{1}{2}}\right)^{\frac{2}{2\gamma+1}}\\ \rightarrow\lambda\propto & n^{-\frac{1}{(1+p)(2\gamma+1 ) } } \end{alignedat}\ ] ] @xmath330\\ = & cn^{-\frac{\gamma}{(1+p)(2\gamma+1)}}+n\cdot6\left(2a\right)^{\frac{1}{2 + 2p}}n^{-\frac{\gamma}{(1+p)(2\gamma+1)}}+n\left(2\theta\right)^{\frac{1}{2}}n^{-\frac{2\gamma(1+p)+p}{2(1+p)(2\gamma+1)}}\\ \leq & cn^{-\frac{\gamma}{(1+p)(2\gamma+1)}}+n\cdot6\left(2a\right)^{\frac{1}{2 + 2p}}n^{-\frac{\gamma}{(1+p)(2\gamma+1)}}+n\left(2\theta\right)^{\frac{1}{2}}n^{-\frac{\gamma}{(1+p)(2\gamma+1)}}\\ = & n^{-\frac{\gamma}{(1+p)(2\gamma+1)}}\left(c+n\cdot6\left(2a\right)^{\frac{1}{2 + 2p}}+n\left(2\theta\right)^{\frac{1}{2}}\right)\\ \leq & q(1+\sqrt{\theta})n^{-\frac{\gamma}{(1+p)(2\gamma+1 ) } } \end{alignedat}\ ] ] where @xmath334 and with probability not greater than @xmath335 .
choose @xmath303 , @xmath336 , @xmath337 , and define @xmath338 , then as in ( [ eq : learning rate bound ] ) , a very similar calculation shows that @xmath339 + 2\eta.\end{alignedat}\ ] ] @xmath339 + 2\eta\\ \leq & c\lambda^{\gamma}+\frac{n}{\sqrt{\lambda}}\left[6\left(\frac{2a}{n}\right)^{\frac{1}{2 + 2p}}+\sqrt{\frac{2\theta}{n}}\right]+\frac{4k^{2}\left(e^{\frac{2\beta+1}{2\beta}}\left(c_{1}\right)^{\frac{1}{2}}\left(2\beta c_{2}\right)^{\frac{1}{2\beta}}+c_{2}\kappa^{\beta}\right)}{b_{1}n^{\frac{\beta}{2\beta+1 } } } \end{alignedat}\ ] ] similarly to case i , choosing @xmath343 minimizes the last bound ( note that the choice of @xmath92 does not depend on @xmath341 ) .
hence that the resulting learning rate is given by i. d. diamond , j. w. mcdonald , and i. h. shah .
proportional hazards models for current status data : application to the study of differentials in age at weaning in pakistan .
_ demography _ , 230 ( 4):0 607620 , 1986 .
antonio eleuteri and azzam f. g. taktak .
support vector machines for survival regression . in elia biganzoli , alfredo vellido , federico ambrogi , and roberto tagliaferri , editors , _ computational intelligence methods for bioinformatics and biostatistics _
, number 7548 in lecture notes in computer science , pages 176189 .
springer berlin heidelberg , 2012 .
n. jewell and m. van der laan .
current status data : review , recent developments and open problems . in n.
balakrishnan and c.r .
rao , editors , _ handbook of statistics , advances in survival analysis _ , number 23 , pages 625642 .
elsevier , 2004 .
f.m . khan and v.b .
support vector regression for censored data ( svrc ) : a novel tool for survival analysis . in _
eighth ieee international conference on data mining , 2008 .
icdm 08 _ , pages 863868 , december 2008 .
v. van belle , k. pelckmans , j.a.k .
suykens , and s van huffel . support vector machines for survival analysi . in _ proceedings of the third international conference on computational intelligence in medicine and healthcare ( cimed2007 )
_ , plymouth ( uk ) , 2007 . | current status data is a data format where the time to event is restricted to knowledge of whether or not the failure time exceeds a random monitoring time .
we develop a support vector machine learning method for current status data that estimates the failure time expectation as a function of the covariates . in order to obtain the support vector machine decision function ,
we minimize a regularized version of the empirical risk with respect to a data - dependent loss .
we show that the decision function has a closed form . using finite sample bounds and novel oracle inequalities , we prove that the obtained decision function converges to the true conditional expectation for a large family of probability measures and study the associated learning rates
. finally we present a simulation study that compares the performance of the proposed approach to current state of the art . and |
recent precision measurements of cosmological quantities , such as the cosmic microwave background temperature and polarization @xcite and the supernova luminosity - to - distance relationship , together with large scale structure and structure formation studies @xcite , allowed to build a _
standard cosmological model _
, that describes a universe spatially flat , homogeneous and isotropic at large scales .
our universe consists of ordinary matter and radiation only for a 4.5% : as shown in figure [ fig : snproject ] , a considerable fraction of the total matter ( 23% ) is in the form of cold dark matter .
the largest fraction ( 73% ) consists of dark energy , acting like a cosmological constant , whose origin is unknown @xcite .
the antimatter content of the universe appears to be only 10@xmath1 of the matter content @xcite .
the aim of ams-02 is to perform precision studies of cosmic ray properties such as composition , production , acceleration and propagation mechanisms .
its large acceptance , energy resolution and particle identification capabilities allow ams-02 to search for primordial antimatter by direct detection of antinuclei , and for dark matter annihilation products independently in different charged and neutral particles spectra .
ams-02 is designed as a single particle spectrometer with large acceptance , high momentum range and efficient particle identification for reliable operation in space , to provide high statistics spectra of charged particles with nuclei and isotope separation , as well as gamma - ray measurements
. it will carry out a 3-years mission taking data above the atmosphere on the international space station .
ams-02 is shown in figure [ fig : ams ] .
it consists of a fully equipped tracker inside a superconducting magnet ( the first ever to operate in space ) providing a 0.8 t m@xmath2 magnetic field .
rigidity is measured up to 3 tev and nuclei and isotopes are separated up to 12 gev / n for z@xmath326 or a@xmath325 with a ring imaging aerogel / sodium - flouride cherenkov detector ( agl / naf rich ) , combined with time of flight ( tof ) and @xmath4 measurements . at 90% positron efficiency ,
a proton suppression by 3 - 4 orders of magnitude is achieved with a lead / scintillating fibre sandwich 3d sampling calorimeter ( ecal ) based on shower shape information and the matching of shower energy with track momentum .
further proton suppression by 2 - 3 orders of magnitude up to 300 gev / c is achieved with a 20 layer fibre fleece xe / co2 proportional wire straw tube transition radiation detector ( trd ) .
the overall 10@xmath5 proton rejection factor is essential for dark matter search .
a weakly interacting massive particle ( wimp ) is the favoured dark matter canditate , and supersymmetric extensions of the standard model such as msugra contain a _ neutralino _ which is stable when r - parity is conserved .
ams-02 is capable of detectig wimps indirectly by measuring decay products of neutralino annihilation processes , as an extra source for positrons , antiprotons , antideuterons and gammas on top of standard model predictions from galactic propagation models . with annihilation cross sections proportional to the square of the neutralino density ,
these measurements are also sensitive to dark matter density fluctuations .
recent measurements of gamma - ray fluxes @xcite and the positron fraction @xcite have shown deviations from standard model expectations which are consistent with supersymmetric dark matter , but controversial within the available statistics and without standard model explanation for the positron excess @xcite .
ams-02 will measure the positron spectrum with substantially improved precision for background and signal , in the energy region where an excess has been claimed . moreover , the simultaneous measurement in other independent channels will increase the sensitivity to dark matter detection .
the spectra of positrons and antiprotons expected in the ams-02 sensitivity region are shown in figure [ fig : dm ] .
ams-02 can also detect gamma rays up to 1 tev either directly in the ecal , or indirectly as an electron - positron pair after conversion in the material above the tracker , which amounts to 40% of a radiation length .
the part of the parameter space from new physics that is within reach of ams-02 gamma measurements in the case of cuspy halo profile or extra enhancements , and even for a standard nfw dark matter halo @xcite , is also shown in figure [ fig : dm ] . in the big bang theory matter and antimatter are created with equal abundances , and the disappearance of antimatter requires barion number violation and another source of cp violation .
antiparticles are indeed produced in collisions between high energy particles , and are observed in the cosmic rays .
for example , @xmath6 however , an anti - helium nucleus has very low probability of being produced in collisions : @xmath7 anti - helium detection in the cosmic rays would be a clear indication of the existence of an antimatter area somewhere in the universe .
ams-02 will collect in three years @xmath8 nuclei with energies up to 2 tev , with a sensitivity that reaches anti - iron . the limit put by the precursor flight ams-01
will be increased of a factor @xmath9 , meaning that , if no antinucleus is observed , there is no antimatter to the edge of the observable universe ( about 1000 mpc ) .
the limit on antimatter presence they can be set by ams-02 is shown in figure [ fig : am ] .
ams-02 will provide a coherent dataset of unprecedented precision for charged cosmic rays and gamma rays up to the tev region . these data will confirm or disprove with high accuracy the excess in the heat positron data in the few gev region .
the multichannel analysis will allow combined fits to the parameters of standard model extensions to establish the nature of dark matter .
several models can be constrained and eventually ruled out .
ams-02 will put a stringent limit on the presence of antimatter : if no antinucleus is observed , the hypothesis of barion asymmetry will be strongly favoured as no antimatter areas are present in the observable universe .
in general , our knowledge of the cosmic rays physics will be improved .
ams-02 will perform an accurate study of composition ( h , he , b / c , @xmath10be/@xmath11be ) and energy spectra , and put significant constraints on galactic propagation model parameters .
d. n. spergel _ et al . _
[ wmap collaboration ] , astrophys .
j. suppl .
* 170 * ( 2007 ) 377 l. page _ et al .
_ [ wmap collaboration ] , astrophys . j. suppl .
* 170 * ( 2007 ) 335 [ arxiv : astro - ph/0603450 ] . c. b. netterfield _ et al . _ [ boomerang collaboration ] , astrophys . j. * 571 * ( 2002 ) 604 a. g. riess _ et al . _
[ supernova search team collaboration ] , astrophys .
j. * 607 * ( 2004 ) 665 s. perlmutter , m. s. turner and m. j. white , phys .
* 83 * ( 1999 ) 670 r. a. knop _ et al . _
[ supernova cosmology project collaboration ] , astrophys . j. * 598 * ( 2003 ) 102 j. alcaraz _ et al . _
[ ams collaboration ] , phys .
b * 461 * ( 1999 ) 387 s. d. hunter _
et al . _ ,
astrophys .
j. * 481 * ( 1997 ) 205 .
j. j. beatty _ et al .
_ , phys .
* 93 * ( 2004 ) 241102 [ arxiv : astro - ph/0412230 ] .
a. w. strong , i. v. moskalenko and o. reimer , astrophys .
j. * 613 * ( 2004 ) 962 [ arxiv : astro - ph/0406254 ] . w. de boer , c. sander , v. zhukov , a. v. gladyshev and d. i. kazakov , phys .
* 95 * ( 2005 ) 209001 [ arxiv : astro - ph/0602325 ] . | ams-02 is a space - borne magnetic spectrometer designed to measure with very high accuracy the composition of cosmic rays near earth . with a large acceptance of 5000 @xmath0 , an intense magnetic field from a superconducting magnet ( 0.7 t ) and a very efficient particle identification
, ams-02 will provide the highest precision in cosmic rays measurements up to the tev region . during a three - years mission on board of the international space station
, ams-02 will achieve a sensitivity to the existence of anti - helium nuclei in the cosmic rays of one part in a billion , as well as provide important information on the origin and nature of the dark matter .
we review the status of the construction of the ams-02 experiment and its remarkable discovery capabilities . |
newtonian gravitation has been shown to describe the motions of bodies with intermediate accelerations ( e.g. solar system bodies ) very accurately .
the newtonian theory of gravity breaks down in the presence of strong gravitational fields , where it has been successfully superseded by general relativity , but there is evidence that there are also discrepancies in the low - acceleration regime .
spiral galaxies , for example , require the invocation of dark matter ( dm ) to reconcile the discrepancy between their rotational velocities and our understanding of newtonian gravity .
another indication is the so - called `` pioneer anomaly '' : radiometric data from pioneers 10 and 11 have shown that both are experiencing an unexplained constant acceleration of @xmath0ms@xmath1 toward the sun ( see * ? ? ? * and references therein ) .
in fact , all spacecraft in the outer solar system seem to be experiencing this anomalous acceleration ( e.g. @xcite ) , potentially indicating a deviation from newtonian gravity at low accelerations .
it has been claimed that one of the leading versions of modified newtonian dynamics ( mond ; * ? ? ?
* ) can predict the kinematic properties of galaxies without invoking dm .
mond predicts a breakdown of newtonian gravity at an acceleration scale @xmath2ms@xmath1 ( see * ? ? ?
* and references therein ) .
this is approximately the acceleration regime where dm becomes necessary to reconcile theoretical velocity profiles of galaxies with observation .
globular clusters ( gcs ) are the ideal test bed for this divergence from newtonian predictions , since they are thought to contain little or no dm .
this is based on evidence from dynamical models ( e.g. * ? ? ?
* ) , _ n_-body simulations ( e.g. * ? ? ?
* ) , observations of gc tidal tails ( e.g. * ? ? ?
* ) , dynamical and luminous masses of gcs ( e.g. * ? ? ?
* ) and the lack of microlensing events from gc - mass dark haloes ( e.g. * ? ? ?
