article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
in the quantum zeno effect @xcite , the transitions of a quantum system are slowed down ( or stopped ) by the fact one is observing the system .
it is the quantum mechanical equivalent of saying that `` a watched pot never boils '' ( i.e. a watched system never makes transitions ) .
however , there is an analogy between observing a system ( projective measurements of its state ) and environmental noise .
indeed one can model a measurement device as a large object with many degrees of freedom interacting with the system .
thus one can ask whether environmental - noise induces a quantum zeno effect , and ( if so ) what its nature is .
in fact , there are a number of noise - induced zeno effects , i.e. noise slows - down quantum transitions in a variety of ( qualitatively different ) ways .
also , certain noise can slightly speed up transitions ( a weak anti - zeno effect ) .
imagine a spin - half which is @xmath9 along the @xmath10-axis , in an @xmath11-axis magnetic - field , so its hamiltonian is @xmath12 if the spin is left to evolve , it precesses around the @xmath11-axis , so that after a time @xmath13 the spin will be @xmath14 .
however what if we observe the spin s z - axis polarization at regular intervals @xmath15 much less than the time @xmath16
? the observation is modelled by projecting out the spin - state along the @xmath10-axis as follows : @xmath17 a simple calculation @xcite shows that for @xmath18 , the probability to find the spin up during all observations up to a time @xmath19 is @xmath20 $ ] .
thus rapid observations greatly reduce the spin s transition - rate .
the anti - zeno effect is the reverse ; transitions are speeded up by the fact one is observing the system .
the circumstance under which this occurs is nicely explained in ref .
@xcite .
, scaledwidth=80.0% ] in 1947 lamb noted that atomic hydrogen s levels are slightly shifted with respect to the predictions of schrodinger ( or dirac ) quantum mechanics .
the same year , bethe explained this shift in terms of the interaction between the electron and the photonic excitations of the vacuum .
similar lamb shifts occur whenever a system is coupled to an environment , and can be thought of in terms of level - repulsion , see fig .
[ fig : lamb ] .
one can interprete this as a weak zeno or anti - zeno effect , in which the spin s coherent oscillations are either slightly slowed - down , or slightly speeded - up by a quantum environment .
there is a very pretty model of dissipative quantum mechanics in which one averages the quantum dynamics over classical white - noise .
the fact the noise is white ( meaning it contains all frequencies ) is equivalent to saying that it is @xmath21-correlated in time .
this means one can time - slice on a timescale much less than that associated with the system dynamics , and then average over the noise in each slice independently .
if the spin evolves under the hamiltonian @xmath22 above , and we turn on noise along the @xmath10-axis , one finds that @xcite @xmath23 where @xmath5 is the spin - polarization along the @xmath10-axis .
note that there is no lamb shift here , so the noise does not modify the off - diagonal terms .
one can use the level - repulsion argument to see why the lamb shift is zero . if the white - noise were generated by an environment of quantum modes , all gaps would be equally represented .
thus there would be equal weight of modes pushing the system gap downwards as upwards , leaving the system s gap unchanged @xcite .
in contrast , for coloured noise the lamb shift would be non - zero , and we would have @xmath24 replaced by @xmath25 $ ] .
it is easy to rewrite eq .
( [ eq : white ] ) to show that @xmath5 is given by the equation of motion of a damped harmonic oscillator ; @xmath26 .
thus it has two regimes , @xmath27 t } + a_- { { \rm e}}^{-[\gamma - { { \rm i}}(b_x^2-\gamma^2)^{1/2}]t}\qquad \\ \hbox { over - damped } ( b_x < \gamma ) : & & s_z(t ) = a_+ { { \rm e}}^{-[\gamma + ( \gamma^2-b_x^2)^{1/2}]t } + a_- { { \rm e}}^{-[\gamma - ( \gamma^2-b_x^2)^{1/2}]t } \qquad\end{aligned}\ ] ] where @xmath28 and @xmath29 are given by @xmath30 boundary conditions . from this
we see that the frequency of coherent oscillations is reduced in the under - damped regime . by increasing @xmath31 one enters the over - damped regime .
as we take @xmath32 , the second term in @xmath5 decays at a rate @xmath33 ; thus strong noise stops transitions of the spin .
consider the spin as initially @xmath9 along the @xmath10-axis , so @xmath34 and @xmath35 , thus for large @xmath31 we have @xmath36 and @xmath37 .
then @xmath38 $ ] , so the transition rate is suppressed by strong noise . in reviewing ref .
@xcite , michael berry @xcite pointed out that this suppression of transitions in the over - damped case was a quantum zeno effect . indeed , in the limit @xmath39
, the result coincides with that of the usual ( observation - induced ) quantum zeno effect if we identify @xmath31 with @xmath40 .
all the above physics ( under - damping and over - damping ) is contained in the bloch - redfield equation , which is applicable whenever the decay time , @xmath41 , is much greater than the environment memory - time ( the memory time is zero for white - noise , so bloch - redfield becomes essentially exact ) @xcite .
however often one makes an additional secular or rotating wave - approximation ( e.g. as in ref .
@xcite ) , which is only justifiable in the strongly under - damped limit ( where one keeps @xmath42$]-terms but neglects @xmath43$]-terms ) . as we see above
, once one is beyond this under - damped limit , level - repulsion is _ not _ the only source of shifts of the coherent oscillation frequency .
the orthogonality catastrophe is a sort of extreme lamb shift , which can only occur in a system strongly coupled to it environment .
it was called `` adiabatic renormalization '' by leggett _ et al _ @xcite , but anderson introduced the idea earlier in the context of a fermionic ( rather than bosonic ) environment @xcite .
each mode of the environment tries to adiabatically follow the system ( spin ) , which means that all environment modes must shift each time the spin flips .
then the transition rate for the spin becomes @xmath44 , where @xmath45 is the state of the @xmath46th environment mode shifted one way or the other ( depending if the spin is up or down ) .
the overlap @xmath47 is slightly less than one , however the product over all overlaps will be _ exponentially small _ when the number of environment modes , @xmath48 .
this argument neglects non - adiabatic effects ( transitions of environment modes ) .
@xcite used renormalization group ( rg ) and find that they too are exponentially small .
we call this a `` super zeno effect '' since one would have to wait an exponentially long time for the system to undergo a transition
. this timescale is of order @xmath49 where @xmath50 is a typical frequency of the environment ( cf . the examples above , where this timescale is of order @xmath51 or @xmath52 ) .
however our recent work has called into question the so - called `` irrelevant '' operators in the rg flow .
we suspected that they summed to give a contribution to the transition rate which was not exponentially small , but instead went like an inverse power of the upper cut - off on the environment spectrum , @xmath53 ( here we restrict our comments to super - ohmic baths ) .
in many cases this cut - off is not so different from other energyscales in the problem , then the transition rate of the system might be dominated by a @xmath54-term , rather than the exponentially suppressed rates found from the adiabatic rg analysis . to check the contribution of these `` irrelevant '' terms , one would have to keep track of their evolution under the rg flow , and then sum them .
this is difficult , so instead we have attacked the problem with a completely different technique .
we take to heart the message of the orthogonality catastrophe and transforming the original hamiltonian ( spin - half coupled to many environment modes ) to a basis of shifted - environment modes .
this powerful and elegant approach is standard for polarons @xcite , and was first applied to the above problem in a number of much neglected papers @xcite .
these works showed that the non - interacting blip approximation of ref .
@xcite is given by a simple weak - coupling analysis of the transformed hamiltonian .
elsewhere , we will present a detailed review of this approach and discuss its regime of validity ( greatly over - estimated in ref .
@xcite ) ; here we simply note the remarkable conclusion that the polaron transformation can map a highly non - markovian environment onto an almost markovian one . treating the transformed problem with a bloch - redfield analysis , we confirm the existence of @xmath55-terms that the rg analysis misses .
they make an important contribution to observables related to the @xmath11- and @xmath56-axis spin - polarizations , as such they are absolutely crucial for analysis of phase information , such as the berry phase @xcite ( note that in earlier versions of @xcite , we used the rg method and so missed the terms that are _ not _ exponentially suppressed ) .
however from our work in progress , it appears that the @xmath54-terms exactly cancel for the quantity considered by leggett _
et al _ ; the @xmath10-axis spin polarization .
we do not yet know if there is a physical reason behind this cancellation .
we also do not know if such a cancellation will occur at higher - order in perturbation theory ( the bloch - redfield analysis involves a born approximation ) .
if this cancellation occurs at all orders , then the transition rate will decay exponentially with @xmath31 for all large @xmath31 .
however if not , we suspect that for extremely large @xmath31 , there will be a cross - over to something that goes like some inverse power of @xmath57 .
@xcite , we show that a noise - induces orthogonality catastrophe can cause a spin to acquires a berry phase when the axis that the noise couples to is rotated around a closed loop .
earlier works @xcite predicted noise - induced berry phases which come from over - damped spin - dynamics , rather than an orthogonality catastrophe . we argue that in our case ( unlike those earlier works ) , the berry phase is relatively weakly affected by decoherence and the leading non - adiabatic phase drops out .
thus we reach the counter - intuitive conclusion that _ environmental noise may make it easier to accurately measure a berry phase_. berry phases occur in many quantum systems @xcite , and have potential applications in both quantum computation @xcite and metrology @xcite .
however in general , berry phases do not appear `` alone '' . if the parameters of a system s hamiltonian complete a closed loop in a time @xmath58 , then the total phase acquired by an eigenstate is @xmath59 where the dynamic phase @xmath60 , the berry phase @xmath61 , and the non - adiabatic ( na ) correction @xmath62 , with @xmath63 being a system energy scale ( such as the gap to excitations ) .
the berry phase is hard to isolate ( and thus hard to use for quantum computation or metrology ) , because it is the second term in this @xmath64-expansion of @xmath65 . to suppress the non - adiabatic terms
, one must make @xmath58 large .
this means @xmath66 , so one must subtract off @xmath67 ( using a spin - echo trick or degenerate states @xcite ) with extreme accuracy .
for example , to get @xmath68 with an error of 0.1% , one must make @xmath69 , which means that @xmath70 .
then since one wants @xmath68 with an error of 0.1% , one has to subtract off @xmath67 with an accuracy of 0.0001% .
in contrast , for the sistuation we consider , the total phase is ; @xmath71 since there is no dynamic phase and @xmath72 is imaginary and so no longer contributes to the phase ( in some cases it is also exponentially small ) .
this can be shown by transforming to a rotating frame @xcite which follows the slowly varying noise - axis
. then we arrive at a problem similar to that discussed in the previous section of the article . in this case to get the berry phase with an error of only 0.1% , one simply needs to ensure that @xmath73 , which requires that @xmath74 .
this requirement is much less strict than that for the convention berry phase above . in conclusion
, we note that we are currently looking at whether such noise can also make non - abelian geometric phases ( wilczek - zee phases @xcite ) .
non - abelian means `` @xmath75 then @xmath76 '' is different from `` @xmath76 then @xmath75 '' . in the above case , it appears possible that if we perform two different loops ( @xmath75 and @xmath76 ) with the noise , we get two different ( su(2 ) ) geometric phases depending on which loop is followed first .
alternatively one can take a bloch - redfield approach , see for example eq .
( 9 ) of ref .
@xcite ( dropping the time - dependence and setting @xmath78 ) .
we see immediately that @xmath79 is real for white - noise ( @xmath80constant ) , thus there is no lamb shift .
jones , v. vedral , a. ekert and g. castagnoli , _ nature _ * 403 * , 869 ( 2000 ) .
a. ekert _ et al _ , j. mod .
opt . * 47 * , 2501 , ( 2000 ) .
pekola , j.j .
toppari , m. aunola , m.t .
savolainen and d.v .
averin , phys .
b * 60 * , r9931 ( 1999 ) .
r. leone , l.p .
lvy and p. lafarge , phys .
100 , 117001 ( 2008 ) .
a. carollo , g.m .
palma , a. ozinski , m. franca santos and v. vedral , phys .
96 * , 150403 ( 2006 ) .
a. carollo and g.m .
palma , laser physics * 16 * , 1595 , ( 2006 ) .
s. dasgupta and d.a .
lidar , j. phys .
b * 40 * , s127 ( 2007 ) .
syzranov and y. makhlin , jetp lett . *
87 * , 390 ( 2008 ) .
|
we briefly review three ways that environmental noise can slow - down ( or speed - up ) quantum transitions ; ( i ) lamb shifts , ( ii ) over - damping and ( iii ) orthogonality catastrophe .
we compare them with the quantum zeno effect induced by observing the system .
these effects are relevant to poor qubits ( those strongly coupled to noise ) .
we discuss berry phases generated by the orthogonality catastrophe , and argue that noise may make it _ easier _ to observe berry phases .
address = institut laue - langevin , 6 rue jules horowitz , b.p .
156 , 38042 grenoble , france .
our presentation for icnf 2009 is based on our work entitled _
`` noise suppressing non - adiabaticity : observing a berry phase alone '' _ @xcite . here , we review the literature about the ways in which environmental noise can change the rate of quantum transitions ( noise - induced quantum zeno effects ) , and then briefly discuss the applications of these ideas to noise - induced berry phases , such as those in ref .
@xcite .
since qubits have been created and studied in experiments , theoretical descriptions of dissipation in two - level system have a renewed relevance .
a `` good '' qubit is a fully controllable two - level system which is sufficiently weakly coupled to environmental noise that it experiences only weak dissipation ( often well - described by the bloch - redfield equation @xcite ) . for quantum computing ,
qubits need to have coherent oscillations with a quality factor of @xmath0 or more ( ten - thousand coherent oscillations before decaying ) .
only then will errors be rare enough that they could be fixed by error correction codes .
in contrast a `` poor '' qubit is a controllable two - level system which is strongly coupled to its environment , and thus experiences strong dissipation .
the theory for this was discussed by leggett _
et al _ @xcite , but there are still many open questions .
it is worth noting that even the best qubits are only `` good '' for certain values of their parameters ( values where the noise couples quadratically rather than linearly to superpositions of the two - levels ) .
the qubits rapidly become `` poor '' away from these special points @xcite .
our aim here is to reveal a little of the rich physics of such poor qubits .
environmental noise ( of classical or quantum origin ) causes dissipation in quantum systems .
we start this brief review with two very general and open questions . * what can a dissipative quantum system do that a non - dissipative system can not ? * could noise be used to _ control _ a quantum system in a manner that can not be achieved using traditional hamiltonian manipulation ? the first question is motivated by the observation that the density matrix of a dissipative two - level system , @xmath1 , is described by _ three _ independent parameters , while a non - dissipative two - level system is described by only _
two_. for example , one can write @xmath2 where @xmath3 , @xmath4 and @xmath5 are real numbers . for non - dissipative ( hamiltonian )
dynamics , all states must have a purity @xmath6=1 $ ] , which means that @xmath7 .
in contrast , for dissipative dynamics we only require that @xmath8 .
this suggests that dissipation makes the dynamics much richer . in this article
we discuss some effects of dissipation , however it is not clear to us whether all possible effects are known
. our second question , which is potentially relevant for quantum information processing , can definitely be answered by `` yes '' !
our presentation at this conference gives one example of this ( see the section below about berry phases and ref .
@xcite ) .
a more obvious example is _ thermalization _ ; which drives the system to a single well - defined state , independent of the system s initial state .
for example , coupling the two - level system to a zero temperature environment will cause it to decay to its ground state , irrespective of what state ( mixed or pure ) it was in before @xcite .
no non - dissipative evolution ( hamiltonian evolution ) can do this .
we wonder if other examples exist ?
|
the top quark was discovered in 1995 by the cdf and d0 collaborations at the fermilab tevatron collider @xcite .
the cdf and d0 collaborations are currently taking data in run 2 of the tevatron .
the increased luminosity and higher collision energy of @xmath4 = 1.96 tev allows for precise measurement of top quark production and decay properties .
the cdf and d0 detectors are described in ref .
@xcite and @xcite .
the top quark is by far the heaviest particle found to date ( @xmath5 gev @xcite ) and is also the quark with the smallest uncertainty on its mass . due to its high mass , it plays a central role in the standard model ( sm ) .
this article focusses on the production cross sections , decay , and properties of the top quark .
the mass measurements are discussed in a separate article in these proceedings @xcite .
the top quark can be pair produced in @xmath6 collisions via the strong interaction and singly via the electroweak interaction in the sm .
the @xmath3 cross section is eleven orders of magnitude lower than the inelastic @xmath6 cross section and several orders of magnitude lower than @xmath7 quark and@xmath8w and @xmath9 boson production , which poses a significant challenge to top quark identification . in the sm , the top quark decays almost exclusively into a @xmath1 boson and a @xmath7 quark , where the @xmath1 boson decays either hadronically or leptonically .
identification of top quarks require the identification of jets , in particular jets from @xmath7 quarks , muons , electrons , and neutrinos .
the identification is based on the fact that decay products have hight transverse momenta and good angular separation in the lab frame .
the principal algorithm used to identify @xmath7 quark jets looks for the presence of charged tracks significantly displaced from the primary vertex coming from the decay of @xmath10 or @xmath11 mesons , which have finite lifetime ( lifetime tagging ) .
top quarks can be produced in pairs via the strong interaction . at the tevatron collision energy of @xmath12
tev the quark - annihilation diagrams make up 85% of the cross section , while gluon fusion accounts for 15% .
theoretical calculations predict a @xmath13 production cross section of @xmath14 pb at a top quark mass of 175 gev @xcite .
the top quark candidate events are classified according to the @xmath1 boson decay mode .
each top quark decays into a @xmath1 boson , which can decay either hadronically or leptonically .
decay channels include di - lepton ( both @xmath1 bosons decay leptonically ) , semi - leptonic ( mixed ) , and all - jets ( only hadronic decays ) .
the @xmath3 production cross section has been measured by the cdf and d0 collaborations in all decay channels .
selected results are presented here .
the di - lepton channel is characterized by the presence of two isolated high @xmath15 leptons , two high @xmath15 @xmath7-jets , and a large missing transverse energy ( @xmath16 ) from the neutrinos .
the background contributions due to instrumental effects are multi - jet production , @xmath1 boson with additional jets ( @xmath1+jets ) , and @xmath17ll events with mis - measured @xmath16 , misidentified jets or misidentified leptons .
these backgrounds are estimated from data .
physics background include z@xmath18 where the @xmath19 leptons decay leptonically and @xmath20/@xmath21 ( di - boson ) production and are estimated from monte carlo ( mc ) simulation .
the di - lepton channel has the advantage of clean lepton identification and therefore high signal to background rations .
however , the branching ratio of 4% ( excluding @xmath19 lepton decay modes ) is low .
an analysis by cdf includes events with all combinations of electron and muon decays of the @xmath1 bosons .
figure [ xsecplot ] shows the number of candidate events in data compared to the expectation from background and signal @xmath3 .
the final selection requires the leptons to have opposite sign charge and a cut on @xmath22 gev .
a preliminary cross section of @xmath23 pb is extracted .
[ xsecplot ] of integrated luminosity .
the top light histogram is the expected contribution from signal with @xmath24 pb as measured in this analysis .
the shaded area is the uncertainty in the total background estimate . *
right : * number of events in the cdf semi - leptonic event selection as a function of jet multiplicity . for events with at least three jets an additional requirement
is made of @xmath22 gev . )
the histograms represent the background expectation .
the black points are the observation in data .
the light - colored histogram is the contribution from @xmath13 normalized to the measured cross section .
, title="fig : " ] of integrated luminosity .
the top light histogram is the expected contribution from signal with @xmath24 pb as measured in this analysis .
the shaded area is the uncertainty in the total background estimate . *
right : * number of events in the cdf semi - leptonic event selection as a function of jet multiplicity . for events with at least three jets
an additional requirement is made of @xmath22 gev . )
the histograms represent the background expectation .
the black points are the observation in data .
the light - colored histogram is the contribution from @xmath13 normalized to the measured cross section .
, title="fig : " ] to increase the efficiency of the identification the d0 collaboration has selected a sample where only one lepton is required to be fully reconstructed allowing for only a track identified from the decay of the other charge @xmath1 boson .
events with one @xmath1 boson decaying into an electron and the other into a muon ( electron - muon events ) are rejected . the signal purity lost by loosening the lepton identification
is recovered by the requirement of at least one @xmath7-jet to be identified ( tagged ) by a lifetime tagger .
the measured cross section is then combined with the cross section from electron - muon events and yields a preliminary @xmath25 pb .
the statistical error in this result is reduced by 15% compared to the d0 result with full lepton identification , keeping the systematic error comparable . in the semi - leptonic decay channel
the identification of the lepton still provides a reasonably high signal to background ratio , although not as clean as the di - lepton channel .
the branching ratio is 29% ( not including the @xmath19+jets branching ratio ) and therefore significantly higher than the di - lepton channel .
this makes the semi - leptonic channel the golden channel for top - quark property measurements .
the dominant background processes are @xmath1+jets and multi - jet production .
cdf measures the cross section in the semi - leptonic channel by selecting a sample of events with an isolated electron ( muon ) with transverse energy ( transverse momentum ) @xmath26 gev , @xmath16>20 gev and jets with @xmath27 gev and @xmath28 .
at least one of the jets is also required to be tagged as a @xmath7-jet by a lifetime tagger .
further separation between signal @xmath3 and @xmath1+jets background is obtained by employing a cut on @xmath29 .
efficiency and physics backgrounds are estimated from mc simulation .
instrumental backgrounds are estimated from data . the selected number of events from 695 pb@xmath30 of data are shown in fig . [ xsecplot ] as a function of the jet multiplicity in the event .
a preliminary @xmath3 cross section is extracted from events with at least three jets and is measured to be @xmath31 pb for a top quark mass of 175 gev . for the @xmath3 cross section measurement in the semi - leptonic channel d0 splits the sample by lepton flavor , jet multiplicity , and number of lifetime tagged jets .
the combined preliminary measurement yields @xmath32 pb from collisions with 360 pb@xmath30 of integrated luminosity . the @xmath3 cross section is also measured in the events where both @xmath1 boson decay hadronically .
this final state is characterized by six high @xmath15 jets , two of which are @xmath7-jets .
the dominant background in this channel is multi - jet production .
although the branching ratio of 46% into this channel is relatively large the background processes have much larger cross sections making the signal hard to identify .
d0 extracts a signal by building invariant masses from two - jet and three - jet combinations in each event .
the jets for the di - jet invariant mass are required not to be tagged by a lifetime tagger ( light jets ) .
for the three - jet invariant mass one of the jets has to be identified as a @xmath7-jet .
figure [ alljet ] shows the di - jet ( left ) and three - jet ( right ) mass distributions .
-tagged jets from the d0 all - jets cross section measurement .
* right : * the three - jet mass spectrum for all combinations of one @xmath7-tagged jet and two non @xmath7-tagged jets .
the green - shaded histogram overlaid on the data points is the distribution from the background sample.,title="fig : " ] -tagged jets from the d0 all - jets cross section measurement . *
right : * the three - jet mass spectrum for all combinations of one @xmath7-tagged jet and two non @xmath7-tagged jets . the green - shaded histogram overlaid on the data points is the distribution from the background sample.,title="fig : " ] [ alljet ] overlaid are the background distributions also obtained from data by assigning a @xmath7-flavor to a jet at random .
the excess of events in the di - jet spectrum reveals the @xmath1 boson , while in the three - jet distribution the excess coming from top quarks can be seen .
the preliminary cross section measured using 360 pb@xmath30 of data is @xmath33 pb . a compilation of cross section measurements by cdf and d0 is shown in fig .
[ topxs ] .
production cross section measurements by d0 ( left ) and cdf ( right).,title="fig : " ] production cross section measurements by d0 ( left ) and cdf ( right).,title="fig : " ] [ topxs ] while top quark pair production via the strong interaction is well established and measured , top quarks may be produced individually via the electroweak interaction .
no evidence for this production channel has not been found to date .
measurement of single top quark production directly access the ckm matrix element @xmath34 and properties of the @xmath35 vertex . at the tevatron
the sm predicts single top quark production to occur from two dominant sets of diagrams , the @xmath36 channel with an expected cross section of @xmath37 pb and the @xmath38 channel diagrams with @xmath39 pb @xcite .
associated production of top quarks is expected to have negligible cross section .
single top quark production is searched for in events with the top quark decaying into a @xmath1 boson and a @xmath7 quark with the @xmath1 boson decaying leptonically .
a lifetime tagger is used to identify the @xmath7 quark jet .
the major challenge of the analyses is separating the single top signal from the @xmath3 and @xmath1+jets background .
multi - variate statistical methods including multi - variate discriminants , artificial neural networks ( ann ) , and decision trees are used by both d0@xcite and cdf .
separate trainings on mc simulated events are performed for the separation of single top signal to @xmath3 and @xmath1+jets .
a likelihood fit is made to the discriminant output as a function of input single top cross section .
signal significance or limits are extracted from the posterior probability density distribution .
no evidence for single top quark production is found by cdf or d0 . with 370 pb@xmath30 of integrated luminosity the limits from d0 ( likelihood discriminant ) at 95% c.l .
are @xmath40 pb for @xmath38-channel and @xmath41 pb for @xmath36-channel .
earlier limits by d0 are published in ref .
@xcite . the limits with 695 pb@xmath30 from cdf ( neural network analysis ) at 95% c.l .
are @xmath42 pb for @xmath36-channel , @xmath43 pb for @xmath38-channel and @xmath44 pb for a combined search .
earlier limits by cdf are published in ref . @xcite .
although single top quark production has yet to be observed , the current limits test models of physics beyond the sm @xcite . figure [ stnp ] shows the allowed region for single top quark production in the 2d @xmath36-channel - @xmath38-channel plane for the d0 neural network analysis .
several model predictions for physics beyond the sm are also shown .
-channel , @xmath38-channel ) plane for the d0 neural network analysis .
several model predictions are also shown in the plot .
, title="fig : " ] [ stnp ] both cdf and d0 continue to refine their analysis techniques to ensure the necessary sensitivity to find evidence for single top quark production with the data available in the very near future .
cdf and d0 measure the ratio @xmath46 . in the sm
this ratio is tightly constrained assuming unitarity of the ckm matrix and exactly three quark generations to the interval 0.9980 - 0.9984 at 90% cl . through the measurement of @xmath47 , the sm and
assumptions can be tested .
in addition , the cross section measurements using lifetime tagging to identify @xmath7-jets also assume @xmath47 to be unity .
a separate determination on @xmath47 can provide a model independent measurement of the cross section .
the measurement is done on a sample similar to the one used for the cross section measurement in the semi - leptonic channel .
@xmath47 is measured from a fit to the relative number of events with zero , one , and two @xmath7-tagged jets .
tagging probabilities also affect these relative numbers and is determined from data control samples .
the cross section information comes from the sum excess of events over the background expectation .
the preliminary result from the simultaneous determination of cross section and @xmath47 on data from 230 pb@xmath30 of integrated luminosity is shown in fig .
[ bratiocharge ] . , @xmath48 ) plane .
the point indicates the best fit to the data . *
right : * the measured value of the top quark charge in 32 events selected by d0 compared to the expected distributions in the sm and exotic case .
, title="fig : " ] , @xmath48 ) plane .
the point indicates the best fit to the data . *
right : * the measured value of the top quark charge in 32 events selected by d0 compared to the expected distributions in the sm and exotic case .
, title="fig : " ] [ bratiocharge ] @xmath47 is found to be @xmath49(stat.+syst . ) and the cross section is @xmath50 pb . using a bayesian method the lower limit on @xmath47 is found to be 0.64 at 95% c.l .
cdf also measures @xmath47 and finds a value of @xmath51 @xcite .
given the @xmath52 structure of the electroweak coupling in the sm and the measured top quark mass , the fraction of @xmath1 bosons from top quark decay with left - handed , longitudinal , and right - handed polarizations are predicted to be @xmath53 and @xmath54 with an uncertainty of order 1% and @xmath55 . a measurement that departs from these values would be a sign of physics beyond the sm .
for example , a @xmath56 term in the @xmath35 coupling would increase @xmath57 but leave @xmath58 unchanged .
top quark decays with the @xmath1 boson decaying into an electron or a muon are used to measure the @xmath1 boson helicity . to measure the helicity of the @xmath1 boson one looks at the distribution of angles between the charged lepton and the top quark ( opposite to the @xmath7 quark ) in the @xmath1 boson rest frame .
this helicity angle can be either directly measured or by using the approximation @xmath59 .
d0 uses the former approach and obtains @xmath60(stat.)@xmath61(syst . ) by combining events from semi - leptonic and di - lepton decays .
a third approach is to use the fact that the transverse momentum of the lepton in the laboratory frame depends on the @xmath1 boson helicity since left - handed @xmath1 bosons tend to emit the lepton opposite of their direction .
cdf combines the result from this approach with the measurement from the @xmath62 approximation and finds @xmath63(stat.+syst . ) or @xmath64 at 95% c.l .
the electric charge is a fundamental quantity characterizing a particle . in the sm
the top quark is defined as having charge + 2/3@xmath65 . in a possible extension to the sm an additional quark doublet @xmath66 with charges ( -1/3@xmath65,-4/3@xmath65 )
is proposed @xcite .
the true top quark in such models is too heavy to be observed at the tevatron and the discovered top quark is indeed @xmath67 .
we can distinguish between the two model by measuring the top quark charge .
the measurement is done on a semi - leptonic sample of @xmath3 candidate events .
the charge of the top quark is measured from the charge of the lepton and the charge of the associated @xmath7-quark .
the lepton charge is measured from the curvature of the charged track in the magnetic field of the detector .
the @xmath7-quark charge is measured from a @xmath15 weighted sum of the charge tracks within jet cone .
one can distinguish between the @xmath7 and @xmath68 on a statistical basis .
figure [ bratiocharge ] shows the reconstructed charges compared to the expectation from sm and -4/3@xmath65 scenario . from a likelihood ratio
fit the data is found to favor the sm and a preliminary exclusion of the -4/3@xmath65 model at 92% c.l can be set .
the top quark lifetime is constrained in the sm to be less than @xmath69 s. however , there is ample experimental room for long - lived top quark in the experimental data .
the top quark lifetime is measured in a sample of @xmath3 candidates selected by identifying an electron / muon , at least three jets , and large missing transverse energy .
one of the jets is required to be identified as a @xmath7 jet from a secondary vertex tag . in such events , the impact parameter ( @xmath70 ) of the lepton to the interaction vertex
is typically expected to be very small .
the vertex location is constrained by the jets .
large impact parameters would arise from longer - lived top quarks leading to displaced @xmath1 boson decay vertexes .
the distributions of lepton impact parameter @xmath70 , only reflecting the resolution of the measurement is predicted using high momentum electron and muon tracks produced in @xmath71 events .
distribution from non @xmath3 backgrounds are evaluated from the mc simulation .
the observed distribution in data is compared to the predicted distribution .
the preliminary result from 318 pb@xmath30 of cdf data shows @xmath70 distributions consistent with expectations and a limit on the top quark lifetime of @xmath72 @xmath73 m at 95% c.l .
the analysis is also sensitive to a new long - lived background to @xmath3 or anomalous top quark production by a long - lived parent particle .
many models beyond the sm predict new particles that couple preferentially to the top quark .
two searches are presented here , a search for a narrow heavy resonance @xmath74 and the search for an additional quark @xmath2 with the same final states as the sm top quark .
a narrow ( @xmath75 ) heavy resonance decaying into @xmath13 pairs is predicted in various topcolor models like `` topcolor assisted technicolor''@xcite . in some models the resonance couples preferentially to third generation quarks and weakly to leptons .
searches from run 1 yield limits in the mass of such a resonance of 480 gev ( cdf ) and 560 gev ( d0 ) .
the search is done by looking at the invariant mass @xmath76 distributions of the @xmath3 pair .
a resonance would show as a peak in the exponentially falling @xmath76 distribution .
cdf uses a matrix element type of analysis . in a sample of semi - leptonic top quark candidate events
a search is performed for @xmath77 gev .
no evidence for resonant @xmath13 production is observed in 680 pb@xmath30 of integrated luminosity .
a small peak previously reported by cdf in the first 320 pb@xmath30 has been diluted .
the corresponding preliminary upper cross section limits as a function of @xmath78 are shown in fig .
[ restprime ] . and @xmath79 standard deviation expected limits . also shown is a line from a theoretical prediction within a leptophobic topcolor model .
* right : * upper limit , at 95% cl , for the production of a hypothetical fourth generation @xmath2 quark as a function of @xmath2 quark mass .
also shown is the a theoretical calculation for the @xmath2 production cross section .
the band represents @xmath80 and @xmath79 standard deviation expected limits . ,
title="fig : " ] and @xmath79 standard deviation expected limits . also shown
is a line from a theoretical prediction within a leptophobic topcolor model .
* right : * upper limit , at 95% cl , for the production of a hypothetical fourth generation @xmath2 quark as a function of @xmath2 quark mass .
also shown is the a theoretical calculation for the @xmath2 production cross section .
the band represents @xmath80 and @xmath79 standard deviation expected limits .
, title="fig : " ] [ restprime ] a lower limit of the existence of a @xmath81 is set to 725 gev at 95% c.l .
d0 also performed a search for a narrow resonance in the @xmath3 invariant mass spectrum . a lower limit of the @xmath81 of 680 gev at 95% c.l .
is set from an integrated luminosity of 370 pb@xmath30 .
several extensions to the sm propose the existence of a heavy fourth generation quark that is not excluded by precision electroweak data or other direct searches @xcite . the new quark is referred to here as @xmath2 , however , it need not to be a standard fourth generation up - type quark . for the presented analysis
it is assumed that this new particle is pair produced via the strong interaction , has a mass greater than the sm top quark , and decays promptly into @xmath82 final states .
the analysis is performed on a sample of candidate events with a lepton and jets in the final state .
the number of expected background events is compared to the number of events selected in data for distributions of the @xmath2 mass , @xmath83 , and @xmath29 .
the background consists of sm @xmath3 and @xmath1+jets .
the mass of the @xmath2 quark is reconstructed from a @xmath84-fit to the kinematic properties of the final state objects .
the significance of a @xmath2 quark signal is extracted from a binned likelihood fit to @xmath29 and @xmath83 .
no evidence for the production of @xmath2 quark is found .
the upper limits for the production cross section as a function of @xmath2 quark mass are shown in fig .
[ restprime ] .
a lower mass limit for @xmath2 can be set at @xmath85 gev at 95% c.l .
eleven years after the top quark discovery , we are now in the position to make precision measurements .
the top quark production cross section via the strong interaction has been measured in all major decay channels and is found to be consistent with the sm expectation .
however , the uncertainties on the cross section still allow exotic decay or production mechanisms .
the electroweak production of top quarks ( single top ) has not been discovered yet .
the current limits are close to the theoretical prediction for the cross section and therefore exclusion or evidence for single top quark production is possible in the near future .
measurements of the properties of the top quark indicate that its charge is consistent with + 2/3@xmath65 and its lifetime @xmath86 shorter than @xmath87 m at 95% c.l . , both consistent with the sm top quark .
one also looks at top quark events to make detailed studies of the electroweak @xmath88 vertex .
branching ratios and @xmath1 boson helicity are found to be consistent with sm expectation .
no evidence for a narrow resonance in top quark decays or additional fourth generation quark is found and limits on the production cross section are set . understanding top quark production and properties will be crucial for success at the large hadron collider ( lhc ) .
what we learn at the tevatron , both in terms of physics and analysis tools , extends directly to the lhc . 99 d0 collaboration , s. abachi _ et al .
_ , phys .
. lett . * 74 * , 2632 ( 1995 ) ; cdf collaboration , f. abe _
et al . _ ,
. lett . * 74 * , 2626 ( 1995 ) . : tevatron electroweak working group ( for the cdf and d0 collaborations ) , hep - ex/0608032 . n. kidonakis and r. vogt , phys .
d * 68 * , 114014 ( 2003 ) , m. cacciari _ et al .
_ , jhep * 404 * , 68 ( 2004 ) .
_ et al . _ ,
instrum . method a * 360 * , 137 ( 1995 ) .
v. abazov _ et al .
_ , physics/0507191 ; fermilab - pub-05/341-e . s. abachi _ et al .
_ , phys .
lett * 95 * , 102002 ( 2005 ) .
b. w. harris _ et al .
_ , phys .
d * 66 * , 054024 ( 2002 ) .
see contribution by shabnam jabeen , these proceedings .
v. abazov _ et al .
_ , phys .
* 622b * , 265 ( 2005 ) . s. abachi _ et al . _ , phys . rev .
d * 71 * , 012005 ( 2005 ) .
see contribution by un - ki yang , these proceedings .
d. chang _ et al .
_ , phys .
d * 59 * , 09153 ( 1999 ) .
t. tait and c. p. yuan , phys . rev .
d * 63 * , 014018 ( 2001 ) . c. t. hill , phys .
* 266b * , 419 ( 1991 ) . c. wagner _
et al . _ , hep - ph/0109097 ; t. han _ et al .
_ , phys . lett . *
563b * , 191 ( 2003 ) ; h .- j .
he , n. polonsky , and s. su , hep - ex/0102144 ; l. okun _ et al .
_ , hep - ph/0111028 .
|
the latest results from the cdf and d0 collaborations on the top - quark pair - production cross section and limits on electroweak production are presented .
included are measurements of properties of the top quark such as charge , lifetime , and the decay branching ratio @xmath0 .
in addition to measurements about the top quark , the selected event samples are used to study the helicity of the @xmath1 boson and to search for additional exotic quarks ( @xmath2 ) and resonances in the @xmath3 invariant mass spectrum .
address = fermilab , p.o .
box 500 , batavia , il 60510 , usa
|
computing drawings of planar graphs is an extensively studied subject . in its most classical
setting , the problem is to design an algorithm that receives as input a planar graph @xmath5 and it produces as output a drawing of @xmath5 such that no two edges of the drawing cross .
aesthetic requirements , such as minimizing the number of bends along the edges , using a small area , or representing the faces as convex or star - shaped polygons can be specified as additional optimization goals .
in addition to these geometric optimization goals , a graph drawing algorithm can receive as input a set of semantic constraints which define placement and/or routing requirements for some of the vertices and/or some of the edges .
for example , a subset of the vertices must be collinear , some of the edges must be horizontal , the size or shape of some edges or vertices is defined in advance ( see , e.g. , @xcite ) .
a well studied graph drawing problem with semantic constraints is the point set emebddability problem .
given a planar graph @xmath5 with @xmath2 vertices and a set @xmath0 of @xmath2 distinct points in the plane , the problem is to compute a crossing - free drawing of @xmath5 such that its vertices are mapped to the points of @xmath0 .
kaufmann and wiese @xcite proved that any planar graph has a point - set embedding with at most two bends per edge on any given point set .
the drawing may however use exponential area , and a polynomial area construction is showed in @xcite .
if no bends along the edges are allowed , cabello @xcite proved that deciding whether a planar graph admits a straight - line point - set embedding is an np - complete problem .
nishat et al . @xcite and moosa and rahman @xcite presented polynomial - time algorithms to test straight - line point - set embeddabiliity of plane 3-trees . at last year
socg , biedl and vatshelle @xcite showed that the problem can be solved in polynomial time for planar graphs with a fixed combinatorial embedding that have constant treewidth and constant face - degree .
they also show that if one of the conditions is dropped ( i.e. , either the treewidth is unbounded or some faces have large degrees ) , the problem becomes np - hard .
gritzman , mohar , pach and pollack @xcite and independently castaeda and urrutia @xcite proved that the outerplanar graphs with @xmath2 vertices are the largest family of graphs admitting a straight - line point - set embedding on any set of @xmath2 points in general position . motivated by the above complexity results and restrictions on the classes of planar graphs that admit a geometric point - set embedding , we study the problem of drawing a planar graph with specified vertex locations from a different angle .
namely , we consider a scenario in which the given locations are fewer than the vertices of the graph . in our setting the algorithm receives as input any planar graph @xmath5 with @xmath2 vertices and any `` sufficiently small '' set @xmath0 of points .
the output is a crossing - free straight - line drawing such that @xmath6 vertices of @xmath5 are mapped to the points of @xmath0 , while the points representing the remaining @xmath7 vertices of @xmath5 are chosen by the algorithm .
we call the output a _ geometric point - subset embedding of @xmath5 on @xmath0_. we stress that @xmath0 can be any set of points , i.e. the drawing algorithm must work for all point sets of `` sufficiently small '' cardinality .
we are interested in making the `` sufficiently small '' cardinality as large as possible .
for example , it is immediate to see that all planar graphs with at least four vertices have a geometric point - subset embedding of @xmath5 on all point sets of cardinality three .
less intuitively , it has been proven in @xcite that all planar graphs with at least six vertices have a geometric point - subset embedding of @xmath5 on all point sets of cardinality four .
the following significantly improves this cardinality .
[ th : general ] let @xmath5 be an @xmath2-vertex plane graph with vertices and let @xmath0 be a set of distinct points in the plane such that @xmath8 .
@xmath5 has a geometric point - subset embedding on @xmath0 , which can be computed @xmath4 time .
we also prove that if the points are in one - sided convex position ( i.e. they form a convex point set such that the leftmost and the rightmost points are adjacent in the convex hull ) , the bound on the cardinality in theorem [ th : general ] can be further increased .
[ th : convex ] let @xmath5 be an @xmath2-vertex plane graph and let @xmath0 be a one - sided convex point set in the plane such that @xmath9{n } \right \rceil$ ] .
@xmath5 has a geometric point - subset embedding on @xmath0 , which can be computed @xmath4 time .
we recall that in @xcite it is proven that there exists a one - sided convex point set @xmath0 of size @xmath10 such that any planar graphs admits a geometric point - subset embedding on @xmath0 .
theorem [ th : convex ] extends this result to all one - sided convex point sets , even if with a slightly smaller cardinality .
[ [ overview - of - the - proof - technique . ] ] overview of the proof technique .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the proofs of theorems [ th : general ] and [ th : convex ] follow a similar approach , although they use different combinatorial results .
the general idea is described in the following steps .
let @xmath5 be an @xmath2-vertex plane graph ( i.e. a planar graph with a given planar embedding ) and let @xmath0 be a set of @xmath11 points with @xmath12 .
step 1 : : : graph @xmath5 is augmented to a maximal plane graph @xmath13 with @xmath2 vertices . step 2 : : : an induced subgraph @xmath14 of size at least @xmath11 is computed .
step 3 : : : a point subset embedding @xmath15 of @xmath16 on @xmath0 is computed .
step 4 : : : a planar embedding preserving straight - line drawing of @xmath13 is constructed by adding vertices and edges to @xmath15 . step 5 : : : the edges of @xmath17 are removed . concerning step 2 , the induced subgraph @xmath16 is outerplanar in the proofs of both theorems . in the proof of theorem
[ th : convex ] , @xmath16 is a biconnected outerplanar graph having at least @xmath18{n } \right \rceil$ ] vertices ; in the proof of theorem [ th : general ] , @xmath16 is an outerpath ( i.e. a biconnected outerplanar graph whose weak dual is a path ) having at least @xmath19 vertices .
the existence of these subgraphs will be proven in theorems [ th : large - induced - outerpath ] and [ th : large - induced - outerplanar ] , where it is also showed how to compute them in @xmath4 time .
the arguments combine a variety of concepts , including schnyder s realizers @xcite , rectangular representations @xcite , stabbing techniques @xcite , 4-block trees @xcite , and mirsky s theorem @xcite .
as for steps 3 and 4 , they rely on different drawing techniques concerning point - set embeddings and straight - line drawings of plane graphs inside star - shaped polygons @xcite .
we conclude this introduction by remarking that theorems [ th : large - induced - outerpath ] and [ th : large - induced - outerplanar ] may be of independent interest .
namely , proving the existence of large induced subgraphs has a long tradition in combinatorial graph theory .
erds , saks , and ss @xcite asked what is the maximum size of an induced tree in a graph and studied the relation between such size and other natural graph parameters .
for example , they give upper and lower bounds on large induced trees in @xmath20-free graphs ; these bounds have been improved by matouek and mal @xcite and by fox , loh , and sudakov @xcite .
restricting to paths , aroha and valencia @xcite showed that every 3-connected planar graph with a large number of vertices has a long induced path .
more specifically , they proved that if a @xmath21-connected graph @xmath5 has a vertex of degree @xmath22 , then @xmath5 contains an induced path of length at least @xmath23 . generalizing this result to non - planar graphs , bhme et al .
@xcite proved that for every positive integers @xmath24 and @xmath25 there exists an integer @xmath26 such that every @xmath11-connected graph of order at least @xmath2 contains either an induced path of length @xmath25 or a subdivision of the complete bipartite graph @xmath27 .
however , their construction , do not seem to yield neither an explicit function for the length of the path nor an algorithm to compute it .
an implication of our results is that every @xmath2 vertex maximal planar graph has an induced path of length at least @xmath28 , which can be computed in @xmath4 time .
the rest of the paper is organized as follows . in section [
se : preliminaries ] we give preliminary definitions .
the techniques behind the proofs of theorem [ th : general ] and [ th : convex ] are described in sections [ se : general ] and [ se : convex ] , respectively .
finally , open problems can be found in section [ se : conclusions ] .
let @xmath5 be a graph .
a _ drawing _ @xmath29 of @xmath5 is a mapping of each vertex @xmath30 of @xmath5 to a point @xmath31 in the plane , and of each edge @xmath32 to a jordan arc connecting @xmath33 and @xmath31 not passing through any other vertex .
@xmath29 is _ straight - line _ if its edges are drawn as straight - line segments .
@xmath29 is _ planar _ if no two edges intersects ( except possibly at common endpoints ) . a planar drawing @xmath29 partitions the plane into topologically connected regions called _ faces _ ; the unbounded region is the _ external _ face and the other regions are the _ internal _ faces .
the boundary of a face is its delimiting circuit .
all the face boundaries of a biconnected graph are simple circuits .
a planar drawing determines a circular ordering of the edges around each vertex .
the cyclic ordering of the edges around each vertex of @xmath29 together with a choice of the external face is a _
planar embedding _ of @xmath5 .
a _ plane graph _ is a graph with a fixed planar embedding .
the boundary of the external face of a plane graph @xmath5 will be also called the _ external boundary _ of @xmath5 . the _ dual graph _
@xmath34 of a given plane graph @xmath5 is a plane graph that has a vertex corresponding to each face of @xmath5 , and an edge joining two vertices corresponding to neighboring faces of @xmath5 .
the _ weak dual _ of @xmath5 is the graph obtained from @xmath35 after removing the vertex that corresponds to the external face of @xmath5 .
a @xmath21-cycle of a graph @xmath5 is also called a _ triangle _ of @xmath5 .
a planar ( plane ) graph is _ maximal _ if each face of the graph is a triangle , thus no edge can be added to it without violating planarity .
a subgraph @xmath16 of a graph @xmath5 is said to be _ induced _ if , for any pair of vertices @xmath36 and @xmath30 of @xmath16 , @xmath37 is an edge of @xmath16 if and only if @xmath37 is an edge of @xmath5 .
let @xmath5 be a graph .
a _ separating @xmath11-ple _ of @xmath5 is a set of @xmath11 vertices whose removal disconnects @xmath5 .
a graph @xmath5 is _ @xmath11-connected _ if it does not contain a separating @xmath38-ple .
every maximal planar graph is @xmath21-connected and every separating @xmath21-ple of @xmath5 ( if any ) is a triangle of @xmath5 .
let @xmath39 and @xmath40 be two separating triangles of a maximal plane graph @xmath5 ; we say that @xmath40 is _ nested _ inside @xmath39 and write @xmath41 if at least one vertex of @xmath40 is inside @xmath39 in the planar embedding of @xmath5 .
a graph is _ outerplanar _ if it admits a planar embedding where all vertices are on the external boundary .
outerpath _ is an outerplanar graph whose weak dual is a path .
let @xmath16 be an outerpath and let @xmath42 and @xmath43 be the two faces corresponding to the endpoints of @xmath44 .
the edges of @xmath42 and @xmath43 that belong to the external boundary of @xmath16 are called the _ extremal edges _ of @xmath16 .
let @xmath45 be an extremal edge on the boundary of @xmath42 , let @xmath46 be an extremal edge on the boundary of @xmath43 .
any path in the external boundary of @xmath16 that connects an endvertex of @xmath45 with an endvertex of @xmath46 is called a _ side path _ of @xmath16 .
it is easy to see that two extremal edges @xmath45 and @xmath46 define two side paths @xmath47 and @xmath48 and that every edge of @xmath16 that is not on the external boundary has an endvertex in @xmath47 and an endvertex in @xmath48 .
in other words both @xmath47 and @xmath48 are induced paths .
let @xmath49 be an extremal edge of @xmath16 ; any side path starting at an endvertex of @xmath49 is also called a _ side path of @xmath16 with respect to @xmath49_. let @xmath5 be a maximal plane graph that contains an induced outerpath subgraph @xmath16 with an extremal edge @xmath49 on the external boundary of @xmath5 and such that there is a side path of @xmath16 with respect to @xmath49 that has @xmath11 vertices .
we say that @xmath5 _ contains an external outerpath @xmath16 with a side path of size @xmath11_. let @xmath0 be a set of @xmath11 points on the plane , @xmath50 and let @xmath5 be a planar graph with @xmath2 vertices , with @xmath51 .
a _ geometric point - subset embedding _ of @xmath5 on @xmath0 is a planar straight - line drawing @xmath29 of @xmath5 such that each point of @xmath0 represents a vertex of @xmath5 .
if @xmath5 is a plane graph and the planar embedding of @xmath5 defined by @xmath29 is the same as the one of @xmath5 , we say that @xmath29 _ preserves _ the planar embedding of @xmath5 .
all the algorithms of this paper compute drawings that preserve the planar embedding of the input graph .
so when we write `` geometric point - subset embedding '' we mean `` geometric point - subset embedding that preserves the planar embedding '' .
in this section we prove theorem [ th : general ] . the main ingredients of the proof are : ( i ) a drawing algorithm that , given a set @xmath0 of @xmath11 points and a maximal plane graph @xmath5 that contains an external outerpath with a side path of size @xmath11 , computes a geometric point - subset embedding of @xmath5 on @xmath0 ( section [ ss : general - embedding ] ) .