* ; * ? ? ?
in addition , gcs at different distances from the galactic centre experience differing gravitational attractions from the galaxy , so that any galactic influence can be ruled out if all exhibit similar behaviour .
furthermore , the accelerations experienced by stars in gcs drop below @xmath3 well inside the tidal radius ( e.g. * ? ? ?
recently , @xcite showed that there may be a flattening of the velocity profiles of several gcs in the galactic halo ( @xmath4 centauri , m15 , m30 , m92 and m107 ) , at or about the acceleration where dm is invoked to maintain stability in rotating galaxies .
on the other hand , @xcite overlaid the @xcite results with a modified gravity ( mog ) model and found little , or no , deviation from newtonian gravity , for gcs with masses less than a few times @xmath5m@xmath6 . because both groups used an identical set of measured velocity dispersions
, there is an obvious discrepancy .
if mond were correct , it would mean that all gravitational interactions would diverge from newtonian gravitation in the predicted regime .
irrespective of mond itself , testing the gravitational interaction at low accelerations is extremely important in the overall understanding of gravity .
this paper continues this test , with four galactic halo gcs , namely m22 ( ngc 6656 ) , m30 ( ngc 7099 ) , m53 ( ngc 5024 ) and m68 ( ngc 4590 ) .
we used the multi - object , double - beam spectrograph ( aaomega ) on the 3.9 m anglo - australian telescope at siding spring observatory in new south wales , australia , to obtain the spectra for this survey .
aaomega has 392 fibres covering a 2 square degree field of view with each fibre capable of obtaining a single spectrum from a single star in this field .
the 1500v grating ( with a resolution of @xmath7 ) was used in the blue arm and the 1700d grating ( @xmath8 ) was used in the red arm , with the central wavelengths set to 5200 and 8640 , respectively .
this configuration was chosen to include the calcium triplet lines at 8498 , 8542 and 8662 , for accurate velocity determination and proxy metallicities , as well as the swathe of iron and magnesium lines around 5200 for accurate metallicity measurements .
the astrometric positions of the targets , taken from the 2mass point source catalogue , have accuracies of @xmath9 .
the observations were taken over two observing runs , on 58 and 1419 june 2008 , with an average seeing of @xmath10 .
we obtained 3@xmath111200-second exposures per fibre configuration , to obtain s / n@xmath12 , with several configurations per cluster ( see table [ observed stars ] for actual numbers of spectra obtained for each cluster note that for m68 , 3 configurations were obtained in june 2008 , and 5 further configurations were obtained during a separate service run in may 2009 ) . in this study
we also made use of data for m30 obtained in a previous observing run in 2006 ( originally published by * ? ? ? * ) .
flat fields , arc lamp exposures and @xmath1325 dedicated sky fibres were used for data reduction and calibration .
fear , cuar , cuhe , cune and thar arc lamps were used to ensure accurate wavelength calibration .
data reduction was performed using the 2dfdr pipeline , designed specifically for the reduction of aaomega data .
the efficacy of the pipeline has been checked with a comparison of individual stellar spectra .
we aimed to observe several hundred stars from each of four gcs , namely m22 ( ngc 6656 ) , m30 ( ngc 7099 ) , m53 ( ngc 5024 ) and m68 ( ngc 4590 ) .
these four targets were selected by several criteria : they are bright ( @xmath14 ) , nearby ( @xmath15kpc ) and have radial velocities @xmath16@xmath17 from the peak velocity given by the besanon galaxy model in that direction , for ease of extracting the cluster members from galactic contaminants .
one cluster ( m30 ) was included specifically because it was one of the targets of @xcite , and it satisfied all other selection criteria .
this allows a comparison with the results of that study .
.clusters selected for the current survey , based on the criteria outlined in the text , with the total number of fibre configurations , and spectra , obtained during the run .
note that the number of spectra quoted here is the total number of spectra obtained before galactic contaminants were removed .
for m68 , 3 configurations were observed on the two observing runs in june 2008 and 5 subsequent configurations were observed during a service run in may 2009 . [ cols="^,^,^",options="header " , ]
one important indicator of large quantities of dm in a dynamical system is a large mass - to - light ratio ( i.e. m / l@xmath181 ) .
furthermore , this dm will cause the stars to have large accelerations , and therefore inherently higher maximal velocities , and hence a larger velocity dispersion . an obvious way to determine whether or not a pressure - supported object like a gc is dark matter dominated is to measure its m / l@xmath19 .
of course , this requires a surface brightness profile , for which we have used those by @xcite .
these profiles were converted to solar luminosities per square parsec .
the density profile was then calculated using ( again , see * ? ? ?
* ) : @xmath20 the m / l@xmath21 profiles are shown in figure [ m / l ] , along with the m / l@xmath21 ratios .
the vertical line shows the value of @xmath22 . because the central regions of gcs are highly concentrated , particularly in post - core - collapsed clusters , it is difficult to accurately determine the mass in the core .
the m / l@xmath21 ratios have , therefore , only been calculated for radii greater than @xmath22 .
@xcite showed that ultra - compact dwarf galaxies ( ucds ) follow the same luminosity velocity dispersion relation as old gcs . since it has also been shown that ucds with a dynamical m / l@xmath21 up to 5 do not require dark matter to explain these mass - to - light ratios ( e.g. * ? ? ?
* ; * ? ? ?
* ) , we see no requirement for any of our clusters , apart from m53 , to contain any dark matter component .
@xcite determined that a m / l@xmath23 for ucds may indicate some dark matter content , and since we estimate m / l@xmath24 for m53 , this may indicate a small dark matter component in this cluster .
note that if we take the lower limit of our uncertainties for the mass - to - light ratio of m53 we reach the regime where dark matter is not required , and none of our clusters show m / l@xmath25 , indicating dm is not dominant .
note that we have produced m / l@xmath21 _ profiles _ rather than simply deriving a single m / l@xmath21 value from the central velocity dispersion and central surface brightness .
we suggest that determining m / l@xmath21 profiles is preferable because it describes the m / l@xmath21 of the entire cluster outside the core , rather than just its core .
this is especially true for post - core - collapse clusters , such as m30 , in which the core masses are highly uncertain .
furthermore , m30 is the only cluster in our sample known to have a collapsed core ( e.g. * ? ? ?
this cluster also has a maximum in its m / l@xmath21 profile at about the scale radius @xmath22 , which may be due to mass segregation .
the most massive stars are known to fall toward the core during the evolution of the cluster ( see * ? ? ? * for a review of mass segregation processes in gcs ) .
the somewhat less pronounced maximum in the m / l@xmath19 profile of m68 may be the first evidence that this cluster is currently undergoing core collapse .
it has been argued that a central concentration parameter @xmath26 may indicate a collapsed core in certain gcs , and for m68 @xmath27 @xcite .
note that , despite our calculation of the m / l@xmath21 ratio being restricted to @xmath28 , the profiles should be accurate for @xmath29 , the core radius , or @xmath30 .
therefore , we argue that the peak in m / l@xmath21 is real , although a larger sample of cluster members for @xmath31 will be required to test this and , therefore , to show whether m68 is truly undergoing core collapse .
we have studied four gcs to determine their velocity dispersion profiles , and found that neither dm nor modified gravity are required to reconcile the kinematic properties of these particular clusters .
m53 may contain a small dm component , indicated by a m / l@xmath23 , but it is not dominant . within the uncertainties ,
the m / l@xmath21 ratio of m53 is still consistent with little , or no , dm content .
the dynamics of all four clusters are well described by a purely analytic @xcite model , which indicates that newtonian gravity adequately describes their velocity dispersions , and shows no breakdown of newtonian gravity at @xmath2ms@xmath1 , as has been claimed in previous studies .
furthermore , the plummer model was used to determine the total mass , scale radius ( the radius at which half the mass is contained ) , and the m / l@xmath19 profile for each cluster .
we find that none of our clusters have m / l@xmath25 , another indication that dm is not dominant . within the uncertainties ,
our estimated cluster masses all match those in the literature ( e.g. * ? ? ?
* ) , and the same is true for the m / l@xmath21 ratios ( e.g. * ? ? ?
we have produced m / l@xmath19 _ profiles _ , rather than quoting a single value based on the central velocity dispersion and central surface brightness .
this method is preferred because it describes the m / l@xmath21 of the entire cluster , rather than only its core ; this is particularly important for post - core - collapsed gcs for which the central mass is highly uncertain .
because the only known of our clusters to have a collapsed core ( m30 ) shows a peak in its m / l@xmath21 profile at approximately @xmath22 , this may be an indication of the resulting mass segregation . if this is the case , then we may have the first evidence that m68 , which also exhibits a m / l@xmath32 peak near @xmath22 and has a central concentration of 1.91 @xcite , is currently undergoing core collapse .
further observations are required , both photometric and kinematic , to confirm this .
another important result from this study is the measured rotations of our clusters . of the four clusters studied here ,
one exhibits an obvious rotation , namely m22 , with an axis of rotation approximately north - south .
m68 may be rotating as well and additional data are required to confirm this .
m53 and m30 do not show any rotation to the level of 0.5kms@xmath33 and 0.8kms@xmath33 respectively .
while these results are strongly indicative of the current picture of globular clusters being dark matter poor , and their dynamics can be explained by standard newtonian theory , we are currently undertaking robust dynamical modelling of gc systems to fully address the question of the influence of non - newtonian physics in explaining our observations .
this project has been supported by the anglo - australian observatory , the australian research council and the hungarian otka grant k76816 .
gregory , p. c. 2005 , bayesian logical data analysis for the physical sciences : a comparative approach with ` mathematica ' support . edited by p. c. gregory .
isbn 0 521 84150 x ( hardback ) ; qa279.5.g74 2005 519.542 dc22 ; 200445930
. published by cambridge university press , cambridge , uk , 2005 | testing newtonian gravity in the weak - acceleration regime is vital to our understanding of the nature of the gravitational interaction .
it has recently been claimed that the velocity dispersion profiles of several globular clusters flatten out at large radii , reminiscent of galaxy rotation curves , even though globular clusters are thought to contain little or no dark matter .
we investigate this claim , using aaomega observations of four globular clusters , namely m22 , m30 , m53 and m68 .
m30 , one such cluster that has had this claim made for its velocity dispersion , was included for comparison with previous studies .
we find no statistically significant flattening of the velocity dispersion at large radii for any of our target clusters and therefore we infer the observed dynamics do not require that globular clusters are dark matter dominated , or a modification of gravity .
furthermore , by applying a simple dynamical model we determine the radial mass - to - light profiles for each cluster .
the isothermal rotations of each cluster are also measured , with m22 exhibiting clear rotation , m68 possible rotation and m30 and m53 lacking any rotation , within the uncertainties .
[ firstpage ] gravitation - galaxy : globular clusters : individual - stellar dynamics |
the supernova remnant g347.30.5 ( rosat catalog source name : rx j1713.73946 ) is one of a growing number of supernova remnants ( snrs ) showing nonthermal x - ray emission attributed to shock accelerated electrons . in some cases ,
the nonthermal emission is evidenced by continuum emission extending to @xmath310 kev or above , usually superimposed with weak lines , or , as with sn1006 , g347.3 - 0.5 , and g266.2 - 1.2 , by a complete lack of line emission in the observed x - ray band . in all cases where
the emission is not associated with a compact object , the nonthermal x - rays have been interpreted as synchrotron emission from shock accelerated tev electrons ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . if this is true , the existence of these tev electrons has important implications , not only for the production of cosmic rays , but also for the thermal properties of the shock heated , x - ray emitting gas in snrs .
here we compare broad - band observations of g347.30.5 with results from a model combining hydrodynamic simulations of snr evolution ( e.g. , * ? ? ? * and references therein ) with nonlinear diffusive particle acceleration ( e.g. , * ? ? ?
the observations combine new radio data from the australia telescope compact array ( atca ) with existing x - ray @xcite and @xmath0-ray data @xcite .
our broad - band comparison of observations and model fits is a powerful way to constrain model parameters , but uncertainties from directly comparing radio , x - ray , and @xmath0-ray observations are unavoidable and should be kept in mind . in an attempt to limit errors from uncertainties in emission volumes , we only consider observations from the northwest portion of [email protected] the center of the tev @xmath0-ray response function coincides with the brightest radio and x - ray emission .
previous work ( e.g. , * ? ? ?
* ) indicates that the steepness of the nonthermal x - ray synchrotron spectrum in shell - like snrs results because these photons are generated by electrons in the steeply falling , high - energy tail of the distribution . the actual maximum electron energy , however , depends on the magnetic field strength , @xmath5 , in the emission region . if these electrons are accelerated by the first - order fermi shock acceleration mechanism as we assume , it is almost certain that the same shocks accelerate ions , but no clear evidence of superthermal ions has yet been obtained ( i.e. , no pion - decay feature has been unambiguously identified at @xmath6 mev in snrs ) . until a pion - decay feature
is observed , all information on the energetic ions must be obtained through modeling . in shocks accelerating particles efficiently , the unseen ions dominate the shock dynamics and largely control the electron emission features ( e.g. , * ? ? ?
* ; * ? ? ?
our main results are : ( i ) a spherically symmetric , wind - shell model with efficient acceleration can produce broad - band continuum spectra in excellent agreement with the observations of g347.3 - 0.5 ; ( ii ) test - particle models of [email protected] ( i.e. , ones where less than a few percent of the total energy ends up in relativistic particles ) can be excluded with good confidence ; ( iii ) the maximum electron energy in [email protected] is well below @xmath7 ev .
thus , [email protected] joins several other snrs now known _ not _ to produce electrons above @xmath7 ev ( e.g. , * ? ? ?
* ) , well below the so - called knee in the galactic cosmic - ray spectrum at @xmath2 ev .
the maximum proton energy depends somewhat on whether or not the ambient magnetic field is compressed by the shock .
if the field is compressed as the density , the maximum proton energy is approximately the same as the electron @xmath8 . in the other extreme , where no compression of the field occurs
, we find that g347.3 - 0.5 could currently be producing protons to @xmath9 tev and fe@xmath10 to @xmath2 ev .
unless substantial magnetic field amplification occurs , [email protected] not produce higher energy protons as it ages ; and ( iv ) our main conclusions concerning maximum particle energies and acceleration efficiency , are relatively insensitive to uncertainties stemming from the morphology of [email protected] .
there are strong arguments for a snr origin of cosmic rays based mainly on ion composition ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and total energy requirements ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the observations of nonthermal x - ray emission showing that tev electrons are produced in snrs adds support for this argument .
however , there is still no direct evidence that ions are generated in particular snrs in a fashion consistent with cosmic - ray observations . the low maximum proton energy inferred for
g347.3 - 0.5 may present a problem for the scenario that snrs are the main sources of galactic cosmic rays below the knee and suggest that the cosmic - ray proton component above @xmath11 ev might come from some subset of snrs substantially different from [email protected] .
alternatively , young snrs which only produce @xmath12 ev particles when they are bright in x - rays , may somehow be able to produce higher energy particles at later times , perhaps by amplifying the magnetic field ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we note that the high magnetic fields inferred for cas a make it likely that this particular remnant _ does _ produce @xmath2 ev protons ( e.g. , * ? ? ?
[email protected] was observed with the _ advanced satellite for cosmology and astrophysics _ ( asca ) on 1997 march 25 for a total of 50 ks using three pointings of 10 ks duration and one ( in the fainter southeast region ) of 20 ks duration @xcite . unlike the line - dominated x - ray spectra typical of snrs
, the spectrum for [email protected] is observed to be featureless and to extend to at least 10 kev , where the asca effective area becomes very small .
a power law spectrum provides an excellent fit to the x - ray data , with some indication that the spectral index varies between the brighter shell regions and the more diffuse interior and eastern regions .
the brightest x - ray emission occurs along the northwest limb of the snr ( see figure [ fig : g347image ] ) , where there is some indication that the remnant may be interacting with a molecular cloud or a stellar wind bubble shell , although the low inferred density for this region more likely indicates some density enhancement caused by a perturbation of the pre - sn wind .
the unabsorbed flux density from this region of the remnant is @xmath13 with a power law photon index of @xmath14 described here are new radio observations of g347.30.5 made with the australia telescope compact array ( atca ) , a six - element synthesis telescope located near narrabri , nsw , australia .
observations at 1.36 ghz were carried out in january , march , and april 1998 , in each case in a different array configuration to give maximal @xmath15 coverage .
the field was imaged in a 10-point mosaic , to ensure that the entire extent of the snr , as well as nearby bright sources , were in the field - of - view .
the interferometric visibilities were edited to remove corrupted data and interference .
calibrations were then applied to solve for the time - variable atmospheric gains above each antenna and for the polarization leakage in the receivers .
an absolute flux density scale was applied to the data using observations of pks b1934638 , for which a flux density of 15.0 jy was assumed at a frequency of 1.4 ghz .
the visibilities from all 10 pointings were then inverted to produce a single image , which was deconvolved using the mosaiced maximum entropy approach described by @xcite .
the final image was then produced by applying a gaussian restoring beam of dimensions @xmath16 .
the sum of the flux from the two bright filaments in the northwest region of the remnant ( indicated in figure [ fig : g347image ] and labeled as arc 1 and 2 in * ? ? ?