( ii ) an algorithm that , given a maximal plane graph @xmath5 , computes an external outerpath of @xmath5 with a side path of size at least @xmath28 ( section [ ss : general - induced ] ) .
let @xmath52 be a polygon on the plane , a _ kernel _
@xmath53 of polygon @xmath52 is the set of all points inside @xmath52 from which all vertices of @xmath52 are visible .
we say that @xmath52 is _ star - shaped _ if @xmath54 .
[ theorem : hongnagamochi ] let @xmath5 be an @xmath2-vertex @xmath21-connected planar graph and let @xmath55 be its external boundary .
every planar straight - line drawing of @xmath55 as a star - shaped polygon can be extended to a planar straight - line drawing of @xmath5 , where each internal face is represented by a convex polygon .
such a drawing can be computed in @xmath4 time . [
le : embedding.1.a ] let @xmath0 be a set @xmath11 distinct points in the plane and let @xmath5 be an @xmath2-vertex maximal plane graph that contains an external outerpath with a side path of size at least @xmath11 .
@xmath5 admits a geometric point - subset embedding on @xmath0 , which can be computed in @xmath56 time .
assume that the points of @xmath0 have distinct @xmath57-coordinates . if not , we can rotate the plane so to achieve this condition
. let @xmath58 be the points of @xmath0 according to their left - to - right order and let @xmath59 be the coordinates of @xmath60 ( @xmath61 ) .
without loss of generality assume @xmath62 , @xmath63 .
if this is not the case we can translate the plane to achieve this condition . let @xmath52 be the side path of the external outerpath @xmath16 of @xmath5 that has size at least @xmath11 .
if @xmath52 has more than @xmath11 vertices we can augment @xmath0 with additional points .
so without loss of generality assume that @xmath52 has exactly @xmath11 vertices .
let @xmath64 be the vertices of @xmath52 in the order they appear along @xmath52 with @xmath65 being the vertex of @xmath52 on the external boundary of @xmath5 .
map @xmath66 to @xmath60 and draw each edge @xmath67 as the segment @xmath68 ( see figure [ fi : drawing1 ] for an illustration of the drawing technique ) .
all vertices of @xmath16 that are not in @xmath52 form a second side path @xmath69 of @xmath16 . since @xmath69 is also an induced path , all the edges of @xmath16 that are not in @xmath52 nor in @xmath69 connect vertices of @xmath52 to vertices of @xmath69 .
let @xmath70 be the vertices of @xmath69 in the order they appear along @xmath69 with @xmath71 being the vertex of @xmath69 on the external boundary of @xmath5 .
we want to place these vertices below the points of @xmath0 in such a way that each point of @xmath0 can be connected to the vertices of @xmath69 without crossing the segments @xmath68 .
let @xmath72 , let @xmath73 , and let @xmath74 .
let @xmath75 be the slope of segment @xmath68 ( @xmath76 ) .
we have @xmath77 for every @xmath76 .
we place vertex @xmath78 at point @xmath79 .
the slope of each segment @xmath80 is at most @xmath81 and therefore each @xmath80 does not cross any segment @xmath68 . also , segments @xmath80 do not cross each other .
namely , let @xmath82 and @xmath83 be two such segments .
since the left - to - right order of the vertices of @xmath52 in the drawing is the order in which they appear along @xmath52 going from the external boundary to the interior , and the bottom - to - top order of the vertices of @xmath69 in the drawing is the order in which they appear along @xmath69 going from the external boundary to the interior , then @xmath84 and @xmath85 .
this means that the slope of @xmath82 is less than the slope of @xmath83 ; since @xmath86 is below @xmath87 , the two segments do not cross each other .
the drawing computed so far is a geometric point - subset embedding of @xmath16 on @xmath0 .
also , the drawing preserves the embedding of @xmath16 .
let @xmath88 be the unique vertex of the external boundary that does not belong to @xmath16 .
in order to complete the drawing we place the vertex @xmath88 at point @xmath89 and draw all the edges connecting it to all its adjacent vertices that have already been drawn . also in this case , the added edges do not cross the existing ones .
so the obtained drawing @xmath90 is a geometric point - subset embedding of the graph @xmath91 induced by the vertices in @xmath16 plus vertex @xmath88 . and also in this case the planar embedding is preserved .
it is also easy to see that all the faces of @xmath90 are star - shaped polygons .
notice that , in the planar embedding of @xmath5 each vertex not yet drawn is inside one of the cycles represented by the faces of @xmath91 .
let @xmath55 be one of these cycles and let @xmath13 be the graph induced by the vertices of @xmath55 plus all the vertices that are inside @xmath55 in the planar embedding of @xmath5 .
all the internal faces of @xmath13 are triangles , while its external boundary is @xmath55 ( which therefore can have more than three vertices ) .
notice that there is no edge connecting two non - consecutive vertices of the external boundary of @xmath13 ; namely if such an edge existed it would have been drawn in @xmath90 and therefore @xmath55 would not have been a cycle of @xmath91 .
an internally triangulated graph is @xmath21-connected if and only if there is no edge connecting two non - consecutive vertices of its external boundary ( see , for example , @xcite ) .
hence , @xmath13 is @xmath21-connected and its external boundary is drawn as a star - shaped polygon . by theorem [ theorem : hongnagamochi ]
the drawing of @xmath55 can be extended to a planar drawing of @xmath13 .
since this is true for all cycles that contain some not yet drawn vertices , the drawing @xmath90 can be completed to a geometric point - subset embedding @xmath29 of @xmath5 on @xmath0 .
since a @xmath21-connected planar graph has a unique embedding once the external face is fixed , the completion of @xmath90 to @xmath29 preserves the embedding of @xmath5 . concerning the time complexity , it is easy to see that the drawing can be computed in @xmath4 time once the input points are sorted from left to right .
such a sorting preprocessing requires time @xmath92 .
[ le : embedding.1.b ] let @xmath5 be an @xmath2-vertex maximal plane graph .
let @xmath65 , @xmath93 , and @xmath94 be the three external vertices of @xmath5 and let @xmath95 , @xmath96 , and @xmath97 be three vertices of an internal face @xmath98 such that there exists three induced paths @xmath99 , @xmath100 , and @xmath101 in @xmath5 ( possibly of zero length ) .
let @xmath102 be a straight - line drawing of @xmath98 on any set @xmath0 of three points .
@xmath5 admits a geometric point - subset embedding on @xmath0 that has @xmath102 as a subdrawing .
such a point - subset embedding can be computed in @xmath4 time .
let @xmath103 , @xmath104 , and @xmath105 be the three points representing @xmath65 , @xmath93 , and @xmath94 , respectively .
let @xmath106 be the bisector of the angle at @xmath60 of triangle @xmath102 ( @xmath107 ) , and let @xmath108 be the half - line contained in @xmath109 that is completely outside @xmath102 ( except for point @xmath60 ) .
let @xmath110 be the vertices of the path @xmath111 ( @xmath107 ) .
place vertex @xmath112 along @xmath108 at distance @xmath113 from @xmath60 ( see figure [ fi : drawing2 ] for an illustration of the drawing technique ) .
since each of the paths @xmath111 is induced , there is no edge connecting non consecutive vertices along the path , and therefore each path can be drawn as a subdivision of a segment .
we also add to the drawing all segments representing edges connecting the vertices of the three induced paths . since the angle between @xmath109 and @xmath108 is less than @xmath114 , and
since the vertices of each path @xmath115 are placed along @xmath108 in the same order they have along the path @xmath115 , the resulting drawing is planar .
moreover , each face of the drawing is drawn as a star - shaped polygon . in the planar embedding of @xmath5
each vertex not yet drawn is inside one of the cycles represented by the faces of the computed drawing .
thus , the drawing can be completed by using theorem [ theorem : hongnagamochi ] as described in the proof of lemma [ le : embedding.1.a ] .
it is immediate to see that the computed drawing preserves the embedding and that the algorithm time complexity in @xmath4 .
we now show that each @xmath2-vertex maximal plane graph contains an external outerpath with a side path of size @xmath116 .
our proof is based on the fact that an @xmath2-vertex maximal plane graph contains either many nested separating triangles or a large @xmath117-connected subgraph .
thus , we first concentrate on graphs with many separating triangles and then on @xmath117-connected graphs . in order to prove that a graph with many separating triangles contains a large induced outerpath , we need to introduce the schnyder realizer @xcite , which will also be used in section [ se : convex ] .
[ [ schnyder - realizers . ] ] schnyder realizers .
+ + + + + + + + + + + + + + + + + + + let @xmath5 be a maximal plane graph with @xmath118 vertices .
a _ schnyder realizer _
@xcite of @xmath5 is an orientation plus a coloring of the internal edges of @xmath5 such that : 1 .
each internal edge receives one of three colors @xmath119 , @xmath120 , and @xmath121 ; 2 .
each internal vertex has exactly one outgoing edge of each color ; 3 .
the counterclockwise order of the outgoing edges around each internal vertex is @xmath119 , @xmath120 , @xmath121 ; 4 . for each internal vertex
, the incoming edges of a color appear counterclockwise between the two outgoing edges of the other two colors .
see figure [ fi : schnyder ] for an example of a schnyder realizer of a maximal plane graph equipped with .
note that conditions @xmath21 and @xmath117 of the definition of a schnyder realizer imply that the counterclockwise circular ordering of the edges around each internal vertex is the one shown in figure [ fi : circularorder ] , i.e. , outgoing colored @xmath119 , incoming colored @xmath121 , outgoing colored @xmath120 , incoming colored @xmath119 outgoing colored @xmath121 , incoming colored @xmath120 , where each of the incoming set can be empty .
it is known that every maximal planar embedded graph has a schnyder realizer .
let @xmath122 , @xmath123 , and @xmath124 be the subgraph induced by the edges of color @xmath119 , @xmath120 , and @xmath121 , respectively .
it is possible to prove each of @xmath122 , @xmath123 , and @xmath124 is a tree containing all inner vertices and exactly one outer vertex .
each edge in each tree is oriented from a vertex to its parent and the single outer vertex of each tree has only incoming edges and therefore is the root of the tree .
we will denote the root of each tree @xmath125 by @xmath126 , @xmath107 .
the three roots @xmath127 , @xmath128 , and @xmath129 are distinct and their counterclockwise order on the external boundary is @xmath119 , @xmath120 , and @xmath121 .
in the following we denote a schnyder realizer @xmath130 of @xmath5 as @xmath131 .
let @xmath5 be a maximal plane graph equipped with a schnyder realizer @xmath131 .
let @xmath30 be an internal vertex and let @xmath132 , @xmath107 be the oriented path of @xmath125 from @xmath30 to @xmath126 .
the path @xmath132 is called the _ @xmath133-path starting at @xmath30_. for @xmath134 ( @xmath135 ) , the two paths @xmath132 and @xmath136 only share the vertex @xmath30 .
thus , for each internal vertex @xmath30 , the three paths @xmath137 , @xmath138 , and @xmath139 divide @xmath5 into three regions @xmath140 , @xmath141 , and @xmath142 , where @xmath143 ( @xmath107 ) denotes the vertices that are inside the cycle @xmath144 ( where indices are taken modulo @xmath21 ) .
notice that the regions @xmath140 , @xmath141 , and @xmath142 do not contain the vertices of @xmath137 , @xmath138 , and @xmath139 , @xmath138 , and @xmath139 .
for our purposes , however , it results more useful to use this different definition . ] .
the following lemmas can be proven by exploiting the properties of schnyder realizer .
[ lemma : schnyder_induced ] let @xmath5 be a maximal plane graph equipped with a schnyder realizer @xmath131 and let @xmath30 be an internal vertex of @xmath5 .
each @xmath132 ( @xmath107 ) is an induced path of @xmath5 .
let @xmath145 be the vertices of @xmath132 in the order they appear walking along @xmath132 from @xmath30 to @xmath126 .
suppose , as a contradiction that @xmath132 is not induced .
this means that @xmath132 has a chord , i.e. , an edge connecting two non consecutive vertices of @xmath132 .
we first observe that the chord must be completely contained either in the region @xmath146 or in the region @xmath147 ( except for the endvertices ) , because otherwise either @xmath126 would not be on the external boundary , or the chord would be crossed by @xmath148 and by @xmath149 .
also , the chord can not be an edge colored @xmath133 , otherwise there would be a cycle in @xmath125 . as a consequence the chord can not be incident to vertex @xmath126 ( all the edges incident to @xmath126 have color @xmath133 except the two on the external boundary ) . for each vertex @xmath150 ( @xmath151 )
the edge @xmath152 is the unique outgoing edge of @xmath150 colored @xmath133 , while @xmath153 is an incoming edge of @xmath150 colored @xmath133 . by the properties of the schnyder realizer the outgoing edge of @xmath150 colored @xmath154 appears between @xmath152 and @xmath153 in the counterclockwise order around @xmath150 , while the incoming edges colored @xmath154 appears between @xmath153 and @xmath152 in the counterclockwise order around @xmath150 . in other words , the outgoing edge of @xmath150 colored @xmath154 and the incoming edges of @xmath150 colored @xmath154
are on different sides of the path @xmath132 . the same is true also for the edges colored @xmath155 .
this implies that there can not be a chord colored @xmath154 or @xmath155 connecting two vertices @xmath150 and @xmath156 with @xmath157 .
the only possibility remaining is that the chord connects @xmath30 to a vertex @xmath150 with @xmath151 and , as already said , it is colored either @xmath154 or @xmath155 .
the chord can not be an outgoing edge of @xmath30 otherwise one of the two paths @xmath148 and @xmath149 would share a vertex with @xmath132 , which is not possible .
so the chord @xmath158 must be an incoming edge of @xmath30 colored @xmath154 or @xmath155 .
suppose it is colored @xmath154 ( the other case is symmetric )
. then @xmath49 must be contained in the region @xmath146 ( it must appear counterclockwise between the outgoing edge of @xmath30 colored @xmath133 and the outgoing edge of @xmath30 colored @xmath155 ) . on the other hand , @xmath49 is also an outgoing edge of @xmath150 colored @xmath154 and therefore it must be contained in the region @xmath147 ( it must appear counterclockwise between @xmath152 and @xmath153 ) .
but , as already said , a chord must be contained in only one region , and therefore we have a contradiction .
[ lemma : schnyder_outerpath ] let @xmath5 be a maximal plane graph equipped with a schnyder realizer @xmath131 and let @xmath30 be an internal vertex of @xmath5 .
the subgraph @xmath159 of @xmath5 induced by @xmath132 , @xmath136 ( @xmath135 ) is an outerpath with the edge @xmath160 on the external boundary of @xmath5 .
also , @xmath132 and @xmath136 are both side paths of @xmath159 with respect to @xmath160 .
let @xmath55 be the cycle @xmath161 .
since @xmath132 and @xmath136 are both induced paths , then every edge of @xmath159 that is not in @xmath55 connects a vertex of @xmath132 to a vertex of @xmath136 . if one of such edges was outside @xmath55 , then it would be crossed by path @xmath162 ( @xmath163 ) . since all edges not in @xmath55 are inside @xmath55 , then @xmath159 is outerplanar .
since each edge not in @xmath55 connects a vertex of @xmath132 to a vertex of @xmath136 , then @xmath159 is an outerpath and both @xmath132 and @xmath136 are its side paths .
clearly , edge @xmath160 is on the external boundary of @xmath5 and both , @xmath132 and @xmath136 are side paths with respect to @xmath160 .
a _ sequence of nested separating triangles _ of a maximal plane graph @xmath5 is a set of @xmath11 separating triangles @xmath164 such that @xmath165 , for @xmath76 .
[ lemma : ordered_triangles ] let @xmath5 be an @xmath2-vertex maximal plane graph .
if @xmath5 contains a sequence of @xmath11 nested separating triangles , then @xmath5 contains an external outerpath with a side path of size at least @xmath166 .
let @xmath167 be the sequence of nested separating triangles of @xmath5 ( see figure [ fig : sep_trianges_2 ] ) . since @xmath168 is a separating triangle , there exists a vertex @xmath36 which lies inside @xmath168 .
let @xmath131 be a schnyder realizer of @xmath5 .
since the three paths @xmath169 , @xmath170 , and @xmath171 of @xmath5 are disjoint , each path visits each triangle @xmath172 ( @xmath61 ) exactly once .
triangles @xmath173 contain in total at least @xmath174 distinct vertices .
therefore at least one among @xmath169 , @xmath170 , and @xmath171 contains at least @xmath166 vertices .
suppose it is @xmath169 . by lemma [ lemma : schnyder_outerpath ] ,
the graph @xmath175 induced by the vertices of @xmath169 and @xmath170 is an external outerpath of @xmath5 with a side path of size at least @xmath166 ( the same is true for the graph @xmath176 induced by the vertices of @xmath169 and @xmath171 ) . the proof that a @xmath117-connected graph contains a large induced outerpath relies on the following result by thomassen @xcite .
a planar drawing of @xmath5 such that all edges ( except four ) are vertical or horizontal straight - line segments and such that the four exceptional edges each consist of a vertical and a horizontal straight - line segment is called _ rectangular representation _ of @xmath5 . [
theorem : thomassen ] let @xmath5 be a @xmath117-connected maximal plane graph and let @xmath177 and @xmath178 be two triangles of @xmath5
. then @xmath34 has a rectangular representation such that @xmath88 corresponds to the unbounded face and such that @xmath36 , @xmath30 , and @xmath179 correspond to rectangles as shown in figure [ fig : thomassen ] .
[ lemma : four_connected ] let @xmath5 be an @xmath2-vertex @xmath117-connected maximal plane graph
. then @xmath5 contains an external outerpath with a side path of size at least @xmath180 .
let @xmath181 be the external boundary of @xmath5 and let @xmath179 be the vertex of @xmath5 , such that @xmath182 is a face of @xmath5 . by theorem [ theorem : thomassen ] , @xmath34 has a rectangular representation such that @xmath88 corresponds to the unbounded face and such that @xmath36 , @xmath30 , and @xmath179 correspond to rectangles as shown in figures [ fig : thomassen ] and [ fig : thomassenconf ] .
i.e , @xmath5 is the dual of a subdivision of a rectangle into @xmath183 rectangles . in
@xcite it is shown that there exists a horizontal or vertical line ( called a _ stabber _ ) that intersects at least @xmath180 rectangles in this subdivision .
the rectangles stabbed by a stabber represent an induced path @xmath52 of @xmath5 . in the following
we show that @xmath5 contains an external outerpath that has @xmath52 as a side path .
we consider two cases based on whether the long stabber is vertical or horizontal .
long stabber is horizontal : : refer to figures [ fig : thomassen_horizontal]- [ fig : thomassenconf_horizontal ] .
path @xmath52 connects @xmath36 to @xmath179 and edges @xmath184 and @xmath37 belong to @xmath5 .
let @xmath55 be the cycle @xmath185 and let @xmath16 be the subgraph induced by the vertices in @xmath55 .
since @xmath52 is an induced path , each edge of @xmath16 that does not belong to @xmath55 is incident to @xmath30 . from the reactangular representation , it is immediate to see that all the edges connecting @xmath30 to @xmath52 lie between @xmath186 and @xmath187 in the clockwise circular order around @xmath30 , i.e. , they are inside the cycle @xmath55 .
thus , @xmath16 is an outerpath ; it has an extremal edge ( i.e. , the edge @xmath37 ) on the external boundary of @xmath5 ; and @xmath52 is a side path of @xmath16 with respect to @xmath37 .
long stabber is vertical : : refer to figures [ fig : thomassen_vertical]- [ fig : thomassenconf_vertical ] . let @xmath188 be the vertex representing the highest rectangle stabbed by the stabber .
edge @xmath189 belongs to @xmath5 ( recall that @xmath88 is represented by the unbounded face ) .
moreover , vertex @xmath88 is not adjacent to any other vertex of @xmath52 except for @xmath30 .
let @xmath55 be the cycle @xmath190 .
since @xmath52 is an induced path and there is no edge connecting @xmath88 to a vertex of @xmath52 distinct from @xmath30 , the subgraph @xmath16 induced by the vertices of @xmath55 coincides with @xmath55 itself , i.e. , @xmath16 is a cycle .
thus , @xmath16 is an outerpath ; it has an extremal edge ( i.e. , the edge @xmath191 ) on the external boundary of @xmath5 ; and @xmath52 is a side path of @xmath16 with respect to @xmath37 .
we are now ready to combine the results of section [ ss : general - embedding ] and [ ss : general - induced ] in order to prove theorem [ th : general ] . to this aim ,
we use the following decomposition tree . [ [ block - tree . ] ] 4-block tree .
+ + + + + + + + + + + + + let @xmath5 be a maximal plane graph .
a _ @xmath117-block tree _
@xcite @xmath192 of @xmath5 is a directed tree containing a vertex for each triangle of @xmath5 ( refer to figure [ fig:4block].a - b ) .
triangle of @xmath5 that corresponds to a vertex @xmath193 of @xmath192 is denoted by @xmath194 .
@xmath192 contains edge directed from its vertex @xmath193 to its vertex @xmath195 if @xmath196 and there is no triangle @xmath197 in @xmath5 such that @xmath198 .
we assume that @xmath192 is rooted at its source i.e. at the vertex which corresponds to the external face of @xmath5 .
note that the leaves of @xmath192 correspond to the internal faces of @xmath5 .
the subgraph of @xmath5 induced by the vertices of @xmath194 and all the vertices of the children of @xmath193 is called the _ pertinent _ graph of @xmath193 and is denoted by @xmath199 ( see figure [ fig:4blocktree_pert ] ) .
[ th : large - induced - outerpath ] in every @xmath2-vertex maximal plane graph @xmath5 there exists a maximal plane subgraph @xmath13 that contains an external outerpath @xmath16 with a side path of size at least @xmath200 . both @xmath13 and @xmath16
can be computed in @xmath4-time .
let @xmath5 be a @xmath2-vertex maximal plane graph and let @xmath192 be its triangle decomposition tree .
let @xmath201 be a function of @xmath2 to be specified later .
we consider two cases based on whether the degree of @xmath192 is bounded by @xmath201 or not .
* the maximum outdegree of @xmath192 is at most @xmath201 . * since @xmath5 is a maximal plane graph it contains exactly @xmath202 faces .
thus , @xmath192 contains @xmath202 leaves and therefore , the height of @xmath192 is at least @xmath203 .
this means that @xmath5 contains a sequence of at least @xmath204 nested separating triangles . by lemma [ lemma : ordered_triangles ]
, @xmath5 contains an external outerpath @xmath16 with a side path of size at least @xmath205 .
* there exists a vertex of @xmath192 with outdegree at least @xmath206 .
* let @xmath193 be such a vertex .
then the pertinent graph @xmath199 of @xmath193 contains at least @xmath206 faces .
since @xmath199 is maximal planar graph , it contains at least @xmath207 vertices . by lemma [ lemma : four_connected ]
, @xmath199 contains an external outerpath @xmath16 with a side path of size at least @xmath208 .
thus , the size of the side path of @xmath16 is @xmath209 by setting @xmath210 , we get that @xmath211 .
we prove now that @xmath13 and @xmath16 can be computed in @xmath4 time .
first , we compute the schnyder realizer of @xmath5 in @xmath4 time @xcite .
then , we compute the longest path of each of the trees @xmath122 , @xmath123 , and @xmath124 and take the longest of the three .
if such a path @xmath114 has length at least @xmath212 we are in the first case above .
the computed path @xmath114 is a path @xmath213 for some internal vertex @xmath36 of @xmath5 .
@xmath16 is the subgraph induced by @xmath213 and one of the other two paths @xmath214 and @xmath215 ( in this case @xmath13 coincides with @xmath5 ) . if @xmath114 is not long enough , we are in the second case above . in this case
we can find @xmath216 by computing the @xmath117-block tree @xmath192 , by performing a visit of @xmath192 in order to find the node with the largest outdegree , and finally by computing the pertinent graph of @xmath193 .
all this can be computed in @xmath4-time @xcite .
once we have @xmath199 , we need to compute a rectangular representation of it with the properties described in the proof of lemma [ lemma : four_connected ] and then to compute the long stabber .
a rectangular representation can be computed in @xmath4 time @xcite and the stabber can also be computed in @xmath4 time @xcite . in this case
@xmath16 is the subgraph induced by the path corresponding to the stabber together with a vertex of the external boundary that is adjacent to both endvertices of the path ( in this case @xmath13 is @xmath199 ) .
[ [ proof - of - theoremthgeneral . ] ] proof of theorem [ th : general ] .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath0 be a set of distinct points in the plane such that @xmath217 and let @xmath5 be an @xmath2-vertex plane graph . if @xmath5 is not maximal we can add dummy edges to it in order to make it maximal ( these edges will be removed at end from the obtained drawing ) . thus , without loss of generality , assume that @xmath5 is maximal .
by theorem [ th : large - induced - outerpath ] , in @xmath5 there exists a maximal plane subgraph @xmath13 of @xmath5 that contains an external outerpath with a side path of size at least @xmath218 .
if @xmath13 coincides with @xmath5 , then , by lemma [ le : embedding.1.a ] , @xmath5 admits a geometric point - subset embedding on @xmath0 .
if @xmath13 does not coincide with @xmath5 , then , by lemma [ le : embedding.1.a ] , @xmath13 admits a geometric point - subset embedding @xmath90 on @xmath0 . since @xmath13 is maximal plane , the external boundary of @xmath13 is a triangle @xmath219 and it is drawn in the plane as a ( geometric ) triangle @xmath102 .
all the other vertices and edges of @xmath13 are drawn inside @xmath102 .
if we remove from @xmath5 all the internal vertices of @xmath13 we obtain a new maximal plane graph @xmath220 ; the triangle @xmath219 is a separating triangle of @xmath5 and is an internal face of @xmath220 .
let @xmath36 be an internal vertex of @xmath13 and let @xmath131 be a schnyder realizer of @xmath5 .
since the three paths @xmath169 , @xmath170 , and @xmath171 of @xmath5 are disjoint and @xmath219 is a separating triangle , each vertex of @xmath219 belong to exactly one of the three paths @xmath169 , @xmath170 , and @xmath171 , which are induced paths by lemma [ lemma : schnyder_induced ] .
it follows that @xmath220 satisfies the conditions of lemma [ le : embedding.1.b ] and therefore it admits a geometric point - subset embedding @xmath221 on @xmath0 that has @xmath102 has a subdrawing .
the union of @xmath90 and @xmath221 is a geometric point - subset embedding of @xmath5 on @xmath0 . by theorem [ th : large - induced - outerpath ] , lemma [ le : embedding.1.a ] , and lemma [ le : embedding.1.b ] , the time complexity of the drawing algorithm is @xmath56 where @xmath11 is the size of @xmath0 . since @xmath222 , we have that @xmath223 which concludes the proof of theorem [ th : general ] .
let @xmath224 be a function such that for every point set @xmath0 of size @xmath224 , any planar graph admits a geometric point - subset embedding on @xmath0 .
angelini et al . proved that @xmath225 , and ask whether @xmath226 . an immediate consequence of theorem [ th : general ] is the following . [ co : infty ] @xmath226 .
furthermore , theorem [ th : large - induced - outerpath ] implies the following corollary which is , for maximal planar graphs , an improvement over the result by aroha and valencia @xcite who showed that every @xmath21-connected planar graph with @xmath2 vertices and maximum vertex degree @xmath22 contains an induced path of length @xmath227 .
[ co : path ] let @xmath5 be an @xmath2-vertex maximal plane graph
. then @xmath5 has an induced path with at least @xmath218 vertices .
a _ one - sided convex point set @xmath0 _ is a set of points in the plane that are in convex position and can be rotated so that the leftmost point and the rightmost point are adjacent in the convex hull of @xmath0 .
this restriction on the point set allows us to increase the size of the set itself .
the structure of the proof of theorem [ th : convex ] is similar to that of theorem [ th : general ] .
illustration of the drawing technique of lemma [ le : embedding.2 ] . ]
[ le : embedding.2 ] let @xmath0 be a one - sided convex point set with cardinality @xmath11 .
let @xmath5 be an @xmath2-vertex maximal plane graph that contains an induced biconnected outerplanar subgraph with at least @xmath11 vertices and one edge on the external boundary of @xmath5 .
then @xmath5 admits a geometric point - subset embedding on @xmath0 , which can be computed in @xmath56 time . if necessary , rotate @xmath0 so that the leftmost and the rightmost point are adjacent in the convex hull of @xmath0 .
let @xmath58 be the points of @xmath0 in the order they appear along the convex hull of @xmath0 , with @xmath103 being the leftmost point and @xmath228 being the rightmost point . without loss of generality assume that each point @xmath60 ( @xmath229 ) is above segment @xmath230 .
let @xmath16 be the induced biconnected outerplanar subgraph of @xmath5 and let @xmath55 be external boundary of @xmath16 .
without loss of generality assume that @xmath16 has exactly @xmath11 vertices ( if not , we can augment @xmath0 with additional points so to maintain @xmath0 to be a one - sided convex position ) .
so without loss of generality assume that @xmath16 has exactly @xmath11 vertices .
let @xmath52 be the path obtained from @xmath55 by removing the edge of @xmath16 that is on the external boundary of @xmath5 .
let @xmath64 be the vertices of @xmath52 in the order they appear along @xmath52 .
map @xmath66 to @xmath60 and draw each edge @xmath67 as the segment @xmath68 ( see figure [ fi : drawing3 ] for an illustration of the drawing technique ) . also add every other edge connecting two of the vertices of @xmath52 . since @xmath0 is convex position , it is immediate to see that the resulting drawing has no crossings .
let @xmath88 be the outer vertex of @xmath5 that does not belong to @xmath16 .
let @xmath231 be the straight line containing the segment @xmath232 and let @xmath233 be the straight line containing the segment @xmath234
. chose a point @xmath235 that is above @xmath231 and above @xmath233 , place @xmath88 on @xmath235 , and draw all edges connecting @xmath88 to the vertices of @xmath16 .
it is easy to see that the obtained drawing @xmath90 has no crossing .
hence @xmath90 is a planar geometric embedding on @xmath0 of the graph induced by the vertices in @xmath16 plus @xmath88 .
moreover , all the faces of the drawing are star - shaped polygons . in the planar embedding of @xmath5
each vertex not yet drawn is inside one of the cycles represented by the faces of the computed drawing .
thus , the drawing can be completed by using theorem [ theorem : hongnagamochi ] as described in the proof of lemma [ le : embedding.1.a ] .
similar to lemma [ le : embedding.1.a ] , the time spent to compute the drawing is @xmath4 , once the points are sorted .
thus , the time complexity is @xmath236 .
we now show that any @xmath2-vertex maximal plane graph contains an induced biconnected outerplanar subgraph with at least @xmath237{n } \rceil$ ] vertices and one edge on the external boundary of @xmath5 . to this aim
we will use the schnyder realizer to define three different partial orders on the vertices of @xmath5 .
let @xmath5 be a maximal plane graph equipped with a schnyder realizer @xmath131 .
for each @xmath238 , we define a partial order on @xmath239 by constructing a graph @xmath240 by means of the following operations ( see figure [ fi : gi ] for an illustration ) .
* we first remove from @xmath5 all the internal edges with color different from @xmath133 and @xmath154 ; * we then reverse the orientation of the edges with color @xmath133 ; * we color the outer edge @xmath241 with color @xmath133 and orient it from @xmath126 to @xmath242 ; * we color the outer edge @xmath243 with color @xmath133 and orient it from @xmath126 to @xmath244 ; * we color the outer edge @xmath245 with color @xmath154 and orient it from @xmath244 to @xmath242 .
the graph @xmath240 is a directed acyclic graph with one source @xmath126 and one sink @xmath242 .
thus , it defines a partial order @xmath246 for the vertices of @xmath5 ; namely , @xmath247 if there is a directed path from @xmath36 to @xmath30 in @xmath240 .
given a set @xmath248 with a partial order @xmath249 ( also called a _ partial ordered set _ ) , a subset @xmath250 of @xmath248 such that every two elements in @xmath250 are comparable with respect to @xmath249 is called a _ chain _ of @xmath248 .
a subset @xmath251 of @xmath248 such that no two elements in @xmath251 are comparable with respect to @xmath249 is called an _ antichain _ of @xmath248 . according to the definition of @xmath246 , a chain @xmath252 of @xmath239 with respect to @xmath246 corresponds to a oriented path @xmath114 in @xmath240 .
we say that @xmath114 is the oriented path of @xmath240 _ associated with _
we prove now some properties of the partial orders @xmath253 . [
le : comparability ] let @xmath5 be a maximal plane graph equipped with a schnyder realizer @xmath131 .
let @xmath36 and @xmath30 be two internal vertices of @xmath5 .
then @xmath36 and @xmath30 are comparable by at least one of the three partial orders @xmath254 , @xmath255 , and @xmath256 .
if @xmath36 belongs to a @xmath133-path staring at @xmath30 for some @xmath133 ( @xmath257 ) , then @xmath36 and @xmath30 are comparable with respect to both @xmath246 and @xmath258 ( indices modulo @xmath21 ) .
namely , the concatenation of the reversed version of @xmath132 with @xmath148 is an oriented path of @xmath259 containing both @xmath36 and @xmath30 , and therefore @xmath36 and @xmath30 are comparable with respect to @xmath253 .
analogously , the concatenation of the reversed version of @xmath149 with @xmath260 is an oriented path of @xmath261 containing both @xmath36 and @xmath30 , and therefore @xmath36 and @xmath30 are comparable with respect to @xmath258 . clearly , the same argument applies if @xmath30 belongs to a @xmath133-path staring at @xmath36 for some @xmath133 ( @xmath257 ) .
consider the case when @xmath36 does not belong to any @xmath133-path staring at @xmath30 and viceversa .
it follows that @xmath36 must be in one region @xmath143 ( @xmath257 ) .
consider the @xmath133-path starting at @xmath36 , @xmath213 . since this path does not contain @xmath30
, it must share a vertex @xmath88 with either @xmath149 or @xmath148 ( see figure [ fi : comparability ] ) .
suppose that @xmath88 belongs to @xmath149 , the case when it belongs to @xmath148 is analogous .
consider the path @xmath52 consisting of the concatenation of : ( i ) the path @xmath214 ; ( ii ) the subpath of @xmath213 from @xmath36 to @xmath88 ; ( iii ) the subpath of @xmath149 from @xmath88 to @xmath30 ; ( iv ) the path @xmath132 .
all edges of @xmath52 colored @xmath155 are oriented towards @xmath244 , while those colored @xmath133 are oriented towards @xmath126 ( see figure [ fi : comparability2 ] ) . thus reversing the edges of @xmath52 colored @xmath155
, we obtain an oriented path @xmath69 that is an oriented path of @xmath262 and that contains both @xmath36 and @xmath30 . thus , @xmath36 and @xmath30 are comparable with respect to @xmath258 .
[ le : comparability2 ] let @xmath5 be a maximal plane graph equipped with a schnyder realizer @xmath131 .
let @xmath36 be an internal vertex of @xmath5 and let @xmath30 be another vertex of @xmath5 such that @xmath263 . then @xmath36 and @xmath30 are not comparable with respect to @xmath253 .
if @xmath263 , the two paths @xmath213 and @xmath132 have a vertex in common @xmath264 , and the paths @xmath215 and @xmath148 have a vertex in common @xmath265 .
thus , in @xmath240 , there are two oriented paths @xmath47 from @xmath264 to @xmath265 : one , call it @xmath47 , containing @xmath36 and the other one , call it @xmath48 , containing @xmath30 .
we prove now that there is no oriented path from @xmath30 to @xmath36 in @xmath240 ( the proof that there is no oriented path from @xmath36 to @xmath30 is analogous ) .
suppose that in @xmath240 there is an oriented path @xmath114 from @xmath30 to @xmath36 .
if @xmath114 does not share a vertex with @xmath266 nor with @xmath267 , then it is completely contained in the ( unoriented ) cycle @xmath55 formed by @xmath47 and @xmath48 .
let @xmath268 be the first edge of @xmath213 ( this edge is the unique outgoing edge of @xmath36 colored @xmath133 in @xmath131 ) and let @xmath269 be the first edge of @xmath215 ( this edge is the unique outgoing edge of @xmath36 colored @xmath154 in @xmath131 ) . since the edges of @xmath114 must be colored @xmath133 or @xmath154 , there should be an edge colored @xmath133 or @xmath154 that follows @xmath268 and precedes @xmath269 in the counterclockwise order of the edges around @xmath36 . but all edges that follows @xmath268 and precedes @xmath269 in the counterclockwise order around @xmath36 are colored @xmath155 .
suppose now that @xmath114 shares a vertex @xmath88 with @xmath266 and/or @xmath267 .
vertex @xmath88 can not belong to path @xmath213 because , in this case , @xmath88 would precede both @xmath36 and @xmath30 according to @xmath253 and therefore there would be a directed cycle in @xmath240 .
analogously , @xmath88 can not be in @xmath148 because , in this case , @xmath88 would follow both @xmath36 and @xmath30 according to @xmath253 and therefore there would be a directed cycle in @xmath240 .
the only possibility is that @xmath88 belongs to either the subpath of @xmath213 from @xmath36 to @xmath264 , or to the subpath of @xmath148 from @xmath30 to @xmath265 ( endpoints excluded ) .
suppose that @xmath88 belongs to @xmath213 .
let @xmath179 and @xmath188 be the vertices that precede and follow @xmath88 along @xmath213 , respectively . in @xmath131 ,
edge @xmath270 is an incoming edge of @xmath88 colored @xmath133 , while edge @xmath271 is the unique outgoing edge of @xmath88 colored @xmath133 . as a consequence
, path @xmath114 can not pass through @xmath88 `` exiting '' from cycle @xmath55 . in order to exit from @xmath55 , in @xmath240 there should be an incoming edge of @xmath88 that follows @xmath270 and precedes @xmath271 in the counterclockwise order around @xmath88 . in @xmath131
this edge should be either a incoming edge of @xmath88 colored @xmath154 or an outgoing edge of @xmath88 colored @xmath133 , but none of the two is possible according to the properties of the schnyder realizer . with a symmetric argument one can show that also when @xmath88 belongs to @xmath148 , path @xmath114 can not pass through @xmath88 `` exiting '' from cycle @xmath55 .
thus , in both cases path @xmath114 can only enter cycle @xmath55 . on the other hand
, path @xmath114 can not be completely outside cycle @xmath55 ( in this case it would cross the common subpath of @xmath213 and @xmath132 , or the common subpath of @xmath215 and @xmath148 , but we have already seen that it is not possible ) .
it follows that path @xmath114 enters @xmath36 from inside @xmath55 , i.e. , there is an edge colored @xmath133 or @xmath154 that follows @xmath268 and precedes @xmath269 in the counterclockwise order of the edges around @xmath36 , but we have already seen that this is not possible .
[ le : comparability3 ] let @xmath5 be a maximal plane graph equipped with a schnyder realizer @xmath131 .
let @xmath36 be an internal vertex of @xmath5 and let @xmath30 be another vertex of @xmath5 that is comparable with @xmath36 by @xmath253 .
if @xmath272 then @xmath273 . if @xmath274 then @xmath275 .
we consider the case when @xmath272 ( the case when @xmath274 can be proven symmetrically ) .
we show that @xmath30 must be in @xmath276 . by lemma [ le : comparability2 ]
, @xmath30 can not be in @xmath277 nor in @xmath214 ( in this case @xmath36 would be in @xmath147 ) .
if @xmath30 is in @xmath215 then @xmath36 ad @xmath30 are comparable with respect to @xmath253 but @xmath274 , so @xmath30 can not be in @xmath215 .
suppose that @xmath30 is in @xmath278 .
if @xmath260 has a vertex in common with @xmath214 , then we are again in the case when @xmath36 is in @xmath147 and thus @xmath36 and @xmath30 are not comparable by lemma [ le : comparability2 ] .
it follows that @xmath260 must have a vertex @xmath88 in common with @xmath279 .
consider the path @xmath52 consisting of the concatenation of : ( i ) the path @xmath279 ; ( ii ) the subpath of @xmath215 from @xmath36 to @xmath88 ; ( iii ) the subpath of @xmath260 from @xmath88 to @xmath30 ; ( iv ) the path @xmath148 .
all edges of @xmath52 colored @xmath133 are oriented towards @xmath126 , while those colored @xmath154 are oriented towards @xmath242 . thus reversing the edges of @xmath52 colored @xmath133
, we obtain an oriented path @xmath69 that is an oriented path of @xmath280 and that contains both @xmath36 and @xmath30 .
thus , @xmath36 and @xmath30 are comparable with respect to @xmath246 .
note however that in this case @xmath274 .
thus @xmath30 can not be in @xmath278 .
the only possibility remaining is that @xmath30 is in @xmath276 . in the next lemma
we show that any maximal chain of @xmath239 with respect to any partial order @xmath246 defines an induced biconnected outerplanar subgraph with one edge on the external boundary of @xmath5 .
[ le : chain - frontier ] let @xmath5 be a maximal plane graph equipped with a schnyder realizer @xmath131 .
let @xmath252 be a maximal chain of @xmath239 with respect to the partial order @xmath246 , @xmath257 and let @xmath114 be the oriented path associated with @xmath252 .
the graph induced by the vertices of @xmath114 is an induced biconnected outerplanar subgraph with one edge on the external boundary of @xmath5 .
first of all , notice that since @xmath240 has a single source @xmath126 and single sink @xmath242 , then every maximal chain of @xmath239 with respect to @xmath246 has @xmath126 as the first element and @xmath242 as the last element . as a consequence @xmath114 starts at @xmath126 and ends at @xmath242 . also , since @xmath252 contains at least one internal vertex , @xmath114 consists of all internal edges .
this implies that it does not coincide with the single edge @xmath241 .
thus , the concatenation of ( the underlying undirected path of ) @xmath114 and of the edge @xmath241 is a simple cycle @xmath55 of @xmath5 .
we prove that the subgraph @xmath16 induced by the vertices in @xmath55 is a biconnected outerplanar graph .
if @xmath55 is the boundary of the external face , then @xmath16 is a @xmath21-cycle and therefore it is a biconnected outerplanar graph .
if @xmath55 is not the boundary of the external face of @xmath5 , in order to prove that is a biconnected outerplanar graph , it is sufficient to show that all edges of @xmath16 that are not in @xmath55 are inside @xmath55 in the planar embedding of @xmath5 . consider an edge @xmath37 of @xmath16 not in @xmath55 .
vertices @xmath36 and @xmath30 must be two non - consecutive vertices of @xmath114 . without loss of generality
assume @xmath274 , i.e. , @xmath36 is encountered first when walking along @xmath114 from @xmath126 to @xmath242 .
since we are considering an edge not in @xmath55 , it must be @xmath281 or @xmath282 ( or both ) .
since @xmath36 and @xmath30 are non - consecutive , there must be a vertex @xmath88 such that @xmath283 . by lemma [ le : comparability3 ]
, @xmath36 is in @xmath284 and @xmath30 is in @xmath285 . as a consequence , if edge @xmath37 is not inside @xmath55 , then it crosses path @xmath286 , which is not possible .
the next lemma shows that the subgraph of lemma [ le : chain - frontier ] has at least @xmath287{n } \rceil$ ] vertices .
[ le : chain ] let @xmath5 be a maximal plane graph equipped with a schnyder realizer @xmath131 .
@xmath239 has a chain of size at least @xmath287{n } \rceil$ ] with respect to one of the three partial orders @xmath254 , @xmath255 , and @xmath256 .
notice that , in order to prove that @xmath239 has a chain of size at least @xmath287{n } \rceil$ ] , it is sufficient to show that @xmath239 has a chain of size at least @xmath288{n}$ ] because the number of vertices in a chain is necessarily an integer number .
consider first the partial order @xmath254 .
if @xmath239 has a chain of size at least @xmath288{n}$ ] with respect to @xmath254 , then the statement is true .
if not , by mirsky s theorem @xcite ( the dual of dilworth s theorem ) , @xmath239 has a partition into at most @xmath288{n}$ ] antichains with respect to @xmath254 . hence ,
one of this antichains must have at least @xmath289{n}}=n^{2/3}$ ] vertices .
let @xmath290 be such a set of vertices .
since the vertices in @xmath290 form an antichain of @xmath239 with respect to @xmath254 , then , by lemma [ le : comparability ] , any two vertices in @xmath290 must be comparable by @xmath255 or by @xmath256 .
consider @xmath255 restricted to the set @xmath290 ; if @xmath290 has a chain of size at least @xmath288{n}$ ] with respect to @xmath255 , then the statement is true .
otherwise , applying again mirsky s theorem , @xmath290 can be partitioned into at most @xmath288{n}$ ] antichains .
one of this antichain must have at least @xmath291{n}$ ] vertices .
let @xmath292 be such a set of vertices .
the vertices in @xmath292 are not comparable with respect to @xmath254 , neither with respect to @xmath255
. thus by lemma [ le : comparability ] they must be comparable by @xmath256 , i.e. , they form a chain of @xmath292 with respect to @xmath256 . [
th : large - induced - outerplanar ] every @xmath2-vertex maximal plane graph @xmath5 contains an induced biconnected outerplanar subgraph with at least @xmath287{n } \rceil$ ] vertices and one edge on the external boundary of @xmath5 .
such a subgraph can be computed in @xmath4-time .
consider a schnyder realizer @xmath131 of @xmath5 and define the three partial orders @xmath254 , @xmath255 , and @xmath256 . by lemma [ le : chain ]
, @xmath239 has a chain of size at least @xmath287{n } \rceil$ ] with respect to one of these partial orders . from lemma [
le : chain - frontier ] it follows that @xmath5 contains an induced biconnected outerplanar subgraph @xmath16 with at least @xmath287{n } \rceil$ ] vertices and one edge on the external boundary of @xmath5 .
the schnyder realizer @xmath131 can be computed in @xmath4 time @xcite .
using @xmath131 , the three graphs @xmath293 , @xmath294 , and @xmath295 can be also computed in @xmath4 time .
computing the longest path in each of these graphs ( which can be done in @xmath4 time ) we find the longest chain of @xmath239 with respect to one of the three partial orders @xmath254 , @xmath255 , and @xmath256 .
the subgraph induced by the vertices in this path is @xmath16 .
[ [ proof - of - theoremthconvex ] ] proof of theorem [ th : convex ] + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath0 be a one - sided convex point set of cardinality at most @xmath287{n } \rceil$ ] and let @xmath5 be an @xmath2-vertex plane graph . if @xmath5 is not maximal we can add dummy edges to it in order to make it maximal ( these edges will be removed at end from the obtained drawing ) .
thus , without loss of generality , assume that @xmath5 is maximal . by theorem
[ th : large - induced - outerplanar ] @xmath5 contains an induced biconnected outerplanar subgraph with at least @xmath287{n } \rceil$ ] vertices and one edge on the external boundary of @xmath5 . by lemma [ le : embedding.2 ]
, @xmath5 admits a geometric point - subset embedding on @xmath0 . by theorem [ th : large - induced - outerplanar ] and lemma [ le : embedding.2 ] the time complexity of the drawing algorithm is @xmath56 where @xmath11 is the size of @xmath0 . since @xmath296{n } \rceil$ ]
, we have that @xmath223 which concludes the proof of theorem [ th : convex ] .
theorem [ th : convex ] can be used to solve a different problem , called the _ allocation problem_. in the allocation problem @xcite the input is an @xmath2 vertex planar graph @xmath5 and a point set @xmath248 of cardinality @xmath2 , and the goal is to compute a planar straight - line drawing of @xmath5 such that as many vertices as possible are represented by points of @xmath248 .
the number of points fitting this purpose is denoted as @xmath297 and @xmath298 , where the minimum is taken over all @xmath2-point sets @xmath248 .
upper and lower bounds on @xmath299 are known for various classes of graphs @xcite .
we define @xmath300 where the minimum is taken over all @xmath248 in one - sided convex position .
theorem [ th : convex ] implies that @xmath301{n } \rceil$ ] for every planar graph @xmath5 .
this improves the best known lower bound on @xmath299 for general planar graphs which is @xmath302{\frac{n}{3}}$ ] @xcite . for every @xmath2-vertex planar graph , @xmath303{n } \rceil$ ]
in this paper we proved that a planar graph with @xmath2 vertices admits a geometric point - subset embedding on every set of @xmath1 points and on every one - sided convex point set of @xmath3{n } \right \rceil$ ] points on the plane .
these are the first non - constant lower bounds for the size of arbitrary ( one - sided convex ) point sets on which planar graphs have geometric point - subsets embedding . in order to achieve this result we proved that every maximal planar graph contains as a subgraph an induced outerpath of size at least @xmath218 and an induced biconnected outerplanar graph of size @xmath3{n } \right \rceil$ ] . to the best of our knowledge ,
these are the first lower bounds for the size of a largest induced outerpath and biconnected outerplanar subgraph in planar graphs .
it is known that not all planar graphs with @xmath2 vertices admit a geometric point - subset embedding on a one - sided convex set containing more than @xmath304 points @xcite . thus closing the gap between lower bounds @xmath1 and
@xmath3{n } \right \rceil$ ] , and upper bound @xmath304 is an interesting open problem . from the graph - theoretical point of view
it would be also interesting to study the upper bounds for the size of maximum induced biconnected outerplanar subgraph and induced outerpath in biconnected planar graphs .
however , it is worth mentioning that an upper bound for the size of induced outerplanar subgraph does not provide us an upper bound for the size of convex point set that represent universal subset for planar graphs .
patrizio angelini , carla binucci , william evans , ferran hurtado , giuseppe liotta , tamara mchedlidze , henk meijer , and yoshio okamoto .
universal pointsubsets for planar graphs . in kun - mao chao ,
tsan sheng hsu , and der - tsai lee , editors , _ isaac _ , volume 7676 of _ lecture notes in computer science _ , pages 423432 .
springer , 2012 .
n. castaeda and j. urrutia .
straight line embeddings of planar graphs on point sets . in f.
fiala , e. kranakis , and j .-
sack , editors , _
canadian conference on computational geometry ( cccg 96 ) _ , pages 312318 , 1996 .
emilio di giacomo and giuseppe liotta .
the hamiltonian augmentation problem and its applications to graph drawing . in md .
saidur rahman and satoshi fujita , editors , _ walcom _ , volume 5942 of _ lecture notes in computer science _ , pages 3546 .
springer , 2010 .
alexander ravsky and oleg verbitsky . on collinear sets in straight - line drawings . in petr kolman and jan kratochvl , editors ,
_ graph - theoretic concepts in computer science _ , volume 6986 of _ lecture notes in computer science _ , pages 295306 .
springer berlin heidelberg , 2011 .
|
it is proven that every set @xmath0 of distinct points in the plane with cardinality @xmath1 can be a subset of the vertices of a crossing - free straight - line drawing of any planar graph with @xmath2 vertices .
it is also proven that if @xmath0 is restricted to be a one - sided convex point set , its cardinality increases to @xmath3{n } \right \rceil$ ] .
the proofs are constructive and give rise to @xmath4-time drawing algorithms . as a part of our proofs ,
we show that every maximal planar graph contains a large induced biconnected outerplanar graphs and a large induced outerpath ( an outerplanar graph whose weak dual is a path ) .
|
the astronomical observations show that our universe is probably approaching de sitter spacetime .
the cosmological constant corresponds to vacuum energy and is usually considered as a candidate of dark energy .
the accelerating universe will evolve into another de sitter phase . in order to construct the entire history of evolution of the universe
, we should have a clear perspective to the classical and quantum properties of de sitter spacetime @xcite .
however , as is well known that in de sitter space there is no spatial infinity and no asymptotic killing vector which is globally timelike .
moreover , black holes in de sitter spacetime can not be in thermodynamic equilibrium in general . because there are multiple horizons with different temperatures for de sitter black holes .
the thermodynamic quantities on the black hole horizon and the cosmological horizon all satisfy the first law of thermodynamics , moreover the corresponding entropies both fulfill the area formulae @xcite .
there have been accumulated interest to the thermodynamics property of de sitter spacetime in recent years @xcite . in this way
, the two horizons can be analyzed independently . besides , one can also take a global view to construct the globally effective temperature and other effective thermodynamic quantities .
no matter which method is used , the total entropy of de sitter black hole is supposed to be the sum of both horizons , namely @xmath1 .
@xcite we think that the correct method may to be so simple because the event horizon and the cosmological horizon are not independent .
there may exist some correlations between them due to the following considerations .
we can take the kerr - de sitter black hole as example .
there are first laws of thermodynamics for both horizons . according to @xcite ,
the thermodynamic quantities on the two horizons satisfy the first law of thermodynamics system , [ 1st ] dm&=&t_+ds_++_+ dj+v_+d , + dm&=&-t_cds_c+_c dj+v_cd , where @xmath2 is the conserved mass in de sitter spacetime , @xmath3 and @xmath4 stand for the angular velocities and angular momentum , and the cosmological constant @xmath5 as a pressure and define the thermodynamically conjugate variable to the thermodynamics volume , which can be different from the geometrical volume , the subscript @xmath6 denotes the black hole horizon and the subscript @xmath7 denotes the cosmological horizon . including the variable @xmath5 ,
the above first laws can have corresponding smarr formulae , which are also given in @xcite .
the two laws in eq.([1st ] ) are not truly independent .
they depend on the same quantities @xmath8 .
all the geometric and thermodynamic quantities for the both horizons can be represented by @xmath8 .
therefore , the size of black hole horizon is closely related to the size of the cosmological horizon , and the evolution of black hole horizon will lead to the evolution of the cosmological horizon .
considering the relation between the thermodynamic quantities on the two horizons is very important for studying the thermodynamic properties of de sitter spacetime @xcite .
considering the correlation or entanglement between the black hole event horizon and the cosmological horizon , the total entropy of the kerr - de sitter black hole is no longer simply @xmath9 , but should include an extra term from the contribution of the correlations of the two horizons . in the ref .
@xcite , we calculate the extra term in the entropy for the spherically symmetric charged black hole in de sitter spacetime . by the method we obtain the reasonable interpretation for the r - n black hole in de sitter spacetime . in this
works , we will expand the method to the axisymmetric rotating black hole in de sitter spacetime .
we can obtain the extra term form and find the relates between the region of stable state and the rotating parameter @xmath0 .
our paper is organized as follows . in the next section
we simple review the thermodynamics quantities of black hole horizon and cosmological horizon of the kerr - de sitter black hole , obtain the condition that the effective temperature of the horizon approaches to zero . in the base that the kerr - de sitter system satisfies the first laws of thermodynamics ,
the entropy and the effect temperature of kerr - de sitter black hole is obtained , we study the condition that kerr - de sitter system satisfies the stable equilibrium of the thermodynamics in sec .
sec.[con ] is devoted to conclusions .
kerr - de sitter black hole is a solution of einstein equation in @xmath10 dimensions with a positive cosmological constant @xmath11 .
it is characterised by parameters , mass @xmath2 and angular momentum @xmath4 .
the spacetime metric of the kerr - de sitter black hole in boyer - lindquist coordinates @xmath12 is given by @xcite [ staticmetric ] ds^2 = - ( dt - d ) ^2 + dr^2 + d^2 + ( adt - d ) ^2 , where the various functions entering the metric are given by @xmath13 [ fr ] _ r ( r ) = ( r^2 + a^2 ) ( 1 - r^2 ) -2mr , the black hole horizon and cosmological horizon satisfies the equation @xmath14 and @xmath15 , here @xmath16 and @xmath17 are the location of black hole horizon and cosmological horizon , respectively .
so we can obtain@xcite @xmath18 [ eq3 ] = , = . in the de sitter spacetime , by regarding respectively the black hole horizon and the cosmological horizon as the thermodynamics systems , we can have @xcite @xmath19 [ eq4 ] t_c = - .
the entropy and energy of the black horizon and the cosmological horizon are respectively , @xmath20 the angular velocity of the black hole horizon and the cosmological horizon are respectively , @xmath21 while the angular momentum and the angular velocity at infinity read @xmath22 taking the cosmological constant @xmath5 as the pressure , the thermodynamics quantities of the both horizons satisfy the first laws of the thermodynamics @xcite , @xmath23 @xmath24 here @xmath25 @xmath26 the black hole and cosmological thermodynamics volumes are respectively @xcite , @xmath27,\ ] ] @xmath28.\ ] ] from the eq.([eq4 ] ) , when @xmath29 , we can obtain the approximate value for the temperature @xmath30},\ ] ] @xmath31}.\ ] ] where @xmath32 .
generally , the temperatures at the black hole event horizon and cosmological horizon are not equal .
thus , the whole kerr - de sitter system can not be in equilibrium thermodynamically . however , there are two special cases for the kerr - de sitter black hole , in which the temperatures at the both horizons are the same .
one case is the so - called nariai black hole , the other is the lukewarm black hole @xcite . for the nariai black hole ,
the event horizon and cosmological horizon coincide _ apparently _ and have the same temperature , zero or nonzero @xcite . considering the connection between the black hole horizon and the cosmological horizon @xcite , we can derive the effective thermodynamic quantities and corresponding first law of black hole thermodynamics as the pressure and then derive the effective volume . in this case
the effective first law should be : @xmath33 .
this has been done in another paper . ] : [ eff1st ] dm = t_effds - p_effdv + _ effdj . here
the thermodynamic volume is that between the black hole event horizon and the cosmological horizon , namely [ vol ] v = v_c - v _ + ( r_c^3 - r _ + ^3 ) = r_c^3 ( 1 - x^3 ) .
the total entropy can be written as [ ent ] s = s _ + + s_c + s_ex= r_c^2 .
when @xmath4 is a constant , from @xmath34 , we can obtain @xmath35 @xmath36 when @xmath29 , from the eqs.([vol ] ) , ( [ eq12 ] ) and ( [ eq13 ] ) , we can obtain @xmath37 from the eq.([eff1st ] ) , we can have the effective temperature of the kerr - de sitter black hole , t_eff & = & ( ) _ j , v + & = & here , @xmath38 .
when @xmath29 , @xmath39 .
we can obtain via neglecting the higher order of @xmath0 , @xmath40},\ ] ] from the eq .
( [ eq3 ] ) , we have @xmath41 considering the expression for the entropy of both horizons ( [ ent ] ) , we can take the entropy of kerr - de sitter black hole as @xmath42 here the undefined function @xmath43 represents the extra contribution from the correlations of the two horizons . then , how to determine the function @xmath43 ? inserting eqs .
( [ eq21 ] ) , ( [ eq20])and ( [ vol ] ) into eq.([eq19 ] ) , we can obtain @xmath44 where @xmath45,\ ] ] @xmath46.\ ] ] neglecting the higher order terms of @xmath0 , we can rewrite the eq.([eq4 ] ) , @xmath30},\ ] ] @xmath47}.\ ] ] from the solution of ( [ eq27 ] ) , when [ arc ] = , the radiation temperature of the both horizons is equal . in general cases ,
the temperatures of the black hole horizon and the cosmological horizon are not the same , thus the globally effective temperature @xmath48 can not be compared with them . however , in the special case , such as lukewarm case @xcite , the temperatures of the two horizons are the same .
we conjecture that in this special case the effective temperature should also take the same value . on the basis of this consideration ,
we can obtain the information of @xmath43 . for kerr - de sitter black hole ,
@xmath49 from the eq.([eq25 ] ) and ( [ eq28 ] ) , we can obtain @xmath50,\ ] ] where @xmath51 is the value of @xmath52 with @xmath53 in eq.([arc ] ) . substuting the eqs.([eq25 ] ) and ( [ eq28 ] ) into the eq ( [ eq29 ] ) , we can obtain @xmath54},\ ] ] inserting eqs.([eq26 ] ) and ( [ eq27 ] ) , we can obtain the equation of @xmath43 [ eq31 ] 2x(1 + x ) + 2x^2f(x ) + ( 1 - x^3)f(x ) + 2[(1 + a^2 / r_c^2 ) ( 1 + x + x^2 ) + a^2 / r_c^2 ] + = 0 in fig.[tex ] , [ fx ] and [ sx ] , we depict the effective temperature @xmath48 , @xmath43 and @xmath55 as functions of @xmath56 .
it is shown that @xmath48 tends to zero as @xmath57 , namely the nariai limit .
although this result does not agree with that of bousso and hawking @xcite , it is consistent with the entropy . as is depicted in fig.[sx ]
, the entropy will diverge as @xmath57 .
the specific heat of kerr - de sitter system can be defined as @xmath58 from fig.[cx ] , when @xmath59 , the specific heat of system is positive , while @xmath60 , it is negative .
this means that the kerr - de sitter black hole with @xmath61 is thermodynamically stable . from the fig.[cx ] , we can find that the region of stable state for kerr - de sitter is related to the angular velocity parameter @xmath0 , i.e. , the region of stable state becomes bigger as the rotating parameters @xmath0 is increases .
in this letter , we mainly studied the entropy of kerr - de sitter black hole .
in addition , we analyzed the thermodynamic stability of this black hole .
firstly , we simply review the thermodynamic quantities of black hole horizon and cosmological horizon of the kerr - de sitter black hole . considering the relation of the black hole horizon and the cosmological horizon ,
we conjecture the total entropy should take the form of eq.([ent ] ) .
secondly , we find that the entropy obtained by eq.([ent ] ) monotonically increases as @xmath56 increased . when @xmath62 , this entropy diverges .
finally , it is found that the kerr - de sitter black hole is unstable due to negative heat capacity in some region for rotating parameter @xmath0 . in a word
, we obtain the corrected entropy of kerr - de sitter black hole , which may make the method more acceptable .
we would like to thank dr hui - hua zhao and yu - bo ma for their indispensable discussions and comments .
this work was supported by the young scientists fund of the national natural science foundation of china ( grant no.11205097 ) , in part by the national natural science foundation of china ( grant no.11475108 ) , supported by program for the innovative talents of higher learning institutions of shanxi , the natural science foundation of shanxi province , china(grant no.201601d102004 ) and the natural science foundation for young scientists of shanxi province , china ( grant no.2012021003 - 4 ) , the natural science foundation of datong city(grant no.20150110 ) .
cai , _ cardy - verlinde formula and thermodynamics of black holes in de sitter spaces _ ,
b 628 , 375(2002 ) .
y. sekiwa , _ thermodynamics of de sitter black holes : thermal cosmological constant _ , phys . rev .
d 73 , 084009(2006 ) ; [ hep - th/0602269 ] .
m. urano and a. tomimatsu , _ mechanical first law of black hole spacetimes with cosmological constant and its application to schwarzschild - de sitter spacetime _ , class .
26 , 105010(2009 ) ; [ arxiv:0903.4230gr - qc ] .
t. katsuragawa and s. nojiri , _ stability and antievaporation of the schwarzschild de sitter black holes in bigravity _ , phys .
d 91 , 084001 ( 2015 ) .
t. roy choudhury and t. padmanabhan , _ concept of temperature in multi - horizon spacetimes : analysis of schwarzschild - de sitter metric _
rel . grav .
39 , 1789 - 1811(2007 ) ; [ arxiv : gr - qc/0404091 ] .
a. m. ghezelbash and r. b. mann , _ entropy and mass bounds of kerr - de sitter space - times _ , phys . rev .
d 72 , 064024(2005 ) ; [ arxiv : hep - th/0412300 ] .
a. gombero , and c. teitelboim , _ de sitter black holes with either of the two horizons as a boundary _ , phys .
d 67 , 104024(2003 ) ; [ arxiv : hep - th/0302204 ] . b. p. dolan , d. kastor , d. kubiznak , r. b. mann , j. traschen , _ thermodynamic volumes and isoperimetric inequalities for de sitter black holes _
, phys . rev .
d 87 , 104017(2013 ) ; [ arxiv:1301.5926hep - th ] k. hajian , _ conserved charges and first law of thermodynamics for kerr de sitter black holes _ , gen .
48 , 114(2016 ) .
d. kubiznak , f. simovic , _ thermodynamics of horizons : de sitter black holes _ ,
arxiv:1507.08630 [ hep - th ] . j. mcinerney , g. satishchandran , j. traschen , _ cosmography of knds black holes and isentropic phase transitions _
33 , 105007(2016 ) ; [ arxiv:1509.02343 hep - th ] s. bhattacharya , a. lahiri , _ mass function and particle creation in schwarzschild - de sitter spacetime _ , the european physical journal c. 73 , 2673(2013 ) ; [ arxiv:1301.4532gr - qc ] t. pappas , p. kanti , and n. pappas , _ hawking radiation spectra for scalar fields by a higher - dimensional schwarzschild de sitter black hole _ ,
phys . rev .
d 94 , 024035(2016 ) .
y. s. myung , _ thermodynamics of the schwarzschild - de sitter black hole : thermal stability of the nariai black hole _
, phys . rev .
d 77 , 104007(2008 ) .
r. zhao , l. c. zhang , h. f. li .
_ hawking radiation of reisser - nordstrom - de sitter space_. gen .
42 , 975(2010 ) .
r. zhao , l. c. zhang , h. f. li , et al .
_ hawking radiation of high - dimensional rotating black hole _ , eur.phys .
j. c. 65 , 289(2010 ) .
l.c . zhang , and r. zhao , m. s. ma , _ entropy of reissner nordstrm de sitter black hole _ ,
b 761 , 74(2016 ) . r. g. cai , y. j. ji , and k. s. soh , _ action and entropy of black holes in space - times with a cosmological constant _ , calss .
15 , 2783(1998 ) .
f. h. li , s. l. zhang , y. q. wu , l. c. zhang , and r. zhao , _ hawking radiation of kerr - newman - de sitter black hole _ , eur .
j. c. 63 , 133(2009 ) .
guo , h. f. li , l.c .
zhang , and r. zhao , _ the critical phenomena of charged rotating de sitter black holes _ , class .
quantum grav .
33 , 135004(2016 ) .
guo , h. f. li , l.c .
zhang , and r. zhao , _ thermodynamics and phase transition of in the kerr - de sitter black hole _
d 91 , 084009(2015 ) .
l.c . zhang , r. zhao , and m.s .
ma _ entropy of reissner - nordstrm - de sitter black hole _ , phys .
b , 761 , 74(2016 ) .
r. bousso , s. hawking , _ pair creation of black holes during inflation _ , phys
d 54 , 6312(1996 ) .
f. mellor , i. moss _
black holes and gravitational instantons _ , class .
quantum grav . 6 , 1379(1989 ) .
l. j. romans _ supersymmetric , cold and lukewarm black holes in cosmological einstein - maxwell theory _ ,
b 383 , 395(1992 ) .
|
based on the consideration that the black hole horizon and the cosmological horizon of kerr - de sitter black hole are not independent each other , we conjecture the total entropy of the system should have an extra term contributed from the correlations between the two horizons , except for the sum of the two horizon entropies . by employing globally effective first law and effective thermodynamic quantities ,
we obtain the corrected total entropy and find that the region of stable state for kerr - de sitter is related to the angular velocity parameter @xmath0 , i.e. , the region of stable state becomes bigger as the rotating parameters @xmath0 is increases .
|
the formation and evolution of galaxies is regulated by their feedback on the surrounding intergalactic medium ( igm ) .
radiative feedback owing to photo - ionization heating by a cosmic uv background or hydrodynamic feedback owing to outflows driven by supernovae or quasars , can suppress the infall of igm gas onto low - mass dark - matter halos ( see , e.g. , @xcite ) .
the feedback introduces a minimum halo mass above which halos can efficiently accrete gas from the igm and host star formation or quasar activity .
infall of gas is suppressed if the gravitational potential well of a halo is shallow , explaining why dwarf galaxies have a much lower abundance than expected for low - mass halos ( e.g. , @xcite ) . in this _
letter _ , we examine a simple observational method to reveal the scars left by hydrodynamic outflows from galaxies on the igm .
powerful outflows from galaxies are expected to push away the neutral intergalactic medium ( igm ) around those galaxies , and produce absorption - free gaps in the ly@xmath0 forest towards a background quasar @xcite .
studies of the ly@xmath0 forest the narrow absorption lines produced by intergalactic neutral hydrogen in quasar spectra had been very successful in mapping large scale structure in the universe at high redshift ( for a review , see @xcite ) .
while the igm dynamics is shaped by gravity alone on large scales ( see , e.g. , @xcite ) , outflows can play an important role on small scales by introducing gaps in the ly@xmath0 forest that can be identified in high resolution observations of quasar spectra .
our model is simple : an outflow from a galaxy evacuates a bubble in the igm , leaving behind little or no neutral hydrogen as the swept - away gas piles up on the bubble surface .
such a bubble would appear as a gap in the ly@xmath0 forest absorption spectrum of a background quasar , i.e. , a flat spectral segment with little or no absorption .
the expected distribution of absorption - free gaps can be calculated from the number density of halos above the minimum mass , as only those halos are capable of hosting supernova or quasar - driven outflows . for the halo mass range of interest , @xmath4@xmath5
, the bubble radius is expected to be almost independent of halo mass at a given redshift ( furlanetto & loeb 2003 ) .
hence , our analytic model has only two free parameters : the minimum halo mass for the production of outflows , @xmath6 , and the characteristic bubble radius @xmath7 for halos above that mass .
there have already been some observational reports about the signatures of outflows around massive starburst galaxies ( with @xmath8 ) at high redshifts .
@xcite discovered large scale outflows with velocities up to @xmath9 by studying @xmath10 lyman break galaxies ( lbgs ) at @xmath11 . using a large sample of quasar spectra which pass by foreground lbgs
, @xcite identified the lack of neutral hydrogen within @xmath12 comoving mpc around lbgs .
recent observations of high redshift , submillimeter galaxies also indicated the existence of large scale outflows ( see , @xcite and reference therein ) . the properties of active bubbles around starburst galaxies may be different than those of mature bubbles around more common galaxies for which the wind material had sufficient time to accumulate on the bubble walls and decelerate after evacuating the bubble interior . in simulations ,
@xcite and @xcite also found absorption - free gaps in their simulated ly@xmath0 spectrum caused by galactic winds .
our goal in this _
letter _ is to characterize the properties of the typical relic bubbles around the most common ( @xmath13 ) galaxies at redshift @xmath14 .
our paper is organized as follows . in
[ sec : calc ] we briefly describe our analytic model and derive its basic equations . in
[ sec : obs ] we compare the model with three high - resolution quasar spectra .
finally , we conclude with a discussion about the assumptions of our model and the implications of our results . throughout the paper , we assume the concordance cosmological model with @xmath15 determined by recent _ wmap _
data @xcite .
in our simple model , a galactic wind sweeps - up the surrounding intergalactic medium and eliminates hi absorption within a bubble radius @xmath7 of order a few hundred kpc ( @xcite ) .
the bubble spends most of the time approaching a fixed comoving distance from its host galaxy , at which the ( decelerating ) expansion speed asymptotes to the local hubble velocity ( see section [ sec : disc ] ) .
such a bubble would produce a gap in the ly@xmath0 forest imprinted on the spectrum of a background quasar .
the absorption - free gap should have a high value of the mean transmission @xmath16 [ defined as @xmath17 where @xmath18 is the optical depth ] that is close to unity .
our goal is to calculate the expected size distribution of such gaps in the ly@xmath0 forest and compare it with observations . given a comoving density of halos per unit mass of @xmath19 and the assumption that the wind from each halo of mass @xmath20 produces bubble of a comoving radius of @xmath21 , the probability of finding an absorption - free gap ( with @xmath22 ) of comoving size in the range between @xmath23 and @xmath24 per infinitesimal comoving path - length , @xmath25 , is given by @xmath26 here , @xmath27 is the cross - section of the bubble for an impact parameter @xmath28 , and @xmath6 is the minimum halo mass into which igm gas accretes efficiently .
the differential cross - section for having a bubble segment of length between @xmath23 and @xmath24 is then , @xmath29 the expected number of absorption - free gaps with a size larger than @xmath23 is @xmath30 since @xmath21 has a rather weak dependence on @xmath20 across the relevant range of halo masses ( see fig . 1 in @xcite ) , we assume for simplicity a single value for @xmath7 when fitting observational data at a given redshift . we adopt the @xcite mass function of halos and
use @xmath6 as the second adjustable parameter .
the concordance cosmological model yields the following values for the cumulative number density : @xmath31 and @xmath32 @xmath33 , for @xmath34 and @xmath35 , respectively .
figure [ f1 ] shows results from equation ( [ eq : cumul ] ) for the three cases of @xmath36 ( dashed line ) , @xmath37 ( dotted line ) , and @xmath38 ( solid line ) .
next , we apply our analytic method to three high - resolution , high signal - to - noise ratio ( @xmath39 ) spectra : q 1422 + 231 at @xmath40 @xcite ; q 0130 - 4021 at @xmath41 and q 0741 + 4741 at @xmath42 ( @xcite ) . these quasar were observed with the keck high resolution echelle spectrometer ( hires ) , at @xmath39 of @xmath43 .
the resolution for q 1422 + 231 is @xmath44 kpc , and for the other two is @xmath45 .
we refer readers to @xcite and @xcite for details of continuum fitting . by fit artificial spectra for 24 qsos
, @xcite measured a typical error in continuum level of 1.2% .
the actually error in our sample should be much smaller due to the higher @xmath39 .
we select a portion of the spectrum around @xmath11 to test our analytic model .
specifically , we consider the observed wavelength band of @xmath46 , corresponding to a comoving path length of @xmath47 mpc . for q 0130 - 4021 and q 0741 + 4741 , only ly@xmath0 absorption presents in this wavelength range , while for q 1422 + 231 , ly@xmath48 will cover part of this range ( but not ly@xmath49 ) and will have quite a bit more aborption .
we split the spectrum into segments with a size of @xmath50 mpc each so that we have a total of 9 segments for the entire sample .
we then normalize each of the 9 segments separately .
we did this by varying the optical depth @xmath18 so that the mean flux @xmath51 , consistent with the observed value at @xmath52 ( @xcite ) .
we did not vary the normalization with wavelength .
we measure gaps in each segment , and the errors are estimated by using the variance between these 9 subsamples .
we directly measure @xmath53 from the observed spectrum .
the flat absorption - free gaps are selected by searching for segments with transmission @xmath54 where @xmath55 is our fiducial threshold .
other choices of the threshold transmission may be considered plausible , since @xcite found that the mean transmitted flux is @xmath56 within 0.5 @xmath57 comoving mpc of the lbgs and possibly higher close to the lbgs .
note that since lbgs are selected by having ongoing starburst activity , their active wind may add some absorption by outflowing material inside their bubbles .
other galaxies are most likely to be surrounded by empty relic bubbles bounded by a thin shell of swept - up gas , long after their powerful starburst activity has ended . however , a low value of @xmath58 is very likely to mix the outflow - driven bubbles discussed here with high - transmission gaps caused by gravitationally - induced voids ( which are underdense but not completely empty ) .
we have chosen the threshold of 0.99 based on a detailed analysis of the results of a numerical simulation from a particle - mesh ( pm ) code involves gravity only ( see .
e.g. , @xcite ) .
we found that the pm simulation provides the same distribution of gaps as the data up to a threshold of @xmath590.97 , and the data shows a clear over - abundance of gaps at a threshold of 0.99 : at @xmath55 the gravity - induced gaps accounts for less than @xmath59 10% of the total gaps found in the observation data . in principle
we could have used the raw spectrum , the noise in each pixel generates artificial gaps with a width of just a few pixels .
instead , we use the fitted spectrum , which smoothes out these random pixel noise . using high - order statistic method
, @xcite found that the coherence scale length of spatial correlation in keck data must be smaller than 0.3@xmath57 mpc , which roughly corresponds to 12 pixels in the spectrum .
so the raw spectrum is smoothed by the boxcar method with a width of 12 pixels to ensure that we can smooth poisson noise in each pixel but preserve the large scale features .
the 1-@xmath60 error bars are obtained directly from poisson statistics .
we fit the observed distribution of gap sizes using equation ( [ eq : cumul ] ) , with two free parameters : @xmath61 and @xmath7 .
we use a least-@xmath62 analysis with @xmath63 ^ 2/\sigma_i^2.\ ] ] here @xmath64 is the expected value of @xmath65 based on equation ( [ eq : cumul ] ) , @xmath66 is the observed value of @xmath65 , and @xmath67 is the 1-@xmath60 error for each data point .
the @xmath62 fit yields the best - fit parameters of @xmath68 and @xmath69 comoving kpc , with a @xmath62 value of 1.2 per degree - of - freedom .
the model fits the observation data well between 0.1 and 1 @xmath57 mpc at @xmath70 implies that the chance for overlap of bubbles with a radius @xmath71 comoving kpc around these galaxies , is small in our data set . ] , as indicated by the red solid line in figure [ f1 ] .
the best - fit value for the minimum halo mass that is capable of hosting outflows ( due to star formation or quasar activity ) , is consistent with the value expected from hydrodynamic simulations and semi - analytic analysis .
the mass of a dark matter halo is related to its circular velocity @xmath72 and collapse redshift @xmath73 ( @xmath74 ) , @xmath75 the infall of gas onto dwarf galaxies is significantly suppressed through photoionization heating of the igm by the cosmic ultraviolet background radiation after cosmological reionization ( @xcite ) .
as soon as the igm is heated to a temperature of @xmath76@xmath77 k , its thermal pressure suppresses gas infall into shallow gravitational potential wells .
the minimum halo mass due to photo - ionization heating corresponds to a virial temperature of @xmath78 k. the reason is that photo - ionization heats the igm to @xmath79 k but when the gas virializes into a halo , its density increases by a factor of @xmath80 and so its temperature rises adiabatically by a factor @xmath81 .
only if this final temperature is below the virial temperature , will the gas condense by this factor of @xmath80 .
the minimum halo mass for galaxies is defined as the threshold where @xmath82% of the baryons that were supposed to condense made it .
this gives a minimum virial temperature of @xmath78 k. the corresponding minimum circular velocity of the halo is @xcite , @xmath83 detailed hydrodynamic simulations ( e.g. , @xcite ) confirm this value , showing that at redshift @xmath52 only @xmath84 of the available gas accretes onto halos with @xmath85 , and the infall of gas onto halos with @xmath86@xmath87 is completely suppressed .
this yields a minimum halos mass of @xmath88 based on equation ( [ eq : minm ] ) , in agreement with the value derived from the statistics of absorption - free gaps in the ly@xmath0 forest .
the typical bubble radius @xmath7 we derived is consistent with both semi - analytic analysis and other observations .
based on a starburst wind model , @xcite showed that at redshift @xmath11 and halos with masses between @xmath89 and @xmath90 , the physical ( proper ) radius @xmath91 ranges between @xmath59 50 to @xmath92 kpc ( see the upper left panel of their fig .
1 ) . for halo masses above @xmath93 , the maximum radius can be achieved by the wind is nearly constant , @xmath92 kpc .
observations indicate the lack of within a radius of 500 comoving kpc around lbgs , which translates to @xmath94 proper kpc at @xmath52 @xcite , although the interpretation may be complicated @xcite .
the gas that is swept - up from the bubble volume by a galactic outflow , is expected to pile - up in a thin shell on the bubble wall , and introduce enhanced absorption at both ends of its absorption - free segment in the ly@xmath0 forest @xcite .
indeed , the spectrum of q 1422 + 231 indicates that almost all of the flat transmission segments have deep absorption features at one or both of their boundaries .
figure [ f2 ] shows four examples of absorption - free segments with enhanced absorption at both ends . in all panels ,
the upper dotted horizontal line show the flux threshold @xmath55 , and the lower horizontal line indicates the mean transmitted flux of @xmath51 at @xmath95 ( bottom ) .
we note that in analogy with supernova remnants @xcite , the shell of the swept - up material may fragment into clumps due to convective or rayleigh - taylor instability .
in such a case , a flat ( absorption - free ) segment would be bounded by a narrow , deep absorption feature only if the line - of - sight happens to cross a clump of gas .
we find that out of 23 absorption - free segments with @xmath96 mpc , 6 are bounded at both ends by deep absorption features extending under @xmath97 , 12 have such a feature at one end , and only 5 are not bounded at all by a flux depression under @xmath97 .
there is tentative evidence for clumpiness of the bubble walls , although the last class of segments may also receive a contribution from ionization ( strmgren ) spheres around bright galaxies .
our analysis assumed that the bubble size can be translated to the spectral width of the absorption - free gaps for a pure hubble flow , and ignored the possible effects peculiar velocities . in reality , the spectral width may be increased due to the bubble expansion relative to the local hubble flow .
this would tend to increase the inferred value of @xmath7 relative to its true physical size .
the local hubble flow at @xmath11 is @xmath98 for bubbles with a radius of @xmath92 proper kpc .
the upper left panel of figure 2 in @xcite shows that for most halos with halo mass between @xmath99@xmath100 the average bubble expansion velocity is around @xmath101 .
since the peculiar velocity , @xmath102 , is much smaller than the hubble flow @xmath103 , we can safely ignore its contribution to the broadening of the flat spectral segments . on the other hand , neutral hydrogen outside the bubble may also have effect in translating the bubble size into the spectral width of the gap , depending on whether the bubble is expanding into an infall region ( surrounding an overdense environment ) or an outflow region ( in a locally underdense region ) : the infall can require the true physical bubble size to be larger ( see , e.g. , @xcite ) . in this _
letter _ we have only attempted to demonstrate the feasibility of our method at @xmath70 .
the statistical significance of our results can be enhanced through the analysis of additional data sets of high - resolution quasar spectra .
we also need better calibrated data , with accurate continuum level , and more realistic simulatoins . in the future
, our method can be used to measure the variation of @xmath104 and @xmath7 with redshift , and to gauge the impact of galaxy feedback on estimates of the dark matter power - spectrum on small scales .
quasar spectra were based on data obtained at the w.m .
keck observatory , a joint facility of the university of california , the california institute of technology , and nasa .
we thank m. rauch and w. salgent for providing keck data of qso 1422 + 231 , and m. white for providing pm simulation data .
t. fang was supported by nasa through the _ chandra _ postdoctoral fellowship award number pf3 - 40030 issued by the _
chandra _ x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of the nasa under contract nas8 - 39073 .
this work was supported in part by nasa grant nag 5 - 13292 , and by nsf grant ast-0204514 ( for a.l . ) .
dt and dk were supported in part by nsf grant ast-0098731 and nasa grants nag5 - 13113 and stsci grant hst - ar-10288.01-a .
|
powerful outflows from star - forming galaxies are expected to push away the neutral intergalactic medium ( igm ) around those galaxies , and produce absorption - free gaps in the ly@xmath0 forest .
we analyze the abundance of gaps of various sizes in three high resolution spectra of quasars at @xmath1 3 3.5 .
the gap statistics agrees well with a model in which galactic halos above a minimum mass scale of @xmath2 produce bubbles with a characteristic radius of @xmath3 mpc .
both numbers are consistent with naive theoretical expectations , where the minimum galaxy mass reflects the threshold for infall of gas out of a photo - ionized igm .
the observed gaps are typically bounded by deep absorption features as expected from the accumulation of swept - up gas on the bubble walls .
= 0.1 in 1.0 in .1 in .1 in
|
several studies have recently been conducted to elucidate the morphology of high redshift objects ( e.g. , glazebrook et al .
@xcite ; driver et al .
@xcite ; abraham et al .
@xcite ; abraham et al .
@xcite ; bouwens et al .
@xcite ; giavalisco et al .
@xcite ) .
near - infrared studies ( @xmath0 2.1 @xmath1 ) probe underlying stellar masses far more accurately than do optical morphologies . in
order to ` penetrate the mask ' and effectively delineate the stellar mass distribution of the older population , optical detectors do not suffice . to dust - penetrate galaxies at high redshifts in their k@xmath2 restframes with the ngst
necessitates the design of detectors sensitive to light whose wavelengths are longward of k@xmath2 ( 2.1 @xmath1 ) . a preliminary description of the ngst facilities are given in gillett and mountain .
bally and morse describe a detector which will operate in the wavelength range 0.6 to 5.3 @xmath1 .
this detector can observe k@xmath2 restframe light of objects out to redshifts of @xmath3 .
the images analysed here were generated following the simulation methodology of takamiya .
the bidimensional fourier method was used to secure information about the morphology ( see , e.g. , puerari and dottori @xcite ; block and puerari @xcite , and references therein ) .
the simulations recreate a given image when moved out to higher redshifts , always in a pre - selected restframe . in this contribution , the images of ngc 922 are simulated at redshifts @xmath4 and @xmath5 in the k@xmath2 restframe
. the original image of ngc 922 was obtained at the nasa infrared telescope facility with nsfcam ( more details can be found in block et al .
@xcite ) .
the bidimensional fourier analysis shows that the near - infrared morphology of ngc 922 remains the same when the galaxy is moved out to redshifts @xmath4 and @xmath5 ( see fig .
the analysis also confirms that only two low - order @xmath6 components ( @xmath7 ) are sufficient to describe the dust - penetrated morphology of ngc 922 ( see fig .
2 ) . image of ngc 922 at its original distance ( top ) .
the middle panels show the fourier spectra when ngc 922 is moved to redshifts @xmath4 ( @xmath8-band ) and @xmath5 ( @xmath9-band ) , where a deceleration parameter @xmath10 is used .
the bottom panels are similar to the middle ones , except that a deceleration parameter of @xmath11 is adopted .
a remarkable similarity in the restframe k@xmath2 images of ngc 922 is found , independent of redshift and of the deceleration parameter assumed . ]
-10pt image .
middle left : simulated 1h exposure ( assuming an 8 m ngst ) , at redshift 0.7 ( l band ) .
bottom left : ngc 922 simulated at redshift z=1.2 ( m band ) with an 8 m ngst .
the right hand panels show contour overlays of the low order @xmath6=1 and 2 inverse fourier transforms . ]
in this paper , we have simulated images of ngc 922 when it is systematically moved out in redshift space , always in restframe k@xmath2 .
our analysis confirms that the morphology of such optically peculiar objects may successfully be described by low - order @xmath6 modes , as for galaxies at @xmath12 in our local universe .
|
images from the hubble deep field ( hdf ) north and south show a large percentage of dusty , high redshift galaxies whose appearance falls outside traditional classification systems .
the nature of these objects is not yet fully understood .
since the hdf preferentially samples restframe uv light , hdf morphologies are not dust or ` mask ' penetrated .
the appearance of high redshift galaxies at near - infrared restframes remains a challenge for the new millennium .
the next generation space telescope ( ngst ) could routinely provide us with such images . in this contribution , we quantitatively determine the dust - penetrated structures of high redshift galaxies such as ngc 922 in their near - infrared restframes .
we show that such optically peculiar objects may readily be classified using the dust penetrated z@xmath00 templates of @xcite and @xcite .
|
study of mg ii absorption line systems in the spectra of qsos has provided a means of detecting distant normal field galaxies which happen to be situated close to the lines of sight to the qsos ( e.g. , bergeron et al . 1991 ; steidel et al .
1994 ) . barring the so called ` associated systems ' ( having @xmath7 )
, the absorption line systems are customarily believed to arise in intervening structures that are wholly unrelated to the background qso .
this view was , however , challenged by the finding that the occurrence rate , @xmath0 , of absorption systems in the spectra of grbs is nearly 4 times the value found for qsos , when strong lines having rest - frame equivalent width @xmath8 are considered ( prochter et al .
later studies have supported this unexpected trend , albeit the excess factor is found to be smaller , 2.1 @xmath9 0.6 ( sudilovsky et al . 2007 ; vergani et al . 2009 ; tejos et al .
more recently , bergeron , boisse & menard ( 2011 , hereinafter bbm ) have examined this issue for the case of blazars which , like grbs , also possess relativistic jets pointed close to our direction , but are considerably less variable . using intermediate - resolution optical / uv spectra of their sample of 45 powerful blazars ( predominantly distant bl lacs ) at @xmath10 , and again excluding ` associated systems ' , these authors have found a factor of @xmath2 2 excess ( 3@xmath11 confidence ) in the incidence of absorption systems , as compared to the systems detected towards qsos .
interestingly , the excess is found both for strong ( @xmath12 ) and weaker ( @xmath13 ) systems , where @xmath14 is the rest - frame equivalent width .
thus , the results pertaining to both grbs and bl lacs hint at a radical premise that the observed occurrence of at least the ( purportedly intervening ) strong mg ii absorption systems is somehow connected with the background source ( bbm ) . [ cols= " <
, < , < , < , < , < , < " , ] origin of the above unexpected result is unclear .
bbm have argued that while dust extinction can lower the apparent incidence of absorbers towards qsos and gravitational lensing can increase it towards grbs and blazars , the expected amplitude of these effects falls short , by at least an order of magnitude , of explaining the afore - mentioned factor of 2 discrepancy between the incidence rates of absorbers found towards blazars / grbs versus normal qsos .
they have further estimated that powerful jet in the blazars are capable of sweeping sufficiently large column densities of gas ( upto @xmath15 ) and accelerating such clouds to velocities of order @xmath16 , thereby possibly accounting for the excess of absorption systems observed in comparison to qsos . from the observational side , evidence is still lacking for such high - velocity outflows of cool material from radio - loud agn , although mildly relativistic outflows of highly ionised gas in the nuclear region now appear to be commonly seen for such sources ( e.g. , holt et al .
based on the observations of fe xxv / xxvi k - shell resonance lines in the x - ray band , outflow speeds of @xmath17 0.15c have recently been inferred for highly ionised gas clouds with column densities of @xmath18 @xmath19 located within the central parsec of the agn ( tombesi et al .
these authors have argued why clouds with even higher ejection velocities could have remained undetected on account of selection effects and instrumental sensitivity limitations
. however independent evidence of such high - velocity outflows of cool material are still essential in view of the recent finding of giustini et al .
( 2011 ) who have detected highly ionized fast ( @xmath20 ) x - ray outflow associated with slower uv outflow ( @xmath21 ) . whilst the above findings lend some credence to the hypothesis that the observed excess of absorption systems towards blazars might have a causal relationship to gaseous outflows triggered by their relativistic jets pointed close to our direction ( bbm ) , it would clearly be desirable to probe this suggestion using an independent sample of agn having powerful
doppler boosted jets . here
we examine this question by using such a sample , but which consists of a different class of powerful agn , namely flat - spectrum radio quasars ( fsrqs , with low optical polarization , hence non - blazar type ) .
compared to the blazar sample studied by bbm , our sample of 115 ( non - blazar ) fsrqs is 2.5 times larger and similar or higher resolution optical / uv spectra are available for all its members , either from the literature or various data archives .
our 4 datasets of fsrqs have been derived from the four publications listed in table 1 .
all but the last of these publications ( narayanan et al . 2007 ) provide samples of radio - selected quasars ; the fourth publication is a mix of radio and optical selection ( table 1 ) . to begin with
, we merged the samples reported in the four publications , obtaining a list of 301 quasars . from this basic list we eliminated the 8 quasars present in more than one of the four samples .
for instance , if a quasars had already occurred in the first sample , it was deleted from all the remaining samples , if present therein . from this
reduced list to 293 mostly radio - selected quasars we then excluded all those not contained in the crates catalog of flat - spectrum radio quasars ( fsrqs ) , which provides for each source the spectral index is defined as : @xmath22 , @xmath23 between 1.4 ghz and 4.8 ghz and contains only sources with @xmath24 and a flux density above 65 mjy at 4.8 ghz ( healey et al . 2007 ) .
this selection filter left us with 201 fsrqs . from this list
we further deleted sources that are either classified as ` hp ' or ` bl ' in the vron - cetty & vron ( 2010 ) catalogue , or as ` bzb ' ( implying bl lac ) , or ` bzu ' ( implying uncertain type ) in the catalog roma - bzcat 2009 ( massaro et al . 2009 ) .
lastly , we excluded all sources which are absent from both these catalogs , treating their nature to be uncertain .
this sequence of selection filters left us with 163 fsrqs ( non - blazars ) , i.e. , 65 , 32 , 54 and 12 fsrqs constituting our dataset-1 , 2 , 3 and 4 , respectively , drawn from the four parent publications mentioned in table 1 .
for the largest dataset-1 ( 65 southern fsrqs ) , the parent publication ( ellison et al . 2004 ) already provides for all but two sources ( j043850.5@xmath25201226/b0436@xmath25203 & j165945.0@xmath26021307/b1654 - 020 ) the entire information relevant for the present study , such as absorber redshift , rest - frame equivalent width ew ( ) and the redshift path values for systems having ew ( ) above the thresholds @xmath27 and @xmath28 ( their table 3 ) . following an unsuccessful search for the optical spectra of the two remaining sources in the uves / vlt archive , both sources were excluded from further analysis . for the sources in the remaining three datasets ( nos . 2 4 ) similar archival search for high resolution
optical spectra was much more crucial , since the requisite information was not available in their parent publications ( table 1 ) .
for the southern dataset-2 , spectra of all 32 fsrqs were found in the uves/[email protected] ] archive .
for the northern dataset-3 , a search was made in the hires / keck archive and spectra for 10 out of the total 54 fsrqs were found .
however , for one of them ( j064204@xmath26675836 ) only raw science frames were available , and calibration file was missing .
for another fsrq ( j185230@xmath26401907 ) , only two exposures of hires / keck observation were available , resulting in low signal - to - noise ratio ( snr @xmath30 ) and hence a much reduced spectral coverage .
we thus ended up with 8 fsrqs with useful spectra , out of the total of 54 fsrqs in the dataset-3 .
lastly , spectra for all 12 fsrqs in our dataset-4 were found in the uves / vlt archive .
this selection process left us with 115 fsrqs ( non - blazar type ) with high - quality optical / uv spectra , contributed by 63 , 32 , 8 and 12 fsrqs out of the dataset-1 , 2 , 3 and 4 , respectively ( table 1 ) . for post - processing of the extracted spectra of the fsrqs belonging to datasets 24 , such as air to vacuum wavelength conversion , heliocentric correction , combining individual exposures to enhance the snr and continuum fitting , we have followed the procedure described in chand et .
al ( 2004 ) .
details of our final sample of 115 fsrqs are given in table [ tab : samp ] .
@rccc ccccc @ & & & @xmath31z significance level . ]
& @xmath32 & @xmath33 & @xmath34 & @xmath35 is used for scaling our computed @xmath0 for the fsrq weak and strong absorption systems , from their respective mean redshifts ( col .
5 ) to the mean redshift of the bbm blazars ( @xmath32 = 0.83 for the weak and 0.82 for the strong absorption systems ) . for details , see the text in the beginning of sect .
& @xmath36 for the weak absorption systems and @xmath37 for the strong absorption systems detected towards the bbm blazars ( their eqs . 5 @xmath38 8) .
] + + weak systems & @xmath39@xmath40 & 53 & 80.23 & @xmath41&@xmath42&@xmath43&0.91&@xmath44 + + strong systems&@xmath45@xmath46 & 22 & 106.02 & @xmath47&@xmath48&@xmath49&0.81&@xmath50 + +
although both uves and hires spectrographs provide a large wavelength coverage , from @xmath2 3000 to @xmath2 10000 , the available spectral coverage varies from quasar to quasar , based on the choice of cross - disperser settings . combining exposures from various settings , therefore ,
resulted sometimes in gaps in the wavelength coverage .
in addition , we systematically excluded the following spectral path lengths from our formal search for absorption systems : ( i ) wavelength region blueward of the ly-@xmath23 emission line , to avoid contamination by the ly-@xmath23 forest ; ( ii ) wavelength regions within 5000 of the emission lines at the quasar redshift , as any absorption lines within this velocity interval have a higher probability of being associated with the quasar itself ( sect .
1 ) ; and ( iii ) spectral regions polluted by the various known atmospheric absorption features ; these were eliminated by eye for any line blending . further , noisy spectral regions got discarded by our considering only the spectral ranges with snr sufficient to detect with @xmath51 significance any absorption line above the adopted ew threshold ( which is 0.3 for weak systems and 1.0 for strong systems ) ; this is also the criterion used in ellison et al .
2004 ( the parent paper for our dataset-1 ) . in searching for absorption systems , within the included redshift path in each quasar spectrum
, we first plotted the normalized spectrum and then plotted over it the same spectrum by shifting the wavelength array by a factor of @xmath52 .
small spectral segments of about 50 were then visually inspected for comparable profile shapes and for a doublet ratio between 1:1 and 2:1 . to further ascertain the detection of the absorption system , we then searched for corresponding metal lines ( e.g. , , , , etc . ) in the velocity plot .
weak doublets that were found within @xmath53 of each other were taken to be part of the same absorption system and therefore classified as one multi - cloud system as is usually done . from the above analysis of the optical spectra of the total 52 fsrqs drawn from the datasets 24
, we detected 29 absorption systems with @xmath54 @xmath40 ( `` weak systems '' ) , and 10 absorption systems with @xmath55 ( `` strong systems '' ) .
thus , including the dataset-1 , the total counts of weak and strong absorption systems in our final sample of 115 fsrqs become 53 and 22 , respectively ( table 4)@xmath56 . the detected absorbing system(s ) have @xmath57 range , median redshift and redshift dispersion , respectively , of @xmath58 , @xmath59 and @xmath60 for the weak systems and @xmath61 , @xmath62 , @xmath63 for the strong absorption systems .
table [ tab : sumary ] summarizes the results for our final sample of 115 fsrqs , giving @xmath64 , where @xmath65 is the number of the observed absorption systems within the redshift path @xmath66 ; for both strong and weak systems .
the errors on @xmath0 are calculated based on poisson statistics for small numbers , with the limits corresponding to 1@xmath11 confidence level of a gaussian distribution , as tabulated by gehrels ( 1986 ) .
it is interesting to note from table [ tab : sumary ] that the @xmath0 for fsrqs and normal qsos is quite similar both for weak and strong absorption systems ( e.g see col .