* ) , corresponding to that from which the x - ray flux above was measured , is @xmath17 . as noted in @xcite ,
while arc 2 ( the right - most arc in figure [ fig : g347image ] ) is askew from the perceived snr shell , it does overlap an x - ray filament .
arcs 1 and 2 have approximately the same radio flux .
g347.3 - 0.5 has also been detected at tev energies with the cangaroo 3.8 m atmospheric imaging erenkov telescope @xcite .
as with the x - ray and radio observations , the @xmath18 detection is from the nw rim of the remnant ( indicated with a circle in figure [ fig : g347image ] ) and has an estimated flux of @xmath19 } \pm 1.6 \mbox{[systematic ] } \times 10^{-12}$ ] photons @xmath20 s@xmath21 at energies @xmath22 tev .
the fluxes from these three detections are shown in figures [ fig : broadband ] and [ fig : vary ] where we have increased the observational error bars on the radio and tev points by a factor of two ( light - weight bars ) to compensate for uncertainties in the relative emission volumes .
in particular , the increased radio error bar will accommodate the possibility that arc 2 is not associated with the snr shell and thus none of our results will change if this is the case .
collisionless shocks are known to accelerate particles efficiently . in the heliosphere , where shocks are observed directly with spacecraft
, there is clear evidence that the quasi - parallel earth bow shock ( with a sonic mach number , @xmath23 ) can place 10 - 30% of the solar wind kinetic energy flux into superthermal particles ( e.g. , * ? ? ?
interplanetary shocks ( ipss ) also accelerate ambient particles but generally with lower efficiencies due to their lower mach numbers ( generally @xmath24 for ipss ) ( e.g. , * ? ? ?
however , on at least two occasions , exceptionally strong ipss have been observed to accelerate particles with high efficiency @xcite .
hybrid plasma simulations showing direct injection and acceleration of ambient particles are consistent with these observations ( e.g. , * ? ? ?
* ) , as are convection - diffusion models ( e.g. , * ? ? ?
compared to shocks in the heliosphere , it might be presumed that the strong forward and reverse shocks in young snrs are much more efficient . as far as we can tell
, there are no important differences in typical plasma parameters ( i.e. , density , magnetic field , temperature , composition ) between the solar wind and the ambient ism .
the only real differences come from the much larger shock spatial and time scales and much higher flow speeds and mach numbers in snrs . in general
, these factors should increase the acceleration efficiency substantially over heliospheric shocks , but some qualifications must be made . at quasi - parallel shocks
( those where the upstream magnetic field makes an angle , @xmath25 , of less than about @xmath26 with the shock normal ) , injection from the shocked heated plasma should occur easily and the fraction of incoming ram kinetic energy flux going into superthermal particles increases with mach number and shock size and/or age ( as measured in particle gyroradii or gyroperiods ) .
if however , the fraction of shocked heated particles injected is small enough , high mach number , test - particle solutions can result ( see * ? ? ?
* ) . in quasi - perpendicular shocks ( ones
where @xmath27 ) , injection and acceleration of shock heated thermal particles is suppressed above some mach number unless strong turbulence is present ( see * ? ? ? * for example ) .
highly oblique interplanetary shocks are observed to accelerate the ambient solar wind ( even without pickup ions , * ? ? ?
* ) , but these are low mach number shocks . as the mach number increases , injection becomes more likely if the turbulence is strong enough for cross - field diffusion in the downstream plasma to be important . while the shock obliquity will surely vary around a snr and some portions of the shocks
may be quasi - perpendicular with low injection rates , other regions should be parallel enough for efficient injection and acceleration to occur .
radio emission is a sure sign that relativistic electrons are present and flux levels , sharp radio edges , etc .
, show that these particles are locally accelerated in young snrs rather than just compressed galactic cosmic - ray electrons ( e.g. , * ? ? ?
the shock acceleration and evolutionary models we use here ignore effects from oblique magnetic fields , other than those that are mimicked by a low injection efficiency .
we do assume , however , that the magnetic field responsible for synchrotron radiation can be compressed by the shock .
for a recent discussion of how the injection efficiency may vary with @xmath25 in snrs , see @xcite .
the nonlinear effects in efficient shock acceleration are of three basic kinds : ( i ) the self - generation of magnetic turbulence by accelerated particles , ( ii ) the modification ( i.e. , smoothing ) of the shock structure by the backpressure of accelerated particles , and ( iii ) the increase in shock compression ratio as relativistic particles are produced and some high - energy particles escape from the system .
briefly , ( i ) occurs when counter - streaming accelerated particles produce turbulence in the upstream magnetic field which amplifies as it is convected through the shock .
this amplified turbulence results in stronger scattering of the particles , and hence to more acceleration , quickly leading to saturated magnetic turbulence levels near @xmath28 in strong shocks .
this is the so - called bohm limit where the mean free path is of the order of the gyroradius , and in this limit , the distinction between parallel and oblique shocks blurs .
the wave - particle interactions produce heating in the shock precursor which may be observable and which lowers the overall acceleration efficiency . effect ( ii ) occurs because the accelerated population presses on the upstream plasma and slows it .
an upstream precursor forms , in which the flow speed ( in the shock frame and in the absence of instabilities ) is monotonically decreasing . since particle diffusion lengths are increasing functions of momentum , high momentum particles sample a broader portion of the flow velocity profile , and hence experience larger effective compression ratios than low momentum particles . consequently , higher momentum particles have a flatter power - law index than those at lower momenta ( producing a concave upward spectral curvature ) and can dominate the pressure in a non - linear fashion .
we call the ratio of densities spanning the entire upstream precursor the overall compression ratio , @xmath29 , to distinguish it from the small scale subshock compression ratio , @xmath30 , where most of the shock heating occurs . for a more complete discussion of the nonlinear effects of diffusive shock acceleration , see @xcite .
the conservation of mass , momentum , and energy fluxes , i.e. , the rankine - hugoniotrelations , yield the standard expression for the compression ratio in shocks where the magnetic field is not dynamically important and an insignificant fraction of energy is placed in superthermal particles ( i.e. , test - particle shocks ) : @xmath31 where @xmath32 is the sonic mach number , @xmath0 is the ratio of specific heats , @xmath33 is the unshocked density , @xmath34 the unshocked pressure , and @xmath35 is the shock speed . in test - particle shocks , @xmath36 .
the ratio of shocked to unshocked pressure follows : @xmath37 if @xmath38 , @xmath39 and @xmath40 for a electron - proton plasma where the downstream temperatures are equal , the post - shock proton temperature in this test - particle limit , @xmath41 , is : @xmath42 where @xmath43 is boltzmann s constant and @xmath44 is the proton mass .
young snr shocks are expected to have high mach numbers and if the test - particle situation is assumed with @xmath45 , we expect @xmath46 .
however , when shocks accelerate particles the overall compression ratio increases above test - particle values ( even in non - radiative shocks ) for two reasons ( effect iii above ) .
first , as relativistic particles are produced and contribute significantly to the total pressure , their softer equation of state makes the shocked plasma more compressible ( @xmath47 ) .
second , and most important , is that particles escape from strong shocks draining energy flux which must be compensated for by ramping up the overall compression ratio to conserve the fluxes .
just as in radiative shocks , this is equivalent to @xmath48 .
for strong injection , it can be shown @xcite that @xmath29 is an ever increasing function of mach number , i.e.,@xmath49 where @xmath50 is the alfvn mach number ( see also , * ? ? ?
* ; * ? ? ?
simultaneously with @xmath29 increasing , the subshock compression ratio , @xmath30 , which is mainly responsible for heating the gas , must drop below the test - particle value .
thus , high compression ratios are accompanied by low post - shock temperatures ( in efficient shocks , the post - shock proton temperature can easily be @xmath51 ) .
if injection is weak enough , however , equation ( [ eq : machsonic ] ) does not apply and high mach number shocks do not accelerate enough particles to cause the total compression ratio to increase above @xmath52 ( see * ? ? ? * for the parameters describing this condition ) .
the most important parameters associated with nonlinear shock acceleration are the mach numbers ( i.e. , the shock speed , @xmath35 , pre - shock hydrogen number density , and preshock magnetic field , @xmath53 ) , the injection efficiency , @xmath54 ( i.e. , the fraction of total protons which end up with superthermal energies ) , and the maximum proton energy produced , @xmath55 .
as described in @xcite , our model includes alfvn heating in the precursor which reduces the efficiency compared to adiabatic heating .
we refer the reader to @xcite for a discussion of our treatment of electron injection and acceleration .
@xcite suggest that the forward shock of [email protected] still within the pre - sn stellar wind bubble with part of it interacting with denser material .
we describe this evolution with a spherically symmetric , wind - shell model with a pre - sn density distribution as shown in the top panel of figure [ fig : presn ] .
the initial density profile in the ejecta has a power law distribution , @xmath56 , with a constant density plateau at small radii , as described in @xcite .
since we only consider the forward shock at fairly late stages in the snr evolution , our results are insensitive to @xmath57 and the position of the plateau .
they are also insensitive to whether a power - law or exponential ejecta mass density distribution ( e.g. , * ? ? ?
* ) is assumed . for concreteness , we use @xmath58 in all of our models . as in @xcite , the density of the pre - sn wind
is given by @xmath59 where @xmath60 , @xmath61 is the pre - sn mass loss rate , and @xmath62 is the constant wind speed . at some distance from the explosion ,
we place a shell of uniform density , @xmath63 our spherically symmetric representation of the density enhancement suggested by @xcite . here
, @xmath64 is the number fraction of helium and @xmath65 is the proton number density in the shell .
while the radio image of [email protected] suggests a shell with the strongest emission coming from the nw rim , the complex morphology makes an exact determination of the diameter somewhat ambiguous .
some radio and x - ray emission extends across @xmath66 , but a smaller @xmath67 diameter ring - like structure is also present . as a compromise , we adopt an angular diameter of @xmath68 , and to further limit parameters , we perform all of our fits assuming a distance of @xmath69 kpc @xcite . with these assumptions ,
the outer shock is at a radius @xmath70 pc and we take ( somewhat arbitrarily ) the inner radius of the shell to be @xmath71 pc , giving a shell thickness approximately equal to the thickness of radio arc 1 , i.e. , @xmath72 .
fortunately , none of our conclusions depend critically on these particular values .
furthermore , as long as @xmath73 , our results are insensitive to @xmath74 and we use @xmath75 gm cm@xmath21 which is consistent with a wind extending to 34 pc .
@xmath77 yr@xmath21 and @xmath78
km s@xmath21 ( giving @xmath75 gm cm@xmath21 ) , @xmath79 would exist in a spherical wind extending to 34 pc . with this @xmath80
, the wind bubble would have taken @xmath81 yr to form and could still not have stagnated in a low density interstellar medium . ]
the evolution is calculated with the virginia hydrodynamics ( vh-1 ) time - dependent hydrodynamic numerical code in one dimension where it is implicitly assumed that dynamic effects from magnetic fields are negligible . for a description of this code ,
see @xcite and references therein . with the shock speed and other parameters obtained with the hydro code ,
we calculate the electron , proton , and helium spectra and with them the synchrotron , bremsstrahlung , inverse - compton , and pion - decay photon continuum spectra .
we then compare the photon emission to the observations and vary parameters until a satisfactory fit is obtained . of necessity ,
our model is approximate in some important ways .
first , we assume the snr is spherically symmetric when , in fact , the emission is strongest from one section of the rim where the shock is likely interacting with denser material .
we compensate for this at the most basic level by defining the emission volume , @xmath82 , as a fraction , @xmath83 , of the shell volume encompassed by the forward shock , i.e.,@xmath84 from figure [ fig : g347image ] we estimate that @xmath85 .
second , we use a one - zone model where the particle spectra and photon emission are assumed to be generated by a shock of constant speed during the time the forward shock is in the shell . we ignore any evolution of the spectra or the shock during the time the shock is in the shell and the shock speed used for acceleration is the average speed in the shell as determined by the hydro code .
furthermore , we neglect any contributions from the reverse shock or reflected shocks which occur when the forward shock enters the shell
. these should be excellent approximations since the forward shock speed does nt vary greatly while in the shell ( middle panel figure [ fig : presn ] ) , particles produced earlier in the low density wind or ejecta have been reduced by adiabatic losses , and reflected shocks in the shell are weak compared to the forward shock .
the evolution of the snr will be influenced by the acceleration process through the increased compression ratios of the forward and reverse shocks and by the escape of high energy cosmic rays .
we have approximated this effect by performing our hydrodynamic calculations with values of the effective ratio of specific heats , @xmath86 , as described in detail in @xcite .
the effect on the forward shock speed is shown for typical parameters in the bottom panel of figure [ fig : presn ] .
the solid curve is for @xmath87 ( @xmath88 ) , the dashed curve is for @xmath89 ( @xmath90 ) , and the dotted curve is for @xmath91 ( @xmath92 ) .
when @xmath87 , the shock reaches 40 pc after @xmath93 yr , for @xmath94 , it reaches 40 pc after @xmath95 yr , and for @xmath91 , 40 pc is reached after @xmath96 yr . while globally changing @xmath97 only approximates the effects of particle acceleration on the evolution ( particle escape is not explicitly included and the evolution of @xmath97 is not modeled , for example ) , we do not believe errors from this approximation are significant since all of the models we discuss below have compression ratios less than @xmath98 when the forward shock is in the shell .
as indicated in figure [ fig : presn ] , at @xmath99 pc @xmath35 varies by less than 25% for @xmath100 . despite this small change in @xmath35 ,
the particle spectra produced by a shock with @xmath101 are still strongly nonlinear and differs importantly from test - particle spectra as shown in the fits below .
finally , our model ignores x - ray _ line _ emission which , for a thermal plasma , can be much more intense than the pure bremsstrahlung component .
for example , the 0.210 kev flux for a @xmath102 kev raymond - smith model @xcite is @xmath103 times higher than for a thermal bremsstrahlung model with the same temperature and emission measure .
this discrepancy decreases with temperature , dropping to @xmath52 at @xmath104 kev and @xmath105 at @xmath106 kev .
if the featureless x - ray spectrum in [email protected] is synchrotron emission , the underlying bremsstrahlungcontinuum which we calculate must be low enough so lines , had they been calculated , would not appear .