7 table [ tab : sumary ] ) . here , to compute @xmath0 for normal qsos at the mean redshift of our fsrqs sample ,
we use the fit of @xmath67 versus @xmath0 for strong absorption system ( @xmath68 ) , as given by prochter et al .
( 2006b ) ( see also eq .
6 of bbm ) and for @xmath69 systems as given by nestor ( e.g. , see eq . 2 of bbm , their pvt . comm . ) , derived in both cases by using the qsos in the sdss ( dr4 ) .
thus , for weak systems toward qsos , @xmath0 at any redshift can be computed by taking the difference between the above two fits ( i.e. , @xmath0 for @xmath69 minus @xmath0 for @xmath68 ) . however , for carrying out the comparison with the result reported by bbm for blazars ( e.g , see col .
9 of table [ tab : sumary ] ) , which has different mean redshift compare to mean redshift of our fsrqs sample ( e.g , see col .
5 of table [ tab : sumary ] ) , we have to first take into account the known cosmological evolution of @xmath0 , by scaling our @xmath0 values found for the fsrqs , to the mean redshift of the absorption systems towards the bbm blazars ( @xmath70 ) . to do this scaling ,
we have assumed that the cosmological evolution of @xmath0 for our fsrq sample is similar to that known for @xmath0 towards qsos , which seem reasonable considering ( i ) the lack of @xmath0 evolution studies for radio - loud quasars and ( ii ) that our @xmath0 estimated for fsrqs appears quite similar to that known for qsos ( see above ) .
the computed scale factors , @xmath35 are listed in col .
8 of table [ tab : sumary ] . by multiplying with these scale factors , our computed values of @xmath0 for the fsrq subsets
are scaled to the ( slightly lower ) mean redshift of the bbm blazar sample ( @xmath70 ) , before the ratio of @xmath0 for our fsrq subsets and the corresponding bbm blazar sets are taken ( col . 9 in table 2 ) .
the most interesting result from table [ tab : sumary ] is that for strong absorption systems , @xmath0 for fsrqs is only about half of the value reported by bbm for blazars , which they showed to be itself nearly twice the value known for optically selected quasars ( qsos , which are mostly radio quiet ) . in other words
, the excess abundance of strong absorption systems seen towards blazars ( and also grbs ) , in comparison to qsos ( bbm and references therein ) , does not seem characteristic of flat - spectrum radio quasars of non - blazar type , even though they too possess powerful doppler boosted jets .
bbm have suggested that the excess seen towards blazars might be due to absorbing gas clouds swept up by the powerful jets and accelerated to mildly relativistic velocities ( sect . 1 ) .
while this might readily explain the excess of the absorption systems relative to normal quasars ( i.e. , optically selected , mostly radio - quiet , qsos ) which lack powerful jets , can this be reconciled with the lack of excess found here for fsrqs which do possess powerful relativistic jets ?
a relevant factor here is the likelihood that , compared to blazars ( and grbs ) , the jets in fsrqs may be less well aligned to the line of sight .
a plausible outcome of such a jet orientation scenario , which underlies a popular class of unification models of radio - loud agn ( e.g. , orr & browne 1982 ; antonucci & ulvestad 1985 ; wills & browne 1986 ) , would be that any gas clouds accelerated by the powerful fsrq jets to extremely high ( even mildly relativistic ) speeds may simply not appear in the foreground of the quasar nucleus and hence escape detection as absorption systems . to make a meaningful comparison we have derived a subset from our fsrq sample , by taking only those 15 fsrqs which satisfy the constraints that ( i ) the emission redshift , @xmath72 , falls within the range @xmath73 of the 33 bbm blazars actually showing absorption systems ( out of the total 45 blazars listed in table 2 of bbm ) ) , and ( ii ) the detected absorption systems satisfy @xmath74 ( for weak systems ) and @xmath75 ( for strong systems ) , being the @xmath57 ranges for the 33 bbm blazars . to begin with ,
histograms of @xmath72 for the subset of these 15 fsrqs and the 33 bbm blazars are compared in fig .
[ fig : zhist ] .
k - s test shows that with @xmath76 probability the two histograms are drawn from the same intrinsic distribution of @xmath72 . in view of the known cosmological evolution of @xmath0 ,
this prior confirmation of redshift matching is important . after the above matching of the redshift range both in @xmath72 and @xmath57 , which led to our subset of 15 fsrqs for comparison with the 33 bbm blazars ,
we are left with 14 weak and 3 strong absorption systems seen towards our 15 fsrqs , to be compared with the 19 weak and 13 strong absorption systems detected towards the 33 bbm blazars .
the two panels in fig .
[ fig : beta ] compare @xmath1 distributions for the redshift matched subsets of our fsrqs and bbm blazars , separately for the weak and strong mg ii absorption systems ( recall that ` associated systems ' having @xmath77 0.017 have already been excluded , sect .
the unpaired t - test probabilities that the @xmath1 distributions for blazars and fsrqs are drawn from the same parent population , are @xmath78 and @xmath79 for the weak and strong absorption systems , respectively . here
unpaired t - test is preferred over the ks - test , in view of the small sample size of the absorption systems , especially the strong systems .
thus , the difference between the strong absorption systems towards blazars and fsrqs appears statistically significant , although larger datasets would clearly be very desirable . as seen from fig .
2 ( right panel ) , the main contributor to the difference between the two @xmath80distributions is the conspicuous presence of strong absorption systems for the bbm blazars at @xmath1 values up to 0.15 .
interestingly , bbm have reported a similar excess of strong absorption systems towards their blazars , in comparison to qsos ( their fig .
4 ) , which they interpret by postulating high - speed outflowing clouds of cool gas accelerated by the powerful blazar jets ( sect . 1 ) . while the fsrqs studied here also possess powerful jets , the deficit of strong absorption systems at lower @xmath1 is still present , in comparison to blazars ( fig . 2 right panel ) .
as noted above , a plausible way to reconcile this with the bbm hypothesis is then to consider the possibility that the fsrq jets are less closely aligned to our direction , compared to the jets in the bbm blazars which are also powerful and expected ( by definition ) to be strongly polarized .
such an inference was indeed drawn by valtaoja et al .
( 1992 ) , based purely on their extensive radio flux monitoring data at centimeter wavelengths , which showed that , on average , low - polarization quasars ( fsrqs ) vary with a smaller amplitude and on longer time scale , compared to blazars ( see , also , orr & browne 1982 ) .
one would then expect that any ultra - fast moving absorbing clouds accelerated by powerful fsrq jets would mostly be out of the line of sight to the quasar nucleus .
this suggestion can be tested by the analysis of absorption systems towards _ steep - spectrum _ quasars whose jets are expected to be even less well aligned to our direction ( e.g. , barthel 1999 ; orr & browne 1982 ) .
another hint favoring such an orientation based explanation comes from the radio properties of the small minority of fsrqs that have been detected with the large area telescope ( lat ) on board the _ fermi gamma - ray space telescope_(see linford et al .
these authors find that the @xmath81ray bright lat fsrqs are extreme sources with higher core brightness temperature and greater core polarization , as well as larger ( apparent ) opening angle of the parsec - scale jets , than their non - lat counterparts .
interpreting these differences , they have suggested that the @xmath81ray loud fsrqs can be explained by doppler boosting , but the jet orientation and/or speeds must be significantly different than for the @xmath81ray quiet fsrqs ( which show weaker polarization ) .
this scenario would be consistent with the above assumption that the jets in our sample of ( weakly polarized , hence non - blazar ) fsrqs may be less closely aligned to the line of sight , in comparison to the similarly powerful jets in the bbm blazars .
although , such a bias may well explain the difference between the occurrence rates of strong absorption towards powerful blazars and fsrqs / qsos , further substantiation of this trend employing larger datasets would greatly strengthen the bbm scenario that strong absorption systems may often be revealing the most extreme velocity outflows of cool gas clouds from powerful agn , underscoring the need to investigate the wide - ranging theoretical implications of this phenomenon .
this research has made use of ( i ) the keck observatory archive ( koa ) , which is operated by the w. m. keck observatory and the nasa exoplanet science institute ( nexsci ) , under contract with the national aeronautics and space administration , by using observations done using hires spectrograph at the keck , mauna kea , hawaii ( ii ) eso science archive facility by using observation done using the uves spectrograph at the vlt , paranal , chile and ( iii ) the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
+ we thank an anonymous referee for the constructive criticism and helpful suggestions .
we gratefully acknowledge the help from t. parasakthi with the processed data for the quasars belonging to our dataset-4 , and from ravi joshi in resolving some issues related to the archival data used here .
antonucci , r. r. j. , & ulvestad , j. s. 1985 , , 294 , 158 barthel , p. d. 1999 , , 162 , 127 bergeron , j. , & boiss , p. 1991
, , 243 , 344 bergeron , j. , boiss , p. , & mnard , b. 2011 , , 525 , a51 bernet , m. l. , miniati , f. , & lilly , s. j. 2010 , , 711 , 380 chand , h. , srianand , r. , petitjean , p. , & aracil , b. 2004 , , 417 , 853 ellison , s. l. , churchill , c. w. , rix , s. a. , & pettini , m. 2004 , , 615 , 118 gehrels , n. 1986 , , 303 , 336 giustini , m. , cappi , m. , chartas , g. , et al .
2011 , , 536 , a49 healey , s. e. , romani , r. w. , taylor , g. b. , et al .
2007 , , 171 , 61 holt , j. , tadhunter , c. n. , & morganti , r. 2008 , , 387 , 639 jorgenson , r. a. , wolfe , a. m. , prochaska , j. x. , et al .
2006 , , 646 , 730 linford , j. d. , taylor , g. b. , romani , r. w. , et al .
2011 , , 726 , 16 massaro , e. , giommi , p. , leto , c. , et al .
2009 , , 495 , 691 narayanan , a. , misawa , t. , charlton , j. c. , & kim , t .- s .
2007 , , 660 , 1093 orr , m. j. l. , & browne , i. w. a. 1982 , , 200 , 1067 prochter , g. e. , prochaska , j. x. , & burles , s. m. 2006a , , 639 , 766 prochter , g. e. , prochaska , j. x. , chen , h .- w .
, et al . 2006b , , 648 , l93 steidel , c. c. , dickinson , m. , & persson , s. e. 1994 , , 437 , l75 sudilovsky , v. , savaglio , s. , vreeswijk , p. , et al . 2007 , , 669 , 741 tejos , n. , lopez , s. , prochaska , j. x. , et al
2009 , , 706 , 1309 tombesi , f. , cappi , m. , reeves , j. n. , et al .
2011 , , 742 , 44 valtaoja , e. , terasranta , h. , urpo , s. , et al .
1992 , , 254 , 80 vergani , s. d. , petitjean , p. , ledoux , c. , et al .
2009 , , 503 , 771 vron - cetty , m .- p . , &
vron , p. 2010
, , 518 , a10 wills , b. j. , & browne , i. w. a. 1986 , , 302 , 56 @lrc c rcrl ccl@ & & & & & & & & + & & & & & & & & & & + j001130.5@xmath26005551 & 2.308620 & -0.142 & 140 & 19.10 & v & 1.1656 & 0 & 1.1656 & 0 & 4 + j001602.4@xmath25001225 & 2.086940 & -0.443 & 645 & 18.36 & v & 1.0823 & 0 & 1.0823 & 0 & 4 + j001708.5@xmath26813508 & 3.366000 & -0.200 & 551 & 16.52 & v & 0.5544 & 0 & 0.5544 & 0 & 3 + j004057.6@xmath25014632 & 1.178000 & -0.017 & 581 & 18.30 & o & 0.9942 & 1 & 0.9942 & 0 & 2 + j004201.2@xmath25403039 & 2.478000 & 0.336 & 299 & 19.93 & p & 0.0114 & 0 & 0.7135 & 1 & 1 + .......... & ..... & ... & ... & ..... & . & ...... & . & ..... & . & .
+ + + + + @lrc rr ccl@ & & & & & & & + & & & & & & & + j004057.6 - 014632 & 1.1780 & 0.6828 & @xmath82 & @xmath250.02 & 581.0 & 22.2 & 2 + j004201.2 - 403039 & 2.4780 & 0.8483 & @xmath83 & @xmath260.34 & 299.0 & ..... & 1 + j005108.2 - 065002 & 1.9750 & 1.4919 & @xmath84 & @xmath250.08 & 841.0 & ..... & 1 + j005108.2 - 065002 & 1.9750 & 1.5698 & @xmath85 & @xmath250.08 & 841.0 & ..... & 1 + j024008.2 - 230916 & 2.2230 & 1.3652 & @xmath86 & @xmath250.46 & 3630.0 & 261.1 & 4 + .......... & ..... & ...... & @xmath87 & @xmath88 .... & ...... & ..... & .
|
the conventional wisdom that the rate of incidence of absorption systems , @xmath0 ( excluding ` associated systems ' having velocity @xmath1c relative to the agn of less than @xmath2 5000 km.s@xmath3 ) is totally independent of the background agn , has been challenged by a recent finding that @xmath0 for strong absorption systems towards distant blazars is @xmath4 times the value known for normal optically - selected quasars ( qsos ) .
this has led to the suggestion that a significant fraction of even the absorption systems with @xmath1 as high as @xmath5 may have been ejected by the relativistic jets in the blazars , which are expected to be pointed close to our direction . here
we investigate this scenario using a large sample of 115 flat - spectrum radio - loud quasars ( fsrqs ) which too possess powerful jets , but are only weakly polarized .
we show , for the first time , that @xmath0 towards fsrqs is , on the whole , quite similar to that known for qsos and the comparative excess of strong absorption systems seen towards blazars is mainly confined to @xmath6 .
the excess relative to fsrqs can probably result from a likely closer alignment of blazar jets with our direction and hence any gas clouds accelerated by them are more likely to be on the line of sight to the active quasar nucleus .
|
the study of possible modifications of einstein s theory of gravity has a long history which reaches back to the early 1920s @xcite .
corrections to the gravitational lagrangian , leading to higher - order field equations , were already studied by several authors @xcite shortly after general relativity ( gr ) was proposed .
developments in the 1960s and 1970s @xcite , partly motivated by the quantization schemes proposed at that time , made clear that theories containing _ only _ a @xmath6 term in the lagrangian were not viable with respect to their weak field behavior .
buchdahl , in 1962 @xcite , rejected pure @xmath6 theories because of the non - existence of asymptotically flat solutions . in recent years
, the effort to give a physical explanation to the today observed cosmic acceleration @xcite has attracted a good amount of interest in @xmath0-gravity , considered as a viable mechanism to explain the cosmic acceleration by extending the geometric sector of field equations without the introduction of dark matter and dark energy .
there are several physical and mathematical motivations to enlarge gr by these theories . for comprehensive review ,
see @xcite .
other issues as , for example , the observed pioneer anomaly problem @xcite can be framed into the same approach @xcite and then , apart the cosmological dynamics , a systematic analysis of such theories urges at short scale and in the low energy limit . while it is very natural to extend einstein s gravity to theories with additional geometric degrees of freedom , recent attempts focused on the old idea of modifying the gravitational lagrangian in a purely metric framework , leading to higher - order field equations . due to the increased complexity of the field equations in this framework , the main body of works dealt with some formally equivalent theories , in which a reduction of the order of the field equations was achieved by considering the metric and the connection as independent objects @xcite .
in addition , many authors exploited the formal relationship to scalar - tensor theories to make some statements about the weak field regime @xcite , which was already worked out for scalar - tensor theories @xcite .
also a post - newtonian parameterization with metric approach in the jordan frame has been considered @xcite . in this paper
, we study the post newtonian limit of @xmath0 in the harmonic gauge .
we are going to focus on the small velocity and weak field limit within the metric approach . in principle , any alternative or extended theory of gravity should allow to recover positive results of general relativity
. it will be very important to check at any level of our modified theory we can cover the outcomes of gr .
the plan of the paper is the following . in the sec.[fieldequations ] , we report the complete scheme of newtonian and post - newtonian limit of field equations for @xmath0-gravity and their formal solutions in the harmonic gauge condition .
general comments about the mathematical properties of equations , their relative solutions ( gauss and birkhoff theorem ) and minkowskian behavior of metric tensor are reported . in the sec .
[ ball_like ] we show the complete solutions ( newtonian and post - newtonian level ) when an uniform mass ball - like source is considered .
the point - like source limit of the newtonian solution is considered and the compatibility of @xmath0-gravity with respect to gr is shown .
concluding remarks are drawn in sec .
[ conclusions ] .
let us start with a general class of higher order theories given by the action @xmath7\end{aligned}\ ] ] where @xmath2 is an unspecified function of curvature invariant @xmath8 .
the term @xmath9 is the minimally coupled ordinary matter contribution . in the metric approach ,
the field equations are obtained by varying ( [ hogaction ] ) with respect to @xmath10 .
we get @xmath11 @xmath12 here , @xmath13 is the the energy - momentum tensor of matter , while @xmath14 is the trace , @xmath15 , @xmath16 and @xmath17 . ] .
the conventions for ricci s tensor is @xmath18 while for the rienman tensor is @xmath19 .
the affinities are the usual christoffel s symbols of the metric : @xmath20 .
the adopted signature is @xmath21 ( see for the details @xcite ) .
the paradigm of post - newtonian limit is starting from a develop of metric tensor ( and of all additional fields in the theory ) with respect to dimensionless quantity @xmath22 .
a system of moving bodies radiates gravitational waves and thus loses energy .
this loss appears only in the fifth approximation in @xmath22 . in the first four approximations ,
the energy of the system remains constant . from this
it follows that a system of gravitating bodies can be described by a lagrangian correctly to terms of order @xmath23 in the absence of an electromagnetic field , for which a lagrangian exists in general only to terms of second order .
we thus find the equations of motion of the system in the next approximation after the newtonian .
to solve our problem we must start with the determination , in this same approximation of the weak gravitational field , of the metric tensor @xmath10 ( for details see @xcite ) @xmath24 the set of coordinates and @xmath25 ; the latin index between @xmath26 and @xmath25 .
] adopted is @xmath27 .
the ricci scalar becomes @xmath28 the @xmath29-th derivative of ricci function can be developed as @xmath30 from lowest order of field equations ( [ fieldequationhog ] ) we have @xmath31 which trivially follows from the above assumption ( [ metric_tensor_ppn ] ) that the space - time is asymptotically minkowskian .
this result suggests a first consideration .
if the lagrangian is developable around a vanishing value of the ricci scalar the relation ( [ ppn - field - equation - general - theory - fr - o0 ] ) will imply that the cosmological constant contribution has to be zero whatever is the @xmath0-gravity theory .
this result appears quite obvious but sometime it is not considered in literature @xcite .
( [ fieldequationhog ] ) and ( [ tracefieldequationhog ] ) at @xmath32 - order ( newtonian level ) become @xmath33\delta_{ij}-f''(0)r^{(2)}_{,ij}=0\\\\ h^{(2)}=-3f''(0)\triangle r^{(2)}-f'(0)r^{(2)}=\mathcal{x}\,t^{(0 ) } \end{array}\right.\end{aligned}\ ] ] where @xmath34 is the laplacian in the flat space , while at @xmath35 - order become @xmath36 the solution for the gravitational potential @xmath37 has a yukawa - like behavior ( @xcite ) depending by a characteristic length on which it evolves . besides the birkhoff theorem at newtonian level
is modified : the solution can be only factorized with a function space - depending and an arbitrary function time depending ( @xcite ) .
still more the corrections to the gravitational potential and the gravito - magnetic effects ( [ ppn - field - equation - general - theory - fr - o3 ] ) are depending on the only first two derivatives of @xmath2 in @xmath38 .
so different theories from the third derivative admit the same newtonian solution . remembering the expressions of christoffel symbols and using the following approximation for the determinant of metric tensor @xmath39+\dots$ ] , at @xmath40 - order we have , @xmath41\\\\\,\,\,\,\,\,\,\,\,\,\,\ , \,\,\,\,\,\,\,\,\,-f'''(0)\biggl[|\nabla r^{(2)}|^2+r^{(2)}\triangle r^{(2)}\biggr]=\mathcal{x}\,t^{(2)}_{tt}\\\\\\ h^{(4)}=-3f''(0)\triangle r^{(4)}-f'(0)r^{(4)}-3f'''(0)\biggl[|\nabla r^{(2)}|^2+r^{(2)}\triangle r^{(2)}\biggr]\\\\\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,+3f''(0)\biggl[r^{(2)}_{,tt}-g^{(2)}_{mn}r^{(2)}_{,mn}- \frac{1}{2}\nabla(g^{(2)}_{tt}-g^{(2)}_{mm})\cdot\nabla r^{(2)}-g^{(2)}_{mn , m}r^{(2)}_{,n}\biggr]=\mathcal{x}\,t^{(2 ) } \end{array}\right.\end{aligned}\ ] ] where @xmath42 is the gradient in the flat space .
note that the propagation of ricci scalar @xmath43 has the same dynamics of previous one .
the complete knowledge of correction at fourth order for the @xmath4-component of ricci tensor fix the third derivative of @xmath2 in @xmath38 .
also at this level there is a degeneracy of @xmath0-theory : different theories for only the first three derivatives admit the same gravitational field without obviously radiation emission . to complete the perturbative scheme and later find the solutions it needs to calculate the ricci tensor components in ( [ ppn - field - equation - general - theory - fr - o2 ] ) , ( [ ppn - field - equation - general - theory - fr - o3 ] ) , ( [ ppn - field - equation - general - theory - fr - o4 ] ) .
after some calculus ( @xcite , @xcite ) one obtains @xmath44 which represent the most general expressions without assuming any gauge condition .
we want to rewrite and generalize the outcome of ( @xcite ) by introducing the green function method ( we remember that the newtonian limit corresponds also to linearization of field equations ) .
let us start from the trace equation .
the solution for the ricci scalar @xmath45 in the third line of ( [ ppn - field - equation - general - theory - fr - o2 ] ) is @xmath46 where @xmath47 and @xmath48 is the green function of field operator @xmath49 .
the solution for @xmath50 , from the first line of ( [ ppn - field - equation - general - theory - fr - o2 ] ) by considering that @xmath51 , is @xmath52 we can check immediately that when @xmath5 we find @xmath53 .
the expression ( [ new_sol ] ) is the `` modified '' gravitational potential ( here we have a factor 2 ) for @xmath0-gravity .
_ a such solution which is the newtonian limit of @xmath0-gravity is also gauge - free_. since we have a linearized version of field equations a such limit corresponds to one of einstein equation and the linear superposition is satisfied .
so the @xmath4-component of energy - momentum tensor is , in this limit , the sum of mass energy volume density of sources : @xmath54 where @xmath55 is the delta function .
as it is evident the gauss theorem is not valid since the force law is not @xmath56 .
the equivalence between a spherically symmetric distribution and point - like distribution is not valid and how the matter is distributed in the space is very important ( @xcite ) . in this limit
the geodesic equation is the lagrangian of material point - like embedded in the gravitational field and its dynamics follows the newtonian law : @xmath57 to simplify the expressions of the components of the ricci tensor ( [ ppn - ricci - tensor ] ) we can use the condition so - called of _ harmonic gauge _ : @xmath58 ( see @xcite ) .
the ricci tensor assumes the following simpler form : @xmath59 from the field equation ( [ ppn - field - equation - general - theory - fr - o3 ] ) we find the general solution for @xmath60 @xmath61 the choice of harmonic gauge enable us to solve with facility the equation ( [ ppn - field - equation - general - theory - fr - o3 ] ) but we lose potential information about the temporal dynamics of @xmath62 .
a such knowledge is very important to obtain at least in perturbative approach some information about the birkhoff theorem . by hypothesizing a perturbative approach ( newtonian - like ) we relegated inevitably eventual temporal dynamics only on the temporal variation of matter source .
in fact in a such hypothesis of work the motion of bodies embedded in gravitational fields develops very slow with respect to motion of matter .
then we have ever an instantaneous readjustment of spacetime .
in other words the motion od bodies is adiabatic and it enables us to factorize the solution and with a time transformation we get a static solution . from second line of ( [ ppn - field - equation - general - theory - fr - o2 ] )
the solution for @xmath63 follows @xmath64\delta_{ij}\nonumber\\\nonumber\\&+&\biggl[\frac{1}{2\pi m^2 ( t,\mathbf{x}')-\frac{2}{3m^2}r^{(2)}(t,\mathbf{x})\biggr]\frac{x_i x_j}{|\mathbf{x}|^2}\end{aligned}\ ] ] where @xmath65 represents the integration volume with radius @xmath66 ( for the details see @xcite ) . by the solutions ( [ new_sol ] ) , ( [ postnew_sol_ti ] ) , ( [ post_new_ij ] ) we can affirm that it is possible to have solution non - ricci - flat in vacuum : _ higher order gravity mimics a matter source_. it is evident from ( [ new_sol ] ) the ricci scalar is a `` matter source '' which can curve the spacetime also in absence of ordinary matter .
then it is clear also that the knowledge of behavior of ricci scalar inside mass distribution is fundamental to obtain the behavior of metric tensor outside the mass . from the fourth order of field equation ,
we note also the ricci scalar ( @xmath43 ) propagates with the same @xmath67 ( the second line of ( [ ppn - field - equation - general - theory - fr - o4 ] ) ) and the solution at second order originates a supplementary matter source in r.h.s . of ( [ fieldequationhog ] ) .
the solution is @xmath68 \nonumber\\\nonumber\\&&-\frac{1}{2}\nabla_{\textbf{x}'}\biggl[g^{(2)}_{tt}(t,\textbf{x}')-g^{(2)}_{mm}(t,\textbf{x } ' ) \biggr]\cdot \nabla_{\textbf{x}'}r^{(2)}(t,\textbf{x } ' ) \biggr\}\end{aligned}\ ] ] where @xmath69 .
also in this case we can have a non - vanishing curvature in absence of matter .
the solution for @xmath70 , from the first line of ( [ ppn - field - equation - general - theory - fr - o4 ] ) , is @xmath71 \nonumber\\\nonumber\\ & & + \frac{1}{4\pi}\biggl[g^{(2)}_{mn}(t,\textbf{x}')g^{(2)}_{tt , mn}(t,\textbf{x}')-g^{(2)}_{tt , tt}(t,\textbf{x } ' ) -|\nabla_{\textbf{x}'}g^{(2)}_{tt}(t,\textbf{x}')|^2-r^{(4)}(t,\textbf{x}')-g^{(2)}_{tt}(t,\textbf{x}')r^{(2 ) } ( t,\textbf{x}')\biggr ] \nonumber\\\nonumber\\ & & + \frac{1}{6\pi m^2}\biggl[\frac{{r^{(2)}}^2(t,\textbf{x}')}{4}-\frac{r^{(2)}(t,\textbf{x}')\triangle g^{(2)}_{tt}(t,\textbf{x}')}{2}+g^{(2)}_{mn , m}(t,\textbf{x}'){r^{(2)}}_{,n } ( t,\textbf{x}')+\triangle r^{(4)}(t,\textbf{x } ' ) \nonumber\\\nonumber\\ & & + g^{(2)}_{tt}(t,\textbf{x}')\triangle r^{(2)}(t,\textbf{x } ' ) + g^{(2)}_{mn}(t,\textbf{x}'){r^{(2)}}_{,mn}(t,\textbf{x}')-\frac{1}{2}\nabla g^{(2)}_{mm}(t,\textbf{x}')\cdot\nabla r^{(2)}(t,\textbf{x}')\biggr]\biggr\}\end{aligned}\ ] ] we conclude this paragraph by having shown the more general solution of field equations of @xmath0-gravity in the newtonian and post - newtonian limit assuming a coordinates transformation for the whose the gauge harmonic condition is verified . in the next paragraph
we shall apply a such scheme to obtain the explicit form of metric tensor for a static and spherically symmetric matter source .
let us consider a ball - like source with mass @xmath72 and radius @xmath73 .
the energy - momentum tensor is ( we are not interesting to the internal structure ) @xmath74 where @xmath75 is the mass density and @xmath76 satisfies the condition @xmath77 , @xmath78 . since ( [ metric_tensor_ppn ] ) the expression ( [ emtensor ] ) becomes @xmath79 the possible choices of green function , for spherically symmetric systems ( _ i.e. _ @xmath80 ) , are the following @xmath81 with @xmath82 .
it notes , for any function of only modulus @xmath83 , that @xmath84 where @xmath85 and @xmath86 is the angle between two vectors @xmath87 , @xmath88 . in
the spherically symmetric case we can choose @xmath89 without losing generality ( the symmetry of system is independent by the angle ) . by making the angular integration
we get @xmath90\end{aligned}\ ] ] an analogous relation is useful also for the green function of newtonian mechanics @xmath91 : @xmath92h(|\textbf{x}'|)\end{aligned}\ ] ] supposing that @xmath93 ( _ i.e. _ @xmath94\,=\,-\,\text{sign}[f''(0)]$ ] ) the ricci scalar ( [ scalar_ricci_sol_gen ] ) , if we hypothesize a matter density as @xmath95 , is @xmath96\theta(\xi-|\mathbf{x}| ) -r_g\,m^2f(\xi)\frac{e^{-m|\mathbf{x}|}}{|\mathbf{x}|}\,\theta(|\mathbf{x}|-\xi)\end{aligned}\ ] ] where @xmath97 is the heaviside function , @xmath98 and @xmath99 is the schwarzschild radius ( otherwise we have to renormalize the coupling constant @xmath100 in the action ( [ hogaction ] ) ) . ]
. the solutions of ( [ new_sol ] ) , ( [ postnew_sol_ti ] ) and ( [ post_new_ij ] ) , given the relations ( [ int2 ] ) and ( [ int3 ] ) , respectively are @xmath101\,\theta(\xi-|\mathbf{x}|)-r_g\biggl[\frac{1}{|\textbf{x}|}+\frac{f(\xi)}{3 } \frac{e^{-m|\mathbf{x}|}}{|\mathbf{x}|}\biggr]\theta(|\mathbf{x}|-\xi)\end{aligned}\ ] ] @xmath102 @xmath103\theta(\xi-|\mathbf{x}|)\nonumber\\\nonumber\\&&\,\,\,\ , + \biggl[\frac{1}{|\textbf{x}|}-\frac{2 } { 3m^2|\textbf{x}|^3}-\frac{f(\xi)}{3}\biggl(\frac{1}{|\mathbf{x}|}-\frac{2}{m|\textbf{x}|^2}-\frac{2 } { m^2|\textbf{x}|^3}\biggr)e^{-m|\textbf{x}|}\biggr]\theta(|\mathbf{x}|-\xi)\biggr\ } \delta_{ij}\nonumber\\\nonumber\\ -&r_g&\biggl\{\biggl[\frac{2(1+m\xi)e^{-m\xi}}{m^2\xi^3}\biggl(\frac{\sinh m|\textbf{x}|}{m|\textbf{x}|}-f(\textbf{x})\biggr)\biggr]\theta(\xi-|\textbf{x}|)\nonumber\\\nonumber\\ & & \,\,\,\,+\biggl [ \frac{2}{m^2|\textbf{x}|^3}-\frac{2f(\xi)}{3}\biggl ( \frac{1}{|\textbf{x}| } + \frac{3}{m|\textbf{x}|^2 } + \frac{3}{m^2|\textbf{x}|^3}\biggr)e^{-m|\textbf{x}| } \biggr]\theta(|\textbf{x}|-\xi)\biggr\}\frac{x_ix_j}{|\textbf{x}|^2}\end{aligned}\ ] ] for fixed values of the distance @xmath66 , the solutions @xmath50 and @xmath63 depend on the value of the radius @xmath73 , then the gauss theorem does not work also if the bianchi identities hold @xcite .
in other words , since the green function does not scale as the inverse of distance but has also an exponential behavior , the gauss theorem is not verified .
we can affirm : _ the potential does not depend only on the total mass but also on the mass - distribution in the space_. by introducing three metric potentials @xmath104 , @xmath105 and @xmath106 ( the dimension is the inverse of length ) we can rewrite ( [ sol_pot ] ) and ( [ gsol2 ] ) as follows @xmath107 and with a fourth function , @xmath108 , ( the dimension is the cubic inverse of length ) the ricci scalar ( [ sol_ric ] ) is @xmath109 the spatial behavior ( [ sol_ric ] ) is shown in fig .
[ plotricciscalar ] .
the metric potentials are shown in figs .
[ plotpontential00 ] , [ plotpontentialij1 ] and [ plotpontentialij2 ] .
it is interesting to note as the function @xmath110 assumes smaller value of its equivalent in gr , then in terms of gravitational attraction we have a potential well more deep . a such scheme can be interpretable or assuming a variation of the gravitational constant @xmath111 or requiring that there is a central greater mass .
these two affirmations are compatible on the one hand with the tensor - scalar theories ( in the which we have a scaling of gravitational constant ) and on the other hand with the theory of gr plus the hypothesis of the existence of the dark matter . in particular , if the mass distribution takes a bigger volume , the potential increases and vice versa . for @xmath112 representing the spatial behavior of ricci scalar at second order .
in gr we would have @xmath113.,title="fig : " ] + vs distance from central mass with @xmath112 .
the dashed line is the gr behavior : @xmath114 \theta(\xi-\textbf{x})-\frac{\theta(\textbf{x}-\xi)}{|\textbf{x}|}$].,title="fig : " ] + vs distance from central mass with @xmath112 .
the dashed line is the gr behavior ( similar to metric potential @xmath110).,title="fig : " ] + vs distance from central mass with @xmath112 . in gr
a such behavior is missing.,title="fig : " ] + in the limit of point - like source , _ i.e. _ @xmath115 and @xmath116 , we get @xmath117\frac{x_ix_j}{|\textbf{x}|^2 } \end{array}\right.\end{aligned}\ ] ] an important point has to be considered .
the ppn - parameters @xmath118 and @xmath119 , in the gr context , are intended to parameterize the deviations from the newtonian behavior of the gravitational potentials .
they are defined according to the standard eddington metric ( in the vacuum ) @xmath120 in particular , the ppn parameter @xmath118 is related with the second order correction to the gravitational potential while @xmath119 is linked with the fourth order perturbation in @xmath22 . since the gauss theorem is not verified for @xmath0-gravity , while the relations ( [ ppn_parameters ] )
satisfy it , we must consider the relations ( [ sol_new_p ] ) and not ( [ sol_pot ] ) , ( [ gsol2 ] ) .
after making the limit @xmath5 we can compare the results with ( [ ppn_parameters ] ) . actually , if we consider this limit for ( [ sol_new_p ] ) , we have @xmath121 which suggest that the @xmath0-gravity is compatible with respect to gr . at last it is interesting to note that also in the case of extended spherically symmetric distribution of matter when we perform the limit @xmath5 the solutions ( [ sol_pot ] ) and ( [ gsol2 ] ) directly converge ( in the vacuum ) in solutions ( [ sol_gr_vacuum ] ) , demonstrating the validity of gauss theorem in gr . other important consideration is about the asymptotic behavior of @xmath2-theory with respect to gr .
in fact increasing the distance from the central mass the gravitational field brings near to one of gr .
a such rejoining is a normal consequence of the hypothesis ( spherically symmetric system and request of minkowskian limit ) .
at last in fig .
[ plotricciscalar ] we report the spatial behavior of ricci scalar ( [ sol_ric ] ) approximating asymptotically the given value in gr .
in fact hypothesizing a @xmath0- theory the ricci scalar acquires dynamics , and in the newtonian limit , we find a characteristic scale length ( @xmath122 ) on the which distance the scalar massive field evolves . only for distances more great than @xmath122
we recover the outcome of gr : @xmath38 . to conclude this section we show in fig .
[ plotforce ] the comparison between gravitational forces induced in gr and in @xmath0-theory in the newtonian limit . obviously also about the force we obtained an intensity stronger than in gr .
-theory with @xmath112 .
the dashed line is the gr behavior.,title="fig : " ] + if we consider @xmath123 ( i.e. @xmath94\,=\,\text{sign}[f''(0)]$ ] ) from ( [ green_function ] ) we can choose the `` oscillating '' green function @xmath124 the ricci scalar ( [ scalar_ricci_sol_gen ] ) and the @xmath4-component of @xmath10 at @xmath32 order ( [ new_sol ] ) become @xmath125\theta(\xi-|\mathbf{x}| ) -2\,r_g\,m^2g(\xi)\frac{\cos m|\textbf{x}|+\sin m|\textbf{x}|}{|\mathbf{x}|}\,\theta(|\mathbf{x}|-\xi)\end{aligned}\ ] ] @xmath126\,\theta(\xi-|\mathbf{x}|)-r_g\biggl[\frac{1}{|\textbf{x}|}-\frac{2g(\xi)}{3 } \frac{\cos m|\mathbf{x}|+\sin m|\mathbf{x}|}{|\mathbf{x}|}\biggr]\theta(|\mathbf{x}|-\xi)\end{aligned}\ ] ] where @xmath127 and @xmath128 , with the properties @xmath129 and @xmath130 .
since we have an oscillating green function ( it is not asymptotically zero ) the `` gravitational potentials '' ( [ sol_pot_oscil ] ) at infinity are zero unless a constant value ( @xmath131 ) .
the spatial behavior of ricci scalar ( [ sol_ric_oscil ] ) and metric component ( [ sol_pot_oscil ] ) are shown in figs .
[ plotricciscalar_oscil ] and [ plotpontential00_oscil ] . with @xmath112 representing the spatial behavior of ricci scalar at second order in the oscillating case.,title="fig : " ] + vs distance from central mass with the choice @xmath112 in the oscillating case .
the dashed line is the gr behavior.,title="fig : " ] + the considerations of preceding subsection hold also for the solutions ( [ sol_ric_oscil ] ) - ( [ sol_pot_oscil ] ) .
the only difference is that now we have oscillating behaviors instead of exponential behaviors .
the correction term to the newtonian potential in the external solution can be interpreted as the fourier transform of the matter density @xmath75 .
in fact , we have : @xmath132 also in this case we conclude the section showing in fig .
[ plotforce_oscil ] the comparison between gravitational forces induced in gr and in @xmath0-theory in the newtonian limit . obviously also in this last case we obtained a force stronger than in gr .
-theory with @xmath112 in the oscillating case .
the dashed line is the gr behavior.,title="fig : " ] + the metric potentials and the function @xmath133 , respectively defined in ( [ pas ] ) and ( [ pas_1 ] ) , satisfy the following properties with respect to derivative of coordinate @xmath134-th ] in the matter @xmath135\,x_l\,=\,\phi_0(\textbf{x})\,x_l \\\\ \psi_{,l}(\textbf{x})\,=\,\biggl[\frac{1}{\xi^3}-\frac{(1+m\xi)}{\xi^3}\,e^{-m\xi}\,\frac{(m^2|\textbf{x}|^2 + 6)\sinh m|\mathbf{x}|+m|\textbf{x}|(m^2|\textbf{x}|^2 - 6)\cosh m|\textbf{x}|}{m^5|\textbf{x}|^5}\biggr]\,x_l\,=\,\psi_0 ( \textbf{x})\,x_l \\\\ \lambda_{,l}(\textbf{x})\,=\,\frac{2(1+m\xi)}{\xi^3}\,e^{-m\xi}\,\frac{(4m^2|\textbf{x}|^2 + 9)\sinh m|\mathbf{x}|-m|\textbf{x}|(m^2|\textbf{x}|^2 + 9)\cosh m|\textbf{x}|}{m^5|\textbf{x}|^5}\,x_l\,=\,\lambda_0(\textbf{x } ) \,x_l\\\\ \xi_{,ln}(\textbf{x})\,=\,\xi_0(\textbf{x})\delta_{ln}+\frac{3(1+m\xi)}{\xi^3}\,e^{-m\xi}\,\frac{(m^2|\mathbf{x}|^2 + 3)\sinh m |\mathbf{x}|-3m|\mathbf{x}|\cosh m|\mathbf{x}|}{m|\mathbf{x}|^5}\,x_lx_n\,=\,\xi_0(\textbf{x})\delta_{ln}+\xi_1(\textbf{x})x_lx_n \\\\
\phi_{,ln}(\textbf{x})\,=\,\phi_0(\textbf{x})\delta_{ln}+\frac{(1+m\xi)}{\xi^3}\,e^{-m\xi}\,\frac{(m^2|\mathbf{x}|^2 + 3)\sinh m |\mathbf{x}|-3m|\mathbf{x}|\cosh m|\mathbf{x}|}{m^3|\mathbf{x}|^5 } \,x_lx_n\,=\,\phi_0(\textbf{x})\delta_{ln}+\phi_1(\textbf{x})x_lx_n \end{array}\right.\end{aligned}\ ] ] and in the vacuum @xmath136\,x_l\,=\,\phi_0(\textbf{x})\,x_l \\\\ \psi_{,l}(\textbf{x})\,=\,\biggl[\frac{m^2|\textbf{x}|^2 - 2}{m^2|\textbf{x}|^5}-\frac{m^3 ( \textbf{x})\,x_l \\\\ \lambda_{,l}(\textbf{x})\,=\,\biggl[\frac{6}{m^2|\textbf{x}|^5}-\frac{m^3|\textbf{x}| ^3 + 4m^2|\textbf{x}|^2 + 9m|\textbf{x}|+9}{m^2|\textbf{x}|^5}\frac{2f(\xi)e^{-m|\textbf{x}|}}{3}\biggr]\,x_l\,=\,\lambda_0(\textbf{x } ) \,x_l\\\\ \xi_{,ln}(\textbf{x})\,=\,\xi_0(\textbf{x})\delta_{ln}-\frac{m^2(m^2|\textbf{x}|^2 + 3m|\textbf{x}|+3 ) } { \\\\
\phi_{,ln}(\textbf{x})\,=\,\phi_0(\textbf{x})\delta_{ln}-\biggl[\frac{3}{| \textbf{x}|^5}+\frac{m^2|\textbf{x}|^2 + 3m|\textbf{x}|+3}{|\textbf{x}|^5}\frac{f(\xi)e^{-m|\textbf{x}|}}{3}\biggr]x_lx_n\,=\,\phi_0(\textbf{x})\delta_{ln}+\phi_1(\textbf{x})x_lx_n \end{array}\right.\end{aligned}\ ] ] obviously when we consider the physics in the matter or in the vacuum we choose the `` right '' quantities @xmath137 , @xmath138 , @xmath139 , @xmath140 , @xmath141 , @xmath142 . the expression of ricci scalar at fourth order , ( [ ricci_quarto ] ) , is @xmath143 \nonumber\\\nonumber\\ & & + \biggl[3\lambda(\mathbf{x}')+\frac{\phi_0(\mathbf{x}')-\psi_0(\mathbf{x}')+\lambda_0(\textbf{x}')}{2}|\textbf{x}'|^2 \biggr]\xi_0(\textbf{x } ' ) + m^2\psi(\textbf{x}')\xi(\textbf{x}')+\xi_1(\textbf{x}')\lambda(\textbf{x}')|\textbf{x}'|^2\biggr\}\end{aligned}\ ] ] from the which we note two contributes .
the first one still depends on the quadratic term ( @xmath144 ) in the action ( [ hogaction ] ) , while the second one is related to cubic term ( @xmath145 ) . by introducing two functions @xmath146 and @xmath147 , the
( [ ricci_quarto_solution ] ) is rewritable as follows @xmath148\end{aligned}\ ] ] an analogous situation is found for the @xmath4-component of metric tensor at fourth order .
in fact the ( [ temp_temp_quarto ] ) becomes @xmath149\nonumber\\\nonumber\\ & & \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \,\,\,\,\,\,\,\,\,\,\,\,\,\,\
, -\frac{1}{2}\biggl[\biggl(3\psi(\mathbf{x}')+\lambda(\mathbf{x}')\biggr)\phi_0(\mathbf{x}')+ \biggl(\psi(\mathbf{x}')+\lambda(\mathbf{x}')\biggr)\phi_1(\mathbf{x}')|\mathbf{x}'|^2\biggr]\biggr\ } \nonumber\\\nonumber\\ & & -\frac{{r_g}^2}{3m^2|\mathbf{x}|}\int_0^\infty d|\mathbf{x}'||\textbf{x}'|\biggl\{||\mathbf{x}|-|\mathbf{x}'||-|\mathbf{x}|-|\mathbf{x}'|\biggr\ } \biggl\{\frac{\xi(\textbf{x}')^2}{4}+\triangle_{\mathbf{x}'}\xi_i(\mathbf{x}')+\xi_1(\mathbf{x}')\lambda ( \mathbf{x}')|\mathbf{x}'|^2 \nonumber\\\nonumber\\ & & \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , + \frac{\xi_0(\mathbf{x}')}{2}\biggr[6\lambda(\mathbf{x}')+\biggl(5\psi_0(\mathbf{x } ' ) + 3\lambda_0(\mathbf{x}')\biggr)|\mathbf{x}'|^2\biggr]\nonumber\\\nonumber\\ & & \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , + \frac{\xi(\mathbf{x}')}{2}\biggr[2m^2\biggl(\phi(\mathbf{x}')+\psi(\mathbf{x}')\biggr)-3\phi_0(\mathbf{x } ' ) -\phi_1|\mathbf{x}'|^2\biggr]\biggr\ } \nonumber\\\nonumber\\ & & + \frac{m^4}{\mu^4}\frac{{r_g}^2}{|\mathbf{x}|}\int_0^\infty d|\mathbf{x}'||\textbf{x}'|\biggl\{||\mathbf{x}|-|\mathbf{x}'||-|\mathbf{x}|-|\mathbf{x}'|\biggr\ } \biggl\{\frac{\xi_{ii}(\textbf{x}')}{2}-\frac{m^2\xi(\textbf{x}')^2+|\textbf{x}'|^2\xi_0(\textbf{x}')^2}{3m^4 } \nonumber\\\nonumber\\ & & \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \,\,\,-\frac{\triangle_{\mathbf{x}'}\xi_{ii}(\mathbf{x}')}{3m^2}\biggr\}\end{aligned}\ ] ] and by introducing other new functions @xmath150 , @xmath151 we have @xmath152\end{aligned}\ ] ] it is useful to note that we have generally four contributes to @xmath153 in ( [ g00_quarto_solution ] ) .
the first one is induced by the non - linearity of the metric tensor even in our static spherically symmetric case .
the product @xmath154 is not zero only in the matter but contributes to determination of @xmath4-component in any point of space .
the second one holds account of the induced contribution , by solution of previous order , to the determination of the @xmath4-component of the ricci tensor at fourth order .
these first two terms are present also in gr . while the second two ones are derived from the modification of theory .
in fact the third contribution depends on the addition of the quadratic term ( @xmath144 ) in the action and finally the fourth one from the addition of the cubic term ( @xmath145 ) .
the choice of free parameter @xmath155 ( which is linked to third derivative of @xmath0 ) is a crucial point in both the expressions ( [ pas_ricci_quarto_solution ] ) and ( [ pas_g00_quarto_solution ] ) to obtain the right behavior . from mathematical interpretation of newtonian limit
one has @xmath156 and if @xmath157 ( _ i.e. _ @xmath94\,=\,-\,\text{sign}[f'''(0)]$ ] , otherwise @xmath158 is not a length ) we have @xmath159 , so we find the constraint @xmath160 . in fig .
( [ plotricciquartoordine ] ) we report the spatial behavior of ( [ pas_ricci_quarto_solution ] ) in the matter and in the vacuum , ( @xmath161 ) , showing that far from source we obtain a spacetime with a vanishing scalar curvature . at newtonian level
the ricci scalar @xmath45 is negative defined ( [ sol_ric ] ) while at post - newtonian limit is positive defined . in fig .
( [ plotg00quartoordine ] ) we report the time - time component of metric tensor , @xmath70 , on the same interval of values of @xmath162 , although the behavior is quite insensitive to changes induced by the contributions of the cubic term in the lagrangian . besides we can observe an important analogy with respect the results of gr . in both cases
we have a potential barrier , but for @xmath0-gravity it is higher ( as in the newtonian limit we found a deeper potential well ) . , representing the ricci scalar at fourth order , where @xmath163 and @xmath112 .
the spatial behavior is shown for @xmath164 ( solid lines ) while the dotted line corresponds to @xmath165-theory.,title="fig : " ] + ( solid lines ) where @xmath163 and of function @xmath166 ( dashed line ) . for @xmath167 ( dotted line )
we have the behavior of @xmath165-theory .
the solid lines are obtained for @xmath164 and for @xmath168.,title="fig : " ] + the found solutions of metric are expressed in isotropic coordinates and often for spherically symmetric problems is conveniently rewritten in standard coordinates ( the usual form in which we write the schwarzschild solution ) . here
the relativistic invariant of metric ( [ metric_tensor_ppn ] ) is @xmath169dt^2-\biggl[1-r_g\psi(\mathbf{x})\biggr]|d\mathbf{x}|^2+r_g\lambda(\mathbf{x})\frac{(\mathbf{x}\cdot d\mathbf{x})^2}{|\mathbf{x}|^2}\end{aligned}\ ] ] from spherically symmetric form of ( [ me0 ] ) it is convenient to replace the position @xmath170 with spherical polar coordinates @xmath171 defined as usual by @xmath172 the proper time interval ( [ me0 ] ) then becomes @xmath173dt^2- \biggl[1-r_g\biggl(\psi(r)+\lambda(r)\biggr)\biggr]dr^2-\biggl[1-r_g\psi(r)\biggr]r^2d\omega\end{aligned}\ ] ] where @xmath174 is the solid angle . to get the metric in the standard form it needs to impose a radial coordinate transformation @xmath175r^2\,=\,\tilde{r}^2\end{aligned}\ ] ]
and we have a new set of coordinates @xmath176 .
the metric ( [ me1 ] ) becomes @xmath177dt^2- \biggl[1-r_g\biggl(\tilde{\psi}(\tilde{r})+\tilde{\lambda}(\tilde{r})\biggr)\biggr]\biggr(\frac{dr}{d\tilde{r } } \biggr)^2d\tilde{r}^2-\tilde{r}^2d\omega\end{aligned}\ ] ] the expression previously desired .
the explicit expression of ( [ metricsolvedstandar ] ) is not displayed because the equation ( [ condit_transf ] ) can not be solved algebraically . however , this technical problem is overcome with the help of numerical methods when we are interesting to test experimentally the theory . in a further work that is taking place , we are considering the main experiments tested for gr and we want retest them with respect to the results obtained .
in this paper , we have generally reformulated the newtonian limit of @xmath0-gravity by applying the green function method . moreover
the post - newtonian limit has been studied in the harmonic gauge condition and the relative spatial behaviors ( yukawa - like and oscillating ) of gravitational potential in the matter and in the vacuum have been shown .
the taylor expansion of a generic @xmath2 has been considered obtaining general solutions in term of the derivatives up to third degree when an uniform mass ball - like source is considered .
all metric potentials , however , depend strictly on the coupling parameters appearing indirectly in the lagrangian of the theory ( the derivatives of @xmath2 in @xmath38 are arbitrary constants ) .
a detailed discussion has been developed for systems presenting spherical symmetry . in this case , the role of corrections to the newtonian potential is clearly evident .
this means that one of the effects introducing a generic function of scalar curvature is to select a characteristic scale length which could have physical interests .
furthermore , it has been shown that the birkhoff theorem is not a general result for @xmath0-gravity .
this is a fundamental difference between gr and fourth order gravity . while in gr a spherically symmetric solution is , in any case , stationary and static , here time - dependent evolution can be achieved depending on the order of perturbations .
hypothesizing a nonlinear lagrangian we obtained a gravitational attraction stronger than in gr .
the hypothesis of dark matter is needed in gr to have more gravitational attraction . here without hypothesis of alternative matter and without modifying the gravitational constant we have qualitatively the same outcome .
this occurrence could be particularly useful to solve the problem of missing matter in large astrophysical systems like galaxies and clusters of galaxies .
in fact dark matter could be nothing else but the effects that gr , experimentally tested only up to solar system scales , does not work at extragalactic scales and then it has to be corrected .
then it is worth pointing out that @xmath0-gravity seem good candidate to solve several shortcomings of modern astrophysics and cosmology .
taking into account the results presented , it is clear that only gr presents directly the newtonian potential in the weak field limit while corrections appear as soon as the theory is non - linear in the ricci scalar . in forthcoming researches
, there is the intention to confront such solutions with experimental data in order to see if large self - gravitating systems could be modelled by them and if the experimental test of gr in the solar system are compatible with @xmath0-gravity .
the author would like to thank prof .
salvatore capozziello for his useful discussions and suggestions .
j. * 116 * , 1009 ( 1998 ) + a.g .
. journ . * 607 * , 665 ( 2004 ) + s. perlmutter _
j. * 517 * , 565 ( 1999 ) + s. perlmutter _
astrophys . * 447 * , 31 ( 2006 ) g. magnano , m. ferraris , m. francaviglia , gen .
* 19 * , 465 ( 1987 ) + g. allemandi , a. borowiec , m. francaviglia , phys .
d * 70 * , 103503 ( 2004 ) + m. amarzguioui , o. elgaroy , d.f .
mota , t. multamaki , astron . and astrophys * 454 * , 707 ( 2006 )
|
a general analytic procedure is developed for the post - newtonian limit of @xmath0-gravity with metric approach in the jordan frame by using the harmonic gauge condition . in a pure perturbative framework and by using the green function method a general scheme of solutions up to @xmath1 order is shown .
considering the taylor expansion of a generic function @xmath2 it is possible to parameterize the solutions by derivatives of @xmath2 . at newtonian order , @xmath3 ,
all more important topics about the gauss and birkhoff theorem are discussed .
the corrections to `` standard '' gravitational potential ( @xmath4-component of metric tensor ) generated by an extended uniform mass ball - like source are calculated up to @xmath1 order .
the corrections , yukawa and oscillating - like , are found inside and outside the mass distribution . at last
when the limit @xmath5 is considered the @xmath0-gravity converges in general relativity at level of lagrangian , field equations and their solutions .
|
shell models of turbulence @xcite serve a useful purpose in studying the statistical properties of turbulent fields due to their relative ease of simulation .
in particular , shell models allowed accurate direct numerical calculation of the scaling exponents of their associated structure functions , including convincing evidence for their universality @xcite . in contrast , for the navier - stokes equations that model actual fluid turbulence simulations are very much harder , and in addition one still does not know whether these equations in 3-dimensions are mathematically globally well posed .
this problem does not exist in shell models @xcite , adding to their numerical attractiveness a possibility to prove various properties and results rigorously @xcite .
consider for example the sabra shell model @xcite which , like other shell models of turbulence , is a truncated description of the dynamics of fourier modes , preserving some of the structure and conservation laws of the navier - stokes equations : @xmath0+f_n \ . \label{sabra}\end{aligned}\ ] ] here @xmath1 , with @xmath2 and the boundary conditions @xmath3 , are the velocity modes restricted to ` wavevectors ' @xmath4 with @xmath5 determined by the inverse outer scale of turbulence .
the model contains one additional parameter , @xmath6 , and it conserves two quadratic invariants ( when the force and the dissipation term are absent ) for all values of @xmath6 .
the first is the total energy @xmath7 and the second is @xmath8 , where @xmath9 .
the scaling exponents are properties of the structure functions . to define the structure function we introduce an average over time according to @xmath10 in practice , however , we take @xmath11 for some @xmath12 large enough , but finite .
our rigorous results also refer to this definition of an average over time . for values of the viscosity @xmath13 small enough , and for a sufficiently large amplitude of the random force @xmath14 there exists a large range of values @xmath15 where numerical simulations show that structure functions follow a power - law behavior .
the low - order structure functions and the associated scaling exponents are @xmath16 the values of the scaling exponents were determined accurately by direct numerical simulations . besides @xmath17 which is exactly unity @xcite , all the other exponents @xmath18 appear anomalous , differing from @xmath19 .
numerical evidence is that the scaling exponents are also universal , i.e. they are independent of the forcing @xmath14 as long as the latter is restricted to small @xmath20 @xcite .
it was shown that for @xmath21 the leading scaling exponents are determined by the cascade of the energy invariant from large to small scales .
for @xmath22 the second invariant @xmath8 becomes important in determining the leading scaling exponents of the structure functions of @xmath1 . in the bulk of this paper we consider the situation with @xmath21 , but return to the other case in sect .
[ 2d ] , in order to clarify the role of invariants in determining the scaling properties . in a recent publication further insight to the anomaly of the exponents of the nonlinear problem ( for the field @xmath1 )
was sought by relating them to the scaling exponents of a _ linear _ model for a field @xmath23 .
the linear model was constructed such that its scaling exponents should be the same as those of the nonlinear problem .
the equations for this field are constructed under the following requirements : ( i ) the structure of the equations is obtained by linearizing the nonlinear problem and retaining only such terms that conserve the energy ; ( ii ) the resulting equation is identical with the sabra model when @xmath24 ; ( iii ) the energy is the only quadratic invariant for the passive field in the absence of forcing and dissipation .
these requirements lead to the following linear model : @xmath25 where the advection term is defined as @xmath26\nonumber\\ & & + k_{n}[(1 - 2\delta)u_{n-1}^*w_{n+1}-(1+\delta)u_{n+1}w_{n-1}^*]\nonumber\\ & & + k_{n-1}[(2-\delta)u_{n-1}w_{n-2}+(1 - 2\delta)u_{n-2}w_{n-1 } ] \ , \label{bilinear}\end{aligned}\ ] ] together with @xmath27 .
observe that when @xmath24 this model reproduces the sabra model , and also that the total energy is conserved because @xmath28 = 0 $ ] .
the second quadratic invariant is not conserved by the linear model ( [ linear ] ) .
finally , both models have the same ` phase symmetry ' in the sense that the phase transformations @xmath29 and @xmath30 leave the equations invariant iff @xmath31 , @xmath32 .
this identical phase relationship guarantees that the non - vanishing correlation functions of both models have precisely the same forms .
thus for example the only second and third correlation functions in both models are those written explicitly in eqs .
( [ s2 ] ) and ( [ s3 ] ) .
the advantage of the linear model is that correlation functions are advanced in time by a linear propagator @xcite .
the linear model possesses statistically preserved structures " ( sps ) which are evident in the decaying problem eq .
( [ lin ] ) with @xmath33 .
these are _ left _ eigenfunctions of eigenvalue 1 of the linear propagators for each order ( decaying ) correlation function @xcite .
for example for the second order correlation function denote the propagator @xmath34 ; this operator propagates any initial condition @xmath35 ( with average over initial conditions , independent of the realizations of the advecting field @xmath1 ) to the decaying correlation function ( with average over realizations of the advecting field @xmath1 ) @xmath36 the second order sps , @xmath37 , is the left eigenfunction with eigenvalue 1 , @xmath38 note that @xmath37 is time independent even though the operator @xmath34 is time dependent .
each order correlation function is associated with another propagator @xmath39 and each of those has an sps , i.e. a _ left _
eigenfunction @xmath40 of eigenvalue 1 .
these non - decaying eigenfunctions scale with @xmath15 , @xmath41 , and the values of the exponents @xmath42 are anomalous .
finally , it was shown that these sps are also the leading scaling contributions to the structure functions of the _ forced _ problem ( [ lin ] ) @xcite .
thus * the scaling exponents of the linear problem are independent of the forcing @xmath14 * , since they are determined by the sps of the decaying problem . to connect the linear model to the nonlinear problem one considers the system of two coupled equations @xmath43 with @xmath44 being a real parameter and @xmath14 and @xmath45 being different realizations of the same random force .
observe that for any @xmath46 , the two equations in ( [ system ] ) exchange roles under the change @xmath47 .
thus if the scaling exponents @xmath42 and @xmath18 of the two fields exist ( i.e. a true scaling range exists ) , they must be the same for all @xmath46 .
for @xmath48 we recover the equations for the nonlinear and the linear models , eqs .
( [ sabra ] ) and ( [ lin ] ) .
in @xcite it was _ assumed _ that the scaling exponents of either field exhibits no jump in the limit @xmath49 .
indeed , ref .
@xcite also presented strong evidence for the validity of this assumption ( see fig .
[ lambdato0 ] ) , but no mathematical proof was provided .
the aim of this paper is to close this gap .
the main result of our paper states that the solutions of the system ( [ system ] ) exist globally in time and depend continuously on the parameter @xmath44 , including the limit @xmath50 .
in particular we will show that _ if _ the structure function of @xmath1 and @xmath23 exhibit the same scaling exponents for any @xmath46 , they will also have the same scaling exponents in the limit @xmath49 ( including @xmath48 ) .
we would like to stress here that our rigorous results are valid for the structure functions , calculated over large , but finite , fixed interval of time , which is consistent with the definition ( [ average ] ) of the long time average .
in addition , we would like to note that the numerical results correspond to the equations with the stochastic implementation of the forcing , while our rigorous proofs deal with the deterministic force that depend on time .
the statement and the proof of the main theorem will be given in section [ section2 ] .
the proof is based on the results on the global existence and uniqueness of solutions of eq .
( [ sabra ] ) , obtained previously in @xcite . in the following section we present the necessary mathematical definitions and formulate the essential statement from @xcite .
following the classical treatment of the navier - stokes and euler equations we re - write the sabra shell model for an infinite vector @xmath51 @xmath52 together with the initial conditions @xmath53 . introduce a hilbert space @xmath54 which is the space of infinite vectors equipped with the scalar product @xmath55 and the corresponding norm @xmath56 defined as @xmath57 for every @xmath58 .
the space @xmath54 is a space of all the velocity vectors having finite energy .
the linear operator @xmath59 , with a domain @xmath60 dense in @xmath54 , is a positive definite , diagonal operator defined through its action on the elements @xmath61 by @xmath62 where the eigenvalues @xmath63 satisfy @xmath64 . using the fact that @xmath59 is a positive definite operator , we can define the powers @xmath65 of @xmath59 for every @xmath66 @xmath67 the space @xmath68 is the domain of the operator @xmath69 and we denote @xmath70 which are hilbert spaces equipped with the scalar product @xmath71 and the norm @xmath72 , for every @xmath73 .
since @xmath74 contains velocity vectors with @xmath75 `` derivatives '' , @xmath76 the case of @xmath77 is of a special interest for us .
we denote @xmath78 a hilbert space equipped with a scalar product and a corresponding norm @xmath79 for every @xmath80 .
in addition the following relation holds @xmath81 in what follows we will need the interpolation inequality for the spaces @xmath82 . *
lemma 1*. let @xmath83 , then for all @xmath84 and @xmath85 @xmath86 _ proof_. the lemma follows by a simple application of hlder inequality .
the bilinear operator @xmath87 is defined above , cf .
( [ bilinear ] ) together with @xmath88 .
in @xcite it was shown that such a definition of the bilinear operator makes @xmath89 and element of @xmath54 whenever @xmath90 and @xmath91 . for any @xmath92 and @xmath93 one proves @xcite
@xmath94 for some positive constant @xmath95 .
in addition the conservation law is written is the present notation as @xmath96 all our rigorous results are valid for the deterministic forcing @xmath97 .
therefore , in order to account for the stochastic implementation of the forcing term we will assume that @xmath97 depends on time , but always stays bounded in the @xmath54-norm . more precisely , define the space @xmath98 , for some @xmath99 , as a space of functions of the time interval @xmath100 with values in @xmath54 and the norm defined as latexmath:[\[\label{inftynorm } what follows , we will assume that the forcing term @xmath97 satisfies @xmath102 . in @xcite
( [ sabra ] ) was studied , and the relevant results can be formulated as the following theorem : * theorem ( clt 06 ) * : the solution of eq .
( [ sabra ] ) exists globally in time for any initial condition @xmath103 in h , and for any @xmath97 in @xmath104 .
moreover , the solutions are unique and the energy of the solution @xmath105 is bounded for all times : @xmath106 where the _ a - priori _ constant @xmath107 depends on all the parameters of the equations , on the forcing and on the initial conditions .
if in addition we assume that the forcing @xmath108 acts on the finite number of shells , namely , if there exists @xmath109 , such that @xmath110 , for all @xmath111 , then the solution @xmath112 has an exponentially decaying spectrum @xmath113 as a function of @xmath15 , and in particular @xmath114 for any @xmath83 and the _ a - priori _ constants @xmath115 depend on all the parameters of the equations , on the forcing and on the @xmath104-norm of the initial conditions ( see definition ( [ inftynorm ] ) ) .
the main statement of this paper is that the system of equations ( [ system ] ) is globally well posed for all real @xmath44 , and that the solutions depend continuously on the parameter @xmath44 . in particular , as @xmath49 the solution of the system converge uniformly on any finite interval of time to the solution of the system with @xmath48 .
this statement is formulated as follows : * theorem 1*. let @xmath116 be given in @xmath54 , and the forces @xmath117 in @xmath118 , h)$ ] . denote by @xmath119 the solutions of the coupled system ( [ system ] )
. then the following holds 1 .
[ thm_main_2 ] the solution of the system ( [ system ] ) exists globally in time for every @xmath44 .
moreover , energy of the solutions @xmath120 and @xmath121 are bounded for all times , where the bounds depend on all the parameters of the equations , @xmath44 , the forcing and the initial conditions .
[ thm_main_4 ] for any @xmath122 , @xmath123 and @xmath124 where @xmath125 are the solutions of ( [ system ] ) with @xmath126 .
[ thm_main_3 ] if in addition we assume that the forces @xmath127 act only on the finite number of shells then for any @xmath128 , and @xmath83 , @xmath129 and @xmath130 where @xmath125 are the solutions of ( [ system ] ) with @xmath126 . _ proof _ : part 1 . of this theorem
follows from defining a new variable @xmath131 . this variable satisfies eq .
( [ sabra ] ) with a forcing @xmath132 .
accordingly theorem clt06 provides the proof that @xmath133 exists globally in time for every @xmath44 .
next observe that the system of equations ( [ system ] ) can be rewritten in the form @xmath134 this form of writing shows that the the fields @xmath135 and @xmath136 satisfy _ linear _ diffusion advection equations advected by a smooth field @xmath137 .
accordingly both fields remain smooth for all time and all @xmath44 .
in addition , using the relation ( [ bound ] ) one is able to derive the bounds for the energies of the solutions . to prove part 2 . of the theorem
we need first + * proposition 1 * : denote by @xmath138 the solution of the sabra model ( [ sabra ] ) with initial data @xmath139 in @xmath54 .
that is also the solution of the first of equations ( [ system ] ) when @xmath126 .
then @xmath140 moreover , if @xmath141 and @xmath142 act only on finite number of shells , then for any @xmath83 we have @xmath143 _ proof _ : let us denote by @xmath144 .
then @xmath145 satisfies the equation @xmath146 take now the inner product in @xmath54 of the above equation with @xmath145 . computing the real part and using eq .
( [ conserve ] ) we find @xmath147 using the cauchy - schwarz inequality and the relation ( [ ineq ] ) we get @xmath148 applying the young s inequality ( @xmath149 ) twice and using the inequality ( [ norms ] ) , we have @xmath150 ) holds for any @xmath151 and the fact that @xmath102 ( see definition ( [ inftynorm ] ) ) , we obtain @xmath152 by gronwall s inequality we conclude @xmath153 therefore , for any @xmath122 , @xmath154 and the first statement of the proposition follows . the second statement of the proposition follows from the boundedness of @xmath155 and @xmath138 in higher order norms after a short transient period of times ,
say @xmath156 , ( provided that the forces @xmath141 and @xmath142 act on the finite number of shells as required by theorem clt06 ) and the interpolation inequality ( [ inter ] ) . finally , we are ready to finish the proof of the main theorem .
let us fix @xmath157 , and show that @xmath158 denote @xmath159 .
the function @xmath160 satisfies the equation @xmath161 we rewrite it in the form @xmath162 and as before , taking the inner product in @xmath54 with @xmath160 , computing the real part and using eq .
( [ conserve ] ) , we obtain @xmath163 applying , subsequently the young s inequality and the inequality ( [ ineq ] ) , we get @xmath164 and integrating over @xmath165 we conclude @xmath166 where we used the fact that @xmath167 and @xmath168 , for some constant @xmath169 , as was stated in the part 1 of the main theorem .
it follows from proposition 1 that @xmath170 , as @xmath49 .
hence we may conclude that @xmath171 by virtue of the interpolation inequality ( [ inter ] ) one can follow steps as in the proof of proposition @xmath172 to show that the higher order norms of @xmath160 also converge to @xmath173 uniformly in the time interval @xmath174 , as @xmath50 .
to finish the proof , on can observe that @xmath175 and from the fact that @xmath176 is globally bounded in time it follows from proposition 1 that @xmath177 using the form of eqs .
( [ system1 ] ) we conclude that for @xmath14 and @xmath45 being different realizations of the same random force , whenever the structure functions of @xmath1 and @xmath23 exhibit the same scaling exponents for all finite @xmath44 , they must also exhibit the same scaling exponents for @xmath48 .
in this section we turn our attention to situations when the nonlinear and the linear models can not exhibit the same scaling exponents . the theorem as stated and proven still holds , but as we shall see this is a situation for which the two fields @xmath1 and @xmath23 can not exhibit the same scaling exponents for all @xmath46 .
a case in point is the same nonlinear sabra model with @xmath22 . in fig .
[ 2d - lam=0 ] we show the second order structure functions @xmath178 and @xmath179 obtained from simulating eqs .
( [ system ] ) for @xmath48 and @xmath180 .
as expected , the scaling of @xmath23 is influenced by the cascade of energy , whereas that of @xmath1 by the cascade of the second invariant . as a result
the scaling exponents are distinctly different .
the same system of equations for @xmath181 is symmetric in @xmath23 and @xmath1 .
indeed , in fig . [ 2d - lam=1 ]
we show the result of simulations for @xmath181 , for which the second order structure function of the two fields is identical .
now however we can not expect that this identity will persist for @xmath49 . in fig ,
[ 2d - lamsmall ] we show the results of simulations for the same system of equations for @xmath182 and @xmath183 . to understand the results of the simulations we note that when @xmath46 the second invariant of the equation for @xmath1 is destroyed , and one could think that the scaling of @xmath1 should be dominated by the energy invariant .
this is certainly true for @xmath181 .
but now when @xmath44 decreases towards zero , we should reconsider the system of equations ( [ system ] ) by renaming @xmath184 .
substituting this re - named variables into eqs .
( [ system ] ) and re - arranging , we read @xmath185 thus the net result of the transformation is that the equations for @xmath1 and @xmath186 are the same , but the forcing of @xmath23 becomes weaker and weaker as @xmath49 . accordingly , while the second invariant is still destroyed as a true invariant for any value of @xmath44 , for small @xmath44 the strength of the term @xmath187 diminishes , allowing a cross - over behavior in the scaling of @xmath1 , precisely as we see in fig .
[ 2d - lamsmall ] .
ip acknowledges useful discussions with l. angheluta , especially concerning the material presented in sec .
this work has been supported in part by the us - israel binational science foundation .
the work of e.s.t . was also supported in part by the nsf grant no .
dms0504619 , and by the isf grant no .
|
in a recent paper it was proposed that for some * nonlinear * shell models of turbulence one can construct a * linear * advection model for an auxiliary field such that the scaling exponents of all the structure functions of the linear and nonlinear fields coincide .
the argument depended on an assumption of continuity of the solutions as a function of a parameter .
the aim of this paper is to provide a rigorous proof for the validity of the assumption .
in addition we clarify here when the swap of a nonlinear model by a linear one will _ not _ work ,
|
hypernuclei are bound systems of neutrons , protons and one or more strange baryons , such as the @xmath0 or @xmath1 hyperons .
understanding the behavior of hypernuclei ( how they are produced , their spectroscopy and decay mechanisms ) has been the subject of intense investigations during the last decades , see e.g. , refs.@xcite .
one of the main goals of such studies is to explore how the presence of the new degree of freedom ( strangeness ) alters and broadens the knowledge achieved from conventional nuclear physics .
several features of the @xmath0 single - particle properties in the nucleus , being essentially different from those of the nucleon , have clearly emerged from these efforts .
it is well accepted nowadays that the depth of the @xmath0-nucleus potential is around @xmath2 mev , which is 20 mev less attractive than the corresponding nucleon - nucleus one .
the spin - orbit splittings of single particle levels in @xmath0 hypernuclei were found to be much smaller than their nucleonic counterparts@xcite , typically more than one order of magnitude .
moreover , the @xmath0 , contrary to the nucleon , maintains its single - particle character even for states well below the fermi surface @xcite indicating a weaker interaction with other nucleons .
studies of the mesonic weak decay of light @xmath0 hypernuclei @xcite have shown that the data @xcite clearly favour @xmath0-nucleus potentials which show a repulsion at short distances .
this seems also to be a characteristic of the @xmath1-nucleus potential for light @xmath1-hypernuclei @xcite , which reproduces the recently measured bound @xmath3 state in @xmath4he with the in - flight @xmath5he@xmath6 reaction @xcite .
this experiment confirms , with new and better statistics , the earlier results from the @xmath5he(stopped @xmath7 reaction @xcite . attempts to derive the hyperon properties in a nucleus have followed several approaches .
a @xmath0-nucleus potential of woods - saxon type reproduces reasonably well the measured @xmath0 single - particle energies of medium to heavy hypernuclei @xcite .
non localities and density dependent effects , included in non - relativistic hartree - fock calculations using skyrme hyperon - nucleon ( @xmath8 ) interactions @xcite , improve the overall fit to the single - particle binding energies .
the properties of hypernuclei have also been studied in a relativistic framework , such as dirac phenomenology @xcite or relativistic mean field theory @xcite .
microscopic calculations , which aim at relating the hypernuclear observables to the bare @xmath8 interaction , are also available .
this approach is especially interesting because it can be used to put further constraints on the @xmath8 interactions , which are not completely determined by the limited amount of scattering data due , essentially , to the experimental difficulties associated with the short lifetime of hyperons and low intensity beam fluxes .
microscopic hypernuclear structure calculations are performed with an effective @xmath8 interaction ( @xmath9-matrix ) obtained from the bare @xmath8 potential through a bethe - goldstone equation .
the comparison with data may therefore further help in constraining the @xmath8 potentials . in several microscopic calculations
a gaussian parametrizations of the @xmath9-matrix calculated in nuclear matter at an average density @xcite was employed .
furthermore , a @xmath9-matrix obtained directly in finite nuclei was used to study the single - particle energy levels of various hypernuclei @xcite or as an effective interaction in a calculation of the @xmath10o spectrum @xcite .
along similar lines , the authors of ref .
@xcite derived microscopically the @xmath0 self energy in @xmath10o , starting from realistic hyperon - nucleon interactions .
the starting point in the latter calculations is a nuclear matter @xmath9-matrix at a fixed energy and density , which is used to calculate the self
energy for the finite nucleus including corrections up to second order .
the obtained self
energy for the @xmath0 is non - local and depends on the energy of the hyperon . solving the schrdinger equation with this self energy it is possible to determine the single - particle energies and wave functions of the bound hyperon .
the approach also provides automatically the real and imaginary part of the hyperon optical potential at positive energies and , therefore , allows to study the hyperon - nucleus scattering properties .
the method was first employed to study the nucleon and @xmath11 properties in nuclei @xcite and later applied to calculate the @xmath12-wave non - local @xmath0 self energy in @xmath10o , from which the single - particle @xmath0 binding energy was obtained @xcite .
the aim of the present work is to extend the calculations of ref .
@xcite in order to derive @xmath12 and @xmath13-wave @xmath0 single - particle energies for a variety of @xmath0 hypernuclei , from @xmath10o to @xmath14pb .
the @xmath1 single - particle energies obtained for different potentials are also discussed .
the @xmath8 potentials employed are the nijmegen soft - core @xcite and the jlich @xcite interactions .
eventual differences in single - particle binding energies may therefore help in further constraining the @xmath8 interactions . in sect .
[ formal ] , we present the formalism which consists of obtaining , first , the nuclear matter @xmath9-matrix at fixed density and starting energy parameter in the center of mass frame and , next , performing a second order calculation to derive the finite nucleus @xmath9-matrix from which one can obtain the hyperon self energy for the different single - particle orbits .
this self energy can in turn be used to define a single - particle potential in a schrdinger equation in order to obtain the corresponding single - particle binding energies .
nuclear matter results and a discussion on the convergence of our method is presented in sect .
[ results1 ] .
an important point raised in the latter section is that the single - particle hyperon energies calculated at first order with a nuclear matter @xmath9-matrix depend quite strongly on the fixed starting energy and density employed .
the @xmath0 single - particle energies for a variety of hypernuclei are shown in sect .
[ results2 ] , where local equivalent woods - saxon potentials to represent our non - local self energy are also derived .
finally , our conclusions are presented in sect .
[ conclusions ] .
in this section we present the formalism to obtain the hyperon single - particle energies in finite hypernuclei using an effective interaction derived microscopically from realistic @xmath8 interactions , for which we take the nijmegen soft core @xcite and jlich@xcite potentials .
although the formalism was already described in ref .
@xcite , part of it will be repeated here in order to set up the notation used later in the description of the results . in the first place
, we solve the @xmath9-matrix in nuclear matter at a fixed density , center of mass momentum and energy .
next , we transform this @xmath9-matrix into an effective interaction for the finite hypernucleus from which we can obtain the hyperon self energy .
finally , we use this self energy in a schrdinger equation to derive the single - particle energies and the corresponding wave functions of the bound states .
the nuclear matter @xmath8 @xmath9-matrix is solved in momentum space and the two - particle @xmath8 states are defined in terms of relative and the center - of - mass momenta , * k * and * k * , given by @xmath15 where @xmath16 and @xmath17 are the nucleon and hyperon momenta , respectively . using an angle - averaged pauli operator ,
we perform a partial wave decomposition of the bethe - goldstone equation , which , in terms of the quantum numbers of the relative and center - of - mass motion ( rcm ) , is written as @xmath18 where @xmath19 is the nuclear matter pauli operator , @xmath20 is the @xmath8 potential and @xmath21 is the nuclear matter starting energy which corresponds to the sum of non - relativistic single - particle energies of the interacting nucleon and hyperon . note that kinetic energies are used in the intermediate @xmath22 states and @xmath23 accounts for the mass difference of the initial and intermediate hyperon .
the variables @xmath24 , @xmath25 , @xmath26 and @xmath27 , @xmath28 , @xmath29 denote relative momenta and angular momenta , respectively , while @xmath30 and @xmath31 are the quantum numbers of the center - of - mass motion .
the total angular momentum , spin and isospin projection of the @xmath8 pair are denoted by @xmath32 , @xmath33 and @xmath34 , respectively . in the brueckner - hartree - fock approach the hyperon single - particle potential @xmath35
is obtained self - consistently by the following sum of diagonal @xmath9-matrix elements : @xmath36 where @xmath37 is the single - particle energy of the nucleon ( hyperon ) . using the partial wave decomposition of the @xmath9-matrix ,
the single - particle potential @xmath38 can be rewritten as @xmath39 where an average over the hyperon spin and isospin ( @xmath40 ) has been performed and the weak center - of - mass dependence of the @xmath9-matrix has been neglected . in eq .
( [ eq : uy ] ) , @xmath24 is the relative momentum of the @xmath8 pair , @xmath41 , @xmath42 is given by @xmath43 and the weight function @xmath44 by @xmath45 ^ 2}{4\xi_{y}(1+\xi_{y})k_{y}k } & \mbox{otherwise } \end{array } \right . \nonumber\ ] ] in the discussion ,
we will associate the binding energy of the hyperon to its single - particle energy : @xmath46 the self energy of the @xmath0 or @xmath1 hyperon in a finite hypernucleus can be obtained , in the hartree - fock scheme , using a finite nucleus @xmath9-matrix as an effective @xmath8 interaction . however , our @xmath9-matrix has been obtained in nuclear matter and , in particular , the intermediate propagator ( that involves the pauli operator and energy denominator ) is very different than the corresponding one in a finite hypernucleus .
one can , nevertheless , find the appropriate finite nucleus @xmath9-matrix , @xmath47 , by relating it to the nuclear matter @xmath9-matrix through the following integral equation written in schematic form : @xmath48 g_{fn } \nonumber \\ & = & g + g \left [ \left ( \frac{q}{e } \right)_{fn } -\left ( \frac{q}{e } \right)_{nm } \right ] g \nonumber \\ & & + g \left [ \left ( \frac{q}{e } \right)_{fn } -\left ( \frac{q}{e } \right)_{nm } \right ] g \left [ \left ( \frac{q}{e } \right)_{fn } -\left ( \frac{q}{e } \right)_{nm } \right ] g + \dots \ , \label{eq : gexpan}\end{aligned}\ ] ] which involves the nuclear matter @xmath9-matrix ( labelled @xmath9 throughout the text ) and the difference between the finite nucleus and nuclear matter propagators .
the latter account for the relevant intermediate states . in the practical calculations we will approximate the expansion up to the second order in the nuclear matter @xmath9-matrix @xmath49 g \ .
\label{eq : gexpan2}\ ] ] therefore , in the evaluation of the hyperon self energy we take into account the diagrams displayed in fig .
[ fig : fig1 ] , where the wiggly interaction lines represent the nuclear matter @xmath9-matrix .
diagram [ fig : fig1](a ) represents the first term on the right hand side of eq .
( [ eq : gexpan2 ] ) which , by analogy to the nuclear case , will be refered to as the hartree - fock contribution . diagram [ fig : fig1](b ) stands for the second order correction , where the intermediate propagator has to be viewed as the difference of propagators appearing in eq .
( [ eq : gexpan2 ] ) .
we will consider the incoming ( outcoming ) hyperon as a plane wave and the nucleon hole states as harmonic oscillator ones , so the two - body wave function will be a mixed representation of both single - particle states given by @xmath50 where @xmath51 and @xmath52 are the quantum numbers of the nucleon hole state and the hyperon state , respectively .
further , @xmath53 is the oscillator parameter appropiate to describe the single - particle wave functions of the bound nucleons in the nuclear core .
it is defined as @xmath54 with @xmath55 chosen as the following function of the mass number @xmath56 typical matrix elements needed in the calculation are @xmath57 for the hartree - fock diagram of fig .
[ fig : fig1](a ) , or @xmath58 appearing in the second - order diagram of fig .
[ fig : fig1](b ) . using eq .
( [ eq : trans1 ] ) , the mixed representation states appearing in eqs .
( [ eq : g1 ] ) and ( [ eq : g2 ] ) can be expressed in terms of momentum and angular momentum variables in the laboratory frame .
then , with appropriate transformation coefficients
@xcite , one can express the two - body states with laboratory coordinates in terms of the variables in the rcm system used in the solution of the @xmath9-matrix and @xmath24 and @xmath59 and @xmath27 . with the notation @xmath60 or @xmath59 we will refer to the quantum numbers of the single - particle state , whereas @xmath27 or @xmath24 without subscripts refer to the coordinates of the relative motion . ] @xmath61 where @xmath62 and @xmath63 are the transformation coefficients from the rcm system to the laboratory system .
we can construct now the expressions for the various diagrams considered in this work .
the first order term of fig .
[ fig : fig1](a ) yields a real and energy - independent contribution to the self energy given by @xmath64 where @xmath65(@xmath66)@xmath67@xmath68@xmath69 are the quantum numbers of the incoming(outcoming ) hyperon . + the computation of the contribution coming from the two - particle - one - hole ( @xmath70 ) diagram of fig .
[ fig : fig1](b ) requires a little more work .
first , we evaluate the imaginary part of the second term in eq .
( [ eq : gexpan2 ] ) .
this term has an an explicit energy dependence .
it reads @xmath71 + where @xmath72 is the energy of the hyperon measured with respect to the hyperon rest mass .
the single - hole energies @xmath73 have been taken equal to the experimental single - particle energies in most of the nuclei studied ( e.g. , @xmath74c , @xmath75o , @xmath76ca ) and have been calculated from a woods - saxon potential with spin - orbit and coulomb terms appropriately fitted in other cases ( e.g. , @xmath77zr and @xmath78pb ) .
the quantities @xmath79 are the quantum numbers of the intermediate @xmath80 states .
next , we obtain the real part through a dispersion relation @xmath81 where @xmath82 means a principal value integral .
finally , we must subtract the @xmath70 correction term coming from the nuclear matter intermediate propagator ( third term in eq .
( [ eq : gexpan2 ] ) ) .
it reads @xmath83 where @xmath19 is the nuclear matter pauli operator and @xmath21 is the nuclear matter starting energy .
this term only contributes to the real part of the hyperon self energy and avoids the double counting over intermediate @xmath22 states contained already in the nuclear matter @xmath9-matrix of the hartree - fock contribution @xmath84 . in summary ,
the self energy of the hyperon reads @xmath85 with the real part given by @xmath86 and the imaginary part by @xmath87 the self energy can then be inserted as a single - particle potential in a schrdinger equation in order to investigate bound and scattering states of a hyperon in a finite nucleus .
the different approximations to the self energy , i.e. , whether we include the @xmath70 contribution or not , result in different single - particle hamiltonians .
we solve the schrdinger equation by diagonalizing the corresponding single - particle hamiltonian in a complete basis within a spherical box of radius @xmath88 following the procedure outlined in @xcite .
this method is especially suitable fon non - local potentials defined either in coordinate or in momentum space @xcite .
in this section we present and discuss results for @xmath0 and @xmath1 hypernuclei using two realistic interactions : nijmegen soft core @xcite and jlich b @xcite . as described before , our method provides the binding energies of the different hyperon orbits in finite hypernuclei starting from a @xmath9-matrix calculated in nuclear matter in the @xmath8 center - of - mass frame at fixed starting energy @xmath21 and fermi momentum @xmath89 . by adding the @xmath70 correction to the hartree - fock term one incorporates , up to second order in the nuclear matter @xmath9-matrix , the correct energy dependence and pauli blocking factor in the finite nucleus .
therefore , the complete calculation ( @xmath90 ) has to be viewed as a hartree - fock approach which uses a effective interaction derived microscopically with the appropriate density and energy dependence of the hypernucleus under study .
this is in contrast to previous calculations @xcite where the determination of the finite hypernucleus effective interaction from the nuclear matter @xmath9-matrix implied a sort of average over the nuclear density . in these
works several local and energy independent effective @xmath8 interactions of gaussian form ( @xmath91 ) were derived by parametrizing the corresponding nuclear matter @xmath9-matrices obtained from various @xmath8 potentials .
the parametrization of the @xmath9-matrix into a local effective interaction @xmath91 to be used in finite hypernuclei calculations required the use of an appropiate value of the fermi momentum @xmath89 .
this value was determined , for each nucleus , by averaging the corresponding nuclear density weighted by the modulus squared of the @xmath0 single - particle wave function of the single - particle level under study .
the parameters of the effective @xmath91 interaction were adjusted to reproduce the @xmath0 potential energy @xmath92 in nuclear matter at the average value of @xmath89 . with these parametrizations , @xmath0 single - particle energies and
excited hypernuclear levels in several @xmath0 hypernuclei were obtained through a shell - model calculation , with the aim of learning about the bare @xmath8 interaction .
it therefore seems appropriate to explore , using our method , how much the hyperon single - particle energy depends on the starting energy ( and density ) of the nuclear matter @xmath9-matrix used in the calculation .
this will allow us to assess how reliable energy independent effective interactions obtained from local density averages might be .
let us first show , in fig .
[ fig : binener ] , the binding energy @xmath93 of a @xmath0 ( curves on the left ) or a @xmath1 ( curves on the right ) in nuclear matter at @xmath94 @xmath95 as a function of the starting energy parameter @xmath96 , where @xmath97 mev is an average of the nucleon binding energy over the fermi sea and @xmath98 .
the long - dashed ( full ) lines are for the nijmegen soft core ( jlich b ) interaction .
an estimate of the self - consistent solution is obtained where the line @xmath96 crosses the calculated values of @xmath93 .
this is indicated by the dotted lines in the figure . in the case of the nijmegen interaction
we obtain @xmath99 mev ( @xmath100 mev ) for the @xmath0 and @xmath101 mev ( @xmath102 mev ) for the @xmath1 , whereas , in the case of the jlich interaction , @xmath103 mev for the @xmath0 ( @xmath104 mev ) and @xmath105 mev ( @xmath106 mev ) for the @xmath1 .
several features emerge from fig .
[ fig : binener ] .
first , the @xmath1 hyperon is unrealistically overbound in nuclear matter by the jlich interaction .
it is therefore necessary to readjust the parameters of this interaction if one wants to use it in shell model calculations of @xmath1 hypernuclei .
secondly , we observe that the energy dependence of @xmath107 is slightly stronger in the case of the nijmegen interaction , especially for the @xmath1 hyperon which is more sensitive to the @xmath108 coupling because the starting energy is closer to the energies of the intermediate @xmath109 states ( which propagate with the kinetic energy spectrum ) .
finally , we observe that the @xmath0 binding energy varies at most by 10 mev in a starting energy range of 80 mev , while the variation of the @xmath1 binding amounts to twice as much .
as we will see below , this has consequences in the results for finite hypernuclei . in tables
[ tab : olambda ] and [ tab : osigma ] we show the binding energy of the @xmath0 and @xmath110 , respectively , in @xmath111o . the columns denoted by @xmath112 correspond to our lowest order calculation ( see eq .
( [ eq : hf ] ) ) which uses , as effective interaction , the nuclear matter @xmath9-matrix calculated in the @xmath8 center of mass frame at fixed energy ( shown in the first column ) and density ( @xmath94 @xmath95 ) .
columns labelled @xmath113 include the @xmath70 corrections ( see eqs .
( [ eq : disprel ] ) and ( [ eq : vc ] ) ) to bring the nuclear matter @xmath9-matrix to the finite nucleus one , with the proper energy and density dependence .
we see that the lowest order results depend quite strongly on the starting energy used , especially in the case of the @xmath1 hyperon as would be expected from the nuclear matter results shown in fig .
[ fig : binener ] .
however , it is worth noticing how , no matter what starting energy is used in solving the nuclear matter @xmath9-matrix , the corrected calculation ( @xmath90 ) ends up giving practically the same result for the hyperon binding energy . particularly stable
are the results for the @xmath0 hyperon .
this weaker energy dependence , seen already in the nuclear matter results of fig .
[ fig : binener ] , is due to the fact that the energies involved in the calculation lie further away from the intermediate @xmath8 states , which propagate with kinetic energy , and therefore the strong @xmath114 coupling is less pronounced . in tables [ tab : dlambda ] and [ tab : dsigma ] we show the binding energy of the @xmath0 and @xmath110 , respectively , in @xmath111o using nuclear matter @xmath9-matrices calculated at several values of the fermi momentum and a fixed value of the starting energy ( @xmath115 for the @xmath0 and @xmath116 for the @xmath1 ) .
the lowest order calculation for the hyperon single - particle energy , shown in the second column , depends quite strongly on the value of @xmath89 .
however , one finds again that , when the @xmath70 correction is included to incorporate the proper intermediate propagator of the finite nucleus , the results nicely converge to practically the same value , no matter what was the density used in the solution of the nuclear matter @xmath9-matrix .
these results are interesting because they confirm that the finite nucleus @xmath47-matrix is already well approximated by the second order term in the expansion in terms of the nuclear matter @xmath9-matrix , see the second and third terms in eq .
( [ eq : gexpan2 ] ) .
the correction , whose size depends on the starting energy or fermi momentum used in the solution of the nuclear matter @xmath9-matrix , already leads to practically the same value for the hyperon single - particle energy .
higher order terms could only help in bringing the results closer than what they already are .
moreover , our results also show that in some cases the correction is quite appreciable , not only for the @xmath110 binding energies shown in tables [ tab : osigma ] and [ tab : dsigma ] , but also for the @xmath0 energies in the case of the nijmegen interaction . therefore ,
if the correction is taken only approximately through an averaged nuclear matter @xmath9-matrix @xcite it may not lead to the proper effective interaction in the finite nucleus one is studying .
these words of caution are particularly relevant in the case of the @xmath1 hyperon where the corrections are very large .
we note that the @xmath0 single - particle energy obtained in the case of the nijmegen soft core interaction is in excellent agreement with that obtained by halderson ( see column 3 in fig
. 7 of ref .
@xcite ) , where the @xmath9-matrix was calculated directly in the finite nucleus for various nijmegen interactions .
our method must be viewed as an alternative way of building up a finite nucleus effective interaction .
it was already shown there that the pauli corrections , which are a source of nucleus dependence , were very large for the nijmegen soft core potential .
this again supports our believe that calculations based on nuclear matter @xmath9-matrices at an average density will carry uncertainties tied to the chosen value of the fermi momentum .
in particular , the whole purpose of using these microscopic calculations to constrain the @xmath8 force can not be achieved if a different value of @xmath89 is used for each @xmath8 interaction when studying the same hypernucleus @xcite . in the studies of single - particle binding energies below
, we will refrain from a study of the @xmath1 binding energy since the results for @xmath117o give single - particle binding energies which are much too attractive .
several analysis of @xmath118 atomic data @xcite suggest a @xmath1 well depth similar to that of the @xmath0 @xcite and more recent analysis @xcite did not discard @xmath1-nucleus potentials showing an inner repulsion .
moreover , @xmath1 hypernuclear spectra from @xmath6 reactions suggest a relatively shallow @xmath1-nucleus potential @xcite .
therefore , although the amount of data is limited , there is no experimental evidence for such strongly bound @xmath1 hyperons .
this clearly points to a weakness in the present @xmath8 interactions , hinting possibly at a too strong @xmath108 coupling in the interactions .
once the method is well established and tested for the specific case of @xmath119o , it is the right moment to study the systematics of the @xmath0 binding energy through the periodic table . to this end , values of the @xmath0 single - particle binding energies obtained in what has been called @xmath112 and @xmath90 approximations
are reported in table [ tab : ener ] together with the available experimental data .
these binding energies have been calculated using the energy - independent version with parameter set b of the jlich potential @xcite .
the results for the nijmegen soft core potential @xcite have not been considered in this section because the corresponding prediction for the @xmath0 binding energy in nuclear matter is @xmath120 mev , about @xmath121 mev weaker than the prediction of the jlich model ( @xmath122 mev ) which agrees well with the extrapolated experimental values of the single - particle @xmath0 binding energies with increasing mass number .
the agreement with the experimental data is rather good . for convenience in the technicalities of the algorithm
we have always considered hypernuclei with a number of nucleons closing a subshell plus a @xmath0 .
unfortunately , experimental data for those nuclei do not always exist and , as indicated in the table [ tab : ener ] , we have taken the closest representative nucleus for which the experimental information is available .
nevertheless , the differences between the calculated and the experimental values should not be associated to this fact but to the approximations used in the calculation or to the potential itself . for the density and starting energy used to calculate the @xmath0-nucleon @xmath9-matrix in nuclear matter , which has been used as effective interaction in our finite nucleus calculation , it turns out that the @xmath70 correction is always attractive . as discussed in the previous section ,
if we had used other starting values for the density or the starting energy we would have ended up with different @xmath70 corrections but with the same value for @xmath90 . in agreement with the experimental information ,
the difference between the @xmath123 and @xmath124 @xmath0 single - particle binding energies associated to these partial waves is very small .
note that the @xmath124 energy is lower than the @xmath123 .
this is a characteristic of the jlich interaction which yeilds too much attraction in the @xmath125 partial wave , as noted already in @xcite where the @xmath126 and @xmath127 states of @xmath128he were calculated and showed to appear in reverse order from the experimental values .
the calculated @xmath0 single - particle energies for @xmath129pb appear clearly overbound with respect to the experimental data .
this is due to the fact that the distortion of the plane wave associated with the nucleon in the intermediate state of the @xmath70 diagram of fig .
[ fig : fig1](b ) , necessary to ensure its orthogonalization to the nucleon hole states , has been taken only approximately .
the orthogonalization procedure is described in ref .
@xcite and has been optimized for the case of @xmath119o .
actually , this feature is already sizable for @xmath130zr and in the case of @xmath129pb leads to a result which is more bound than a @xmath0 in nuclear matter .
traditionally , the systematics of @xmath0 single - particle binding energies has been studied by using a phenomenological woods - saxon potential @xmath131 } } , \label{eq : ws}\ ] ] with a fixed diffusivity @xmath132 and depth @xmath133 , and a radius @xmath134 .
a good parameterization of the experimental data is obtained with @xmath135 mev , @xmath136 fm and @xmath137 fm @xcite .
a more refined analysis of the same authors allows for a smooth dependence of @xmath138 in @xmath139 and a slightly shallower potential ( @xmath140 mev ) with a larger radius , @xmath141 ) fm , provides a better agreement with the experimental @xmath0 binding energies .
we have performed a similar analysis for the calculated @xmath0 binding energies . in principle , the calculated self
energy is non - local both in @xmath24-space and in @xmath142-space . however , in a previous work @xcite we have shown that one can generate a local representation of the self
energy by performing an appropriate average of the non - local self
energy @xmath143 , where @xmath53 indicates the quantum numbers of the single - particle state , over the coordinate @xmath144 .
this local representation might , in first approximation , be characterized by the shape of a woods - saxon potential . instead of doing this average , a possible strategy is to assume a woods - saxon shape , fix the depth and the diffusivity independent of the mass number and determine the radius @xmath145 by requiring the woods - saxon potential to reproduce the same eigenvalue than the microscopic non - local energy dependent self energy . a reasonable value for the depth @xmath133 is the @xmath0 binding energy in nuclear matter , which is taken to be @xmath122 mev , and for the diffusivity @xmath137 fm .
the resulting values of @xmath145 when we apply this procedure to the @xmath12 deepest state of @xmath146c , @xmath10o and @xmath147ca are 2.25 fm , 2.53 fm and 3.82 fm respectively .
fitting these three values with a functional form similar to the one used in ref .