= 0 most of the parameters used in our models are described in @xcite and/or shown here in table [ table-1 ] .
briefly : by 1 the explosion energy , @xmath107 , largely determines the overall dynamics and is typically assumed to be near @xmath108 erg .
the ejecta mass , @xmath109 , is unimportant for the ages and parameters we consider as long as it is less than about 3 @xmath77 .
if @xmath110 , the reflected shock produced when the forward shock enters the shell can interact with the density peak behind the reverse shock producing complicated flow structures . to avoid these complications , we limit consideration to @xmath111 and do not list this parameter in the table . by 1 the magnetic field affects
the particle acceleration and largely determines the shape of the broad - band photon emission through its influence on synchrotron emission .
the field morphology is certain to be complex in any real snr and will vary with location ( see * ? ? ?
* for an excellent 2-d mhd study of shock - cloud interactions ) . to consider a range of possibilities in our idealized , spherically symmetric model , we assume the downstream field , @xmath5 , is either equal to the upstream ( unshocked ) field , i.e. , @xmath112 , or that @xmath5 is compressed along with the gas , i.e. , @xmath113
. the actual situation should lie somewhere in between .
we assume the unshocked proton temperature , @xmath114 , where @xmath115 @xmath116 is the upstream electron ( helium ) temperature .
our solutions are insensitive to the upstream temperature as long as it is below @xmath117 k. the shocked electron to proton temperature ratio , @xmath118 , is a free parameter which has not yet been deduced from first principles for collisionless shocks .
the electron temperature is an important parameter for thermal bremsstrahlungand x - ray line emission , but is not strongly constrained for [email protected] because no lines are seen . in any case , we assume @xmath119 in all of our models but note that if electron - proton equilibration timescales are long compared to dynamic timescales , @xmath118 could be considerably less than one , lowering the thermal bremsstrahlung to synchrotron ratio . by 1
the injection efficiency , @xmath54 , is the fraction of thermal protons that end up as accelerated particles .
this parameter influences the overall acceleration efficiency . by 1 the electron to proton ratio at relativistic energies , @xmath120 ,
is an arbitrary parameter in our models but is expected to lie between 0.01 and 0.05 if @xmath120 in galactic cosmic rays is typical of that produced by the strong shocks in young snrs .
this factor influences the normalization of the photon emission from electrons and is important for determining the pion - decay contribution to tev @xmath0-rays relative to inverse - compton .
the maximum energy cosmic rays obtain depends on the scattering mean free path , @xmath121 , which is assumed to be , @xmath122 where @xmath123 is a constant , @xmath124 is the gyroradius in si units , @xmath125 is the gyroradius at the maximum momentum , @xmath126 , and @xmath127 is a constant parameter .
small values of @xmath123 imply strong scattering and allow higher maximum proton energies in a given system .
we use the bohm limit ( @xmath128 ) in all of our models and note that while good fits to the observations can be obtained with larger @xmath123 , these models will have similar acceleration efficiencies with _
lower _ proton maximum energies than the models we show . by 1
finally , @xmath65 is the proton number density in the shell which models the density enhancement presumed to exist in the nw rim .
@xcite conclude , from the lack of thermal emission , that the mean density around the remnant is quite low . with large uncertainties
, they estimate upper limits on the ambient density of @xmath129@xmath130 @xmath131 , with a higher upper limit ( @xmath132 @xmath131 ) for the northwest rim .
crrrrrrrrrrr input & a & b & c & + @xmath107 [ @xmath108 erg ] & 2 & 3 & 4 & + @xmath53 [ @xmath133 g ] & 2.4 & 2.4 & 16 & + @xmath54 & @xmath134 & @xmath135 & @xmath134 & + @xmath120 & 0.03 & 0.05 & 0.04 & + @xmath65 [ @xmath131 ] & 0.015 & 0.02 & 0.008 & + output & & + @xmath35 [ km s@xmath21 ] & 980 & 1100 & 2100 & + @xmath138 [ yr ] & 9000 & 8000 & 5000 & + @xmath139 [ yr ] & 5000 & 4700 & 2200 & + @xmath140 & 60 & 70 & 130 & + @xmath50 & 27 & 35 & 6.3 & + @xmath29 & 6.1 & 4.1 & 4.3 & + @xmath30 & 3.0 & 4.0 & 3.4 & + @xmath97 & 1.4 & 1.65 & 1.6 & + @xmath5 [ @xmath133 g ] & 15 & 9.6 & 16 & + @xmath141 [ kev ] & @xmath142 & @xmath143 & @xmath144 & + @xmath145 [ kev ] & @xmath146 & @xmath143 & @xmath147 & + @xmath55 [ tev ] & 14 & 16 & 70 & + @xmath148 [ tev ] & 14 & 16 & 32 & + @xmath149 & 0.55 & 0.03 & 0.20 & + @xmath150 & @xmath151 & @xmath152 & @xmath153 & + flux & & + @xmath83 & 0.16 & 6.8 & 0.19 & + [ numbers ] the output values are also given in table [ table-1 ] .
the shock speed , @xmath35 , is the average obtained from the hydro simulation over the shell thickness between @xmath154 and 40 pc .
the age of [email protected] is unknown , but @xcite have used emission measures to estimate a range @xmath155 yr . the value in the table is that necessary to yield an outer shock radius of @xmath156 pc consistent with an angular diameter @xmath157 at a distance of 6 kpc .
the time the shock spends between 34 pc and 40 pc is @xmath139 .
given @xmath35 and the density , temperature , and magnetic field in the shell , the sonic and alfvn mach numbers are determined .
the above parameters allow the nonlinear shock model to determine @xmath29 , @xmath30 , and @xmath97 , as well as the proton , electron , and helium spectra .
note that consistent values of @xmath29 and @xmath97 are found by iterating the hydro simulation with the particle acceleration calculation .
the shocked magnetic field is @xmath5 , and the shocked proton temperature is @xmath158 . the shocked proton temperature for the corresponding test - particle shock , @xmath41 , is also given .
the particle momentum distributions , @xmath159 , are calculated as in @xcite with the turnover at the highest energies described by : @xmath160 } \ .\ ] ] here , @xmath127 is the same as in eqn .
( [ eq : mfp ] ) and @xmath126 is determined by setting the acceleration time equal to @xmath139 , or by setting the diffusion length of the highest energy particles equal to 1/10 @xmath161 , whichever gives the lowest @xmath126 ( see * ? ? ?
it turns out that a relatively slow turnover ( i.e. , @xmath162 ) is required to obtain the proper x - ray slope and to join the radio and x - ray observations .
a similar effect was required to match the intensity and shape of the x - ray observations in sn1006 @xcite , and we take @xmath163 in all of our models . the electron and proton energies corresponding to @xmath126 are @xmath148 and @xmath55 respectively and are listed in the table .
the acceleration efficiency , @xmath149 , is defined as , @xmath164 where @xmath165 and @xmath166 are the fractions of energy flux in relativistic protons and escaping protons , respectively . the spectral index at 2 ghz , where @xmath167 , is shown in the next to last row of table [ table-1 ] .
radio data has been collected at 2.4 ghz from [email protected] but has not yet been analyzed .
when the 2.4 ghz flux is combined with the existing 1.4 ghz flux , the inferred spectral index can be used as a check of our models and may help constrain parameters . finally , the last row in the table is the fraction of the shell volume , @xmath83 , producing the emission seen from the nw limb . based on equation ( [ eq : emission ] ) ,
models with @xmath168 are unacceptable .
the broad - band emission from [email protected] allows us to constrain the parameters of the shock model to fairly narrow ranges . in the top panel of figure [ fig : broadband ] ( model @xmath169 ) , we show a satisfactory fit to the broad - band observations with @xmath170 .
the parameters for this model and two others discussed below are listed in table [ table-1 ] .
the four components of the emission are labeled and it s clear that the synchrotron emission produces the radio and x - ray emission , while the tev point is fit with inverse - compton photons from electrons scattering off the 2.73 k background radiation .
the value @xmath171 is consistent with the morphology and @xmath172 @xmath131 is consistent with the upper limit estimate of @xcite .
we note that using a lower @xmath107 would force a smaller @xmath65 and a larger @xmath83 ; @xmath173 erg is clearly favored .
using model @xmath169 as a reference , we investigate how sensitive the fit is to particular parameters . in figure
[ fig : vary ] we show variations in @xmath53 , @xmath65 , @xmath107 , and @xmath54 . in each panel
we keep all parameters except the one labeled fixed to the values used in model @xmath169 . for all models
, we have normalized the synchrotron emission to the flux in the x - ray band and plot the sum of the four photon components .
the solid curve in each panel is model @xmath169 . in the top panel ,
the dotted curve is for @xmath174 @xmath133 g , the solid curve is for @xmath175 @xmath133 g , and the dot - dashed curve is for @xmath176 @xmath133 g . in all cases ,
the number in parentheses is @xmath83 .
the dependence on @xmath53 is quite strong and the shape of the broad - band emission clearly sets upper and lower limits on @xmath53 .
furthermore , the @xmath174 @xmath133 g result can be excluded because @xmath177 ; a low @xmath178 produces too little flux .
the changes produced by varying @xmath65 are shown in the second panel of figure [ fig : vary ] .
the dotted curve is for @xmath179 @xmath131 , the solid curve is for @xmath172 @xmath131 , and the dot - dashed curve is for @xmath180 @xmath131 . clearly , the @xmath181 @xmath131 result is too flat in the x - ray band , while the @xmath180 @xmath131 result is somewhat too steep .
the value @xmath182 for @xmath180 @xmath131 , shows that this high density example produces too little flux .
this occurs because @xmath35 is low in this case ( 520 km s@xmath21 ) producing a low @xmath148 and forcing a higher normalization to match the high energy x - rays . in the third panel , we vary @xmath107 from @xmath183 erg ( dotted curve ) to @xmath184 erg ( solid curve ) to @xmath185 erg ( dot - dashed curve ) .
the low explosion energy gives an unacceptable fit because of the shape in the x - ray band and because @xmath186 .
however , for @xmath187 erg the shape is still reasonably good considering the uncertainties in the model and the normalization ( @xmath188 ) while low , can not be excluded . in the bottom panel of figure [ fig : vary ] we show results varying the injection efficiency .
the dotted curve is for @xmath189 , the dot - dashed curve is for @xmath190 , and the solid curve is model @xmath169 with @xmath191 .
the shape of the sum of the emission is virtually identical for @xmath190 and @xmath192 .
however , the intensity of the emission varies with @xmath54 and for @xmath190 , @xmath193 , while for @xmath191 , @xmath194 .
the constraint that the emission is only observed from a small fraction of the shell implies that @xmath195
. this will be true unless @xmath107 is considerably greater than @xmath184 erg .
the @xmath189 result , which is what would be expected if the shock _ did not _ accelerate particles efficiently , is unacceptable because of the overall shape and low emissivity ( i.e. , @xmath196 ) . to further investigate the limits on the acceleration efficiency ,
we choose @xmath54 low enough so only a few percent of the total energy flux ends up in relativistic particles and freely adjust other parameters to obtain a fit . with @xmath197 ( yielding @xmath198 )
we are able to obtain an approximate match to the shape of the observed emission with model @xmath178 in figure [ fig : broadband ] and table [ table-1 ] .
note that this model requires @xmath199 erg ; lower energies give poorer fits .
while the shape of the continuum emission is a reasonable match to the observations , the normalization requires @xmath200 , well above an acceptable value .
furthermore , the synchrotron emission from the test - particle shock is somewhat too steep to simultaneously match the radio and x - ray fluxes , and the ratio of synchrotron to bremsstrahlung continuum emission in the x - ray range is less than a factor of 3 .
if we further decrease @xmath54 , the synchrotron to bremsstrahlung ratio will drop and at some point x - ray emission lines would stand above the continuum flux in conflict with the observations .
this conclusion depends , of course , on @xmath118 and could be weakened by assuming @xmath201 . however , the relative acceleration efficiency of electrons and protons is also likely to scale as @xmath118 @xcite making it difficult to produce the observed radio and tev flux if @xmath118 is too small .
furthermore , if a flatter synchrotron slope is required to give a better match between the radio and x - rays ( as in model @xmath169 ) , this forces the shock to be more efficient ( to produce a larger @xmath29 ) and implies a larger @xmath54 . the overall flux , the lack of x - ray emission lines , and the radio to x - ray flux ratio all point toward efficient shock acceleration .
using this model we can get a firm lower limit on the acceleration efficiency by increasing @xmath54 until @xmath202 .
we find @xmath203 for @xmath204 , yielding a lower limit @xmath205 , i.e. , even with @xmath199 erg , the shock must place at least 25% of its kinetic energy flux in relativistic particles .
note that we took @xmath206 for this example .
a lower @xmath120 would imply a higher acceleration efficiency . in order to investigate the maximum proton energies that [email protected] can generate , we have produced model @xmath207 ( bottom panel of figure [ fig : broadband ] ) where we have chosen parameters ( including taking @xmath112 ) with the intention of maximizing @xmath55 .
very generally , diffusive shock acceleration yields the same maximum energy per _ charge _ for all particles ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , so the only way the proton and electron spectra can have different maximum energies is for losses ( synchrotron and/or inverse - compton ) to be important for the electrons . unless we have neglected an abundant supply of photons in addition to the 2.73 k background radiation , inverse - compton losses are not important and only when the magnetic field is large can @xmath55 be much greater than @xmath148 .
however , a large field produces a lower alfvn mach number which weakens the acceleration and , in order to fit the broad - band spectrum , a higher shock speed is required .
this can be achieved with a larger @xmath107 and/or a lower @xmath65 , and we have taken @xmath208 erg and @xmath209 @xmath131 for model @xmath207 . the observed radio flux sets the strength of the magnetic field in the emission region combined with the relativistic electron density , i.e. , the flux is loosely proportional or the density produces changes in the acceleration model and/or the snr evolution .
] to @xmath210 .
furthermore , the tev inverse - compton emission comes from the same electrons that produce the x - rays .
since the inverse - compton emission is independent of @xmath178 , it constrains the number of relativistic electrons .
the combination of these constraints allows us to limit @xmath55 and we find in our most optimistic case that @xmath211 tev . in figure
[ fig : spectra ] we show the momentum distribution functions , @xmath159 , for models @xmath169 , @xmath178 , and @xmath207 , where we have plotted the dimensionless quantity , @xmath212 , to emphasize the spectral curvature . in this representation ,
the test - particle result of @xmath213 is a horizontal line ( model @xmath178 ) .
the concave shape of @xmath159 ( at superthermal energies and below the cutoff ) is evident for models @xmath169 and @xmath207 , as is the large difference between the relativistic particle populations in these shocks with efficient acceleration compared to the inefficient shock @xmath178 . as noted above , a flat electron spectrum ( at energies greater than @xmath214 ) produced by a large @xmath29 ( @xmath215 in model @xmath169 )
allows a good match between the radio and x - ray observations .
the arrows in figure [ fig : spectra ] indicate the positions of @xmath55 and @xmath148 as determined from equation ( [ eq : expo ] ) with @xmath216 . in all cases
, @xmath159 starts to fall off exponentially at momenta less than @xmath126 and much below @xmath217 .
furthermore , if g347.3 - 0.5 continues to evolve uniformly with no magnetic field amplification , it will not produce higher energy particles as it ages ( for a more complete discussion of the time particles take to obtain their maximum energies , see * ? ? ?
* ; * ? ? ?
since the causes of magnetic field amplification are still uncertain , it remains possible that higher fields and therefore higher energy particles could be produced in the future .
magnetic field amplification apparently occurs in cas a ( e.g. , * ? ? ?
* ; * ? ? ?
* ) but if it is due to rayleigh - taylor instabilities in this young remnant as commonly assumed , it is unlikely that this mechanism would occur in [email protected] , which is much older .
g347.3 - 0.5 is one of a growing number of shell - type snrs showing nonthermal x - rays believed produced by synchrotron emission from tev electrons accelerated by the remnant shocks . in an attempt to have a consistent , broad - band spectrum , we have combined recent x - ray and tev @xmath0-ray observations of the remnant with new radio observations and have restricted all of the observations to the bright northwest rim of the shell . comparing a spherically symmetric , wind - shell model of snr evolution and efficient diffusive shock acceleration against the observations shows that the observed emission can be well fit across @xmath218 decades in energy with a single set of parameters all close to expected supernova and ism values .
the densities we find for the emission region ( @xmath219 protons @xmath131 ) are consistent with the upper limits found earlier by @xcite , and are below those expected for either a molecular cloud or a swept - up shell produced as the pre - sn wind interacts with the interstellar medium ( or with a slower red supergiant wind ) .
this suggests the shock is still within the low density bubble but has encountered a density enhancement perhaps caused by perturbations in the pre - sn wind .
we put limits on the acceleration efficiency and find that [email protected] is currently putting at least @xmath220% ( and more likely @xmath221% ) of the forward shock ram energy flux into relativistic particles ( i.e.,cosmic rays ) .
recently , @xcite have inferred efficient shock acceleration in snr e0102.2 - 7219 in the small magellanic cloud from the shock speed , determined from expansion measurements , and the postshock electron temperature , determined from x - ray line models ( unlike [email protected] , lines are visible above the synchrotron continuum in the post - forward shock gas ) .
they concluded that the extremely low postshock temperature can only be reconciled with the high shock speed if a sizable fraction ( @xmath221% ) of the shock energy goes into cosmic rays .
our results are fully consistent with this and with the important point that efficient particle acceleration causes a strong coupling between the broad - band emission from relativistic particles and the properties of the shocked heated x - ray emitting gas ( see * ? ? ?
we have verified that sedov models ( not shown ) with no density enhancement produce satisfactory nonlinear fits to the observed spectral shape with input parameters similar to those given here , but are inconsistent with the observed remnant diameter and distance .
( or even @xmath222 ) at @xmath223 kpc . only if @xmath224 kpc could standard models produce acceptable fits .
the wind - shell model allows the shock to propagate for large distances in the low density wind before encountering the higher density emission region .
] we conclude that the nonlinear aspects we infer , in particular the required high acceleration efficiency , do not depend significantly on the geometric details of the wind - shell model while the overall normalization does .
furthermore , we find that the maximum energies of the cosmic - ray electrons and protons are well below @xmath225 ev . while it has been known for some time that snrs emitting nonthermal x - rays do not produce _ electrons _ to energies near @xmath225 ev ( e.g. , * ? ? ?
* ) , the nonlinear effects we model make it possible to infer the proton distribution from photons emitted by electrons . for no magnetic field compression in the shock and other parameters adjusted to produce the highest possible @xmath55 and still give a reasonable fit to the observations , we find a proton maximum energy @xmath226 tev .
more realistic parameters with magnetic field compression yield @xmath227 tev . however , snr shocks will also accelerate heavier ion species such as he , c , o , and fe , and the maximum energy in diffusive shock acceleration scales as charge .
therefore , this snr could produce a power - law fe@xmath10 spectrum to @xmath2 ev . to illustrate
, we have estimated the fe@xmath10 spectrum produced by our model @xmath207 and placed it in the bottom panel of figure [ fig : spectra ] .
the normalization is determined by the cosmic abundance of iron ( @xmath228 iron nuclei for every proton ) plus the enhancement expected because iron is injected and initially accelerated when it is trapped in dust grains .
in fact , the normalization of iron relative to hydrogen in figure [ fig : spectra ] is just that shown in figure 8 of @xcite at @xmath229 kev / nucleon .
cosmic - ray observations in the @xmath230 ev range typically measure the total energy of the particles and have large uncertainties in differentiating individual ion species ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
nevertheless , a number of authors have produced models of this transition energy range adding together the fluxes from different ion species with cutoffs that scale as charge ( e.g. , * ? ? ?
* ; * ? ? ?
our high @xmath53 results are in general agreement with this work .
the authors thank j. blondin and the referee a. decourchelle for helpful comments .
this work was supported in part by the national aeronautics and space administration through contract nas8 - 39073 and grant nag5 - 4803 .
the australia telescope is funded by the commonwealth of australia for operation as a national facility managed by csiro . b.m.g .
acknowledges the support of nasa through hubble fellowship grant hf-01107.01 - 98a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa under contract nas 526555 . | the supernova remnant g347.30.5 emits a featureless power - law in x - rays , thought to indicate shock - acceleration of electrons to high energies .
we here produce a broad - band spectrum of the bright nw limb of this source by combining radio observations from the australia telescope compact array ( atca ) , x - ray observations from the advanced satellite for cosmology and astrophysics ( asca ) , and tev @xmath0-ray observations from the cangaroo imaging erenkov telescope .
we assume this emission is produced by an electron population generated by diffusive shock acceleration at the remnant forward shock .
the nonlinear aspects of the particle acceleration force a connection between the widely different wavelength bands and between the electrons and the unseen ions , presumably accelerated simultaneously with the electrons .
this allows us to infer the relativistic proton spectrum and estimate ambient parameters such as the supernova explosion energy , magnetic field , matter density in the emission region , and efficiency of the shock acceleration process .
we find convincing evidence that the shock acceleration is efficient , placing @xmath1% of the shock kinetic energy flux into relativistic ions . despite this high efficiency , the maximum electron and proton energies , while depending somewhat on assumptions for the compression of the magnetic field in the shock , are well below the observed ` knee ' at @xmath2 ev in the galactic cosmic - ray spectrum . |
optical clocks , with natural linewidth at the millihertz level , have demonstrated great improvements in stability and accuracy over the microwave frequency standards .
the research on optical atomic clocks have achieved remarkable progress in the past several years especially in single - ion optical clocks . all - optical atomic clock referenced to the 1.064 petahertz transition of a single trapped @xmath2hg@xmath1 ion
has been realized in 2001 @xcite .
two al@xmath1 ion optical clocks , operated at @xmath3s@xmath4 to @xmath5p@xmath4 clock transition with frequency near 1.121 petahertz and narrow natural linewidth of 8 mhz , have been constructed with fractional frequency inaccuracy of the order of magnitude @xmath6 @xcite .
although these single - ion optical clocks have reached unprecedented stability and accuracy , there are still some problems to be solved .
the observed linewidth of the clock transition is limited by the linewidth of the probe laser @xcite .
thus , narrow linewidth laser light source becomes the key factor of the performance of single - ion optical clocks .
since the proposal of active optical clock @xcite , a number of neutral atoms with two - level , three - level and four - level at thermal , laser cooling and trapping configurations have been investigated recently @xcite .
the potential quantum - limited linewidth of active optical clock is narrower than mhz , and it is possible to reach this unprecedented linewidth since the thermal noise of cavity mode can be reduced dramatically with the mechanism of active optical clock @xcite .
therefore , a laser light source based on active ion optical clock will be favorable to single - ion optical clocks .
one candidate is @xmath0yb@xmath1 ion . with about @xmath7 @xmath0yb@xmath1 ions in a paul trap @xcite ,
the population inversion can be realized as shown in fig . 1 .
the cooling laser and repumping laser at @xmath8s@xmath9 to @xmath8p@xmath10 and @xmath8s@xmath10 to @xmath8p@xmath9 transitions are used as pumping laser .
first , during the cycling @xmath11 to @xmath12 transition , there is a branching probability of @xmath13 for ions decaying to @xmath8d@xmath14 sublevel @xcite .
second , when the ions that leaked to @xmath8s@xmath10 are repumped to @xmath8p@xmath9 , ions will decay to @xmath8d@xmath15 sublevels .
then the population inversion is established between @xmath8d@xmath14 and @xmath8s@xmath10 sublevels .
the ions at @xmath8d@xmath16 can be repumped with a 935 nm laser at transition from @xmath8d@xmath16 to @xmath5d[3/2]@xmath9 and then decay to @xmath8s@xmath9 by spontaneous transition very quickly .
so this channel can be ignored when calculating the energy level population .
small leakage channels to other sublevels can be closed with additional repumping lasers if necessary .
the calculations of dynamical process of population inversion of @xmath0yb@xmath1 ions in a paul trap and lasing process with a fabry - perot type resonator are presented in detail .
we calculated the quantum - limited linewidth of active @xmath0yb@xmath1 ion optical clock and got it should be narrower than 1 mhz .
other proper candidate ions for active optical clock and ion laser are also discussed and compared with @xmath0yb@xmath1 in this paper .
considering the spontaneous decay rate of @xmath8p@xmath17 is @xmath18 mhz , it is reasonable to set the pumping rate @xcite from @xmath8s@xmath9 to @xmath8p@xmath10 @xmath19 .
so the decay rate to @xmath8d@xmath14 sublevel is @xmath20 .
then we can set the repumping rate of @xmath8s@xmath10 to @xmath8p@xmath9 transition @xmath21 to be of the order of magnitude @xmath22 and the repumping rate of @xmath8d@xmath16 to @xmath5d[3/2]@xmath9 sublevel to be @xmath23 .
the effective power is reduced to one third if only the 0 - 0 clock transition is effective . in order to depopulate ions at @xmath8d@xmath14 ( @xmath24 ) zeeman sublevels , each transition from these two zeeman sublevels @xmath8d@xmath14 ( @xmath24 ) to @xmath8s@xmath10 is also coupled with one mode of ion laser cavity .
based on the assumption that the ions have been trapped in a paul trap , and the theories of the interaction between the ions and light , the density matrix equations for @xmath0yb@xmath1 ions interacted with the cooling light and the repumping light can be written in rwa - approximation as follows : @xmath25 the subscript numbers of the density matrix correspond to different energy levels as shown in fig . 1 .
the diagonal elements mean the number of ions in corresponding states and off - diagonal elements mean coherence between two states .
the rabi frequency is @xmath26 , where @xmath27 is the electric dipole matrix between the @xmath8s@xmath9 state and @xmath8p@xmath10 state , @xmath28 is the electric strength of the cooling light .
@xmath29 , with the electric dipole matrix @xmath30 between the @xmath8s@xmath10 state and @xmath8p@xmath9 state and the electric strength @xmath31 of the repumping light .
@xmath32 and @xmath33 are frequency detunings of cooling and repumping light on the transition frequencies .
they are both set to be 0 .
@xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 , @xmath39 and @xmath40 are decaying rates related to the lifetimes of the corresponding energy levels described in fig . 1 .
@xmath41 is the cavity loss rate and @xmath42 is the photon number if there exists a cavity .
@xmath43 is the laser emission coefficient @xcite where @xmath44 is ion - cavity coupling constant and @xmath45 is the interaction time . without the cavity , @xmath41 and @xmath46
are both 0 . the effective decay rate of the channel , from 6p @xmath8p@xmath17 to 5d @xmath8d@xmath47 then repumped with 935 nm laser to @xmath5d[3/2]@xmath9 and decay to @xmath8s@xmath17 by spontaneous transition , is ignored in calculations since it is much smaller than @xmath34 .
for the convenience of reading , here we summarize all of the relevant parameters mentioned above in table 1 as follows .
* table 1 .
parameters related to + active ion optical clock * [ cols="^,^,^,^",options="header " , ] @xmath48 another value for discussion is @xmath49 .
@xmath50 without the cavity , the value is 0 . and
@xmath51.,width=283 ] without the cavity , the numerical solution results for the above equations are shown in fig . 2 .
assuming there are @xmath52 ions in the paul trap and a quarter of them are in the @xmath8s@xmath10 state at the beginning , the population inversion between @xmath8d@xmath14 and @xmath8s@xmath10 states is built up at the time scale of @xmath53s . from fig .
2 , it is obvious that under the action of cooling laser , the number of ions in the @xmath8s@xmath9 state decrease rapidly .
the lifetime of @xmath8p@xmath10 and @xmath8p@xmath9 states are much shorter than that of @xmath8d@xmath14 state , so ions accumulate in @xmath8d@xmath14 state and the population inversion is built up .
if we keep other conditions unchanged , the number of ions at the steady - state in @xmath8d@xmath14 sublevel is 10 times as much as that in @xmath8s@xmath10 sublevel for @xmath54 , and 20 times for @xmath55 . an optical resonant bad cavity
, whose linewidth of cavity mode is much wider than the linewidth of gain , could be applied to the lasing transition between the inverted states @xmath8d@xmath14 and @xmath8s@xmath10 , as the population inversion occurs .
the oscillating process could start up once the optical gain exceeds the loss rate .
the corresponding density matrix equations together with the equations of emitted photons(@xmath42 ) from the total ions(@xmath56 ) inside the cavity can be written as above .
we use a fabry - perot resonator with mode volume about @xmath57 and the cavity loss rate @xmath58 khz .
the laser emission coefficient is @xmath59 .
3 describes the solution of the photon number equation . from fig .