@xcite for the analysis of the experimental data one obtains @xmath148 fm . in order to visualize the quality of these fits ,
we show in fig .
[ fig : ener ] the binding energies for the @xmath12 and @xmath13 waves of @xmath149c , @xmath10o , @xmath147ca and @xmath150zr calculated with our non - local self energies ( triangles ) together with the values obtained with a woods - saxon potential with the parameters just defined above ( solid lines ) . as the spin - orbit splitting is so small
we have reported the average value of the @xmath123 and @xmath124 energies obtained from the non - local self energies and have not considered any spin - orbit term in the adjusted woods - saxon potential .
the results of @xmath14pb have not been included in the plot because , as mentioned before , the @xmath12 wave binding energy was larger than the binding energy in nuclear matter which we have taken as the depth of the woods - saxon potential .
the calculated binding energies are well reproduced by the woods - saxon shape and , as expected , both partial waves extrapolate to the binding energy for nuclear matter
. of course the binding energies are not enough to characterize the single - particle states since potentials giving rise to the same binding energies can generate substantial differences in the corresponding wave functions .
therefore , in order to analyze the microscopically calculated self energy we should also study the single - particle wave functions . to have a measure of the goodness of the wave functions generated by the woods - saxon potential , we calculate their overlap with the wave functions obtained by solving the schrdinger equation using the self energy .
the overlaps for @xmath146c , @xmath10o , @xmath147ca and @xmath150zr are 0.9917 , 0.9869 , 0.9924 and 0.9853 respectively , which are not close enough to 1 to guarantee the equality of the wave functions .
this is visualized in fig .
[ fig : funcs ] , where the wave function for the @xmath12 wave in @xmath10o obtained with the woods - saxon potential ( dashed - line ) is compared with the one obtained directly from the self energy ( solid line )
. another possibility would be to keep the diffusivity fixed and adjust the radius @xmath145 and the depth @xmath133 to reproduce the eigenvalue and to maximize the overlap with the eigenfunction provided by the self energy .
the values of @xmath133 by applying this procedure to the @xmath12 wave of the three lighter nuclei considered above are respectively @xmath151 , @xmath152 and @xmath153 mev whereas the values of the radius @xmath145 are 2.92 , 3.32 and 4.39 fm . with these values of @xmath133 and @xmath145 the overlaps are 0.9999 for the three nuclei considered .
the eigenfunction obtained by this procedure for the @xmath10o is also drawn in fig .
[ fig : funcs ] ( dot - dashed line ) and shows a large overlap with the self energy eigenfunction ( solid - line ) . in conclusion ,
the single - particle energies of closed - shell nuclei with one @xmath0 are well reproduced by using both the microscopic self energy or the simpler parametrization of woods - saxon type in the schrdinger equation . however , the wave functions provided by the microscopic self energy differ from the ones originated by a woods - saxon with a fixed depth and diffusivity and a @xmath139-dependent radius . it is important to note that the mean square radius of the self energy eigenfunction is larger than that from the corresponding woods - saxon wave function
. this can have important consequences in the study the mesonic decay of these @xmath0 hypernuclei .
indeed , it has been observed that the mesonic decay rates of light hypernuclei , such as @xmath154h , @xmath154he and @xmath155he , could be better reproduced if the @xmath0 wave function was pushed out to the surface by the effect of a repulsive hyperon - nucleus potential at short distances .
this would favour the mesonic decay of these hypernuclei because the @xmath0 would be exploring smaller nuclear density regions and the pauli blocking effects , which prevent the mesonic decay from occuring , would be less pronounced .
the mesonic decay rates of light hypernuclei have been calculated using repulsive @xmath0-nucleus potentials at short distances that have been obtained either phenomenologically @xcite , from a quark based bare @xmath8 interaction @xcite or from a microscopic @xmath91 effective interaction folded with an extremely compact @xmath5he density @xcite . at present
, no calculation exists that combines the use of an @xmath8 effective interaction with an appropiate density treatment of the host nucleus .
our method provides such ingredients and has been shown to produce @xmath0 wave functions that are pushed out to the surface .
this is a consequence of the non - localities of the self energy and might not be related to a repulsive character of the @xmath0-nucleus potential at short distances .
the implications of our results on the mesonic decay of @xmath0 hypernuclei will be explored in a future work .
we have analyzed a method to obtain the effective hyperon - nucleon interaction in finite nuclei based on an expansion over a @xmath9-matrix calculated in nuclear matter at fixed density and starting energy .
the purpose of this study is to set up a reliable frame for hypernuclear structure calculations with the aim of obtaining information about the hyperon - nucleon interaction , complementary to that provided by hyperon - nucleon scattering experiments .
we have shown , by explicit calculation of the @xmath0 and @xmath1 single - particle energies in @xmath111o , that truncating the expansion over the nuclear matter @xmath9-matrix at second order gives results that are very stable against variations of the density and starting energy used in the @xmath9-matrix . moreover ,
both first and second order terms depend quite strongly on those parameters .
this is an indication that the density dependent effects considered when treating explicitly the finite size of the nucleus are very important and , therefore , they might not be well approximated by energy independent and local effective interactions which start from a parametrized nuclear matter @xmath9-matrices evaluated at an average density .
we note that the use of local and density averaged effective interactions can be extremely useful in detecting similarities and differences among the various hyperon - nucleon potentials .
however , if the aim is to fine - tune the bare @xmath8 interactions to reproduce the spectroscopic data of hypernuclei , an appropiate effective interaction for the hypernucleus under study , as the one provided by our method , is in order . in particular ,
the two interactions employed in the present work give rise to very attractive @xmath1 binding energies , while the @xmath156 atomic data and @xmath6 spectra seem to indicate @xmath1-nucleus potential depths similar to that of the @xmath0 or even shallower .
although the method can be viewed as an alternative way of building up a finite nucleus effective interaction , it provides also the complete energy dependence of the hyperon self energy .
this allows in turn for a study of not only the bound states , as done here , but also the scattering states .
this is especially of interest in the analysis of hypernuclear production reactions which yield a large amount of quasifree hyperons .
we have obtained local woods - saxon @xmath0-nucleus potentials that reproduce the @xmath0 single - particle energies of several hypernuclei .
however , the wave functions obtained from our non - local self energy are far more extended and can be simulated only when we allow the woods - saxon potential to have an @xmath139-dependent depth and a relatively larger radius .
this can have important implications on the weak mesonic decay of @xmath0 hypernuclei , which will be explored in a future work .
this work is partially supported by the dgicyt contract no .
pb95 - 1249 ( spain ) and by the generalitat de catalunya grant no .
grq94 - 1022 .
one of the authors ( i.v . )
wishes to acknowledge support from a doctoral fellowship of the ministerio de educacin y cultura ( spain ) .
a. gal , adv .
( 1977 ) 1 .
b. povh , ann .
nucl . part .
28 ( 1978 ) 1 .
h. band , k. ikeda , t. motoba , y. yamada and t. yamamoto , prog .
suppl . 81 ( 1985 ) .
dover , d.j .
millener and a. gal , phys .
184 ( 1989 ) 1 .
e. oset , p. fernndez de crdoba , l.l .
salcedo and r. brockmann , phys .
reports 188 ( 1990 ) 79 .
j. cohen , prog .
phys . 25 ( 1990 ) 139 .
h. band , t. motoba and j. ofka , int .
j. mod.phys . a5 ( 1990 ) 4021 .
gibson and e.v .
hungerford , phys .
reports 257 ( 1995 ) 349 .
y. akaishi and t. yamazaki , prog .
39 ( 1997 ) 565 .
e. oset and a. ramos , prog .
41 ( 1998 ) , in press .
w. brckner et al . , phys .
lett . 79b ( 1978 ) 157 .
pile et al , phys .
66 ( 1991 ) 2585 .
t. hasegawa et al , phys . rev .
c53 ( 1996 ) 1210 .
t. motoba , h. band , t. fukuda and j. ofka , nucl .
a534 ( 1991 ) 597 .
u. straub , j. nieves , a. faessler and e. oset , nucl . phys .
a556 ( 1993 ) 531 .
i. kumagai - fuse , s. okabe and y. akaishi , phys .
b345 ( 1995 ) 386 .
szymanski et al . , phys .
c43 ( 1991 ) 849 .
t. harada , s. shinmura y. akaishi and h. tanaka , nucl .
a507 ( 1990 ) 715 . t. nagae et al , phys .
. lett . 80 ( 1998 ) 1605 .
hayano et al , nuovo cimento 102a ( 1989 ) 437 ; phys .
lett . b231 ( 1989 ) 355 .
a. bouyssy , j. hfner , phys .
64b ( 1976 ) 276 ; a. bouyssy , phys .
84b ( 1979 ) 41 . c. dover , l. ludeking and g.e .
walker , phys .
c22 ( 1980 ) 2073 .
t. motoba , h. band , r. wnsch and j. ofka , phys .
c38 ( 1988 ) 1322 .
y. yamamoto , h. band and j. ofka , prog .
theor . phys .
80 ( 1988 ) 757
millener , c.b .
dover and a. gal , phys .
c38 ( 1988 ) 2700 .
f. fernndez , t. lpez - arias and c. prieto , z. phys .
a334 ( 1989 ) 349 .
lanskoy and y. yamamoto , phys .
c55 ( 1997 ) 2330 .
r. brockmann and w. weise , nucl .
a355 ( 1981 ) 365 .
m. chiapparini , a.o .
gattone and b.k .
jennings , nucl . phys .
a529 ( 1991 ) 589 . j. mare and j. ofka , z. phys .
a33 ( 1989 ) 209 . j. mare and b.k .
jennings , phys .
c49 ( 1994 ) 2472 .
lombard , s. marcos and j. mare , phys .
c51 ( 1995 ) 1784 .
glendenning , d. von - eiff , m. haft , h. lenske and m.k .
weigel , phys .
c48 ( 1993 ) 889 . f. ineichen , d. von - eiff and m.k .
weigel , j. phys .
g22 ( 1996 ) 1421 .
y. sugahara and h. toki , prog .
92 ( 1994 ) 803 .
z. ma , j. speth , s. krewald , b. chen and a. reuber , nucl .
a608 ( 1996 ) 305 . y. yamamoto and h. band , prog .
81 ( 1985 ) 9 ; k. tsushima , k. saito and a. w. thomas , phys .
b411 ( 1997 ) 9 ; k. tsushima , k. saito , j. haidenbauer and a. w. thomas , nucl .
a630 ( 1998 ) 691 .
y. yamamoto and h. band , prog .
( 1990 ) 254 .
y. yamamoto , a. reuber , h. himeno , s. nagata and t. motoba , czec.jour .
phys . 42 ( 1992 ) 1249 .
y. yamamoto , t. motoba , h. himeno , k. ikeda and s. nagata , prog .
suppl . 117 ( 1994 ) 361 .
d. halderson , phys .
c48 ( 1993 ) 581 .
j. hao , t.t.s .
kuo , a. reuber , k. holinde , j. speth and d.j .
millener , phys . rev
. lett .
71 ( 1993 ) 1498 .
m. hjorth - jensen , a. polls , a. ramos and h. mther , nucl .
a605 ( 1996 ) 458 .
m. borromeo , d. bonatsos , h. mther and a. polls , nucl .
phys . a539 ( 1992 ) 189 .
m. hjorth - jensen , h. mther and a. polls , phys .
c50 ( 1994 ) 551 .
maessen , t.a .
rijken , and j.j .
de swart , phys .
rev.c40 , 2226 ( 1989 ) .
b. holzenkamp , k. holinde , and j. speth , nucl .
a500 , 485 ( 1989 ) .
kung , t.t.s .
kuo and k.f .
ratcliff , phys .
rev.c19 ( 1979 ) 1063 c.w . wong and d.m .
clement , nucl .
a183 ( 1972 ) 210 g. backenstoss , t. bunacin , j. egger , h. koch , a. schwitter and l. tauscher , z. phys .
a273 ( 1975 ) 137 c. j. batty et al , phys . lett .
74b ( 1978 ) 27 c. j. batty , phys .
lett . 87b ( 1979 )
324 c. j. batty , nucl .
a372 ( 1981 ) 433 c. j. batty , e. friedman and a. gal , phys .
( 1994 ) 273 ; prog .
suppl . 117
( 1994 ) 227 j. mare , e. friedman , a. gal and b. k. jennings , nucl .
a594 ( 1995 ) 311 c| cc |cc @xmath94 @xmath95 & & @xmath72 & @xmath112 & @xmath157 & @xmath112 & @xmath157 ( mev ) & ( mev ) & ( mev ) & ( mev ) & ( mev ) @xmath158 & @xmath159 & @xmath160 & @xmath161 & @xmath162 @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 @xmath168 & @xmath169 & @xmath170 & @xmath171 & @xmath172 c| cc |cc @xmath94 @xmath95 & & @xmath72 & @xmath112 & @xmath157 & @xmath112 & @xmath157 ( mev ) & ( mev ) & ( mev ) & ( mev ) & ( mev ) @xmath173 & @xmath174 & @xmath175 & @xmath176 & @xmath177 @xmath178 & @xmath179 & @xmath180 & @xmath181 & @xmath182 @xmath183 & @xmath184 & @xmath185 & @xmath186 & @xmath187 c| cc |cc @xmath115 mev & & @xmath89 & @xmath112 & @xmath157 & @xmath112 & @xmath157 ( @xmath95 ) & ( mev ) & ( mev ) & ( mev ) & ( mev ) @xmath188 & @xmath189 & @xmath190 & @xmath191 & @xmath192 @xmath193 & @xmath194 & @xmath195 & @xmath196 & @xmath167 @xmath197 & @xmath169 & @xmath170 & @xmath171 & @xmath172 c| cc |cc @xmath116 mev & & @xmath89 & @xmath112 & @xmath157 & @xmath112 & @xmath157 ( @xmath95 ) & ( mev ) & ( mev ) & ( mev ) & ( mev ) @xmath188 & @xmath198 & @xmath199 & @xmath200 & @xmath201 @xmath193 & @xmath202 & @xmath203 & @xmath204 & @xmath205 @xmath197 & @xmath174 & @xmath175 & @xmath176 & @xmath177 .
@xmath0 binding energies in the @xmath206 , @xmath207 and @xmath208 single - particle orbits for different nuclei .
the available experimental data , indicating the hypernucleus for which they have been measured , are taken from the compilation of @xcite supplemented by new measures reported in @xcite and @xcite .
all the results have been derived from the jlich b interaction . [
cols="^,^,^,^,^",options="header " , ]
|
single - particle energies of @xmath0 and @xmath1 hyperons in several nuclei are obtained from the relevant self energies .
the latter are constructed within the framework of a perturbative many - body approach employing present realistic hyperon - nucleon interactions such as the models of the jlich and nijmegen groups .
the effects of the non - locality and energy - dependence of the self energy on the bound states are investigated .
it is also shown that , although the single - particle hyperon energies are well reproduced by local woods - saxon hyperon - nucleus potentials , the wave functions from the non - local self energy are far more extended .
implications of this behavior on the mesonic weak decay of @xmath0 hypernuclei are discussed .
|
in the course of the last three decades , spontaneous parametric down - conversion ( spdc ) has proven to be a valuable tool in the experimental investigation of fundamental properties of the electromagnetic field in the quantum domain @xcite , including nonclassical correlations , entanglement and nonlocality .
the fact that spdc is capable of generating pairs of photons in a wide range of frequencies and wave vectors , in addition to the fact that these photons may be entangled in a number of different degrees of freedom , qualifies spdc also as a unique tool in the demonstration of quantum information procedures and protocols @xcite . in many applications , especially in the pioneering ones ,
a basic knowledge of a few properties of the two - photon states generated by spdc is enough to explain the effects discussed .
as the applications become more sophisticated , for example in those that combine correlations in more than one degree of freedom , a more comprehensive approach is necessary in order to enhance the capabilities of quantum state engineering . in this direction ,
the use of fourier optics concepts in two - photon optics represents a significant step .
an important application field of spdc is the construction of entangled photon state sources with high fidelity and purity @xcite . in order to obtain efficient and dependable two - photon sources ,
a number of questions must be addressed , as for example , phase compensation , mode coupling , pair collection efficiency , maximum photon flux , etc .
to be optimized , all these points require a detailed knowledge of the spdc process .
another context in which a more precise model is crucial refers to the issue of conservation of orbital angular momentum ( oam ) of light in the process of spdc @xcite .
oversimplifications or inconsistent assumptions about the two - photon state generated have led to conflicting conclusions on whether oam is conserved or not in spdc . in this work , we present a detailed account of the two - photon states generated by spdc in both type i and type ii phase matching , including the effects of anisotropy of the nonlinear medium and the frequency spread of the down - converted fields . as an exact approach to this problem presents a considerable level of difficulty , we are forced to restrict ourselves to some approximations .
we adhere to the commonly adopted simplified procedure for the quantization of the electromagnetic field in the nonlinear medium by just multiplying the @xmath0 vectors by the corresponding refractive indices and keeping the vacuum expressions for the field operators .
the frequency spread of the down - converted fields is considered to be small compared to their central frequencies .
we also work in the paraxial approximation , in which the propagation of the fields is much easier to deal with .
fortunately , these approximations , are good enough to encompass a great number of practical cases .
we admit , however , that while the model discussed here is helpful in far field applications , it may not be appropriate in discussions of the fundamental interactions inside the nonlinear medium . whenever possible , the results reported here are compared with experimental data , obtained in our labs or reported in the literature . it should be mentioned that other authors have reported their contributions to this field , in different levels of detail and scope @xcite . from our point of view , a comprehensive description of spdc that accounts for all features observed so far in two - photon states , with expressions ready to use in fourier optics , is still missing .
it is in this context that we present our work .
the interaction hamiltonian describing the optical processes of parametric down - conversion in nonlinear birefringent crystals is usually written in terms of a simplified field quantization in matter @xcite .
the field modes participating in this process are coupled by the second - order susceptibility tensor @xmath1 . due to the phase matching conditions ,
the coupling occurs only for some sets of polarizations , known as type i and type ii . in negative uniaxial crystals ( @xmath2 ) such as bbo ( @xmath3bab@xmath4o@xmath5 ) and lithium iodate ( liio@xmath6 ) , the down - conversion process with type i phase matching
can be summarized as @xmath7 , meaning that one photon of the pump beam with extraordinary polarization is converted into two photons with ordinary polarization . in type
ii phase matching , the process is represented by @xmath8 or @xmath9 . in a perturbative approach ,
the ( post - selected ) two - photon state generated by spdc is written as @xcite @xmath10 where @xmath11 and @xmath12 are the wave vectors of the down - converted fields , @xmath13 is the wave vector of the pump field , whose frequency @xmath14 is assumed to be well defined .
@xmath15 indicates the polarization of each down - converted field , that can be ordinary ( @xmath16 ) or extraordinary ( @xmath17 ) , and @xmath18 represent a one - photon state in the plane wave mode @xmath19 .
considering that the nonlinear crystal is a rectangular block of sides @xmath20 , @xmath21 , @xmath22 , with two faces parallel to the plane @xmath23 , the amplitude @xmath24 is given by @xmath25 where @xmath26 represents the set of indices @xmath27 , @xmath28^{\frac{1}{2}}$ ] , @xmath29 is the quantization volume , @xmath30 is the interaction volume ( @xmath31 ) , @xmath32 is the refractive index corresponding to the mode @xmath19 , @xmath33 , @xmath34 is the interaction time , @xmath35 is related to the second - order nonlinear susceptibility tensor @xcite , @xmath36 ( @xmath37 ) are the cartesian components of the polarization vectors , @xmath38 is the pump field amplitude in the mode @xmath39 , and @xmath40 . in order to simplify expression ( [ ampli1 ] ) , it is convenient to make the following approximations : ( _ a _ ) the pump frequency @xmath14 is well defined and the interaction time is long , so that the term @xmath41 is significant only when @xmath42 .
the frequencies @xmath43 and @xmath44 can therefore be written as @xmath45 this assumption can be justified by the use of a moderate power continuous - wave pump laser so that the time interval between two down - conversions is large compared to the detection resolving time .
( _ b _ ) the frequency spread of the detectable down - converted fields is small compared to the central frequency ( @xmath46 ) , so that the dispersion of the refractive indices around the central frequency @xmath47 is small and a linear approximation can be used .
this assumption is justified by the use of narrow - band interference filters in front of the detectors .
for example , for a 100 nm wavelength spread centered at 700 nm , @xmath48 lies in the interval ( @xmath49 ) .
( _ c _ ) the terms @xmath50 and @xmath35 are slowly - varying functions of @xmath51 , so that they may be taken as constants in the intervals considered for @xmath51 .
( _ d _ ) the pump beam propagates along the @xmath52 axis and the crystal is large enough in the @xmath53 and @xmath54 directions to contain the whole pump beam transverse profile . in this case ,
@xmath20 and @xmath21 can be extended to infinity and the last term in expression ( [ ampli1 ] ) , the integral @xmath55 , is proportional to @xmath56 where @xmath57 locates the center of the crystal .
( _ e _ ) the quantization volume is large enough to justify the replacement of summations in @xmath0 by integrals .
( _ f _ ) the pump beam contains only extraordinary polarization .
it is implicit in this assumption that we are dealing with negative birefringent crystals . under the above assumptions , eq .
( [ hong ] ) is written as @xmath58 where @xmath59 represents a one - photon state in the mode defined by the transverse ( @xmath23 ) component @xmath60 of the wave vector , the frequency @xmath61 and by the polarization @xmath15 .
the amplitude @xmath24 is now reduced to @xmath62 where @xmath63 is a coupling constant , which depends on the nonlinear susceptibility tensor , @xmath64 is the spectral function defined by the narrow bandwidth filters placed in front of the detectors , and @xmath65 is the plane wave spectrum of the pump beam @xmath66 , with @xmath67 replaced by @xmath68 .
the integral in @xmath52 is discussed in what follows .
the function @xmath64 may be centered at zero in frequency degenerate configurations or at some fixed small detuning @xmath69 .
most of the properties of the down - converted fields is determined by the longitudinal wave vector mismatch @xmath70 if the anisotropy of the medium is neglected ( which is not always convenient ) , @xmath71 ( monochromatic approximation ) , and the crystal is cut for collinear phase matching , @xmath72 has a simple expression in terms of @xmath73 and @xmath74 , with the @xmath52 components replaced by @xmath75 . in the paraxial approximation , @xmath76 can be approximated by @xmath77 .
this leads , for @xmath78 , to @xmath79 which appears in the amplitude reported in ref .
this result , however , is a good approximation only when the crystal is very thin , the diffraction ( rayleigh ) length @xmath80 of the pump beam is large , and the detection area is small so that the sinc function can be approximated by 1 . in more general conditions ,
the anisotropy of the crystal must be taken into account . from this point on
, we will refer to the arrangement depicted in fig.[fig : slab ] .
let us consider a monochromatic electromagnetic field in a plane wave mode @xmath81 propagating through a nonmagnetic uniaxially birefringent medium with ordinary and extraordinary refractive indices @xmath82 and @xmath83 , respectively . considering that the optic axis lies on the plane @xmath84 , making an angle @xmath85 with the @xmath52 direction , and that the plane wave has extraordinary polarization , maxwell s equations require that the components of the wave vector @xmath86 , whose cartesian components are ( @xmath87 ) , satisfies the _ ray surface _
equation @xcite geometry of the problem .
the uniaxial medium is assumed to be a slab of thickness @xmath22 , having its faces parallel to the @xmath23 plane .
the optic axis lies in the @xmath84 plane , at an angle @xmath85 with the @xmath52 direction . ]
@xmath88 solving eq .
( [ eq : elipsoiderodada ] ) for @xmath89 , we find @xmath90 where @xmath91 @xmath92 @xmath93 @xmath94 with @xmath95 the ordinary and extraordinary refractive indices @xmath82 and @xmath83 are obtained from the sellmeier equations .
for bbo , @xmath82 and @xmath83 are obtained from @xcite @xmath96 with @xmath97 given in @xmath98 m . for lithium iodate
we use @xcite @xmath99 within the paraxial approximation , @xmath100 the term @xmath101 represents a linear displacement in the @xmath53 direction , also known as the _ walk - off_. it reaches its maximum value in the neighborhood of @xmath102 .
the terms @xmath103 and @xmath104 account for the deviation of the curvatures of the ray surface from a spherical surface in the @xmath53 and @xmath54 directions , respectively . differently from @xmath101 ,
the terms @xmath103 and @xmath104 have little influence in the phase matching .
the term @xmath105 is the refractive index for a plane wave with extraordinary polarization , propagating along the @xmath52 direction , whose wavenumber is @xmath106
[ alpha ] , [ beta ] and [ gamma ] show the values of @xmath101 , @xmath103 and @xmath104 as functions of the phase matching angle @xmath85 for 300 nm and 600 nm in bbo and lithium iodate .
( color online ) values of @xmath101 for for 300 nm and 600 nm in bbo and lithium iodate as functions of the phase matching angle @xmath85 .
the curves are ordered from top to bottom , according to the legend . ]
( color online ) values of @xmath103 for for 300 nm and 600 nm in bbo and lithium iodate as functions of the phase matching angle @xmath85 .
the curves are ordered from top to bottom , according to the legend . ]
( color online ) values of @xmath104 for for 300 nm and 600 nm in bbo and lithium iodate as functions of the phase matching angle @xmath85 .
the curves are ordered from top to bottom , according to the legend . ]
let us consider first the case of type i phase matching , where the pump field has extraordinary polarization and the down - converted fields have ordinary polarization ( @xmath7 ) . in this case , @xmath107 @xmath108 @xmath109 where @xmath110 and @xmath111
. since we are assuming @xmath112 , the refractive indices @xmath113 and @xmath114 can be written in a linear approximation as @xmath115 where @xmath116 is the ordinary refractive index at the frequency @xmath47 , and factor @xmath117 is given by latexmath:[\[\label{defa } a=\frac{\omega_{p}}{2\bar{n}_{o}}\left.\frac{dn_{o}}{d\omega}\right in order to calculate @xmath72 , which in type i phase matching will be referred to as @xmath119 , we have to write @xmath120 replacing @xmath121 and @xmath122 in eq .
( [ k0z ] ) by @xmath123 and @xmath124 , respectively . after some manipulation
, we find @xmath125,\nonumber\\\end{aligned}\ ] ] where @xmath126 , @xmath127 , and @xmath128 .
the subscript @xmath129 means that @xmath101 , @xmath103 and @xmath104 refer to the pump field .
@xmath130 is the type i _ collinear index mismatch _ , defined as @xmath131 in the derivation of eq .
( [ fi ] ) , terms in @xmath132 were neglected with respect to @xmath48 . for type
i collinear ( @xmath133 ) phase matching in bbo pumped with @xmath134 nm , @xmath135 , @xmath136 , @xmath137 , and @xmath138 . for lithium iodate , @xmath139 , @xmath140 , @xmath141 and @xmath142 .
sometimes , instead of the transverse components of the wave vectors , one is interested in the output angles of the down - converted photons .
in general , the output fields of interest propagate at small angles with respect to the @xmath52 axis , so that one can approximate the output angles by [ xiq ] @xmath143 where @xmath144 , and we define the vectors @xmath145 and @xmath146 as shown in fig .
[ xi ] .
angular components of the down - converted fields . ] in the far field , @xmath147 and @xmath148 can be written in terms of the transverse coordinates @xmath149 and @xmath150 of the detectors as [ xirho ] @xmath151 where @xmath152 is the @xmath52 coordinate of the detection plane ( parallel to the @xmath23 plane ) . in terms of the angular variables @xmath147 and @xmath148 , the amplitude @xmath153 , defined in eq .
( [ ampli2 ] ) , is written as @xmath154\nonumber\\ & & \times { \ensuremath{\mathrm{sinc}}}\left[\frac{kl_{z}}{2}\ f_{oo}({\ensuremath{\boldsymbol{\xi}}}_{1},{\ensuremath{\boldsymbol{\xi}}}_{2},\nu)\right]\nonumber\\ & & \times \exp \left[-i kz_{c}\ f_{oo}({\ensuremath{\boldsymbol{\xi}}}_{1},{\ensuremath{\boldsymbol{\xi}}}_{2},\nu)\right],\end{aligned}\ ] ] where @xmath155 is a normalization constant and @xmath156\nonumber\\ & & -\frac{1}{4\bar{n}_{o } } \big{\ { } ( 1+\nu ) \xi_{1}^{2 } + ( 1-\nu ) \xi_{2}^{2}\nonumber\\ & & -\frac{b}{2}\left[(1+\nu ) \xi_{1x}+(1-\nu ) \xi_{2x}\right]^{2}\nonumber\\ & & -\frac{g}{2}\left[(1+\nu ) \xi_{1y}+(1-\nu ) \xi_{2y}\right]^{2}\big{\}}.\end{aligned}\ ] ] the single count rate as a function of @xmath147 ( @xmath148 ) and @xmath48 can be obtained from eq .
( [ phiang ] ) by integration in @xmath148 ( @xmath147 ) . to show that eqs .
( [ phiang ] ) and ( [ foo ] ) describe the correct dependence of @xmath24 on frequency and output angles , the output angle @xmath157 is plotted as a function of the down - converted wavelength in fig .
[ spectrum ] for a 15mm - long lithium iodate crystal pumped by a 325 nm laser beam for three different values of the phase matching angle @xmath85 .
results for @xmath158 ( collinear ) and @xmath159 are in good agreement with experimental data reported by bogdanov _
@xcite for @xmath160 and @xmath161 , respectively .
the difference of @xmath162 in the noncollinear phase matching angle is probably due to the fact that the noncollinear condition is obtained by tilting the crystal .
output angle in the @xmath84 plane as a function of the down - converted wavelength for a 15mm - long lithium iodate crystal pumped by a 325 nm laser beam , for different values of the phase matching angle @xmath85 .
these results were obtained from eqs .
( [ phiang ] ) and ( [ foo ] ) integrating in @xmath148 with @xmath163 .
the ( nonlinear ) horizontal scale was adjusted to allow a direct comparison with experimental data reported by bogdanov _
_ @xcite . ]
the two - photon frequency spectral content of @xmath153 for a given pair of selected output angles @xmath164 and @xmath165 can also be derived from eq .
( [ phiang ] ) .
experimentally , this spectrum can be directly measured @xcite , or inferred from the hong - ou - mandel dip when the measurement is taken with broadband filters .
the expected profile of the coincidence detection probability as a function of the path length difference @xmath166 in a hong - ou - mandel interferometer is given by @xcite @xmath167 the integrals are taken from @xmath168 to @xmath169 .
[ fighom1 ] shows a plot of @xmath170 calculated with eqs .
( [ phiang ] ) and ( [ hom1 ] ) , for a bbo crystal with @xmath171 mm in type i phase matching , pumped by a 351 nm laser , without interference filters .
the collection angles are such that both down - converted beams are centered at 702 nm .
hong - ou - mandel dip for a 1mm - long bbo crystal cut for type i phase matching , with @xmath172 pumped by a 351 nm laser , calculated with eqs .
( [ phiang ] ) and ( [ hom1 ] ) .
it is assumed that no interference filters are used . ] in the collinear detection configuration ( @xmath173 ) , the spectral profile of the two - photon state depends on the phase matching angle @xmath85 .
figures [ match1 ] and [ match2 ] show the down - converted wavelengths for collinear type i phase matching as a function of @xmath174 , the deviation from @xmath175 for lithium iodate and from @xmath176 for bbo , with @xmath177 mm and @xmath178 mm , both crystals pumped with @xmath134 nm . ( color online ) down - converted wavelengths for collinear ( @xmath173 ) type i phase matching as a function of @xmath174 , the deviation from @xmath175 for lithium iodate with @xmath177 mm ( left ) and @xmath178 mm ( right ) , pumped with @xmath134 nm
. the solid line corresponds to the condition @xmath130=0 . ]
( color online ) down - converted wavelengths for collinear ( @xmath173 ) type i phase matching as a function of @xmath174 , the deviation from @xmath176 for bbo with @xmath177 mm ( left ) and @xmath178 mm ( right ) , pumped with @xmath134 nm .
the solid line corresponds to the condition @xmath130=0 . ]
another useful representation of @xmath153 is obtained when the following new coordinates are used : [ newvar ] @xmath179 now , eq.([phiang ] ) is written as @xmath180\nonumber\\ & & \times \exp \left[-i kz_{c}\ f_{oo}({\ensuremath{\boldsymbol{\xi}}}_{s},{\ensuremath{\boldsymbol{\xi}}}_{d},\nu)\right],\end{aligned}\ ] ] where @xmath155 is a normalization constant and @xmath181,\nonumber\\\end{aligned}\ ] ] where @xmath182 was neglected with respect to @xmath48 .
many of the interesting features of @xmath153 are retained when the anisotropy is neglected in second order , that is , when the parameters @xmath117 and @xmath183 are both approximated by 1 . in this case , @xmath184 in the frequency degenerate case ( @xmath71 ) , @xmath185 let us analyze the behavior of eqs .
( [ phiang2 ] ) and ( [ fooapproxdeg ] ) in two different measurement schemes : ( a ) when the detectors are scanned in opposite directions in the far field , so that @xmath186 .
the probability of coincidence detection is proportional to @xmath187\right|^{2}.\ ] ] eq .
( [ cone1 ] ) describes a circular profile in the variable @xmath188 , with a radius @xmath189 and a half - width @xmath190 . under these detection conditions ,
the coincidence transverse profile maps the well - known down - converted light cones .
\(b ) when the detectors are scanned in the same direction in the far field , so that @xmath191 .
this condition means @xmath192 eq .
( [ transf ] ) illustrates the possibility of transferring the angular spectrum @xmath193 from the pump beam to the down - converted field @xmath194 @xcite , and its dependence on the crystal length @xmath22 .
the sinc function has a half - width @xmath195 in the @xmath53 direction .
depending on the spatial bandwidth of the pump beam , this factor may limit the transfer of its angular spectrum to the two - photon field . considering that the pump beam is gaussian
, the transfer will be satisfactory only when its waist @xmath196 is much larger than @xmath197 . for a bbo of @xmath198 1 mm , @xmath199 m .
this `` clipping '' in the angular spectrum transfer , caused by the walk - off term @xmath200 , has consequences in the anisotropy of entanglement in spatial variables of the two - photon state @xcite
[ clip1 ] and [ clip2 ] show a comparison between experimental results obtained in our labs and the predictions of eq .
( [ transf ] ) for a bbo crystal with @xmath177 mm in collinear type i phase matching , pumped by a 405 nm laser with @xmath201 m . in fig .
[ clip1 ] , the two detectors are scanned in the same sense , along the @xmath53 direction .
the effect of the walk - off term is evident .
the dashed line shows the expected angular profile of the pump beam . in fig .
[ clip2 ] , the two detectors are scanned in the same sense , along the @xmath54 direction . in both cases ,
interference filters of fwhm=10 nm were used in front of the detectors .
( color online ) comparison between experimental results ( @xmath202 ) and the prediction of eq .
( [ transf ] ) ( solid line ) for a bbo crystal with @xmath177 mm in collinear type i phase matching , pumped by a 405 nm laser with @xmath201 m . the two detectors are scanned in the same sense , along the @xmath53 direction .
the dashed line shows a gaussian profile corresponding to the pump laser beam . ]
( color online ) comparison between experimental results ( @xmath202 ) and the prediction of eq .
( [ transf ] ) ( solid line ) for a bbo crystal with @xmath177 mm in collinear type i phase matching , pumped by a 405 nm laser with @xmath201 m .
the two detectors are scanned in the same sense , along the @xmath54 direction .
the gaussian predicted by eq.([transf ] ) coincides with the expected profile of the pump beam . ]
in type ii phase matching , the calculation is similar , except for the fact that one of the down - converted beams has extraordinary polarization ( @xmath8 or @xmath9 ) , that is , @xmath203.\end{aligned}\ ] ] now , the @xmath89 component of the @xmath17-polarized down - converted wave has to be written according to expression ( [ eq : aprox_paraxial_kz ] ) , that is , for the case @xmath8 , @xmath204 @xmath108 @xmath205 where @xmath206 and @xmath207 .
the factor @xmath208 is defined analogous to eq . ( [ defa ] ) , replacing @xmath82 by @xmath105 .
@xmath209 , @xmath210 , @xmath211 and @xmath212 are calculated with @xmath82 and @xmath83 taken at @xmath47 .
the amplitudes @xmath213 and @xmath214 have expressions analogous to eq .
( [ ampli2 ] ) , with @xmath215 and @xmath216 given by [ deltaoe ] @xmath217,\qquad \\ \delta_{z}^{eo}&=&k\mu_{eo}+ ( \alpha_{p}-\bar{\alpha})q_{1x}+\alpha_{p}q_{2x}\nonumber\\ & & + \frac{1}{\bar{n}_{o}k}\big[\frac{b}{2}(q_{1x}+q_{2x})^{2}+ \frac{g}{2}(q_{1y}+q_{2y})^{2}\nonumber\\ & & -\bar{b}q_{1x}^{2}-\bar{g}q_{1y}^{2}-q_{2}^{2}+(q_{2}^{2}- \bar{b}q_{1x}^{2}-\bar{g}q_{1y}^{2})\nu\big],\end{aligned}\ ] ] where @xmath218 @xmath127 , @xmath128 , @xmath219 , @xmath220 , and @xmath221 . in eqs .
( [ deltaoe ] ) , terms in @xmath117 and @xmath208 were neglected with respect to 1 , and @xmath132 was neglected with respect to @xmath48 .
note that @xmath222 and @xmath223 have a linear dependence on @xmath48 , differently from @xmath130 , whose dependence is on @xmath132 . in terms of the output angles @xmath147 and @xmath148
, we have @xmath224\nonumber\\ & & \times { \ensuremath{\mathrm{sinc}}}\left[\frac{kl_{z}}{2}\ f_{oe}({\ensuremath{\boldsymbol{\xi}}}_{1},{\ensuremath{\boldsymbol{\xi}}}_{2},\nu)\right]\nonumber\\ & & \times \exp \left[-i kz_{c}\ f_{oe}({\ensuremath{\boldsymbol{\xi}}}_{1},{\ensuremath{\boldsymbol{\xi}}}_{2},\nu)\right],\end{aligned}\ ] ] where @xmath155 is a normalization constant and @xmath225^{2}\nonumber\\ & & -\frac{g}{2}\left[(1+\nu ) \xi_{1y}+(1-\nu ) \xi_{2y}\right]^{2}\big{\}}.\end{aligned}\ ] ] @xmath214 has an expression similar to ( [ amplioe ] ) , with @xmath226^{2}\nonumber\\ & & -\frac{g}{2}\left[(1+\nu ) \xi_{1y}+(1-\nu ) \xi_{2y}\right]^{2}\big{\}}.\end{aligned}\ ] ] fig .
[ figconeii ] shows density plots of single counts obtained from eqs .
( [ amplioe]-[feo ] ) integrated in one of the output angles and in @xmath48 over a bandwidth of 0.025 ( corresponding to a 10 nm filter ) for a bbo crystal pumped by a 407 nm laser , cut for @xmath227 , with @xmath171 mm ( left ) and @xmath228 mm ( right ) .
both plots are in good agreement with experimental data reported by lee _
_ @xcite .
density plot of single counts obtained from eq .
( [ cone2 ] ) integrated in one of the output angles and in @xmath48 over a bandwidth of 0.025 ( which corresponds to a 10 nm filter ) for a bbo crystal pumped by a 407 nm laser , cut for @xmath227 , with @xmath171 mm ( left ) and @xmath228 mm ( right ) . ]
the linear dependence of @xmath222 and @xmath223 with @xmath48 leads to a frequency spectrum of the two - photon state quite different from the type i case .
figure [ match3 ] shows the down - converted wavelengths for collinear ( @xmath173 ) type ii phase matching as a function of @xmath174 , the deviation from @xmath229 for bbo with @xmath177 mm and @xmath178 mm , pumped with @xmath134 nm . both plots were obtained from eqs ( [ amplioe]-[feo ] ) .
( color online ) down - converted wavelengths for collinear ( @xmath173 ) type ii phase matching as a function of @xmath174 , the deviation from @xmath229 for bbo with @xmath177 mm ( left ) and @xmath178 mm ( right ) , pumped with @xmath134 nm .
the solid lines correspond to the conditions @xmath222=0 and @xmath223=0 . ]
[ fighom2 ] shows the hong - ou - mandel dip for a 1mm - long bbo crystal cut for _ beamlike _ type ii phase matching ( @xmath230 ) , pumped by a 351 nm laser .
the solid line corresponds to the case in which interference filters of 20 nm bandwidth are used , in good agreement with the data reported by kim @xcite .
the dashed line corresponds to the case in which no interference filters are used .
notice the difference between figs .
[ fighom1 ] and [ fighom2 ] . in type
ii phase matching , the hong - ou - mandel dip is much larger than in type i , indicating a narrower frequency spectrum in type ii .
( color online ) hong - ou - mandel dip for a 1mm - long bbo crystal cut for type ii phase matching , with @xmath230 pumped by a 351 nm laser , calculated with eqs .
( [ hom1 ] ) and ( [ amplioe ] ) .
the solid line corresponds to the case in which interference filters of 20 nm bandwidth are used .
the dashed line corresponds to the case in which no interference filters are used . ]
it is also possible to write @xmath213 and @xmath214 in terms of the variables @xmath231 and @xmath232 defined in eq .
[ newvar ] , but the exact expressions are too long to be of practical use . however ,
when @xmath183 , @xmath233 , @xmath234 and @xmath235 are all approximated by 1 , we arrive at the useful expressions [ foe ] @xmath236 when the detectors are scanned in opposite directions in the far field , so that @xmath186 , the probability of coincidence detection @xmath237 is proportional to @xmath238 , where [ cone2 ] @xmath239,\\ \tilde{\mathcal{e}}(0)|^{2}\nonumber\\ & & \times\,{\ensuremath{\mathrm{sinc}}}^{2}\big [ \frac{kl}{4\bar{n}_{o}}(r_{eo}^{2}-|{\ensuremath{\boldsymbol{\xi}}}_{d}- { \ensuremath{\boldsymbol{c}}}_{eo}|^{2})\big ] , \ \ \
\end{aligned}\ ] ] @xmath240,\\ r_{eo}^{2}&=&2\bar{n}_{o } \mu_{eo } + { \bar{n}_{o}^{2 } \bar{\alpha}^{2 } } \left [ \frac{1}{4 } + \nu\left(\frac{\alpha_{p}}{\bar{\alpha } } -\frac{1}{2}\right)\right],\end{aligned}\ ] ] @xmath241 { \ensuremath{\mathbf{\hat{x}}}},\\ { \ensuremath{\boldsymbol{c}}}_{eo}&=&\bar{n}_{o}\bar{\alpha}\left [ -\frac{1}{2 } + \nu\left(\frac{\alpha_{p}}{\bar{\alpha } } -\frac{1}{2}\right)\right ] { \ensuremath{\mathbf{\hat{x}}}},\end{aligned}\ ] ] and @xmath242 is the unit vector in the @xmath53 direction .
expressions ( [ cone2 ] ) describe two circular profiles , centered at @xmath243 and @xmath244 , with radii @xmath245 and @xmath246 , respectively .
collinear down - conversion occurs when @xmath247 and @xmath248 .
the so - called _ beamlike _ type ii down - conversion occurs in the degenerate case ( @xmath71 ) when @xmath85 is such that @xmath249 .
let us analyze the case when the detectors are scanned in the same direction in the far field , so that @xmath250 . then , [ transf2 ] @xmath251}\big{\}},\ \ \ \\ { \ensuremath{\boldsymbol{\xi}}}_{s})|^{2}\nonumber\\ & & \times\,{\ensuremath{\mathrm{sinc}}}^{2}\big{\{}\frac{kl_{z}}{2}\big{[}\mu_{eo } + \big{(}\alpha_{p } - \frac{\bar{\alpha}}{2}\big{)}\xi_{sx}\big{]}\big{\}}.\ \ \\end{aligned}\ ] ] note that in type ii the transfer of angular spectrum described by eqs .
( [ transf2 ] ) is affected by the detuning @xmath48 through the parameters @xmath222 and @xmath223 .
the detuning has the effect of displacing the sinc function by an amount @xmath252 in the @xmath53 axis , in opposite directions for @xmath253 and @xmath254 . in the limit of long crystals ,
the sinc@xmath255 functions are narrow enough to allow us to write the coincidence detection probability as @xmath256,\end{aligned}\ ] ] where @xmath257 . in type
ii phase matching , due to the linear dependence of @xmath222 and @xmath223 on @xmath48 , the angular spectrum of the pump field may be transferred to the two - photon state if the bandwidth of the detection filters is broad enough , even for long crystals
[ clip3 ] and [ clip4 ] show comparison of the predictions of eq .
( [ transf2 ] ) with experimental data for a 5mm - long bbo crystal pumped by a 405 nm laser with @xmath201 m and detection filters with a bandwidth of @xmath258 nm . in both figures , there is a good agreement between experimental data and the expected gaussian profile of the pump beam .
the experimental conditions are the same of figs .
[ clip1 ] and [ clip2 ] , except for the phase matching type .
notice the difference between the two cases when the detectors are scanned in the @xmath53 direction ( figs .
[ clip1 ] and [ clip3 ] ) . in the case of fig .
[ clip3 ] , the 10 nm filter allows a complete transfer of the pump beam profile to the coincidence detection .
the dashed line shows the expected curve for the monochromatic case , that is , @xmath259 ( zero bandwidth filter ) .
( color online)comparison between experimental results ( @xmath202 ) and the prediction of eq .
( [ transf2 ] ) ( solid line ) for a bbo crystal with @xmath177 mm in collinear type ii phase matching , pumped by a 405 nm laser with @xmath201 m .
the two detectors are scanned in the same sense , along the @xmath53 direction .
the gaussian predicted by eq.([transf2 ] ) coincides with the expected profile of the pump beam .
the dashed line shows the expected curve for the monochromatic case . ]
( color online)comparison between experimental results ( @xmath202 ) and the prediction of eq .