3 we can conclude that a stable laser field has been built up within laser cavity on condition that the population inversion is preserved .
the steady - state value of photon number @xmath42 is 160 and the power of this @xmath0yb@xmath1 ion laser lasing from @xmath8d@xmath14 to @xmath8s@xmath10 is about 37pw . .
there are @xmath52 ions in the cavity .
the steady - state value of photon number @xmath42 is 160.,width=283 ] the population inversion depends on the rabi frequency @xmath21 and @xmath60 . here
we consider the effect of @xmath21 .
the enhancement of @xmath21 will increase the population inversion , thus increasing the probability of lasing transition between the inverted states .
the steady - state value of photon number for @xmath55 is larger than that of @xmath54 in the case of the same total ions number(see fig . 4 ) .
given @xmath55 , the largest value of @xmath21 we can get under general experimental condition , the steady - state value of photon number @xmath42 is 336 .
the power of this @xmath0yb@xmath1 ion laser is about 77 pw .
varies with the total ions number @xmath56 in the cavity .
the red line is for @xmath54 , the blue line is for @xmath55.,width=283 ] as the number of ions in the cavity increases , the steady - state value of photon number is expected to rise accordingly .
the steady - state value of photon number @xmath42 varying with the total ions number @xmath56 in the cavity is also shown in fig . 4 . considering the actual experimental condition , we set the largest number of total ions in the cavity to be @xmath52 .
compared with the traditional laser , the primary three conditions are fulfilled for the active optical frequency standards based on paul trap trapped ions just proposed in this paper .
the very difference is that the lifetime of the lasing upper energy level is so long that the natural linewidth of the laser field is much narrower than the cavity mode linewidth , which is 500 khz .
besides , the temperature of cold ions trapped in paul trap decreases the doppler broadening linewidth . therefore , active ion optical clock operating at the condition of bad cavity will be a very stable narrow - linewidth laser source since the cavity pulling effect will dramatically reduce the cavity length noise due to johnson thermal effect .
it is suitable for single - ion optical frequency standards due to its higher stability .
although conventional single - ion optical clocks have reached unprecedented stability and accuracy , the observed linewidth of the clock transition is limited by the linewidth of the probe laser @xcite . active ion optical clock , as which we presented in this paper , can offer a new laser light source for single - ion optical clocks because of its narrow linewidth and high stability .
after laser cooling , the @xmath61 hz natural linewidth clock transition between @xmath8s@xmath10 and @xmath8d@xmath14 states has been measured with a fourier - limited linewidth of 30 hz recently @xcite .
based on the mechanism of active optical clock @xcite , to reduce the effect caused by the thermal noise of cavity , the ion laser cavity should be bad cavity , which means the linewidth of the cavity ( or the cavity loss rate ) should be much wider than the linewidth of the gain ( the broaden ion transition linewidth after considering the doppler broadening , collision broadening and light shift _ etc . _ ) . given the linewidth of ion gain to be 30 hz as measured , one can set the cavity mode linewidth as @xmath62 khz
. then the cavity related noise is reduced by 4 orders of magnitude to below 1 mhz with the best cavity design .
the ion - cavity coupling constant @xcite can be set around @xmath63 hz .
the quantum - limited linewidth of active ion optical clock @xcite @xmath64 is narrower than 1 mhz .
there are still some effects that will shift and broaden the linewidth of active ion optical clock , like light shift caused by the repumping laser .
for @xmath0yb@xmath1 active ion optical clock , the largest light shift effect is caused by the repumping laser of @xmath8s@xmath10 to @xmath8p@xmath9 transition . by adjusting this repuming laser at red detuning to @xmath8p@xmath9
, its light shift can be greatly reduced by suitable blue detuning to @xmath8p@xmath10 transition .
the repumping laser can be stabilized to an uncertainty of @xmath65 hz .
considering that the spontaneous decay rate of @xmath8p@xmath17 is @xmath18 mhz , far greater than 100 hz , we can write the light shift as @xmath66 . given the repumping rate @xmath55 , the value of light shift is 0.69 mhz .
other ions such as ba@xmath1 , ca@xmath1 and sr@xmath1 are also potential candidates for active optical clock and narrow linewidth ion laser with the same mechanism . compared with most clock transitions of traditional passive ion optical frequency standards , which are between @xmath8s@xmath17 and @xmath8d@xmath67 states like ba@xmath1 , ca@xmath1 and sr@xmath1
, we recommend the lasing from @xmath8d@xmath68 to @xmath8s@xmath17 states for active optical clock and narrow linewidth ion laser . in this case , here taking the even isotopic @xmath69sr@xmath1 as an example , it can be a natural three - level laser configuration with the cooling laser used as repumping laser at the same time , thus simpler than traditional passive ion optical clock . during the 422 nm cooling procedure , ions spontaneously decay to @xmath8d@xmath68 state and accumulate at this state with 0.4s lifetime @xcite
. then , coupled with a bad cavity via stimulated emission , the ions transit from @xmath8d@xmath68 to @xmath8s@xmath17 state in the configuration of three - level ion laser . with a magnetic field along the cavity axis , the laser light will be circularly polarized . in summary
, we propose a scheme of active ion optical clock , i.e. narrow linewidth bad cavity ion laser with detailed pumping method , lasing states , output power , linewidth and light shift reduction .
we especially study the @xmath0yb@xmath1 ions in a paul trap and propose to utilize a fabry - perot resonator to realize lasing of active optical frequency standards .
the population inversion between @xmath8d@xmath14 and @xmath8s@xmath10 states can be built up at the time scale of @xmath53s .
the steady - state value of photon number @xmath42 in the fabry - perot cavity increases with @xmath21 and the total ions number @xmath56 . if @xmath55 and @xmath70 , the steady - state value of photon number is @xmath71 and the laser power is 77 pw .
the quantum - limited linewidth of active @xmath0yb@xmath1 ion optical clock is narrower than 1 mhz .
other ions like ba@xmath1 , ca@xmath1 and sr@xmath1 are suitable for active optical clock .
active optical clock and narrow linewidth ion laser may also be realized with these candidate ions .
this work is supported by the national natural science foundation of china ( grant no .
10874009 ) . | in this paper , we propose a scheme of active ion optical clock with detailed pumping method , lasing states , output power , linewidth and light shift .
using @xmath0yb@xmath1 ions in a paul trap we propose to utilize a fabry - perot resonator to realize lasing of active optical frequency standards .
the quantum - limited linewidth of active @xmath0yb@xmath1 ion optical clock is narrower than 1 mhz .
based on the mechanism and advantages of active optical clock at ion optical clock transition frequency , this new laser light source as a stable local oscillator , will be beneficial to the single - ion optical clock , which currently is the most accurate clock . |
the dynamics of impurities jiggling in classical and quantum liquids has tantalized many generations of physicists since the early studies on brownian motion .
in particular , the interaction of a quantum system with an external environment strongly affects its dynamics @xcite .
because of this coupling , the motion of a quantum particle is characterized by a renormalized mass , decoherence , and damping .
polarons @xcite , originally studied in the context of slow - moving electrons in ionic crystals , and impurities in @xmath0he @xcite are two prototypical examples in which the bath is bosonic and fermionic , respectively .
these problems have been at the center of great interest for many decades in condensed matter physics .
recent advances in the field of cold atomic gases @xcite have made it possible to observe and study these phenomena from a different perspective and hence to disclose new aspects not addressed so far .
it is indeed possible to accurately tune the coupling between a quantum particle and the bath and to modify the many - body nature of the bath itself .
furthermore , the dynamics of the dressed particle can be studied in real time , thus giving direct access to both mass renormalization and damping .
this problem becomes of particular relevance if the bath is one - dimensional ( 1d ) . in this case interactions
strongly affect the excitation spectrum of the bath @xcite and therefore the effective dynamics of the coupled system .
the dynamics of impurities in cold atomic gases has attracted a great deal of experimental @xcite and theoretical @xcite attention in recent years . in particular ,
et al_. @xcite have recently studied experimentally the dynamics of k atoms ( the impurities " ) coupled to a bath of rb atoms ( the environment " ) confined in 1d atomic wires " .
motivated by ref .
we perform a time - dependent density - matrix renormalization group ( tdmrg ) @xcite study of the dynamics of the breathing mode in a 1d bath of interacting bosons . at a first sight
one might think that the problem under consideration reduces to the study of a particle coupled to a luttinger liquid @xcite . as we will discuss in the remainder of this work , it turns out that the nonequilibrium impurity dynamics eludes this type of description .
this is the reason why we choose to tackle the problem with an essentially exact numerical method .
several features of the experimental data in ref . are also seen in our simulations . as we will discuss in the conclusions , however , a detailed quantitative account of the data in ref
. may require additional ingredients and is outside the scope of the present work . here
, we highlight a number of distinct signatures of the impact of interactions on the breathing motion of an impurity , which are amenable to future experimental testing . ,
in units of @xmath1 ( @xmath2 is the lattice spacing ) , is plotted as a function of time @xmath3 ( in units of @xmath4 ) .
data labeled by empty symbols represent tdmrg results corresponding to two opposite values of the impurity - bath coupling constant @xmath5 .
the strength of interactions in the bath has been fixed to @xmath6 .
the data for @xmath7 ( squares ) are shifted downward by @xmath8 .
the black solid lines are fits to the tdmrg data based on eq .
( [ eq : motion_final]).[fig : one ] ]
we consider a 1d bath of interacting bosons coupled to a single impurity confined in a harmonic potential .
the bath is modeled by a bose - hubbard hamiltonian with hopping @xmath9 and on - site repulsion @xmath10 : [ eq : bath ] _
b = - j_1_i ( b^_ib_i+1 + h.c . ) + u_1_i n^2_i + _ iw_in_i . here
@xmath11 ( @xmath12 ) is a standard bosonic creation ( annihilation ) operator on the @xmath13-th site . to avoid spurious effects due to quantum confinement along the 1d system
, we consider a nearly - homogeneous bath : the external confining potential @xmath14 is zero in a large region in the middle of the chain ( @xmath15 ) and raises smoothly at the edges ( see sec .
[ appendix : mapping ] ) .
the local density @xmath16 is thus essentially constant in a region of length @xmath17 . in ( almost all ) the results shown below we fix the number of particles in the bath to @xmath18 and distribute them over @xmath19 sites , thus keeping the average density to a small value , @xmath20 . for this choice of parameters
the lattice is irrelevant and the model ( [ eq : bath ] ) is ideally suited to describe a continuum .
indeed , in the low - density limit , eq . ( [ eq : bath ] ) reduces to the lieb - liniger model @xcite ( a mapping between the coupling constants of the two models is summarized in sec .
[ appendix : mapping ] ) .
the hamiltonian describing the impurity dynamics is @xmath21 with @xmath22 the impurity density operator and @xmath23 .
( [ eq : impurity ] ) includes a kinetic term and a harmonic potential centered at @xmath24 , whose strength @xmath25 depends on time , mimicking the quench performed in the experimental study of catani _ et al .
_ @xcite . because the on - site impurity density @xmath26 is low ,
also in this case the model ( [ eq : impurity ] ) well describes the corresponding continuum hamiltonian ( sec .
[ appendix : mapping ] ) . in this work
we have fixed @xmath27 to take into account the mass imbalance between the impurity and bath atoms . for future purposes we introduce @xmath28 , with @xmath29 and @xmath30 , _
i.e. _ the harmonic - potential frequencies before and after the quench , respectively .
the full time - dependent hamiltonian @xmath31 contains a further density - density coupling between bath and impurity @xmath32 .
the quench in @xmath33 excites the impurity _ breathing _ mode ( bm ) , _ i.e. _ a mode in which the width @xmath34^{1/2}$ ] , associated with the impurity density @xmath35 , oscillates in time @xcite .
this quantity is evaluated with the tdmrg .
in fig . [ fig : one ] we illustrate the time evolution of the impurity width @xmath36 dictated by @xmath37 @xcite . different sets of data refer to two values of the impurity - bath interaction @xmath38 .
time @xmath3 is measured in units of @xmath39 , where @xmath40 is the period set by the harmonic - confinement frequency @xmath33 after the quench .
the tdmrg results ( empty symbols ) have been obtained by setting @xmath41 and @xmath42 for @xmath43 and @xmath44 for @xmath45 @xcite .
the black solid lines are fits to the tdmrg data based on the following expression : [ eq : motion_final ] = 1 + _
i = x , p_i ^2(t-_i ) , where @xmath46 , @xmath47 , @xmath48 , @xmath49 ^ 2 -1 $ ] , and @xmath50 ^ 2 - 1 $ ] .
( [ eq : motion_final ] ) is the prediction for the bm width obtained by solving a quantum langevin equation for the impurity position operator @xmath51 in the presence of ohmic damping and a random gaussian force with colored spectrum ( see sec [ appendix : quantum_breathing ] ) : ^2_tx(t)+ 2 _ tx(t ) + ^2x(t ) = ( t ) .
the three parameters @xmath52 , @xmath53 , and @xmath54 ( respectively the asymptotic width at long times , the frequency of the breathing oscillations , and the friction coefficient ) have been used to fit the data .
the initial width @xmath55 is extracted from numerical data for the ground - state width at @xmath56 .
note that , in the limit in which the impurity - bath interaction is switched off ( @xmath57 ) , @xmath36 must oscillate at the frequency @xmath58 , since only even states ( under exchange @xmath59 ) of the harmonic - oscillator potential are involved in the time evolution of a symmetric mode ( like the bm ) .
this implies @xmath60 in the limit @xmath61 , where eq .
( [ eq : motion_final ] ) reproduces the exact non - interacting dynamics . in fig .
[ fig : two ] we plot the values of the frequency @xmath53 as a function of @xmath5 and for different values of @xmath62 .
several features of the data in this figure are worth highlighting : i ) the behavior of @xmath53 is dramatically different when the sign of interactions is switched from attractive to repulsive , except at weak coupling , for a tiny region of small @xmath5 values ; ii ) the behavior becomes more symmetric with respect to the sign of @xmath5 as the bath is driven deeper into the tonks - girardeau ( tg ) regime , _
i.e. _ , for @xmath63 ( in passing , we notice that our results in this limit are relevant in the context of the so - called fermi polaron " problem @xcite ) ; iii ) the renormalization of the frequency @xmath53 is reduced on increasing the strength of repulsive interactions in the bath @xmath62 .
it becomes almost independent of @xmath5 for @xmath64 . ) as a function of the impurity - bath lieb - liniger parameter @xmath5 , for different values of the bath dimensionless coupling constant @xmath62 .
all the data in this figure have been obtained by setting @xmath42 for @xmath43 and @xmath44 for @xmath65 .
error bars refer to the fitting procedure .
solid lines are guides to the eye.[fig : two ] ] fig .
[ fig : three ] illustrates the dependence of the damping rate @xmath54 on @xmath5 , for different values of @xmath62 .
three features of the data are remarkable : i ) @xmath54 displays a strong asymmetrical behavior with respect to @xmath66 away from the weak - coupling limit ; ii ) @xmath54 decreases with increasing @xmath62 , saturating to a finite result in the tg limit ; and , finally , iii ) @xmath54 depends quadratically on @xmath5 for @xmath67 .
the non - monotonic behavior of @xmath54 on the attractive side can be explained as following . for @xmath68
the damping rate must be small .
increasing @xmath69 the damping rate increases because the coupling of the impurity to the bath increases .
however , upon further increasing @xmath69 another effect kicks in .
we have indeed discovered ( data not shown here ) that @xmath54 decreases monotonically with decreasing frequency . as shown in fig .
[ fig : two ] , on the attractive side @xmath53 decreases rapidly as @xmath69 increases , thereby reducing the damping rate .
the non - monotonic behavior of @xmath54 does not occur for @xmath70 because on the repulsive side @xmath53 changes slightly with respect to @xmath5 . , but for the friction coefficient @xmath54 ( in units of @xmath71 ) .
note that @xmath54 vanishes quadratically for weak impurity - bath couplings ( @xmath72 ) and saturates to a finite value in the limit @xmath73 .
the inset illustrates the dependence of @xmath54 [ in units of @xmath74 on @xmath71 ( in units of @xmath75 ) . for each value of @xmath71 ,
the tdmrg data ( green symbols ) have been obtained by performing a quench corresponding to a value of @xmath76 .
the other parameters are : @xmath77 , @xmath78 ( @xmath79 ) , @xmath80 , and @xmath81 .
the solid line represents the prediction @xmath82 , based on the abraham - lorentz model with the value of @xmath83 corresponding to @xmath84 .
the dashed line at @xmath85 is the result of a best fit to the data.[fig : three ] ] the asymptotic width @xmath52 , shown in fig .
[ fig : four ] , fairly agrees with the equilibrium value at the frequency @xmath71 that one can calculate numerically .
this finding seems to suggest that the impurity has nearly thermalized " with the bath over the time scale of our simulations . ,
but for the asymptotic value @xmath52 at long times of the width @xmath36 ( in units of @xmath86 ) .
here the lines are _ not _ guides to the eye , but represent the equilibrium value @xmath87 for the width in the harmonic potential with frequency @xmath71 ( @xmath88 for @xmath57).[fig : four ] ]
we now discuss which features in figs . [
fig : two]-[fig : four ] can ( or can not ) be explained by employing a low - energy luttinger - liquid description of the bath .
the hamiltonian of a single impurity of mass @xmath89 , described by the pair of conjugate variables ( @xmath90 ) , coupled to a bath of harmonic oscillators ( the bosonic excitations of the luttinger liquid ) with dispersion @xmath91 , is @xcite : [ eq : ll ] ( t ) = + v(x , t ) + _ k0 _ k ^_k _ k + g_12(x ) , where @xmath92 ( the identification with the lattice model fixes @xmath93 with @xmath2 the lattice spacing ) and @xmath94 is a coupling constant playing the role of @xmath5 in the discrete model . in eq .
( [ eq : ll ] ) @xmath95 ( @xmath96 ) is the creation ( annihilation ) operator for an acoustic - phonon mode with wave number @xmath97 and @xmath98 is the bath density operator .
the sound velocity @xmath99 is related to the luttinger parameter @xmath100 of the lieb - liniger model by galilean invariance @xcite ; @xmath100 is in turn defined by the relation @xmath101 , where @xmath102 is the compressibility of the bath @xcite .
the parameters @xmath100 and @xmath103 , which completely characterize the luttinger liquid , can be expressed in terms of the coupling constants of the model in eq .
( [ eq : bath ] ) ( see sec . [
appendix : luttinger_impurity ] ) .
we now observe that the sign of the impurity - bath coupling @xmath94 can be gauged away from the hamiltonian ( [ eq : ll ] ) by the canonical transformation @xmath104 .
this means that if the bath was truly a luttinger liquid , impurity - related observables such as @xmath53 and @xmath54 should _ not _
depend on @xmath94 being attractive or repulsive .
this low - energy description seems to apply only in a tiny region around @xmath57 .
all the deviations from this prediction seen in figs .
[ fig : two]-[fig : four ] have to be attributed to physics beyond the luttinger - liquid paradigm .
the heisenberg equation of motion induced by the hamiltonian ( [ eq : ll ] ) reads ( see sec . [
appendix : luttinger_impurity ] ) : [ eq : llkernel ] m^2_tx(t)&+ & m ^2_2x(t ) + m _ 0^t dt ( t , t)_tx(t ) + & = & - g_12 _ x ( x , t)|_x = x(t ) , where [ eq : memorykernel ] ( t , t ) = _ k0
e^ikx(t)e^-ikx(t ) , with @xmath105^{1/2}|k|e^{-|k|/2k_c}$ ] , is the memory kernel @xcite ( @xmath106 is an ultraviolet cut - off ) . if the dynamics of the impurity is slow with respect to the speed @xmath99 of propagation of information in the bath , then retardation effects " can be neglected and we can approximate the operator @xmath107 with the following c - number @xmath108}/(m\omega^2_k)$ ] . in this limit it is possible to show ( see sec . [
appendix : luttinger_impurity ] ) that eq .
( [ eq : llkernel ] ) reduces to a quantum langevin equation with an _ abraham - lorentz _
( al ) term , which describes the reactive effects of the emission of radiation from an oscillator @xcite .
this is a term of the form @xmath109 with a characteristic time " @xmath110 .
remarkably , neglecting the well - known runaway " solution @xcite and keeping only the damped solutions , we find that the quantum langevin equation with the al term yields an expression for @xmath36 after the quench which is identical to eq .
( [ eq : motion_final ] ) with @xmath111 .
the full functional dependence of @xmath112 on @xmath71 is reported in sec .
[ appendix : quantum_breathing ] .
note that @xmath112 is proportional to @xmath113 .
this is in agreement with the tdmrg results shown in fig .
[ fig : three ] in the weak - coupling @xmath114 limit .
moreover , @xmath112 is proportional to @xmath115 ( in our case @xmath116 from galilean invariance @xcite ) and proportional to @xmath117 .
the former statement implies a fast saturation of the friction coefficient to a finite value in the tg limit ( @xmath118 ) . this is in agreement with the tdmrg data shown in fig .
[ fig : three ] .
the quadratic dependence of the damping rate on @xmath71 is also well displayed by the tdmrg data , at least for @xmath119 , as shown in the inset to fig .
[ fig : three ] .
it is instructive to compare our findings with the experimental data of ref . .
the latter show that , in a sizable range of interaction strength @xmath5 , the frequency of the breathing mode does not vary appreciably while the damping coefficient increases up to @xmath120 . as shown in figs .
[ fig : two ] and [ fig : three ] , we do observe the same behavior for @xmath121 . moreover , as in the experiment , we do see that the width of the breathing mode reduces upon increasing @xmath5 .
a detailed quantitative comparison with the experiment is , however , not possible at this stage : i ) one notable difference is that our calculations are carried out at @xmath122 , while temperature effects seem to be important in ref . ;
ii ) furthermore , the impurity trapping frequency in our calculations is considerably larger than in the experiment @xcite . extending the current calculations to take into account these differences lies beyond the scope of this work . in summary
, we have shown that , in the dynamics of an impurity coupled to a 1d bosonic bath , a luttinger - liquid description of the bath is applicable only in a very small region of parameter space , where the impurity suffers an al radiation - reaction friction .
among the most striking features we have found , we emphasize the non - monotonic behavior of the damping rate and the large renormalization of the oscillation frequency for attractive impurity - bath interactions .
in our simulations we consider a 1d bath of interacting bosons coupled to a single impurity confined in a harmonic potential .
the bath is modeled by a bose - hubbard hamiltonian @xmath123 , eq .
( 1 ) in the main text , with hopping @xmath9 and on - site repulsion @xmath10 . the external confining potential @xmath14 has the following explicit form : @xmath124 & \hspace*{-1 mm } 1 < i < \delta\ , , \\ 0 & \hspace*{-1mm}\delta \leq i \leq l-\delta\ , , \\
\frac{w}{2 } \big [ 1+\tanh \big ( \frac{3(i-(2l-\delta)/2)}{2\sqrt{(i - l+\delta)(l - i ) } } \big ) \big ] & \hspace*{-1 mm } l-\delta < i < l\ , , \\ w & \hspace*{-1mm}i = l\ , . \end{cases } \label{eq : external}\ ] ] the parameter @xmath125 is the length in unit of lattice sites of the left and right boundary regions , where the potential goes from 0 to a finite value @xmath126 .
the interpolation between the two values is as smooth as possible , since the potential , as a continuous function of @xmath127 ( @xmath2 being the lattice spacing ) , has zero derivatives of all orders at the joining points in eq .
( [ eq : external ] ) .
we set @xmath128 lattice sites , while we used @xmath129 for @xmath130 and @xmath131 ( see fig .
[ external_potential ] ) , @xmath132 for @xmath78 and @xmath133 , where @xmath134 is the length of the ( two ) lattices used in our simulations , and @xmath135 the number of particles in the bath . in the latter case @xmath136 is smaller since the on - site density @xmath137 in the middle of the chain is smaller and a weaker potential is used . with our choice of parameters ,
the local density @xmath16 is essentially constant in a region of length @xmath17 , and kept to a low value @xmath138 everywhere . at such low densities and for not too strong repulsive interaction ( @xmath139 ) , the lattice is irrelevant and the model can be mapped to a lieb - liniger hamiltonian @xcite describing 1d bosons of mass @xmath140 interacting through a contact ( repulsive ) two - body potential .
the upper bound for the parameter @xmath141 can be understood as follows : as the interaction between bosons is increased , the healing length of the lieb - liniger gas gets smaller , and , when it is comparable with @xmath2 , lattice effects becomes relevant .
an extensive discussion of this point can be found in refs . .
other relevant parameters of the continuum model are the density @xmath142 ( where @xmath143 is the on - site density taken in the central region where the bath is homogeneous ) and the dimensionless lieb - liniger parameter @xmath144 , where @xmath145 is the strength of the contact repulsion between bosons .
the hamiltonian @xmath146 describing the impurity is written in eq .
( 2 ) in the main text . given the low impurity density @xmath26 , also in this case the lattice model well describes the continuum hamiltonian of a particle of mass @xmath147 moving in a parabolic potential @xmath148 ( centered without loss of generality at @xmath149 ) . the impurity mass is fixed to @xmath150 ( thus corresponding to a ratio between the two lattice hoppings @xmath151 ) , such to take into account the mass imbalance between rb and k atoms , as experimentally done in ref . .
the total hamiltonian @xmath152 contains a further density - density coupling between bath and impurity , @xmath153 , which in the continuum limit corresponds to a @xmath2-function of strength @xmath154 .
in this section we derive a quantum langevin equation @xcite for a single particle , described by the conjugate variables @xmath155 , that is coupled to a luttinger liquid by a density - density interaction .
the relevant hamiltonian @xcite is given by eq .
( [ eq : ll ] ) where the first two terms describe the impurity hamiltonian , the potential @xmath156 representing an harmonic confining trap of frequency @xmath157 for the impurity , while the third one denotes the quadratic luttinger hamiltonian , in which @xmath158 ( @xmath159 ) is the creation ( annihilation ) operator for an acoustic - phonon mode with wave vector @xmath97 and dispersion @xmath160 .
apart from the coupling constant @xmath94 , the last term , which contains an ultraviolet cut - off ( @xmath106 ) , defines the bath density operator @xmath161 of the luttinger liquid @xcite , and is controlled by the parameter @xmath100 .
this , in turn , controls the speed of `` sound '' @xmath99 by virtue of galilean invariance : @xmath162 .
repulsive interactions in the bath enter the problem through the dependence of @xmath99 and @xmath100 on the lieb - liniger parameter @xmath163
@xcite . in eq .
( [ eq : ll ] ) the phonon modes couple linearly to the particle position , the latter entering through the exponential @xmath164 . as a consequence , the harmonic excitations of the luttinger liquid can be integrated out leaving an effective dissipative equation ( quantum langevin equation @xcite ) for the impurity degree of freedom .
we first switch from annihilation and creation operators to ( complex ) position and momentum , by defining @xmath165 and @xmath166 , with @xmath167 = i\hbar\delta_{kk'}$ ] .
notice that , unlike the usual position and momentum , they are complex and their adjoint are @xmath168 and @xmath169 . using these new variables one
gets @xmath170 where the coefficients @xmath171 have been defined according to ref . .
usually , after completing the square , some additional terms depending only on @xmath172 are left out and their effect is to renormalize the potential @xmath173 .
given the peculiar non - linear nature of the coupling @xmath174 , this does not happen in our case : the terms completing the square turn out to be independent of @xmath175 , therefore the potential @xmath176 is not renormalized .
the heisenberg equations of motion dictated by eq .
( [ eq : ll2 ] ) are given by @xmath177 where @xmath178 .
the solution of the second equation can be immediately written down : @xmath179 \ , e^{-ik{\hat x}(t ' ) } \ , .\end{aligned}\ ] ] integrating by parts and substituting into eq .
( [ eq : heisenberg1 ] ) , we get @xmath180 \ ] ] with a memory kernel function @xmath181 that depends _ separately _ on @xmath3 and @xmath182 ( and not only on @xmath183 ) , due to the presence of the two non - commuting operators @xmath51 and @xmath184 evaluated at different times . upon the substitution @xmath185 , the right hand side is just @xmath186 . in order to see this , simply rewrite @xmath187 and @xmath188 using the corresponding annihilation and creation operators and compare to the density - density coupling in the hamiltonian ( [ eq : ll ] ) . as discussed in ref .
, one can forget about the term @xmath189 in the right hand side of eq .
( [ eq : rad_damping_first ] ) , since it has no effect when taking averages on the equilibrium state of the bath coupled to the particle at @xmath190 ( as a consequence the stochastic term in the quantum langevin equation ( [ eq : rad_damping_first ] ) has zero average : @xmath191 ) .
when evaluating the noise correlator one runs into difficulties since @xmath192}$ ] . in the limit when the particle has a small velocity with respect to @xmath99
, we can however perform the following approximation : @xmath193 since in this case the product of the first two exponentials is assumed to be slowly varying with respect to the third one . therefore in this limit the noise correlator @xmath194^{\dagger } \partial_x \hat\rho({\hat x}(t'),t')\rangle$ ] reads : @xmath195 where we passed in the continuum by substituting the series over @xmath97 with an integral over the frequencies @xmath157 . in the same approximation ,
the memory kernel ( [ eq : kernel ] ) entering the third term in the left hand side of eq .
( [ eq : rad_damping_first ] ) reads @xmath196 notice that , when the cut - off @xmath197 goes to infinity , this kernel basically reduces to the second derivative of a delta function : @xmath198 where @xmath199 is the proper time scale .
the relation @xmath200 = m\hbar \omega\coth(\beta\hbar\omega/2)\re \big[\widetilde\gamma(\omega)\big]$ ] states the fluctuation - dissipation theorem @xcite , and holds between the real parts of the fourier transforms of the noise correlator , @xmath201 , and of the memory kernel , @xmath202 .
this ensures the consistency of the _ slow - particle approximation _
, see eq .
( [ eq : slowpart ] ) . within this approximation ,
the quantum langevin equation takes the linear form @xmath203 the fact that the position operator appears on the right hand side as an argument of the bath density does not spoil linearity , since the noise correlator is independent of the difference @xmath204 in the slow - particle approximation .
considering @xmath205 as a _ classical _ variable , the @xmath97-sum in the third term on the left of eq .