( [ transf2 ] ) ( solid line ) for a bbo crystal with @xmath177 mm in collinear type ii phase matching , pumped by a 405 nm laser with @xmath201 m .
the two detectors are scanned in the same sense , along the @xmath54 direction .
the gaussian predicted by eq.([transf2 ] ) coincides with the expected profile of the pump beam . ]
we have discussed in detail the two - photon state generated by spontaneous parametric down - conversion in bulk crystals , taking into account the effect of crystal anisotropy , in both in type i and type ii phase matching .
our discussion was based on the perturbative approach introduced by hong and mandel , in the context of fourier optics , in which the fields are treated by means of their plane - wave expansion .
the frequency spectrum of the down - converted fields was also considered , in the approximation of small detuning and monochromatic pump beams .
extension of the theory to cover pulsed pump beams seems straightforward .
several approximations were made in order to provide simple expressions that still exhibit the main features of the frequency and spatial spectral properties of the two - photon states .
these expressions were shown to be in good agreement with experimental data for some selected situations .
in particular , the spectral content of the two - photon state and the transfer of angular spectrum from the pump beam to the two - photon state were analyzed , and a significant differences of these features in type i and type ii phase matching were discussed .
the results presented here are a contribution to the fourier optics of two - photon states and may be helpful to improve the understanding and further development of entangled state sources for quantum information and quantum communication .
|
we present a detailed account of the two - photon states generated by spdc in both type i and type ii phase matching , including the effects of anisotropy of the nonlinear medium and the frequency spread of the down - converted fields .
accurate as well as simplified expressions are derived for type i and type ii phase matching in the context of fourier optics .
the main results are compared with experimental data available in the literature , showing good agreement in all cases .
|
our understanding of the origin of gamma - ray bursts ( grbs ) has continued to advance rapidly in the years since the first x - ray @xcite , optical @xcite and radio @xcite afterglows were discovered .
in particular the nearby afterglow from has solidified the grb - supernova connection @xcite , and provided the first grb with a well determined expansion rate @xcite .
this event has presented a unique opportunity to test afterglow models @xcite , and to explore the environment around a grb .
afterglow models invoke gas - rich environments around at least some of the grbs .
several basic properties of this circumburst medium are presently unknown , and need to be addressed in order to better understand the afterglow and its evolution . for example , in the simplest emission models , the density of the ambient medium is related to the afterglow flux density @xcite .
especially , if grbs exist in both gas - poor and gas - rich environments , that could explain why afterglow is absent in some grbs .
direct observations of the circumburst medium associated with grbs would thus provide an important test of the fireball model .
another issue is the density profile of the medium into which the grb ejecta expands .
current models postulate a medium in which density is governed by the mass - loss of the progenitor star ( @xmath6 ) . at present
however , some observations instead point to a uniform density in order to explain the evolution of the afterglow lightcurve @xcite . in a few cases ,
strong damped ly@xmath7 absorption has been found indicating column densities as high as 5 @xmath8 10@xmath9 cm@xmath1 @xcite . during
the expansion of the grb ejecta , the circumburst medium is likely to go through different stages of ionization .
perna & loeb ( 1998 ) suggest that as a consequence ( optical ) absorption lines will vary in their ew with time .
the rate of this variation can constrain the size of the absorbing region . similarly , in the radio the fraction of molecular and atomic gas should be governed by the effective ionization , and might therefore vary in a similar way .
one way to investigate the circumburst medium is via atomic or molecular absorption studies in the radio .
another way is to measure the deceleration of the expanding fireball . in this paper
we present both late time very long baseline interferometry ( vlbi ) observations of the afterglow of and similar epoch and oh absorption observations taken with the arecibo telescope . assuming a lambda cosmology with @xmath10 km / s / mpc , @xmath11 and @xmath12 , the angular - diameter distance of at @xmath13 is @xmath14mpc , and 1 milliarcsec corresponds to 2.85 pc .
lrrrrrc 20 jun 2003 & 83 & 8.409 & 138 & 50 & 2 & y27 + 1 nov 2003 & 217 & 8.409 & 138 & 50 & 2 & y27 + 1 nov 2003 & 217 & 8.409 & 165 & 32 & 2 & vlba+eb+y27+wb+ar+mc+nt + 30 nov 2003 & 247 & 1.216 & 30 & 12 & 2 & ar + 30 nov 2003 & 247 & 1.428 & 50 & 12 & 2 & ar + 30 nov 2003 & 247 & 1.473 & 30 & 12 & 2 & ar + the vlbi observations were taken at 8.4 ghz on 2003 november 1 , 217 days after the burst , with a global array including the very long baseline array ( vlba ) of the nrao .
other telescopes used were the effelsberg 100-m telescope , the phased vla , the green bank telescope ( gbt ) , the 305 m arecibo telescope , the westerbork ( wsrt ) tied array , and the noto and medecina telescopes of the consiglio nazionale delle ricerche . in all some 56 individual and combined antennas
each of 25 m or more in diameter were employed with a combined collecting area of 0.12 km@xmath15 .
most antennas were on - source for a period of 5.5 hours .
all stations recorded with 256 mbps with 2 bit sampling in dual circular polarization with the exception of noto which had only a right - circular polarization receiver available .
the observations were correlated at the joint institute for vlbi in europe ( jive ) . the nearby ( 1.5@xmath16 ) source j1051 + 2119 was used for phase - referencing with a 2:1 minute cycle on source : calibrator . the weak calibrator j1048 + 2115
was observed hourly to check on the quality of the phase referencing .
self - calibration with a 1 hour solution interval was used to further refine the calibration and remove some slow - changing atmospheric phase errors .
the final image has an rms noise of 24 @xmath17jy / beam .
this is substantially higher than the expected thermal noise of 8@xmath17jy / beam , in part because of partial loss of signals from arecibo and westerbork for reasons not completely understood .
we fit a symmetric , two - dimensional gaussian to the measured visibilities on and find a size of 0.176 @xmath18 0.08 mas . as in @xcite
the error of the size is estimated from signal - to - noise ratios and from monte - carlo simulations of the data using identical ( @xmath19,@xmath20 ) coverage , similar noise properties , and a gaussian component of known size added .
the standard deviation of the recovered sizes , modelfitted in the same way as we treat the observations , was found to be 0.053 mas .
we also obtain a position for of r.a .
10@[email protected]@xmath23 and dec .
[email protected] with an uncertainty of 0.2 mas in each coordinate . solving for proper motion using all the high frequency vlbi observations to date , we derive @xmath24 mas yr@xmath25 and @xmath26 mas yr@xmath25 , or an angular displacement over the first 217 days of [email protected] mas ( fig . 1 ) .
these observations are consistent with those reported by @xcite , and impose an even stronger limit on the proper motion .
this limit argues against the cannonball model for grbs proposed by @xcite .
single dish observations of grb 030329 were carried out with the l - band wide receiver of the 305 m arecibo radio telescope on 2003 november 30 , for a total of 2 hr .
the simple position - switched observations utilized all four interim correlator boards to observe various l - band frequencies .
the bandwidth of each board was 12.5 mhz .
while the first two boards were set to observe the frequency of the redshifted @xmath2721 cm h i line , recording each linear polarization with 2048 spectral channels , the other two boards were set to observe simultaneously the orthogonal polarizations at the frequencies of the redshifted @xmath2718 cm mainline oh transitions ( 1665 & 1667 mhz ) and the 1720 mhz satellite oh transition with 1024 spectral channels per polarization .
following the editing out of data suffering from radio frequency interference , the total on - source integration time for the redshifted @xmath2721 cm h i and the 1720 mhz satellite oh transitions was 30 min each .
for the redshifted @xmath2718 cm mainline oh transitions , the on - source integration time was 50 min .
figures 2 , 3 , and 4 show pairs of total - intensity spectra of the grb 030329 centered at the optical heliocentric velocity that corresponds to the frequency of the redshifted @xmath2721 cm h i line , the redshifted @xmath2718 cm oh mainlines , and the redshifted @xmath2718 cm oh satellite line at 1720 mhz , respectively .
the top spectra in figs . 2 , 3 , and 4 ,
are hanning - smoothed with spectral resolutions of 3.52 km s@xmath25 ( 12.2 khz ) , 5.99 km s@xmath25 ( 24.4 khz ) , and 5.80 km s@xmath25 ( 24.4 khz ) , respectively . the rms noise level in each of these
spectra is 1.02 , 0.32 , and 0.57 mjy beam@xmath25 , respectively .
the bottom plot in each figure is a five - channel smoothed version of the respective top spectra .
the velocity resolution in these spectra is 8.80 , 14.98 , and 14.50 km s@xmath25 , and the rms noise level of each spectrum is 0.482 , 0.224 , and 0.282 mjy beam@xmath25 , respectively .
no emission or absorption is seen in any of the arecibo spectra .
we can derive a 3@xmath28 limit on the opacity of @xmath29 with a velocity resolution of 8.8 km s@xmath25 .
this corresponds to a column density limit of @xmath2 cm@xmath1 assuming a spin temperature of 100 k , and uniform coverage .
if the spin temperature is higher or the line is shallow and wide , then this limit could be higher . given the extreme energies involved in the grb , enough to cause a sudden ionospheric disturbance in the earth s atmosphere @xcite some 740 mpc away , it could well be that nearly all the along the line - of - sight in the host galaxy has been ionized and not yet recombined .
further research in this area is needed . in a similar manner
we can place a 3@xmath28 upper limit on the 1667 mhz oh mainline opacity of @xmath30 with a velocity resolution of 15.0 km s@xmath25 .
this corresponds to an oh column density limit of @xmath31 cm@xmath1 assuming a uniform covering factor , and an excitation temperature of 10 k. we can further deduce a limit on molecular hydrogen using the relation @xmath32
@xmath33 10@xmath34 @xmath35 @xcite , of @xmath36 cm@xmath1 .
the continuum flux density of the grb 030329 at 1.4 ghz at the time of these observations was @xmath37 mjy .
the various spectra reported here show a higher continuum flux density , because of a 15 mjy continuum source located at an angular distance of about 3 arcmin from the grb 030329 , i.e. , within the primary beam of the arecibo radio telescope .
our measured size of 0.176 @xmath18 0.08 mas is quite close to the size of 0.172 @xmath18 0.043 found by @xcite , indicating a possible slowing of the burst at late times .
this trend was already apparent from the average expansion velocities derived from the angular size measurements on april 22 and june 20 .
the entire history of expansion for is shown in fig .
the first measurement at 15 days comes from a model - dependent estimate of the quenching of the scintillation @xcite .
the late time curvature may indicate that between 83 and 217 days , has transitioned into a non - relativistic expansion .
alternatively , it is possible ( but somewhat contrived ) that the intrinsic surface brightness profile has changed in a way to compensate for the expansion . for all epochs
we fit a two - dimensional gaussian model since our resolution does not permit us to discriminate between a gaussian , ring , disk , or more complex profile .
a gamma - ray burst drives a relativistic blast wave into a circumburst medium of density @xmath38 whose radius @xmath39 is related to the energy of the explosion approximately by @xmath40 , where @xmath0 is the bulk lorentz factor of the fireball .
more complete treatments are given by @xcite .
the density profile of the circumburst medium is generally taken to be either a @xmath41 wind , or a constant density ism , such as one might find beyond the termination shock of the progenitor s stellar wind @xcite .
the deceleration at late times currently favors the ism model of @xcite which shows a break in the expansion rate ( see their figures 4 and 6 ) . unfortunately , the large uncertainty in our measurement of the size of at late times does not permit us to discriminate strongly between the various models .
@xcite place the termination shock of the wolf - rayet wind at a distance of 0.4 pc from the progenitor . beyond
that they predict a fairly constant density out to the red supergiant shell at a radius of 1.7 pc . with a current diameter of 0.5 pc for ( radius of 0.25 pc ) at day 217
it is near the termination shock , especially if the progenitor had a shorter lifetime , or a relatively high density ism has stalled the shock . from 8.4 ghz vlba observations on april 6
@xcite derived a 3@xmath28 limit on the linear polarization of 0.16 mjy / beam , corresponding to a limit on the fractional polarization of @xmath51.0% . in a contemporaneous optical observation @xcite
measure a polarization of 2.2+/-0.3% .
the decrease in polarization at lower frequencies has been explained as the result of the source being optically thick at 8.4 ghz at these early times since the maximum degree of linear polarization of an optically thick synchrotron source is 12% while the maximum polarization of an optically thin synchrotron source is about 80% @xcite . to look for any change in polarization with time as the source transitions to optically thin at 8.4 ghz we have analyzed the very sensitive , late time phased vla observations for polarimetry .
we find no detection at either epoch and place 3@xmath28 limits of @xmath51.8% and @xmath54.7% on 2003 jun . 20 and 2003 nov . 1 respectively .
the total intensity measured with the vla on these epochs is 3.11 @xmath18 0.03 and 0.75 @xmath18 0.02 mjy respectively . the emission mechanism for the grb afterglow is widely accepted to be synchrotron radiation , which is intrinsically linearly polarized if there is an ordered component to the magnetic field , or if the magnetic field is generated at the internal shocks @xcite .
the recent detection of 80 @xmath18 20% polarization in the prompt @xmath0-ray emission from grb 021206 @xcite , although controversial @xcite , has led to an increased interest in modeling the time and frequency dependent behavior of the polarization from the afterglow @xcite . to have an observed polarization of @xmath55% in the late time afterglow ( see 2.2 )
could be explained as the result of highly disordered magnetic fields @xcite .
propagation effects can also reduce the intrinsic linear polarization below detectable levels . a faraday screen produced by ionized
gas and magnetic fields can cause gradients in the observed polarization angle across the source , leading to depolarization if the resolution element of the telescope , or the size of the source if unresolved , is large compared to the gradients .
the rotation measures ( @xmath42 ) can be related to the line - of - sight magnetic field , @xmath43 , by @xmath44 where @xmath43 is measured in mg , @xmath45 in @xmath46 , d@xmath47 in pc , and the upper limit of integration , @xmath48 , is the distance from the emitting source to the end of the path through the faraday screen along the line of sight . we can estimate the minimum magnetic field needed to produce a gradient across the source . to produce a 90@xmath16 rotation at 8.4 ghz
requires a rm of 1200 rad m@xmath1 .
assuming a density within the red supergiant shell of 0.2 @xmath46 @xcite and a path length of 2 pc , then the magnetic field strength required is @xmath49 mg .
this field strength is similar to the equipartition field strengths @xcite have found from modeling the radio emission from colliding - wind wolf - rayet ( wr ) binary systems , although the field strengths in these systems have probably been enhanced by the collision .
faraday screens in the grb environment are considered in more detail by @xcite .
while no detection of atomic ( ) or molecular ( oh ) material is found towards , the limits of @xmath2 cm@xmath1 are not particularly constraining .
observations of more shielded molecules like nh@xmath50 towards future bursts at early times would be of interest , and given the expected strength of a nearby afterglow of @xmath3350 mjy or more , could yield detections , or place interesting limits on the amount of material in the grb environment .
although grb 030329 has faded considerably , it may still be detectable with vlbi techniques .
even crude estimates of the size could differentiate between the predictions of wind and constant - density environments at these late times .
radio re - brightening of has been predicted by @xcite , and @xcite estimate a level of 0.6 mjy 1.7 years after the burst .
this re - brightening might occur as the counterjet becomes non - relativistic and therefore radiatively isotropic .
this brightening will be accompanied by a temporary rapid growth in the size of the source as two , well - separated jets become visible , and by a change in the light centroid @xcite .
if such a re - brightening occurs then a precise size estimate at late times becomes readily achievable with existing facilities .
rather surprisingly , we find no detectable linear polarization from at cm wavelengths to limits as low as 1%
. this could indicate less order than expected in the magnetic fields of the external shock that drives the afterglow , or a faraday screen that depolarizes the radio emission .
all of these limits are from 8 days or more after the burst . given the rhessi result @xcite of large polarization from the @xmath0-rays , and predictions of some fireball models , it would be well worth searching for polarization from the prompt optical and radio emission ( _ e.g. _ , granot & taylor 2005 ) .
in the near future the _ swift _ satellite should dramatically increase the number of grbs with measured redshifts , revealing some that are nearby . in future vlbi studies of grb afterglows at redshifts less than 0.1
it should be possible to image the structure of the afterglow .
fireball models of heating by a single relativistic shock front predict that at late times the fireball should look like a ring @xcite .
gbt thanks the kavli institute for particle astrophysics and cosmology for hospitality and support .
we thank jonathon granot and avi loeb for useful discussions .
this research has made use of nasa s astrophysics data system .
|
we present very long baseline interferometry ( vlbi ) observations 217 days after the @xmath0-ray burst of 2003 march 29 .
these observations provide further measurements of the size and position of grb 030329 that are used to constrain the expansion rate and proper motion of this nearby grb .
the expansion rate appears to be slowing down with time , favoring expansion into a constant density interstellar medium , rather than a circumstellar wind with an r@xmath1 density profile .
we also present late time arecibo observations of the redshifted and oh absorption spectra towards grb 030329 .
no absorption ( or emission ) is seen allowing us to place limits on the atomic neutral hydrogen of @xmath2 cm@xmath1 , and molecular hydrogen of @xmath4 cm@xmath1 .
finally , we present vla limits on the radio polarization from the afterglow of @xmath52% at late times .
|
the system ie 0657 - 56 of two colliding clusters ( termed the bullet " ) , is particularly interesting among such systems .
x - ray observations of this system show a prominent bow - shock generated by the supersonic motion ( mach number of @xmath3 ) of the sub - cluster ( the bullet ) relative to the gaseous component of the main cluster . the estimated speed of the bullet relative to the main cluster is @xmath4 , while the relative line - of - sight velocities between the groups of galaxies associated with the two components is only @xmath5 . therefore , the relative motion between the two clusters is almost entirely perpendicular to the line - of - sight ( barrena et al .
2002 ; markevitch et al .
2004 ) .
the current dynamical state of the system must be consistent with that evolved in an expanding universe .
therefore , all matter belonging to the system must originate from a region of vanishing size as we approach the initial singularity ( @xmath6 ) .
the positions of objects in the system can be uniquely moved back - in - time from their current positions and velocities .
the masses of these objects can then be tuned so that the cosmological constraint at @xmath7 is satisfied .
for a two component system , this is traditionally known as the timing argument and it has been used for studying the system of the galaxy and m31 ( e.g. kahn & woltjer 19969 ; peebles 1993 ) . timing arguments
have also been applied for the the bullet system by zhao ( 2007 ) .
this study has not considered systems embedded in an over - dense region of a larger scale .
further , it differs in the details from the model presented here .
the outline of this paper is as follows . in
2 we briefly present the lensing measurements and constrain the masses of the main and sub components assuming nfw forms ( see below ) for their density profiles . in
3 we describe our model and present the results for the mass estimates .
we conclude in 4 with a general discussion .
we assume that the density profiles of dark matter halos of the main and the sub components follow the form given by navarro , frenk & white ( 1996 ) ( hereafter , nfw ) as @xmath8 where @xmath9 is the concentration paramater , @xmath10 is the distance from the halo center in units of the virial radius , and @xmath11 is the critical mean density ( @xmath12 where @xmath13 is the scale factor ) . here
we assume that the mean density within the virial radius is @xmath14 .
this gives @xmath15 to be @xmath16 $ ] the profile ( [ eq : nfw ] ) depends on the halo concentration , @xmath9 , and its virial radius , @xmath17 .
we determine these two parameters for each halo by matching the combined weak and strong lensing estimates of the projected mass density as given in figure ( 5 ) of brada et el .
( 2006 ) . for convenience ,
we borrow the results for the projected mass in that figure and show them in figure ( [ fig : lensem ] ) here .
points extending to 0.5 and @xmath18 correspond to the main and sub components , respectively . attached to the points are @xmath19 error - bars , taken as @xmath20 and @xmath21 of the mass for the main and the sub - cluster , respectively ( brada et el .
we fit nfw profiles to these data by minimizing the quantity @xmath22 with respect to @xmath23 and @xmath9 . here
the summation is over all data points , @xmath24 and @xmath25 are projected masses taken from the lensing measurements and the nfw profile , and @xmath26 are @xmath19 error - bars at each data point .
this procedure yields multiple solutions for @xmath17 and @xmath9 for both dark components .
the results are represented in figure ( [ fig : lensec ] ) as contour plot of the quantity @xmath27 as a function of @xmath17 and @xmath9 for the main ( top panel ) and sub ( bottom ) components .
there is a clear degeneracy between @xmath17 and @xmath9 .
the difference between two solutions with similar @xmath27 is illustrated in figure ( [ fig : lensem ] ) where , in addition to the data , we show the projected mass corresponding to nfw profiles with @xmath28 & 2.6 mpc ( for the main ) and @xmath29 & 2.3 mpc ( for the sub ) , as indicated in the figure .
these values of @xmath17 , respectively for the main and the sub , represent the lower and upper limits of what we consider as acceptable match to the mensing measurements .
therefore , the largest acceptable ratio between the virial radii of the main and sub components is @xmath30 .
the mass of a halo within a sphere of radius @xmath17 is @xmath31 .
therefore , the largest mass ratio allowed by these considerations is @xmath32 .
we will see in the next section that this ration is also consistent with cosmological initial conditions if the bullet system resides in a region of a mild over - density of 10 times the background density .
we give here the equations governing the evolution of the two dark components of the system ie 0657 - 56 .
we neglect the gravity of gas in the system .
we consider nfw profiles truncated at the virial radii for both components .
the system is assumed to be embedded in region of density @xmath33 where @xmath34 is the mean cosmic matter density and @xmath35 is the density contrast with a time dependence derived from the spherical top - hat model . we assume a flat universe with a cosmological constant @xmath36 to the density from the dark energy .
the values of the density parameters are @xmath37 and @xmath38 .
we work in the system of the center of mass of the two clusters .
the motion of the center of mass , @xmath39 , of the component @xmath40 is determined by @xmath41{\bf r}_{\rm i}\ ] ] where @xmath42 is the mass of each component .
for a given value of @xmath43 today , the function @xmath44 is obtained from the spherical collapse model .
the force , @xmath45 , the halo @xmath46 applies to @xmath40 , is computed for nfw profiles for both components .
for the mass , @xmath47 , of the main component we consider the range @xmath48 .
this corresponds to @xmath49 . according to figure ( [ fig : lensec ] ) , @xmath50 is ruled out by lensing for all values of the concentration parameter , but we explore it anyways and show that such masses are not favored by the timing argument as well . for the mass , @xmath51 of the sub , we consider the range @xmath52 , corresponding to @xmath53 .
the lower end of this range is inconsistent with lensing .
the concentration parameters , @xmath9 , are chosen to be those giving the best match to the lensing data ( see fig .
[ fig : lensec ] ) . given a choice of @xmath47 and @xmath51 we numerically solve the equations ( [ fig : eom ] ) back - in - time for three values of the current relative velocity , @xmath54 : @xmath55 , @xmath56 and @xmath57 .
we assume a distance of @xmath58 between the mass centers of the two components . for simplicity
we assume that the relative speed lies along the separation between the centers of mass of the cluster components .
figure ( [ fig : delr ] ) shows the separation , @xmath59 , between the two components at @xmath6 , as contour plots in the plane of @xmath60 and @xmath61 . the top , middle and bottom rows , respectively , @xmath62 , @xmath56 and @xmath63 .
the column to the left shows solutions obtained for a cluster embedded in mean cosmological density , while the one to the right , to a nonlinear density contrast of @xmath64 at the measured redshift of the system .
each panel in the middle and bottom rows contains more than one set of contours .
the set corresponding to the smallest masses represent the relevant solutions in which the separation reaches a maximum value only once within a hubble time .
the separations at @xmath65 are sensitive to the assumed relative speed , @xmath54 .
for example , according to the left panels , for @xmath66 and @xmath67 , the separation is zero for @xmath68 , while it is @xmath69 for @xmath70 . for @xmath71
we have seen in the previous section that the lensing analysis gives 3:1 as the largest ratio between the mass of the main and the sub - components .
inspection of figure ( [ fig : delr ] ) reveals that this is also consistent with the requirement of cosmological initial conditions if the system is embedded in a region which 10 times denser than the background ( top - right panel ) .
the inferred total mass from this figure is @xmath72 which is close to the mass inferred from lensing for a 3:1 mass ratio .
the effect of having an over - dense region engulfing the system is most pronounced for the largest relative speed ( cf the three rows to the right ) and is almost unnoticeable for the smallest speed .
the reason is as follows .
the existence of this region has little effect on the dynamics when the cluster components are still close to each other .
as we go sufficiently backward in time , their separation increases and so their mutual gravity decreases . as this happens , the density of the region decreases , in accordance with the top - hat model . for large relative velocities ,
large separations are reached while the surrounding region is still at a significant over - density and thus can affect the dynamics .
the relative velocity between the two cluster components is large compared to the usual virial speeds ( @xmath73 ) .
this triggered discussions on whether or not the system if consistent with the currently viable models for structure formation based on the cold dark matter scenario ( cdm ) with a cosmological constant , @xmath74 , i.e. the @xmath75 model .
modified newtonian dynamics ( mond ) has been invoked ( angus , famaey , & zhao 2006 , angus & mcgaugh 2007 ) to explain the estimated relative speed .
other explanations relied on long range scalar interactions in the dark sector to aid gravity in boosting the relative speed ( farrar & rosen 2007 ) .
nonetheless , like in the newtonian case both of these suggestions need to be assessed in a cosmological context .
scalar interactions modify large scale structure formation in a desirable way ( nusser , gubser & peebles 2005 ; farrar & peebles 2004 ) .
still , proper cosmological simulations need to be performed to assess whether or not they could easily reproduce a bullet system " . as for mond - like modifications ( bekenstein 2004 ) , it is still unclear how they can be implemented in a cosmological context .
the simplest cosmological adaptation had been incorporated in a cosmological simulation and lead to large scale structure formation similar to @xmath76 , but with mond s acceleration parameter which is smaller by an order of magnitude that the standard value ( nusser 2000 ) .
we argue here that even with a large value for the relative speed , gravitational force alone is consistent with cosmological initial conditions , i.e. the timing argument .
the total mass we derive is @xmath77 , if the system resides in a region of a density contrast of about 10 .
hayashi & white ( 2006 ) performed a likelihood analysis on the millennium simulation to conclude that the bullet system can be accommodated in the lcdm cosmogony .
they adopt masses of @xmath78 and @xmath79 for the main and sub components , respectively .
our mass estimate is slightly larger than theirs but we still could be consistent with the @xmath76 model normalized to @xmath80 .
however , the mass ratio we advocate here , 3:1 , is quite different from the value adopted by hayashi & white ( 2006 ) . zhao ( 2007 ) also used the timing argument to constrain the mass of the bullet system without taking into account dynamical effects of an over - dense surrounding environment .
the mutual force in that study is also computed differently from this work .
zhao s estimate of the total mass is as twice as the value derived here .
springel & farrar ( 2007 ) used non - cosmological simulations of two colliding clusters of a mass ratio of 10:1 and found that the relative speed between the two dark components could be as small as 2700 km / s , while the shock speed is 4500 km / s .
as mentioned before , a 10:1 mass ratio is inconsistent with lensing . of course , the physical effects leading to velocity lag of dark mass relative to the shock could still be important and they need to be quantified with simulations having a mass ration of 3:1 . nonetheless , for low relative speeds the timing argument yields masses ( see bottom - left panel in fig .
[ fig : delr ] ) which are too small to be consistent with the lensing measurements .
we have made some attempts , the details of which are not presented here , at including dynamical friction and tidal stripping according to the recipes in nusser & sheth ( 1999 ) .
these effects reduce the relative speed between the two components , therefore , for a given final relative speed they increase corresponding masses required to match the timing constraint .
the analysis presented here neglects important processes which might affect the dynamics of the system .
each of the two components have formed by merging of smaller halos .
therefore , sometime in the past , the matter making each component was distributed in more than one clump . at that time , the mutual force determining the relative motion of the centers of mass of the two components should depend on the spatial distribution of their progenitors . if , however , major merging and accretion activities in both components ceased when the separation between them was large enough .
then the monopole term alone , which we use here for the mutual force , should be a reasonable description to the dynamics of the system .
we have tuned the masses so that the solution back - in - time gives vanishing separation near @xmath65 .
this has been done by solving the initial value problem of given current separation and relative speed as input to the numerical solution .
alternatively , we could have solved a boundary value problem where the first boundary condition is current separation and the second is a constraint which guarantees the cosmological constraint near @xmath65 ( e.g. peebles 1993 , nusser & branchini ) .
this could be done using the least action principle , as proposed by peebles ( 1989 ) .
this approach yields a prediction for the current relative velocity for a given mass choice .
this predicted velocity could then be compared with the observed speed .
this approach will be employed in future work .
this research is supported by the german - israeli foundation for development and research and the asher space research fund .
|
we present a simplified dynamical model of the `` bullet '' system of two colliding clusters .
the model constrains the masses of the system by requiring that the orbits of the main and sub components satisfy the cosmological initial conditions of vanishing physical separation a hubble time ago .
this is also known as the timing argument .
the model considers a system embedded in an over - dense region .
we argue that a relative speed of @xmath0 between the two components is consistent with cosmological conditions if the system is of a total mass of @xmath1 is embedded in a region of a ( mild ) over - density of 10 times the cosmological background density .
combining this with the lensing measurements of the projected mass , the model yields a ratio of 3:1 for the mass of the main relative to that of the subcomponent .
the effect of the background weakens as the relative speed between the two components is decreased . for relative speeds
lower than @xmath2 , the timing argument yields masses which are too low to be consistent with lensing .
h # 1*martin : # 1 * # 1*adi : # 1 * # 1#1 # 1#1 # 1#1 cosmology : theory , large - scale structure of the universe clusters
|
bridson asked the question of whether a quasi - isometry exists between thompson s group @xmath0 and the group @xmath4 .
burillo @xcite provides an example of a quasi - isometric embedding of @xmath5 into @xmath0 as possible evidence addressing this question .
we state his example in [ sec : examples ] below . while investigating this question , we came across some interesting examples of quasi - isometric embeddings into @xmath0 which we describe below . these quasi - isometric embeddings all have simple geometric interpretations , which are often easier to express than the corresponding algebraic or group theoretic definitions .
we present these examples to illustrate the beautiful geometry evident in thompson s group @xmath0 .
our examples are based on the interaction between the finite and infinite presentations of @xmath0 and the representation of elements of @xmath0 as pairs of binary rooted trees .
these embeddings use shift maps of @xmath0 and provide concrete geometric realizations of subgroups of @xmath0 of the form @xmath6 which are known to be quasi - isometricly embedded by work of burillo @xcite and guba and sapir @xcite
thompson s group @xmath0 has both finite and infinite presentations ; it is usually presented finitely as @xmath7,[x_0x_1^{-1},x_0^{-2}x_1x_0 ^ 2 ] \rangle\ ] ] and infinitely as @xmath8 the infinite presentation provides a set of normal forms for elements of @xmath0 .
namely , each @xmath9 can be written @xmath10 where @xmath11 , and @xmath12 and @xmath13 . to obtain a unique normal form for each element
, we add the condition that when both @xmath14 and @xmath15 occur , so does @xmath16 or @xmath17 , as discussed by brown and geoghegan @xcite
. we will always mean unique normal form when we refer to a word @xmath18 in normal form . we give a brief introduction to @xmath0 below ; for a more detailed and comprehensive description , we refer the reader to @xcite and @xcite .
we interpret elements of @xmath0 geometrically as pairs of finite rooted binary trees , each with the same number of exposed leaves , as described by cannon , floyd and parry in @xcite .
let @xmath19 be a rooted binary tree .
an _ exposed leaf _ of @xmath19 ends in a vertex of valence @xmath20 , and the exposed leaves are numbered from left to right , beginning with @xmath21 . a node together with two leaves
is called a _ caret_.
a caret @xmath22 may have a _
right child _ , that is , a caret @xmath23 which is attached to the right leaf of @xmath22 .
we can similarly define the _ left child _
@xmath24 of the caret @xmath22 . in a pair @xmath25 of rooted binary trees ,
the tree @xmath26 is called the _ negative tree _ and @xmath27 the _ positive tree_. this terminology is explained further in [ sec : exponents ] below .
the _ left side _ of @xmath19 is the maximal path of left edges beginning at the root of @xmath19 .
similarly , we define the _ right side _ of @xmath19 .
a pair of trees is _ unreduced _ if both @xmath26 and @xmath27 contain a caret with two exposed leaves numbered @xmath28 and @xmath29 .
there are many tree pair diagrams representing the same element of @xmath0 but each element has a unique reduced tree pair diagram representing it .
when we write @xmath25 to represent an element of @xmath0 , we are assuming that the tree pair is reduced .
the tree pair diagrams for the generators @xmath30 and @xmath31 are given in figures [ fig : x0tree ] and [ fig : x1tree ] .
there is a bijective correspondence between the tree pair diagrams described above and the normal form of an element .
the _ leaf exponent _ of an exposed leaf numbered @xmath32 in @xmath26 or @xmath27 is defined to be the length of the maximal path consisting entirely of left edges from @xmath32 which does not reach the right side of the tree , and is written @xmath33 .
note that @xmath34 for an exposed leaf labelled @xmath32 which is a right leaf of a caret , as there is no path consisting entirely of left edges originating from @xmath32 . in figure
[ fig : exptree ] , the exponents of the left - hand tree @xmath26 are as follows : @xmath35 and @xmath36 . of @xmath0 , in the presentation @xmath37 . ] of @xmath0 , in the presentation @xmath37.,title="fig : " ] + once the exponents of the leaves in @xmath26 and @xmath27 have been computed , the normal form of the element @xmath38 is easily obtained .
the positive part of the normal form of @xmath18 is @xmath39 where @xmath28 is the number of exposed leaves in either tree , and the exponents are obtained from the leaves of @xmath27 .
the negative part of the normal form of @xmath18 is similarly found to be @xmath40 where the exponents are now computed from the leaves of @xmath26 . note that many of the exponents in the normal form as given above may be zero .
[ fig : w ] .,title="fig:",width=144 ] + fordham @xcite introduces a method of calculating , directly from the tree pair diagram @xmath38 , the word length @xmath41 of @xmath18 with respect to the word metric arising from the finite presentation @xmath37 . throughout this paper
we use @xmath41 to denote word length of a word @xmath18 in this word metric . in @xcite
we use fordham s method to estimate this word length in terms of the number of carets @xmath42 in either tree of a tree pair diagram representing @xmath9 .
[ thm : bound ] let @xmath38 and @xmath42 be the number of carets in @xmath26
. then @xmath43 fordham s method of determining word length in @xmath0 is based on a classification of the types of carets found in a finite rooted binary tree @xmath19 .
a rough classification divides these carets into three types : _ left carets _ ,
which have a left edge on the left side of @xmath19 , _ right carets _ , which are not the root caret and have a right edge on the right side of @xmath19 , and _
interior carets_. fordham specializes further by defining seven distinct types of carets , as follows . 1 .
the first caret on the left side of the tree , with caret number @xmath21 .
every tree has exactly one caret of type @xmath44 .
2 . @xmath45 .
any left caret other than the one numbered @xmath21 .
3 . @xmath46 .
an interior caret which has no right child .
an interior caret which has a right child .
any right caret numbered @xmath49 with the property that caret @xmath50 is an interior caret .
@xmath51 . a right caret which is not an @xmath48 but for which there is a higher numbered interior caret
@xmath52 . a right caret with no higher - numbered interior carets .
the root caret is always considered to be a left caret and will be of type @xmath45 unless it has no left children , in which case it would be the @xmath44 caret .
the carets in @xmath19 are numbered according to the infix ordering of nodes .
caret @xmath21 is a left caret with an exposed left leaf numbered @xmath21 in the leaf numbering .
according to the infix scheme , we number the left children of a caret before the caret itself , and number the right children after numbering the caret . examples of trees whose carets are numbered in this way can be found in @xcite .
fordham @xcite proves that the word length @xmath41 of @xmath38 can be computed from knowing the caret types of the carets in the two trees via the following process .
we number the @xmath50 carets in each tree according to the infix method described above , and for each @xmath53 with @xmath54 we form the pair of caret types consisting of the type of caret number @xmath53 in @xmath26 and the type of caret number @xmath53 in @xmath27 .
the single caret of type @xmath44 in @xmath26 will be paired with the single caret of type @xmath44 in @xmath27 , and for that pairing we assign a weight of 0 . for all other caret pairings , we assign weights according to the following table . [ cols="^,^,^,^,^,^,^",options="header " , ] the main result of fordham @xcite is the following theorem .
[ thm : blake ] given a word @xmath9 described by the reduced tree pair diagram @xmath25 , the word length @xmath55 is the sum of the weights of the caret pairings in @xmath25 .
group multiplication of @xmath38 and @xmath56 is accomplished by creating temporary unreduced representatives @xmath57 of @xmath18 and @xmath58 of @xmath59 in which @xmath60 .
then the product @xmath61 is defined to be the tree pair @xmath62 which may be unreduced .
this method is used to compute the distance between elements @xmath18 and @xmath59 , namely , @xmath63 .
let @xmath64 and @xmath65 be metric spaces , and @xmath66 and @xmath67 be constants . a _ @xmath68-quasi - isometric embedding _
@xmath69 is a map satisfying the following property : @xmath70 for all @xmath71 , where @xmath72 ( resp .
@xmath73 ) represents the metric in @xmath64 ( resp .
@xmath65 ) .
when considering a quasi - isometric embedding between groups , we use the word metric on each group induced by a particular set of generators .
a quasi - isometric embedding is a _ quasi - isometry _ if there is a constant @xmath74 so that @xmath75 .
the definition of a quasi - isometry does not include any algebraic requirements , in fact , a quasi - isometry need not be a continuous map .
in addition , a quasi - isometry is not required to preserve any algebraic structure at all .
below we give several examples of quasi - isometric embeddings of @xmath76 into @xmath0 , for nonnegative integers @xmath28 and @xmath32 , which , though homomorphisms , are algebraically cumbersome .
however , they are easily described geometrically by the way they alter the tree pair diagram of an element @xmath9 .
we call these _ geometric _ quasi - isometric embeddings .
let @xmath77 be a finitely generated group and @xmath78 a finitely generated subgroup of @xmath77 .
one can ask if the inclusion of @xmath78 into @xmath77 is a quasi - isometric embedding .
the inclusion will be a quasi - isometric embedding if and only if the distortion function of the group is bounded .
define the _ distortion function _
@xmath79 of @xmath78 in @xmath77 by @xmath80 we will use the distortion function to show that the examples we give below are quasi - isometric embeddings .
we begin with an example of a quasi - isometric embedding @xmath81 which is easily understood either algebraically or geometrically .
there is a _ shift map _ @xmath82 defined on thompson s group in the infinite presentation @xmath83 which increases the index of each generator by @xmath20 .
for example , if @xmath84 , then @xmath85 .
thus if @xmath18 is in normal form , so is @xmath86 .
we show that any power of @xmath1 is a quasi - isometric embedding .
let @xmath93 , and @xmath94 .
we first describe the relationship between the trees @xmath95 and @xmath96 .
the smallest possible index of a generator in @xmath97 is @xmath32 .
thus , the leaf exponents of all leaves numbered @xmath21 through @xmath98 in @xmath96 must be @xmath21 .
the trees @xmath96 are now easily deduced from @xmath95 .
the tree @xmath99 must have an empty left subtree , @xmath98 right carets with exposed left carets beginning at the root , and then the tree @xmath26 .
similarly , @xmath100 is the tree having an empty left subtree , @xmath98 right carets with no left subtrees to the right of the root , and then the tree @xmath27 . figure [ fig : phi2w ] gives an example of he tree pair diagram of @xmath101 , using the word @xmath18 depicted in figure [ fig : exptree ] .
we now show that @xmath92 is a quasi - isometric embedding .
recall from theorem [ thm : bound ] that @xmath102 where @xmath42 is the number of carets in @xmath26 .
since @xmath103 , we obtain the same inequality for @xmath104 : @xmath105 combining the two above inequalities yields the desired bound : @xmath106 since @xmath32 is a constant , it follows that the distortion function is bounded and thus @xmath92 is a quasi - isometric embedding .
given the simple geometric representation of the shift map @xmath1 , it is natural to define a `` reverse shift map '' , which we denote @xmath107 , in which the original trees become left subtrees of the root caret rather than right subtrees .
see , for example , figure [ fig : psi2w ] , which depicts the tree pair diagram for @xmath108 .
geometrically , it is completely natural to consider @xmath107 as well as @xmath1 . however , in the literature , @xmath107 is rarely mentioned , likely because it is algebraically cumbersome , unlike @xmath1 , in the following way .
if @xmath9 is written in normal form , then the normal form of @xmath86 is easily determined , whereas for @xmath107 the situation is quite different .
let @xmath38 and @xmath109 . then leaves with exponent @xmath21 which are the exposed left leaves of right carets in @xmath26 or @xmath27 will have leaf exponent @xmath20 in @xmath99 and @xmath100 , respectively , and thus cause new generators to appear in the normal form of @xmath110 .
it is difficult to predict solely from the normal form of @xmath18 which additional generators will appear in the normal form of @xmath110 . for example if @xmath111 then @xmath112 . to apply the method of proof used above in theorem [ thm : phi ]
, we give the generators of a subgroup of @xmath0 isomorphic to @xmath115 . once we do this
, it is clear from the tree pair diagrams for @xmath18 and @xmath116 defined above that @xmath96 has @xmath32 more carets than @xmath95 , and we obtain inequalities for @xmath117 identical to those for @xmath104 given in the proof of theorem [ thm : phi ] .
thus @xmath118 will be a quasi - isometric embedding of @xmath0 into @xmath0 .
it is easily checked that @xmath107 is a group homomorphism , and thus @xmath115 is generated by @xmath119 and @xmath120 . after forming the tree pair diagrams representing these generators , using the exponent notation of [ sec : exponents ]
, we see that @xmath121 and @xmath122 which gives an algebraic description of @xmath107 .
similarly , we can compute @xmath123 substituting this formula into the normal form of @xmath9 , we obtain an expression for @xmath116 in the infinite generating set which can be simplified to obtain the normal form for @xmath116 .
the maps @xmath92 and @xmath118 are shown to be quasi - isometric embeddings because an element and its image are represented by tree pair diagrams which differ by a finite number of carets .
this idea can be generalized as follows .
let @xmath124 denote the composition @xmath125 where each @xmath126 is either @xmath1 or @xmath107 and we compose from left to right , so that @xmath127 .
a _ clone subgroup _ is the image of @xmath0 under @xmath124 , that is , @xmath128 .
the simplest examples of clone subgroups are @xmath89 and @xmath115 .
clone subgroups can be understood in a number of ways .
let @xmath124 be as above , and consider the clone subgroup @xmath128 .
we can describe this subgroup using a binary address .
a node in a binary tree is given an address using the following inductive method .
the root node has the empty label .
given a node with label @xmath132 , the left child of the node is labelled @xmath133 and right child of the node labelled @xmath134 .
for example , the right child of the right child of the left child of the right child of the root has address @xmath135 . given @xmath124 as above , and @xmath136 ,
let @xmath137 . from the definitions of @xmath138 and @xmath124
, we know that the tree @xmath26 will be a subtree of @xmath99 , and @xmath27 will be a subtree of @xmath100 .
we consider the address @xmath139 of the root caret of @xmath26 as a subtree of @xmath99 and use @xmath132 to give an address " for the clone subgroup @xmath128 .
namely , we can uniquely refer to @xmath128 as @xmath140 .
for example , the subgroup @xmath141 is the image @xmath142 . we can also describe clone subgroups by their representation as piecewise linear homeomorphisms of the unit interval .
each dyadic subinterval of the form @xmath143 $ ] is equivalent to the standard unit interval @xmath144 $ ] by an affine map with dyadic coefficients . for a fixed clone subgroup @xmath140 where @xmath132 has length @xmath145
, there is a proper subinterval @xmath146 of @xmath147 $ ] which contains the @xmath148- and @xmath149- coordinates of all the breakpoints of elements of @xmath140 , and such that all elements of @xmath140 are the identity map outside of the interval @xmath150 .
the length of @xmath150 is @xmath151 and the endpoints of @xmath150are @xmath152 and @xmath153 .
the endpoints can be computed easily from the address @xmath132 . furthermore , any element of @xmath0 whose breakpoints all lie in the interval @xmath154 $ ] will be an element of @xmath140 .
for the example subgroup @xmath141 , the dyadic interval containing the all breakpoints of its elements is @xmath155 $ ] .
it is easy to see that any clone subgroup @xmath156 is isomorphic to @xmath0 .
if @xmath38 and @xmath137 , we know by definition that the tree pair @xmath25 is reduced . from the definitions of @xmath1 and @xmath107 , and thus @xmath124 , no additional reduction can occur when the tree pair @xmath157 is formed .
thus each element of @xmath0 produces a unique element of @xmath140 , and it is clear that @xmath129 and @xmath131 must generate @xmath140 . the clone subgroup @xmath140 is determined by @xmath158 where each @xmath126 is either the map @xmath1 or @xmath107 .
we know from theorems [ thm : phi ] and [ thm : psi ] that @xmath1 and @xmath107 are @xmath159-quasi - isometric embeddings of @xmath0 into @xmath0 .
geometrically , we see that if @xmath93 and @xmath160 , the trees @xmath161 and @xmath162 differ by @xmath32 carets .
thus , we again get that @xmath163 for all @xmath18 in @xmath0 , which shows that @xmath124 is also a quasi - isometric embedding , with constants @xmath164 .
thus @xmath140 is quasi - isometrically embedded .
we now show that @xmath0 contains a family of quasi - isometrically embedded subgroups of the form @xmath2 for @xmath165 .
these examples include those of burillo @xcite , who presents a family of subgroups of @xmath0 of the form @xmath166 which are quasi - isometrically embedded .
guba and sapir @xcite , using the approach of diagram groups , also show that @xmath2 occur as quasi - isometrically embedded subgroups of @xmath0 and furthermore show that all abelian subgroups and centralizers of elements are quasi - isometrically embedded .
we present an alternate geometric proof that some concrete geometric realizations of the subgroups @xmath2 are quasi - isometrically embedded , using the shift maps described above .
burillo proves these results by bounding the distortion function using a technique he develops to estimate the length of @xmath18 in the word metric corresponding to the generating set of the presentation @xmath37 as a function of the normal form of @xmath18 .
we present an alternate geometric proof of this theorem and corollary before presenting the proof of theorem [ thm : products ] .
we note that guba and sapir @xcite show that many subgroups of the form @xmath2 are quasi - isometrically embedded using methods for diagram groups .
consider the quasi - isometric embedding @xmath174 of theorem [ thm : burillo ] .
we identify geometrically the subgroup @xmath4 of @xmath0 generated by the elements @xmath175 and @xmath167 .
drawing the tree pairs representing @xmath176 and @xmath167 , we see that they generate a clone subgroup @xmath177 of @xmath0 .
the tree pair diagrams representing these generators are given in figure [ fig : fxz ] .
an element of @xmath177 , when viewed as an element of the larger group @xmath0 , has no instances of @xmath178 or @xmath179 in its normal form , and therefore lies in the subgroup @xmath180 .
we also know that @xmath180 is generated by @xmath181 and @xmath182 .
it is easily seen that @xmath183 , and we draw the tree pair diagram @xmath157 representing this element as follows .
the form of the tree @xmath100 is clear , and in determining @xmath99 , we see that the left subtree of the root caret is empty , and the right subtree of the root caret has as string of @xmath184 carets of type @xmath47 , a single caret of type @xmath46 and a single caret of type @xmath52 .
we are really interested in the fact that the right subtree of the right child of the root caret in @xmath95 is empty .
this can be easily determined by applying the following lemma from @xcite to the substring @xmath185 of @xmath186 , since this word determines the right subtree of the root of @xmath99 . since @xmath190 , we see that the right subtree of the right child of the root caret of @xmath26 is empty , as is the corresponding subtree in @xmath27 .
this fact guarantees that an element in @xmath191 and @xmath192 commute , as is easily seen from the definition of group multiplication of pairs of trees , or simply by noting that subintervals where these elements are different from the identity map are disjoint .
we apply the method of counting carets used above in [ sec : phi ] and [ sec : psi ] to give an alternate proof of theorem [ thm : burillo ] .
we first note that @xmath177 is the image @xmath180 it is easily seen that @xmath193 and @xmath194 .
the two parts of @xmath18 are in different subtrees , excepting an overlap of two carets at the root and the right child of the root .
so when we count carets , we get @xmath195 . thus @xmath196 and @xmath197 thus the distortion function is bounded and the map @xmath174 is a quasi - isometric embedding . to extend this proof to a geometric interpretation of corollary [ cor : burillocor ] , notice that the first @xmath98 terms listed in the corollary , @xmath198 , commute with each other .
this is easily seen from the tree pair diagrams representing these elements or simply by noting that these elements lie in clone subgroups that intersect only in the identity .
the last two generators of the corollary , @xmath199 and @xmath200 , generate a clone subgroup @xmath201 , whose address is @xmath202 one s .
_ proof of theorem [ thm : products ] . _ fix a pair of integral @xmath3 .
we now produce generators for subgroups of the form @xmath2 of @xmath0 and show that these subgroups are quasi - isometrically embedded .
each copy of @xmath0 comprising the @xmath203 will be given by a distinct clone subgroup .
we can let @xmath204 be binary addresses of nodes in a binary rooted tree , subject to the condition that pairwise , no @xmath205 is a prefix of @xmath206 .
for a fixed @xmath28 there are infinitely many choices of prefix sets satisfying this requirement .
then for each @xmath53 with @xmath207 , we have that @xmath208 is a clone subgroup of @xmath0 , and by construction , no two of these clone subgroups intersect in non - identity element .
thus the product @xmath209 is isomorphic to @xmath203 .
the final factor clone subgroup @xmath210 will contain the abelian factors .
let @xmath211 be the product of maps @xmath1 and @xmath107 corresponding to the binary address @xmath212 , and thus @xmath213 .
using the elements of corollary [ cor : burillocor ] , we see that @xmath214 will generate a subgroup of @xmath0 isomorphic to @xmath173 .
first we note that all the factors have been chosen to lie in clone subgroups which intersect only in the identity , so they commute .
since each factor lies in a clone subgroup @xmath208 there are constants @xmath217 so that @xmath218 , where @xmath219 is product of @xmath217 maps . similarly , from the tree pair diagrams , it is easy to see that @xmath220 . thus @xmath221 combining the above inequalities with theorem [ thm : bound ] , we obtain the following upper bound .