( [ eq : rad_damping_first ] ) can be easily performed , thus obtaining the following classical langevin equation : @xmath206 = -g_{12}\partial_{x}{\rho}({x}(t ) , t)\ , .
\end{split}\ ] ] from here it is apparent that the particle is interacting at point @xmath207 and time @xmath3 with the phonons ( density fluctuations in the luttinger liquid ) emitted by itself at point @xmath208 in a past time @xmath182 .
this is similar to radiation damping of the motion of a charge particle , where the role of the electromagnetic field is now played by the luttinger liquid .
the problem of damping due to the emission of radiation is a very old one and the classical version of eq .
( [ eq : lin_rad_damping ] ) has been known in this context for a long time as the abraham - lorentz equation @xcite . the quantum version of eq .
( [ eq : rad_damping ] ) is more complicated , due to its operatorial character ( in this case indeed @xmath192}$ ] ) . in the specific , eq .
( [ eq : lin_rad_damping ] ) has been obtained under the assumption ( [ eq : slowpart ] ) , thus it is expected to be valid only when the particle does not move too fast with respect to the phonons .
this can be quantified by considering a small value for the ratio @xmath209 between the maximum velocity of the impurity distribution when it expands , @xmath210 ( here @xmath211 is the harmonic oscillator length ) , and the velocity of sound @xmath103 .
such ratio is independent of the trap frequency , and is fixed only by the initial squeezing through the uncertainty principle if the uncertainty in position is @xmath55 then the uncertainty in velocity is @xmath212 . in our simulations
this parameter is actually not so small .
an estimate can be given as follows : the velocity of sound @xmath99 is a fraction of the fermi velocity for a gas of free fermions at the same density , say @xmath213 , the initial squeezing is @xmath214 and the density @xmath215 , so @xmath216
so we expect corrections to eq .
( [ eq : lin_rad_damping ] ) due to retardation effects embodied in eq . ( [ eq : rad_damping_first ] ) .
in this section we derive the function used to fit the tdmrg data [ see eq .
( 3 ) in the main text ] .
we show how it can be obtained starting both from the quantum langevin equation for an _ ohmic _ bath @xcite : @xmath217 with noise correlator @xmath218 \ , , \nonumber\end{aligned}\ ] ] and from the langevin equation ( [ eq : lin_rad_damping ] ) in the previous section , with a noise correlator given by eq .
( [ eq : noise_corr ] ) that describes a _ superohmic _ bath @xcite .
the predictions for the oscillation frequency and the damping coefficient are different in the two cases .
however it is possible to treat both equations in the same way , because their respective green functions @xmath219 have a similar pole structure in the frequency domain .
both have two poles at complex frequencies @xmath220 with negative imaginary part ( @xmath221 ) : @xmath222 these are the physically relevant ones and correspond to exponentially decaying solutions .
it is essential for our purposes that @xmath223 .
we point out that eq .
( [ eq : lin_rad_damping ] ) is of third order and its green function has a third pole in the upper half of the complex plane that corresponds to an unphysical diverging solution , also called `` run - away '' solution @xcite , which will be discarded in the following .
note that , while for the ohmic case @xmath53 is equal to the trap frequency @xmath71 , and @xmath54 is exactly the coefficient that appears in eq .
( [ eq : langevin_ohmic ] ) , for the langevin equation in ( [ eq : lin_rad_damping ] ) one has a more complex dependence on the trap frequency @xmath71 and on the time scale @xmath83 . in the specific , the three roots of the _ cubic _ characteristic equation @xmath224 read @xmath225\,,\end{gathered}\ ] ] with @xmath226 note that @xmath220 have always negative imaginary part , while @xmath227 has positive imaginary part and zero real part .
using eq . , and the fact that @xmath231 , we can evaluate some asymptotic averages which turn out to be useful to calculate the impurity breathing mode .
some of them are null : @xmath232 while other ones , such as @xmath233 and @xmath234 , can be implicitly written in terms of the spectral function @xcite , whose form depends on the nature of the bath . finally , call @xmath235 and note that
@xmath236 with all these results at hand , the average of @xmath237 can be calculated at an arbitrary time @xmath3 .
suppose that the particle starts its motion at @xmath238 in a squeezed harmonic potential .
at @xmath239 it will have equilibrated with the bath , and expectation values are taken on the equilibrium state of the _ whole _
system , with the particle at rest in the squeezed harmonic potential .
then the frequency of the harmonic confinement is suddenly quenched to a new value @xmath71 .
the width of the impurity density distribution as a function of time , for @xmath240 , can be explicitly calculated by taking the square of eq .
( [ eq : sol ] ) on the global state : @xmath241 . using eqs .
( [ eq : avgzero1])- , one ends up with the following expression : @xmath242 \,,\end{gathered}\ ] ] with @xmath243 .
it is not necessary to calculate explicitly the average @xmath244 , since it has to cancel exactly the other terms in eq . when @xmath245 and @xmath246
. therefore the final result is @xmath247 \bigg\ } \
, , \end{gathered}\ ] ] where the definitions below have been used : @xmath248 we employ eq . in our fitting procedure as follows .
@xmath249 and @xmath250 are fixed to their non - interacting values @xmath251 while @xmath52 , @xmath53 , @xmath54 are used as fitting parameters .
the initial width @xmath55 is provided directly by the numerical data .
this choice allows reliable fits that smoothly interpolate with the exact non - interacting result : @xmath252 surprisingly , the same functional form of eq .
( [ eq : motion2 ] ) holds for two quite different models .
this is due to the form of the two relevant complex frequencies [ see eq .
( [ eq : roots ] ) ] that applies to both of them
. however the two models provide distinct predictions for the modulus @xmath253 and the damping coefficient @xmath54 , with some appreciable differences . for an ohmic bath one
has @xmath254 , so that the frequency is not renormalized in this case .
moreover @xmath54 is a fixed time constant independent of the trap frequency .
for the superohmic case of eq .
( [ eq : lin_rad_damping ] ) , for @xmath255 it can be shown that @xmath256 is a function of @xmath257 , the same holds for @xmath258 ( see fig . [ varpi - gamma ] ) . notice also that @xmath220 always has a non - zero real part , therefore there is no overdamped solution for any value of @xmath83 .
finally we emphasize that the functional form given in eq .
is essential to extract the parameters @xmath53 , @xmath54 and @xmath52 from the numerical data .
indeed we have verified that the numerical fits performed using the functional form @xcite @xmath259 with @xmath260 being the oscillation amplitude ( grey lines in fig .
[ fig : two_bis ] ) , are significantly different from the ones that are obtained employing eq .
( [ eq : motion2 ] ) ( black lines in fig .
[ fig : two_bis ] ) .
for example , the fit done with eq .
( [ eq : naive ] ) often overestimates the asymptotic with @xmath52 .
77 u. weiss , _ quantum dissipative systems _
( world scientific , singapore , 2008 ) .
gardiner and p. zoller , _ quantum noise _
( springer , berlin , 2004 ) .
a.o . caldeira and a.j .
leggett , http://www.sciencedirect.com/science/article/pii/0378437183900134[physica a * 121 * , 587 ( 1983 ) ] ; a.j .
leggett , s. chakravarty , a.t .
dorsey , m.p.a .
fisher , a. garg , and w. zwerger , http://rmp.aps.org/abstract/rmp/v59/i1/p1_1 [ , 1 ( 1987 ) ] .
feynman , _ statistical mechanics : a set of lectures _ , ( addison wesley , reading ma , 1972 ) .
, edited by k.h .
benneman and j.b .
ketterson ( wiley , new york , 1978 ) .
i. bloch , j. dalibard , and w. zwerger , http://rmp.aps.org/abstract/rmp/v80/i3/p885_1 [ , 885 ( 2008 ) ] .
t. giamarchi , _ quantum physics in one dimension _
( oxford university press , oxford , 2004 ) .
j. catani , l. de sarlo , g. barontini , f. minardi , and m. inguscio , http://pra.aps.org/abstract/pra/v77/i1/e011603 [ , 011603 ( 2008 ) ] ; s. palzer , c. zipkes , c. sias , and m. khl , http://prl.aps.org/abstract/prl/v103/i15/e150601 [ , 150601 ( 2009 ) ] ; b. gadway , d. pertot , r. reimann , and d. schneble , http://prl.aps.org/abstract/prl/v105/i4/e045303[_ibid._ * 105 * , 045303 ( 2010 ) ] ; s. schmid , a. hrter , and j.h .
denschlag , http://prl.aps.org/abstract/prl/v105/i13/e133202[_ibid._ * 105 * , 133202 ( 2010 ) ] ; c. zipkes , s. palzer , c. sias , and m. khl , http://www.nature.com/nature/journal/v464/n7287/full/nature08865.html[nature * 464 * , 388 ( 2010 ) ] .
f.m . cucchietti and e. timmermans , http://prl.aps.org/abstract/prl/v96/i21/e210401 [ , 210401 ( 2006 ) ] ; k. sacha and e. timmermans , http://pra.aps.org/abstract/pra/v73/i6/e063604 [ , 063604 ( 2006 ) ] ; m. bruderer , a. klein , s.r .
clark , and d. jaksch , http://pra.aps.org/abstract/pra/v76/i1/e011605[_ibid_ * 76 * , 011605 ( 2007 ) ] ; m. bruderer , a. klein , s.r .
clark , and d. jaksch , http://iopscience.iop.org/1367-2630/10/3/033015[new j. phys .
* 10 * , 033015 ( 2008 ) ] ; j. tempere , w. casteels , m.k .
oberthaler , s. knoop , e. timmermans , and j.t .
devreese , http://prb.aps.org/abstract/prb/v80/i18/e184504 [ , 184504 ( 2009 ) ] ; a. privitera and w. hofstetter , http://pra.aps.org/abstract/pra/v82/i6/e063614 [ , 063614 ( 2010 ) ] ; d.h .
santamore and e. timmermans , http://iopscience.iop.org/1367-2630/13/10/103029[new j. phys .
* 13 * , 103029 ( 2011 ) ] ; w. casteels , t. van cauteren , j. tempere , and j.t .
devreese , http://www.springerlink.com/content/03183tq382v60j16/[laser phys .
* 21 * , 1480 ( 2011 ) ] ; w. casteels , j. tempere , and j.t .
devreese , http://pra.aps.org/abstract/pra/v83/i3/e033631 [ , 033631 ( 2011 ) ] and http://pra.aps.org/abstract/pra/v84/i6/e063612[*84 * , 063612 ( 2011 ) ] ; w. casteels , j. tempere , and j.t .
devreese , http://www.springerlink.com/content/068j68416q1n355t/[j . low temp
. phys . * 162 * , 266 ( 2011 ) ] ; t.h .
johnson , s.r .
clark , m. bruderer , and d. jaksch , http://pra.aps.org/abstract/pra/v84/i2/e023617[*84 * , 023617 ( 2011 ) ] .
j. catani , g. lamporesi , d. naik , m. gring , m. inguscio , f. minardi , a. kantian , and t. giamarchi , http://pra.aps.org/abstract/pra/v85/i2/e023623 [ , 023623 ( 2012 ) ] .
see _ e.g. _ u. schollwck , http://www.sciencedirect.com/science/article/pii/s0003491610001752[ann .
( ny ) * 326 * , 96 ( 2011 ) ] .
see , for example , a.h .
castro neto and a.o .
caldeira , http://prb.aps.org/abstract/prb/v46/i14/p8858_1 [ , 8858 ( 1992 ) ] ; http://prb.aps.org/abstract/prb/v50/i7/p4863_1[_ibid._ * 50 * , 4863 ( 1994 ) ] ; http://prb.aps.org/abstract/prb/v52/i6/p4198_1[_ibid._ * 52 * , 4198 ( 1995 ) ] ; a.h .
castro neto and m.p.a .
fisher , http://prb.aps.org/abstract/prb/v53/i15/p9713_1[_ibid._ * 53 * , 9713 ( 1996 ) ] ; j. bonart and l.f .
cugliandolo , http://arxiv.org/abs/1203.5273[arxiv:1203.5273 ] .
cazalilla , r. citro , t. giamarchi , e. orignac , and m. rigol , http://rmp.aps.org/abstract/rmp/v83/i4/p1405_1 [ , 1405 ( 2011 ) ] . here
@xmath35 is a shorthand for @xmath261 , @xmath262 being the state of the system at time @xmath3 .
the initial condition is given by the ground state of @xmath263 at times @xmath43 .
this implies an oscillation period @xmath264 , to be compared with @xmath265 in ref . .
see , for example , a. schirotzek , c .- h .
wu , a. sommer , and m.w .
zwierlein , http://prl.aps.org/abstract/prl/v102/i23/e230402 [ , 230402 ( 2009 ) ] ; a. sommer , m. ku , and m.w .
zwierlein , http://iopscience.iop.org/1367-2630/13/5/055009[new j. phys .
* 13 * , 055009 ( 2011 ) ] ; c. kohstall , m. zaccanti , m. jag , a. trenkwalder , p. massignan , g.m .
bruun , f. schreck , and r. grimm , http://arxiv.org/abs/1112.0020[arxiv:1112.0020 ] nature ( 2012 ) , published online ; m. koschorreck , d. pertot , e. vogt , b. frhlich , m. feld , and m. khl , http://arxiv.org/abs/1203.1009[arxiv:1203.1009 ] nature ( 2012 ) , published online .
haldane , http://prl.aps.org/abstract/prl/v47/i25/p1840_1 [ , 1840 ( 1981 ) ] .
jackson , _ classical electrodynamics _
( wiley , new york , 1998 ) . m.a .
cazalilla , http://pra.aps.org/abstract/pra/v67/i5/e053606[*67 * , 053606 ( 2003 ) . ] m.a .
cazalilla , http://pra.aps.org/forward/pra/v70/i4/e041604[*70 * , 041604(r ) ( 2004 ) . ] c. kollath , u. schollwck , j. von delft , w. zwerger , http://pra.aps.org/abstract/pra/v71/i5/e053606[*71 * , 053606 ( 2005 ) ] .
ford , j.t .
lewis , and r.f .
oconnell , http://pra.aps.org/abstract/pra/v37/i11/p4419_1 [ , 4419 ( 1988 ) ] . | by means of time - dependent density - matrix renormalization - group ( tdmrg ) we are able to follow the real - time dynamics of a single impurity embedded in a one - dimensional bath of interacting bosons .
we focus on the impurity breathing mode , which is found to be well - described by a single oscillation frequency and a damping rate .
if the impurity is very weakly coupled to the bath , a luttinger - liquid description is valid and the impurity suffers an abraham - lorentz radiation - reaction friction . for a large portion of the explored parameter space , the tdmrg results fall well beyond the luttinger - liquid paradigm . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.