@xmath222 @xmath223 @xmath224 @xmath225 where @xmath226
. the lower bound @xmath227 is obtained analogously , for the appropriate constant @xmath228 depending on @xmath229 .
thus the map @xmath230 is a quasi - isometric embedding .
since there are infinitely many choices for @xmath229 , we have produced an infinite family of quasi - isometrically embedded subgroups of @xmath0 of the form @xmath2 for any integral pair @xmath231 with @xmath232 .
|
we use geometric techniques to investigate several examples of quasi - isometrically embedded subgroups of thompson s group @xmath0 . many of these are explored using the metric properties of the shift map @xmath1 in @xmath0 .
these subgroups have simple geometric but complicated algebraic descriptions .
we present them to illustrate the intricate geometry of thompson s group @xmath0 as well as the interplay between its standard finite and infinite presentations .
these subgroups include those of the form @xmath2 , for integral @xmath3 , which were shown to occur as quasi - isometrically embedded subgroups by burillo and guba and sapir .
|
* in this supplemental material we discuss the preparation of our fermionic quantum gas ( s1 ) , the detection after tof and _ in situ _ ( s2 ) and the data analysis ( s3 ) . the theoretical model is introduced in ( s4 ) and the linearized calculation is presented in ( s5 ) . for the tensor expansion in the @xmath0 system see ( s6 ) . *
we sympathetically cool about @xmath76 spin - polarized @xmath14k atoms in the state @xmath15 and @xmath77 to @xmath78 using @xmath2rb in a magnetic trap .
afterwards we transfer the atoms to a crossed circular - elliptical optical dipole trap with @xmath79 .
the @xmath80 radii are @xmath81 for the circular beam and @xmath82 and @xmath83 of the elliptical beam , where the tight focus is in the vertical direction . using rf - pulses and rf - sweeps , an equal mixture of the two hyperfine states @xmath84 and @xmath85
is prepared and evaporatively cooled in the trap by a @xmath86 exponential intensity ramp .
the final trapping frequencies are typically @xmath87 with @xmath88 particles at temperatures of @xmath89 .
to obtain the experimental data , we used two different detection protocols , which have complementary advantages . in most of the experiments , we applied time of flight imaging , where we switch off all optical potentials , and perform a stern - gerlach separation of the different spin states within the free expansion time of @xmath90 .
the atoms are detected via resonant absorption imaging and the center - of - mass is calculated for each of the separated clouds individually .
this method has the advantage that all spin - components can be detected simultaneously , but the disadvantage that all spatial modes but the dipole mode are effectively washed out , since the tof mixes momentum and spatial components of these modes . to obtain more information on the spatial modes of the excited spin - waves we employed an _ in situ _ detection protocol , allowing us to observe , e.g. , the breathing dynamics in fig .
2(b ) . instead of separating the different spins by a stern - gerlach procedure in tof , we use microwave pulses at @xmath91 , to transfer all but one single component to the @xmath92 manifold of @xmath14k , which is off - resonant to the detection light , and
consequently the transferred atoms do not appear in the absorption images .
afterwards we switch off all magnetic fields and optical potentials for @xmath93 to detect the atomic sample .
to record the time evolution of all four components , it is therefore necessary to repeat the full measurement four times .
for every time step of the spin - wave dynamics , we determine the center - of - mass of each of the spin components .
we extract the oscillation frequencies @xmath94 by fitting an exponentially damped cosine of the form @xmath95 with oscillation amplitude @xmath96 , damping rate @xmath97 , a phase shift @xmath98 and a constant offset @xmath99 . for the dipole and octupole fraction in fig .
4(c ) , we use a slightly different method .
we take the spatial average of the absolute value of the projected multipoles @xmath100 for each time step as @xmath101 where @xmath66 is the pixel index of the image and @xmath102 is the number of pixels . by this procedure ,
the sign - information of the projection are lost and therefore , we fit the resulting time series by @xmath103 and report the amplitude @xmath96 .
we describe the multi - component fermionic system using the single - particle wigner - function , with its time evolution given by the semiclassical kinetic equation @xmath104\nonumber\\+\frac12\left\lbrace \partial_pw(z , p),\partial_zv(z)\right\rbrace \label{eq:1}\end{aligned}\ ] ] with the short notation @xmath105 for the single - particle part of motion .
we do not explicitly include the magnetic field gradient , since it is used for the creation of the spin - wave but not necessary for its propagation once created .
the derivation of this equation is outlined in previous works @xcite .
one basically starts from the full mean - field dynamics of the wigner function and applies a semiclassical approximation by neglecting terms involving higher order derivatives with respect to position and momentum . to leading ( zeroth ) order ,
the mean - field interaction gives rise to the commutator and the next ( first ) order is given by the anticommutator .
we have neglected here even higher orders terms .
the important term here is the commutator which is absent in the spinless case and describes interaction driven coherent spin dynamics .
it is also dominant with respect to the anticommutator and we neglect the latter in our analytical studies ( while we keep it in the numerical simulations ) .
the kinetic equation ( [ eq:1 ] ) is nonlinear , since the mean - field potential depends on the wigner - function itself and its matrix elements are defined as @xmath106 with the coupling constants defined as @xmath107 for arbitrary spin @xmath108 with the corresponding clebsch - gordan coefficients @xmath109 . here
@xmath110 denotes the interaction strength for the scattering of two particles with total spin @xmath111 . for a real spin-3/2 system there
are only scattering channels for @xmath112 , but in the numerical simulations of ( [ eq:1 ] ) we take into account the values of @xmath113 for spin 9/2 in the sub - manifold of @xmath114 , which also depend on @xmath115 @xcite . since in the experiment spin - waves along the @xmath65-axis are created , an effective one - dimensional kinetic equation is sufficient to describe these spin - waves .
this is obtained by integrating out the transversal degrees of freedom from the three - dimensional case .
the coupling constants for the 3d case , @xmath116 with s - wave scattering lengths @xmath117 , are modified by the local transversal density profile as @xmath118 .
this modification depends on @xmath65 but only weakly so we approximate it with the central density @xmath119 .
frequencies for small amplitude spin - waves in the linear regime , such as close to the minimum in fig .
3(a ) , can be obtained by linearizing the kinetic equation ( [ eq:2 ] ) . for this , we consider small changes with respect to the stationary state @xmath120 , namely @xmath121 .
thus we investigate the spin - waves for short times and small amplitudes . the mean field , as a function of density likewise expands as @xmath122 and we substitute both expressions into ( [ eq:1 ] ) without the anti - commutator , to obtain the linearized kinetic equation @xmath123+\left[w^0,\delta v,\right]\right ) \label{eq:3}\end{aligned}\ ] ] our next step to explicitly study different kinds of spin - wave modes is the so called _ moment method _ @xcite .
the thinking behind it is to look at the time evolution of moments of the position and momentum operators .
the @xmath35-th moment of such an operator in the phase - space representation is the expectation value @xmath124 . expanding the kinetic equation in moments means taking into account only different modes of spin - waves up to a certain order ,
e.g. dipole modes for @xmath51 , breathing dynamics for @xmath53 and further .
first we need to find a suitable expression for the stationary state @xmath125 , the state of the system before a spin - wave is excited by applying a gradient . with the preparation scheme in mind ,
considering the preparation pulse to be infinitely short in time , we approximate @xmath126 as a product of spin and orbital degrees of freedom @xmath127 where @xmath128 is a matrix in spin space determined by the preparation pulses and @xmath129 is the phase - space distribution for a two - component fermi gas onto which the pulse is applied .
this distribution is given for a non - interacting gas in a harmonic trap in local - density approximation as @xmath130 . in principle , corrections to this distribution arise due to interactions but in our case the mean - field energy is small compared to kinetic and potential energy and we neglect them .
the mean field @xmath131 then depends on the density distribution @xmath132 , and we introduce the short notation @xmath133 . for the frequencies of the dipole modes we expand @xmath134 and @xmath135 into moments of position and momentum up to first order only @xmath136 we take the zeroth and first moments of @xmath137 of @xmath134 in ( [ eq:4 ] ) and obtain their relationship to @xmath138 @xmath139 with particle number @xmath140 , as well as @xmath141 and @xmath142 .
we substitute expressions ( [ eq:6 ] ) into the linearized equation ( [ eq:3 ] ) , then take again the zeroth and first moments of @xmath137 to obtain three equations for the matrices @xmath143 respectively @xmath144+\left[a,\tilde m\right]\right)\,,\\ \label{eq:8 } \partial_t b - m\omega_z^2 c=&\frac{i_1}{i\hbar}\left(\left[m,\tilde b\right]+\left[b,\tilde m\right]\right)\,,\\ \label{eq:9 } \partial_t c+\frac1 m b=&\frac{i_2}{i\hbar}\left[c,\tilde m\right]\,,\end{aligned}\ ] ] with coefficients @xmath145 , @xmath146 and @xmath147 .
the first equation is decoupled from the others and does not lead to spatial dynamics , so we discard it . equations ( [ eq:8 ] ) and ( [ eq:9 ] ) describe dipole oscillations of all spin components of the wigner function and the frequencies can be obtained by a fourier transform @xmath148 and solving the eigenvalue equations , similar to the procedure for a spin 1 bose gas in refs .
@xcite . for the sake of simplicity
we now demonstrate this for the special case of su(n)-symmetry ( @xmath149 ) assuming all scattering lengths to be equal . for the @xmath150 subsystem of @xmath14k considered in the experiments ,
our numerical comparison shows only small differences in the spin - wave behavior between this case and the true scattering parameters . in the su(4)-symmetric case with
all scattering lengths equal , @xmath151 , the coupling constants are of the particularly simple form @xmath152 .
for all matrices with a tilde in equations ( [ eq:7 ] ) , ( [ eq:8 ] ) and ( [ eq:9 ] ) we obtain the simple expression @xmath153 .
further , the r. h. s. of eqs .
( [ eq:7 ] ) and ( [ eq:8 ] ) are zero and the equations for @xmath154 , written in matrix form reduce to @xmath155\end{aligned}\ ] ] which decouple trivially . after a fourier transform @xmath148 we obtain the eigenvalue equation @xmath156 \label{eq:10}\ ] ] where we have introduced the mean - field frequency @xmath157 .
for all ( nematic ) initial spin states @xmath100 considered here the solutions of ( [ eq:10 ] ) are either the trivial case @xmath158 that corresponds for instance to oscillations of the entire atomic cloud in the trap or @xmath159 which describe the spin - wave propagation .
the methods described thus far in this supplemental material are valid for any value of the spin .
we now focus on the @xmath0 case in order to demonstrate the decoupling of the spin - nematic component . in previous studies of the @xmath1 case @xcite the kinetic equation and wigner - function
were rewritten in terms of the pauli matrices @xmath160 which form a complete basis for hermitian @xmath161-matrices and allow to identify the vector - part of the mean field as an effective magnetic field .
for @xmath0 we define a similar basis @xmath162 comprising the identity matrix @xmath163 , the three spin operators @xmath164 , five nematic operators @xmath165 and seven octupole operators @xmath166 . for
the nematicity we select all traceless symmetric matrices quadratic in @xmath167 , this set of 16 hermitian matrices forms an orthonormal basis set with respect to the trace @xmath170 so we can expand the wigner function into @xmath171 with coefficients @xmath172 . in this basis
the coupling constants of the mean field ( [ eq:2 ] ) have the form @xmath173 where the coefficients @xmath174 depend on the coupling constants @xmath110 . in a real @xmath0 system , @xmath175 , which is a result of the so(5 ) symmetry of the system @xcite .
analogous to ( [ eq : w_tensor ] ) the mean field can be expanded as @xmath176 and we find , that in this basis each tensor component @xmath177 of the mean field is proportional to the respective component of the wigner function with the coefficients @xmath174 @xmath178 we now insert equations ( [ eq : w_tensor ] ) , ( [ eq : v_tensor ] ) and ( [ eq : v_tensor_prop ] ) into the kinetic equation ( [ eq:1 ] ) , neglecting the anticommutator for simplicity and obtain @xmath179\ , .
\label{eq:11 } \end{split}\ ] ] the time evolution for each tensorial component @xmath180 of the wigner function is obtained from ( [ eq:11 ] ) by taking the trace with respect to the @xmath162 basis as in ( [ eq : trace ] ) .
on the right hand side we define @xmath181\right)\ , , \label{eq : lambda}\ ] ] such that ( [ eq:11 ] ) becomes @xmath182 equation ( [ eq : lambda ] ) simply states the decomposition of the commutators of tensor operators in the tensor basis itself .
apart from the trivial case that @xmath183 commutes with all other tensors we know that by construction @xmath184=i\sqrt{5}\sum_k\epsilon_{ijk}t_1^k$ ] and thus @xmath185 . formulating an explicit expression for all values of @xmath186 is not straightforward
so we restrict ourselves to provide those values of @xmath187 for which the @xmath186 are non - zero for some values of @xmath188 : @xmath189{rl } l \vline & ( l ' , l'')\\ \hline 0 \vline & ( 0,0)\\ 1 \vline & ( 1,1 ) , ( 2,2 ) , ( 3,3)\\ 2 \vline & ( 1,2 ) , ( 2,1 ) , ( 2,3 ) , ( 3,2)\\ 3 \vline & ( 2,2 ) , ( 3,3 ) , ( 1,3 ) , ( 3,1 ) \end{tabular } \label{eq : tab}\ ] ] these relations show , which tensor components @xmath190 contribute to the time evolution of the tensor component @xmath35 . to illustrate the these findings in a matrix representation we introduce the symbolic notation @xmath191 suppressing the @xmath40-indices , and insert the relations ( [ eq : tab ] ) into the equations of motion for @xmath192 .
this brings us to equation ( 2 ) of the manuscript @xmath59 + [ w_2,v_2 ] + [ w_3,v_3]\right)\,,\\ \partial_t w_2 & \cong \partial_0 w_2 + \frac{1}{i\hbar } \left([w_2,v_1+v_3 ] + [ w_1+w_3,v_2]\right)\,,\\ \partial_t w_3 & \cong \partial_0 w_3 + \frac{1}{i\hbar } \left([w_3,v_1 ] + [ w_1+w_3,v_3 ] + [ w_2,v_2]\right ) , \end{split}\ ] ] where the decoupling of the nematic spin tensor observed in the experiment is evident , since no term of the form @xmath193 $ ] exists on the right hand side of the equation for @xmath194 .
see online supplemental material for details on the quantum gas preparation , detection and data analysis , as well as in - depth discussion of the theoretical model , the method of moments and the tensor decomposition .
|
we report on the detailed study of multi - component spin waves in an @xmath0 fermi gas where the high spin leads to novel tensorial degrees of freedom compared to @xmath1 systems .
the excitations of a spin - nematic state are investigated from the linear to the nonlinear regime , where the tensorial character is particularly pronounced . by tuning the initial state we engineer the tensorial spin - wave character , such that the magnitude and the sign of the counterflow spin currents are effectively controlled .
a comparison of our data with numerical and analytical results shows good agreement .
spin - interaction driven phenomena are crucial for the behavior of many quantum systems , e.g. , ferromagnets @xcite and high - temperature superconductors @xcite and they are also relevant in spintronics applications @xcite .
apart from condensed matter systems with an electronic spin of @xmath1 , dilute atomic gases show a wealth of novel spin excitations , where the spin is provided by the internal hyperfine structure of the atoms .
pioneering experiments with hydrogen @xcite and helium @xcite showed the existence of transverse spin waves , which arise from intrinsic spin - exchange interactions @xcite .
longitudinal spin waves in two - component mixtures have been observed in non - condensed bosonic @xmath2rb gases @xcite . for weakly interacting fermions ,
slow spin currents were reported near the zero - crossing of a feshbach resonance @xcite and the interaction - induced damping of dipole oscillations was studied @xcite .
prominent examples for spin dependent phenomena in strongly interacting fermionic systems are the miscibility of spin mixtures @xcite and the quest for itinerant ferromagnetism @xcite . in contrast to conventional two - component systems ,
the hyperfine structure of many atoms also allows for spinor gases with @xmath3 , which offer a whole new set of possibilities to study spin - dependent phenomena @xcite .
this includes spin - changing collisions @xcite , hidden interaction symmetries @xcite , spontaneous domain formation @xcite , the existence of spin - nematic states @xcite , novel superfluid phases @xcite and su(n ) degenerate ground states @xcite . for fermionic atoms , @xmath0 constitutes the simplest realization of a high - spin system and has been thoroughly studied theoretically , being a model system for all higher spins @xcite . in this letter
we demonstrate the controlled generation of spin waves in a quantum degenerate fermi gas with pseudo - spin @xmath0 .
we experimentally study the properties of these fundamental collective spin excitations for a wide range of parameters .
the results are explained within a generalized semiclassical mean - field theory ( smft ) for fermionic atoms with a high spin of @xmath4 , being an extension of the collisionless boltzmann equation used to describe conventional @xmath1 systems @xcite .
spin waves in such high - spin systems are predicted to exhibit very complex and novel properties , which can be most intuitively understood in the language of irreducible spherical tensors ( for bosonic gases , see @xcite ) .
while for @xmath1 it is sufficient to use the identity and the three spin matrices @xcite , the description of higher spins additionally requires higher - order tensors , such as the the nematic and octupole tensor for @xmath0 .
we have investigated spin - wave excitations from the linear regime , where the oscillation frequency is minimal , to the non - linear regime , where the spin - wave frequency strongly depends on the excitation amplitude .
the use of the tensor basis allows to directly observe the effect of the nonlinear mode - coupling , which leads to the excitation of breathing modes in the spin - nematic component .
moreover , we demonstrate the controlled manipulation of the spin - wave composition by engineering the coherences of the initial state . in that way
, the spin current for two of the four components can be reversed changing the spin - wave character from spin - octupole to spin - vector .
our results illustrate the high degree of control that can be exerted on spin waves in high - spin fermi gases .
the good agreement with the theoretical results shows , that our smft well describes high - spin fermi gases in the quantum degenerate regime .
the combined experimental and theoretical findings pave the way towards novel schemes for atom spintronics using the intrinsic high spin . and @xmath5 ( left ) and the resulting coherent superposition of all four components after a resonant rf - pulse ( right ) .
diagonal elements @xmath6 are real and represent the populations .
off - diagonal elements are complex numbers @xmath7 and include the phase @xmath8 between different components .
plotted is only the absolute value @xmath9 ( b ) sketch of the local phase across the fermi gas after pulses with different magnetic field gradients .
( c ) @xmath10 component of the @xmath11 tensor operators @xmath12 for @xmath0 in comparison to the corresponding spherical harmonics @xmath13.,width=325 ] our measurements are performed in a quantum degenerate gas of @xmath14k in the @xmath15 hyperfine manifold .
we initially evaporate a balanced mixture of @xmath16 and @xmath17 to quantum degeneracy in an elongated , spin - independent optical dipole trap @xcite .
the final trapping frequencies are @xmath18 . at this point
, we apply a radio - frequency ( rf ) pulse to create a coherent superposition with the states @xmath19 [ fig .
[ fig1](a ) ] .
we initialize the spin waves by applying a small magnetic field gradient up to a few @xmath20 for @xmath21 , which leads to a phase spiral for coherent superpositions of different spin components as sketched in fig .
[ fig1](b ) .
while these coherent superpositions are initially still spin - polarized locally , the phase - twist allows for interactions in a trapped gas where the external potential induces spatial dynamics @xcite . in general , the resulting mean - field interaction couples the spin degrees of freedom to different modes of the external trap leading to the emergence of spin waves .
we detect the spin current using absorption imaging either _ in situ _ or after @xmath22 time - of - flight ( tof ) with a stern - gerlach separation of the spin components @xcite . in fig .
[ fig2](a ) we show a typical example for an @xmath0 spin wave initialized by a @xmath21 gradient of @xmath23 .
the measurements reveal oscillatory spin currents in all four spin components .
we observe a time - independent total density , meaning that the spin waves constitute counterflow spin currents without an accompanying net mass transport .
in particular , note the inverted flow direction of @xmath24 ( @xmath5 ) with respect to @xmath25 ( @xmath26 ) , which is a clear indication of the new tensorial degrees of freedom as discussed later . pulse with a magnetic field gradient of @xmath23 . shown
are the column densities at different times after the excitation .
( b ) deviation from the initial population of the @xmath27 component of the vector ( @xmath28 ) , nematic ( @xmath29 ) and octupole ( @xmath30 ) component .
the vector and octupole component show spatial dipole oscillations , while the nematic component clearly exhibitsbreathing dynamics .
( c , d ) numerical calculation for the parameters of ( a , b).,width=325 ] for the theoretical description of high - spin fermi gases , we generalize a 1d smft @xcite previously used to explain spin - wave phenomena in thermal fermionic and bosonic systems with effective spin @xmath31 @xcite and to predict the spin - wave dynamics in thermal bosonic @xmath32 gases @xcite .
the multi - component system is described in a mean - field fashion by a single - particle density matrix ( spdm ) in the form of a 1d wigner function @xmath33 with spin indices @xmath34 and @xmath35 .
the semiclassical equations of motion take the form of a boltzmann - equation in the collisionless regime . to leading order they read @xmath36\ , , \label{eq : p1}\ ] ] assuming a spin - independent external harmonic trap . here
, @xmath37 is the effective mean - field potential with the spin - dependent coupling constants @xmath38 @xcite , @xmath39 captures the time evolution due to the harmonic trap and the kinetic energy , @xmath40 is the mass of @xmath14k , and @xmath41 $ ] indicates the commutator in spin space . in the simulations , also higher - order terms of the mean - field interactions are taken into account , leading to very small deviations only @xcite . for an @xmath0 system ,
the wigner function in spin space is a @xmath42 spdm , where the diagonal elements @xmath6 represent the absolute population of the spin components and the off - diagonal elements @xmath7 represent the single - particle coherences between different components . to induce a time evolution of the populations @xmath6 , it is sufficient to spatially vary the phases @xmath8 of the off - diagonal elements @xmath43 , since both are coupled via the commutator in ( [ eq : p1 ] ) .
figure [ fig2](c ) shows numerical results for the exact experimental parameters , which are in good agreement with the measured results .
this demonstrates the capability of the smft to quantitatively describe interacting high - spin fermi gases in the quantum degenerate regime . .
negative amplitudes denote an inverted initial direction .
solid lines are numerical data for each component .
all error bars solely correspond to fit errors , representing one standard deviation .
the experimental amplitudes which are taken after tof and the numerical amplitudes calculated _ in situ _ are rescaled onto each other by a global factor .
( c ) sketch of the phase windings across the atom cloud for different gradient pulses.,width=325 ] to obtain deeper insight into the underlying physical processes , let us at this point briefly recall the description of spin in the language of irreducible spherical tensors @xmath12 , which simplify the equations of motion drastically .
the @xmath12 transform invariantly under rotations and therefore can be ordered by a total spin @xmath35 and a magnetic quantum number @xmath44 .
most common are the spherical harmonics for orbital angular momentum [ fig .
[ fig1](c ) ] and the spin - vector @xmath45 where @xmath46 are the pauli matrices . decomposing the wigner function ( mean field ) in the tensor basis as @xmath47 [ @xmath48 ,
the @xmath10 components describe the occupations whereas all other components describe coherences .
any ( pseudo ) @xmath1 system can be conveniently described by the identity matrix ( @xmath49 ) describing the total density , and the spin - vector @xmath50 ( @xmath51 ) describing the magnetization and the coherences @xcite . to describe the physics of larger spins
it is necessary to include higher - order tensors . in a spin @xmath52 system , as discussed here , the spin - nematic tensor ( @xmath53 ) and the spin - octupole tensor ( @xmath54 ) must be included ( see fig .
[ fig1](c ) and @xcite ) .
figures [ fig2](b ) and ( d ) show the time evolution of the @xmath10 component of the @xmath55 tensors for the experimental and numerical data of fig .
[ fig2](a ) and ( c ) , respectively .
note the predominantly breathing dynamics of the spin - nematic component , qualitatively different from the spatial dipole oscillations in the spin - vector and spin - octupole component .
this results from a linear decoupling of the nematic component due to the rotational symmetry which can be understood by inserting the decomposition @xmath56 ( @xmath57 ) into eq .
( [ eq : p1 ] ) .
the rotational symmetry of the interactions leads to the particular simplification that @xmath58 . omitting the @xmath40-index for simplicity the equations of motion
read @xcite : @xmath59 + [ w_2,v_2 ] + [ w_3,v_3]\right)\,,\\ \partial_t w_2 & \cong \partial_0 w_2 + \frac{1}{i\hbar } \left([w_2,v_1+v_3 ] + [ w_1+w_3,v_2]\right)\,,\\ \partial_t w_3 & \cong \partial_0 w_3 + \frac{1}{i\hbar } \left([w_3,v_1 ] + [ w_1+w_3,v_3 ] + [ w_2,v_2]\right ) .
\label{eq : p2 } \end{split}\ ] ] the structure of eqs .
( [ eq : p2 ] ) together with the relation @xmath58 has several important consequences .
first , the total density @xmath60 is not altered by the phase spiral , since its time derivative does not depend on the off - diagonal elements ; it remains constant as we observed in the experiment .
second , the time derivative of the nematic tensors @xmath61 is proportional to the vector and octupole components ( @xmath62 and @xmath63 ) , but does not depend on a term @xmath64 $ ] . this is a result of time - reversal symmetry and leads to a linear decoupling of the nematic component , in the sense that a purely nematic state does not support nematic excitations to first order . in the nonlinear regime
, however , where vector and octupole excitations possess a large amplitude , nematic excitations are created via nonlinear mode - coupling .
this leads to the weak breathing dynamics of the nematic component visible in fig .
[ fig2 ] , where a purely nematic state was initially prepared .
the discussion above demonstrates the improved insight into high - spin spin waves granted by the irreducible spherical tensor description . to analyze the behavior of the system for different excitation amplitudes
, we applied different gradient strengths during the initialization of the spin wave @xcite .
this corresponds to a change of the initial phases @xmath8 in the spdm , while the initial coherence amplitudes @xmath9 are kept constant . in fig .
[ fig3 ] experimental results are compared to numerical calculations and show good agreement : for small gradients , the frequency is amplitude - independent and the amplitude rises approximately linearly with the gradient strength . for large gradients ,
the frequency approaches the trapping frequency and is again only weakly dependent on the excitation amplitude . for intermediate gradients
, the system shows a strongly nonlinear behavior which results in an amplitude - dependent oscillation frequency . in the regime of small gradients , corresponding to small excitation strengths one can linearize eqs .
( [ eq : p2 ] ) and describe excitations in terms of their leading moments in @xmath65 and @xmath66 @xcite , which corresponds to pure spatial dipole oscillations @xcite . their oscillation frequency for the present initial state can be derived to be @xmath67 , where @xmath68 is the mean - field interaction energy as defined in ref .
@xcite .
this frequency is determined by a competition between the trap - induced spatial oscillations and the mean - field induced rotation of the spin . for our parameters , we calculate @xmath69 in good agreement with the experimental results .
the linearized equations of motion also confirm the pure vector and octupole character of the dipole excitations for a perfectly nematic initial state as discussed above . for large gradients ,
the frequency slowly approaches the trapping frequency : in this regime , the phase spiral is averaged out dynamically on timescales @xmath70 @xcite such that the mean - field potential no longer affects the subsequent oscillation .
solid lines show the initial spin - wave amplitude extracted from numerical calculations .
( b ) exemplary spdms for different initial coherences . ( c )
amplitude of the dipole and octupole tensor components @xcite .
( d ) vector and octupole tensors evaluated in ( c ) .
all error bars solely correspond to fit errors , representing one standard deviation .
the experimental amplitudes which are taken after tof and the numerical amplitudes calculated _ in situ _ are rescaled onto each other by a global factor.,width=325 ] all measurements discussed so far were performed with the same purely nematic initial state . by modifying the rf - pulse sequences used for the preparation of the initial state we can control the amplitude of the coherences @xmath71 and populations @xmath6 in the spdm . by this the multipole decomposition of the initial state
can be widely controlled and allows for the initialization of e.g. pure vector or nematic initial states , which in turn results in different spin and spatial characteristics of the emerging spin wave .
following this direction , we performed a second set of experiments , where we engineered the spin - wave excitations by keeping the population of all four spin components constant but changing the initial coherence amplitudes @xmath9 [ fig .
[ fig4](b ) ] .
note , that this is complementary to the results shown in fig .
[ fig3 ] , where we changed the phase @xmath8 of the coherence by using different gradient strengths .
figure [ fig4](a ) shows the resulting oscillation amplitude for all four spin components depending on @xmath72 . at @xmath73
the system changes its qualitative behavior where the @xmath74 components interchange their oscillation direction . using the tensor notation ,
the spin wave at small @xmath75 is dominated by the spin - octupole , where neighboring spin components have an inversed sign and therefore oscillate in opposed directions . at large @xmath75 ,
the oscillation becomes spin - vector dominated , where spin components with the same sign of magnetization oscillate in the same direction [ fig .
[ fig4](c ) ] .
the anew increase of the octupole amplitude at large @xmath75 is due to higher - order spatial excitations , possible in the nonlinear regime , where the measurements were performed . at @xmath73 ,
the vector and octupole component contributions mutually cancel each other , leading to a vanishing spin current in the @xmath74 components .
again the numerical calculations describe the engineered spin waves very well . in conclusion
, we have thoroughly investigated the physics of collective spin waves in a high - spin fermi gas . comparing experimental and numerical results
, we showed that high - spin fermi gases in the quantum degenerate regime can be well described using a smft .
we have analyzed the spin - wave excitation spectrum for different excitation strengths ranging from the linear deep into the nonlinear regime . by employing irreducible spherical tensors
, the smft allows to intuitively explain the novel emerging spin - wave characteristics in a high - spin system .
we find a linear decoupling of the spin - nematic component , which in turn allowed us to directly observe nonlinear mode - coupling in the spin - wave dynamics . finally , we demonstrated how to control the multipole character of spin waves which leads to a reversal of the resulting counterflow spin current of two spin components .
our results constitute the first experimental investigation of coherent many - body dynamics of a high - spin fermionic quantum gas .
they demonstrate the controlled manipulation of atomic spin currents which , together with the theoretical understanding , paves the way towards novel schemes for spintronics in ultracold atomic gases , using the intrinsic high spin as a valuable resource .
we acknowledge financial support by dfg via grant no .
for801 , from spanish micinn ( fis 2008 - 00784 , aaii - hubbard , fpi - fellowship ) , erc grant quagatua and the spanish foundation universidad.es .
|
astronomical masers such as h@xmath0o and sio are unique tools for studying mass - loss activities in high mass star formation .
these maser species are basically excited in strong shocks on a working surface between a protostellar outflow and a dense envelope ( e.g. , @xcite ; @xcite ) .
the masers allow us to directly observe an intrinsic outflow driven by individual high mass young stellar objects ( hmysos ) within 10@xmath6 au scale , minimizing any contamination from a mass entrainment and/or outflows driven by nearby cluster members .
outflow properties such as size , velocity and morphology , predicted from interferometric maser studies vary source to source , e.g. , collimated jet ( @xcite ) , disk wind ( @xcite ; @xcite ) , wide angle outflow ( @xcite ) , expanding shell ( @xcite ; @xcite;@xcite ) , equatorial outflow ( @xcite ) or combination of jet and wide angle outflow ( @xcite ) .
such divergences are possibly related to a number of factors such as stellar mass , evolutionary stage , geometry of surrounding envelope and driving mechanism of host outflows .
@xcite ( hereafter paper i ) focused on a highly variable h@xmath0o maser source associated with a southern hmyso , g353.273 + 0.641 ( hereafter g353 ) .
they have found two intermittent maser flares during two years , combining a frequent monitoring by very long baseline interferometry ( vlbi ) and single - dish observation . in paper
i , they proposed that such a flare activity is recurrently excited by episodic shock propagations caused by a high - velocity protostellar jet .
if this is the case , one can extract time - domain information of a jet - launching region from a maser variability , since bright maser emission allows us easy and frequent monitoring even with a small size radio telescope .
such a region is expected to be so small ( @xmath1 a few au or less , e.g. , @xcite ; @xcite ) , compared to current observational resolutions .
g353 is embedded in the high mass star - forming region ngc6357 .
@xcite determined its photometric distance as 1.74 @xmath7 0.31 kpc from the sun .
exact yso mass is still unclear , however , association of a class ii ch@xmath8oh maser emission suggests that the source is already a high mass object ( @xcite , hereafter cp08 ) .
g353 is known as the dominant blue - shifted maser ( dbsm ) source that is a special class of a 22 ghz h@xmath0o maser sources ( cp08 ) .
dbsms have a highly asymmetric spectrum , where almost all the flux is concentrated in highly blue - shifted components . the peak line - of - sight ( los ) velocities
are ranged from -60 to -45 km@xmath9 against the systemic velocity ( @xmath10 ) of @xmath1 -5.0 km s@xmath9 in the case of g353 , adopted from the peak velocity of the class ii ch@xmath8oh maser emission ( cp08 ) .
cp08 proposed that dbsms are excited by a well - collimated jet in a nearly face - on geometry .
@xcite have actually detected both of a faint radio jet and an extremely high - velocity thermal sio jet , supporting cp08 .
paper i also found a signature of maser acceleration along the los in two velocity components ( @xmath1 -53 km s@xmath9 and 73 km s@xmath9 ) accompanied by a significant flare - up .
such an acceleration was observed in both of the single - dish and vlbi data , excluding instrumental errors . despite a similar acceleration rate ( @xmath1 -5 km s@xmath9 yr@xmath9 )
, these two components were spatially independent .
this fact confirmed that observed acceleration was not an apparent phenomenon caused by christmas - tree effect and/or any variation of internal structures of individual maser features , but a real one driven by global momentum supply related to shock propagations . in this paper , we report on further monitoring observations of the h@xmath0o maser at 22.23508 ghz ( @xmath11 = @xmath12@xmath13 ) towards g353 .
we have studied variability and recurrence in spatial distributions of maser features between each flare , employing denser vlbi monitoring intervals than our previous work .
new single - dish data validated the systematic acceleration of the masers .
the vlbi monitoring was performed by the vlbi exploration of radio astrometry ( vera ) .
typical observing interval was @xmath1 45 days that is roughly half of that in paper i. we had total 15 epochs between 2010 september to 2012 august .
all observations were made in the dual beam mode , where target and phase calibrator were observed simultaneously ( @xcite ; @xcite ) .
total on - source time was @xmath1 5 hours in each epoch .
the phase tracking center for g353 was ( @xmath14,@xmath15)@xmath16 = ( , ) .
we used a bright quasar , j1717 - 3342 ( @xmath1 700 mjy ) for the dual - beam phase referencing .
the separation angle between g353 and j1717 - 3342 is with a position angle of - .
nrao530 and/or pks1510 - 089 was observed once per hour for delay and bandpass calibration .
antenna gains were corrected based on measured system noise temperatures ( @xmath17 ) of each stations .
typical @xmath17 are ranged 200 400 k for mizusawa and iriki stations and 400 800 k for ogasawara and ishigaki stations .
details of backends and correlation processes were already described in paper i. we observed single left - hand circular polarization with a data rate of 1024 mb s@xmath9 ( 2 bit , 512 mhz sampling ) .
the total bandwidth was 240 mhz ( 16 mhz @xmath3 15 if ) for calibrators and 16 mhz ( 1 if ) for g353 .
we cut out an 8-mhz sub - if from the maser if and divided it into 512 spectral channels in initial two epochs , as it was in paper i. the total velocity coverage is 108 km s@xmath9 with spectral resolution of 0.21 km s@xmath9 . on the other hand , we used two 8-mhz sub - ifs in latter 13 epochs , in order to cover a larger velocity range in g353 .
central frequencies of these sub - ifs are separated 4 mhz , as both including maser emission peaks .
this gives a velocity coverage of 162 km s@xmath9 with the same spectral resolution .
data reduction was carried out in the standard reduction procedure for vera data ( see @xcite ) , using the astronomical imaging processing system ( aips ) package developed by national radio astronomy observatory ( nrao ) .
we first determined an astrometric frame in each epoch from phase - referenced positions of bright maser spots with respect to j1717 - 3342 , and then , fringe - fitting and self - calibration were performed using the brightest maser channel in each epoch .
we finally searched all maser spots from the self - calibrated data cube with a detection limit of 6 @xmath18 .
searched area was @xmath3 from the phase tracking center .
an interval of phase solution was @xmath12 min for a fringe - fitting using j1717 - 3342 , and down to 15 sec in final self - calibration .
only a maser emission that appeared in successive three or more spectral channels in at least one observing epoch was treated as a real detection .
each set of spatially successive maser spots was defined as a maser feature that means individual gas clump , showing a single - peaked velocity profile .
velocities and positions of each feature were determined by the flux - weighted averaging .
we included thermal image noises , baseline solution errors and tropospheric zenith delay residuals as astrometric error sources .
the third term was estimated within a typical range of @xmath7 3 cm , by the image - optimization method @xcite .
this estimation , however , seemed to be incomplete in some epochs , since image dynamic ranges were still limited even after the correction .
we deduced overall errors of the image optimization , using two distinct sub - if data which were available in most epochs .
we analyzed each dataset independently and compared astrometric positions of maser spots detected in both data .
these positions were basically consistent within an error of thermal image noises , but some epochs show a significant positional offset up to 3 milli - arcsecond ( mas ) . to be conservative ,
we included it in total astrometric errors as a systematic error of our analysis , since frequency difference of only 4 mhz is unlikely to cause such a large offset .
the relative positional accuracies in the self - calibrated data , which were dominated by thermal image noises , are better than 0.03 and 0.1 mas for the right ascension and declination , respectively .
table [ tab : obs ] summarizes all vera observations including that reported in paper i. we note that all tables in this paper are presented in appendix section .
it contains observing dates , synthesized beams ( @xmath19 ) , beam position angles ( pa ) east of north , typical image noise levels ( @xmath20 ) , astrometric errors ( @xmath21 ) , residual systematic errors ( @xmath22 ) and any special comments .
table [ tab : feature ] listed only features detected in three or more epochs , which were used for kinematic analyses in the following sections .
all data of detected maser features are available in supplementary table [ tab : obs ] .
single - dish monitoring was performed by the hokkaido university tomakomai 11-m telescope @xcite .
follow - up monitoring was started from 2010 november and completed 2013 march .
observing intervals were denser than bi - weekly , except for maintenance sessions during summer seasons . the beam size and aperture efficiency were 4.2 arcmin and 49 per cent , respectively .
detailed procedure of observations and data reductions using the newstar software package developed at nobeyama radio observatory were already described in paper i. the total velocity coverage and final spectral resolution are 216 and 0.21 km s@xmath9 , respectively .
the rms noise levels were @xmath1 2 - 3 jy in winter seasons and increased up to 15 jy in the worst case in summer seasons .
figure [ fig : light ] shows the light curve of the main velocity component ( -65 -45 km s@xmath9 ) that contains almost all maser flux in g353 .
typical single - dish spectra in both flare - up and quiescent phases were also plotted in figure [ fig : spectrum ] .
we have detected five maser flares during five years , and four except for the last one were also monitored by vera .
the durations of these four flares were defined as , 1st : 300 500 , 2nd : 800 950 , 3rd : 1050 1300 , 4th : 1400 1690 , using the relative day counted from january 1st , 2008 .
the target showed intermittent flare - up at a typical interval of @xmath1 1.5 years , although no clear periodicity was seen as reported in paper i. a higher velocity component ( @xmath23 -70 km s@xmath9 ) , which showed a significant flare - up in the initial flare , has been staying quiescent state after the second flare .
we show a spectral movie made by all single - dish data in supplementary figure 1 .
( 90mm,90mm)fig1.eps ( 90mm,90mm)fig2.eps ( 134mm,134mm)fig3.eps figure [ fig : dynamic ] presents a dynamic spectrum with a velocity range of -80 to -40 km s@xmath9 .
the spectrum contains both of the single - dish and vlbi data .
the velocity widths were sometimes narrower in the latter data , because of resolved - out flux ( e.g. , day @xmath1 840 ) .
there is a clear signature of a systematic acceleration in several velocity components .
as already mentioned , accelerations of two velocity components in the 1st flare - up ( @xmath1 -53 and -73 km s@xmath9 ) were already reported in paper i. these spatially independent components have shown a similar acceleration rate ( @xmath1 -5 km s@xmath9 yr@xmath9 ) .
after the 2nd flare - up , the newly appeared component around -47 km s@xmath9 had been continuously accelerated up to -63 km s@xmath9 during 3 years .
the linearly - fitted acceleration rate of this new component is -5.4 @xmath7 0.18 km s@xmath9 yr@xmath9 , where velocity dispersions were used as a fitting weight .
this is consistent with that of the two components in the 1st flare - up .
these accelerations along the los were also recognized in four individual maser features .
we accepted only an acceleration that exceeds a velocity dispersion of each maser feature , in order to avoid apparent accelerations that can be caused by an internal structure of features .
figure [ fig : acc_feature ] presents observed accelerations in the four features and table [ tab : acceleration ] summarizes linearly - fitted acceleration rates .
the averaged acceleration rate well agrees with that derived from the dynamic spectrum .
it should be noted that no significant acceleration is found in proper motions .
this is probably because the acceleration of -5 km s@xmath9 yr@xmath9 is relatively small , and also , an accelerating force is mainly along the los .
we will discuss about proper motions and an inclination angle of the maser outflow in the following sections .
these accelerations along the los probably contribute to the short maser lifetime in g353 , decreasing the coherent path for maser amplification .
( 150mm,150mm)fig4.eps ( 90mm,90mm)fig5.eps figure [ fig : map ] presents the overall distribution of maser features detected in all the vera observations , including that reported in paper i. the @xmath24 , @xmath25 and color show the right ascension offset ( @xmath24 = @xmath26cos@xmath15 ) , declination offset ( @xmath25 = @xmath27 ) and los velocity ( @xmath28 ) , respectively .
the coordinate origin is the position of the brightest maser spot detected in paper i , i.e. , ( @xmath14,@xmath15)@xmath16 = ( , ) .
most of the maser features are concentrated near the origin , belonging to the main velocity components ( -65 -45 km s@xmath9 ) .
all of other significantly blue - shifted features ( -120 -70 km s@xmath9 ) spread eastward . on the other hand , slightly red - shifted components ( @xmath1 0 km s@xmath9 ) were appeared at the western end during the latter 7 epochs .
this is the first vlbi detection of red - shifted components in g353 .
we note that the most red - shifted component of + 87 km s@xmath9 found in cp08 was out of our velocity coverage .
we compared distributions of the central maser cluster appeared in each flare - up phase in figure [ fig : flare ] .
here 1st to 4th flares correspond epoch 2 4 , 8 9 , 11 15 and 18 23 , respectively .
we always detected filamentary maser distributions in all flares , suggesting a shock propagation .
although all the flares showed clearly different morphologies each other , the maser filaments always appeared in 100 @xmath3 100 au@xmath4 region around the origin .
( 140mm,140mm)fig6.eps we have successfully traced several relatively - stable maser features during 200 days in the 4th flare , thanks to the short observing interval .
we , hence , tried to measure absolute proper motions and annual parallax of these maser features .
we only used features that were successively detected in more than five epochs .
some features were also detected in epoch 17 , which was observed by only three stations because of the antenna trouble in mizusawa station .
however , data in this epoch were not included for the astrometric analysis , because all features showed large and systematic jump ( @xmath1 3 mas ) of absolute positions .
there were five maser features available in these criteria ( see table [ tab : feature ] ) .
it should be noted that we only used @xmath24 data for a parallax fitting , since @xmath25 data contributed not at all .
this is because the synthesized beams were significantly elongated along the n - s direction , causing large errors in @xmath25 direction ( see table [ tab : obs ] ) .
in addition , expected amplitude of a sinusoidal motion in @xmath25 direction was five times smaller than that in @xmath24 , because of a small ecliptic latitude ( @xmath1 -11@xmath29 ) .
we performed @xmath30-fitting of the common annual parallax ( @xmath31 ) and linear absolute proper motions ( @xmath32 ) of each features .
we included both of total astrometric errors in table [ tab : obs ] and positional dispersions in table [ tab : feature ] as a fitting weight .
figure [ fig : parallax ] shows the best - fit parallax with all data points after the subtraction of linear proper motions . individual fitting results for each maser feature are shown in supplementary figure 2 . the best - fit @xmath31 was 0.59 @xmath7 0.05 .
the formal error was estimated as the reduced @xmath33 square to be unity .
unfortunately , a short lifetime of the maser features prevented us to trace a full parallactic cycle .
our parallax fitting can contains a systematic error that can not be evaluated directly .
we deduced such a systematic error via the montecarlo analysis , where 10@xmath34-times simulated fittings were performed , using the same observing epochs and the parallax of 0.59 mas .
positional errors in each epoch were randomly added , following a gaussian distribution with standard deviations same as the astrometric errors in table [ tab : obs ] .
figure [ fig : prob ] presents a probability distribution of simulated @xmath31 values .
the best - fit 1-@xmath18 deviation of 0.035 mas was adopted as a systematic error of our fitting .
we finally obtained a parallax of 0.59 @xmath7 0.06 , considering a root sum square of the formal fitting error and the systematic error .
the newly derived distance is 1.70 @xmath5 kpc .
this is consistent with the commonly used photometric distance @xcite .
we , thus , conclude that our parallax measurement is well reliable and new distance of 1.7 kpc were employed in this paper .
the best - fit absolute motions are listed in table [ tab : absolute ] , where absolute proper motions in @xmath25 direction were fitted by a linear motions under the fixed @xmath31 .
we also determined absolute motions of three maser features , which had been excluded from the parallax fitting , in the same way .
( 90mm,90mm)fig7.eps ( 90mm,90mm)fig8.eps all the absolute proper motions determined in previous section includes a common systemic velocity that consists of the solar motion with respect to the lsr , the differential of the galactic rotation and the unknown peculiar motion of the system .
therefore , we also estimated relative proper motions with respect to the reference maser features , in order to discuss intrinsic 3d kinematics of the masers .
we employed four distinct reference features , since no identical feature survived throughout all the observing epochs .
data were , hence , divided into four durations ( epoch 2 7 , 8 11 , 12 15 and 16 23 ) , depending on a reference feature .
we estimated relative proper motions for each duration , using only a feature that detected in three or more epoch .
all the features used for relative motion measurements are also listed in table [ tab : feature ] .
internal proper motions without any systemic motion were derived using these relative motions , as follows .
an relative proper motion vector of i - th feature ( @xmath35 ) can be expressed as , @xmath36 here @xmath37 and @xmath38 indicates an internal proper motion vector of i - th and reference maser feature , respectively .
one can reconstruct @xmath37 , if both of @xmath39 and @xmath38 are avilable .
for this purpose , we deduced @xmath38 , assuming a symmetric distribution of @xmath37 in each duration , i.e. , @xmath40 here @xmath41 is total number of maser features except for a reference feature . in this case , equation ( 2 ) simply gives , @xmath42 we listed originally measured @xmath43 and reconstructed @xmath37 in table [ tab : internal ] .
it should be noted that the assumption of the equation ( 2 ) is adequate in our case .
this was examined by following procedures , comparing @xmath37 with that independently deduced from the absolute proper motions ( @xmath44 ) in table [ tab : absolute ] .
we first estimated the systemic motion using the reference feature d8 in the final duration as , @xmath45 obtained @xmath46 is 0.47 @xmath7 0.07 and 0.99 @xmath7 1.04 mas yr@xmath9 in @xmath24 and @xmath25 direction , respectively .
then @xmath44 were converted to @xmath37 , subtracting the @xmath46 .
figure [ fig : propers ] presents a comparison of @xmath37 determined by two different ways .
a good correlation between two data suggests that the equation ( 2 ) is well applicable in g353 .
figure [ fig : arrow ] shows both relative and internal motion vectors , superposed on the maser distribution .
the highly blue - shifted features in the east region and the red - shifted features at the west end were moving away each other , suggesting a bipolar motion along the e - w direction . on the other hand
, the central cluster basically showed westward motions in all the durations .
we note that this is only a projected motion and most of the 3d vectors in the cluster are nearly along the los ( see section 4 ) .
( 80mm,80mm)fig9.eps ( 90mm,90mm)fig10.eps
table [ tab : inclination ] contains 3d velocities and inclination angles , which were calculated from the internal proper motions and los velocities assuming the distance of 1.7 kpc .
almost all the blue - shifted features have 3d motions nearly along the los .
this fact confirmed that the h@xmath0o masers in g353 were associated with a face - on protostellar jet , as proposed by cp08 .
the westward expansions seen in the central cluster can be explained by a finite opening angle and face - on geometry of the jet .
figure [ fig : inclination ] shows the relation of the inclination angles to the internal proper motions in @xmath24 direction .
the features moving eastward are more closely along the los than that moving westward .
the averaged inclination angles are 17@xmath29 and 34@xmath29 from the los for the eastward and westward features , respectively .
if the masers are excited on the working surface between the jet and cavity wall having a constant opening angle , the jet axis is rather inclined ( @xmath1 8@xmath29 ) to the west from the los .
another possible explanation is that the masers are associated with two distinct outflows having a different collimation as in the case of cepa hw2 @xcite , i.e. , collimated jet and wide angle ( or equatorial ) outflow .
if this is the case , the eastward features , which have rather higher 3d velocities , may trace the collimated jet .
a jet inclination of @xmath1 17@xmath29 is simply expected from the averaged inclination of the eastward masers . on the other hand ,
a wide angle outflow traced by the westward features would have a large opening angle up to @xmath1 80@xmath29 from the jet axis .
schematic views of these two cases are showed in figure [ fig : schematic ] .
although the red - shifted features at the west end strongly suggest a bipolar morphology along the e - w direction , this position angle seems to be perpendicular to that of the radio jet in the arcsecond scale @xcite .
this can be explained by the projection effect , or the radio jet may trace an ionized equatorial wind reported in @xcite and @xcite .
these red - shifted features are moving almost along the celestial plane , contrary to the blue - shifted features .
we suggest that the blue - shifted masers trace more intrinsic outflow kinematics at the root region , since the peak position of the radio jet indicates that the host object is located at the almost same position as the central blue - shifted cluster ( see figure [ fig : flare ] ) .
if the red - shifted features are excited by the same driving source , the jet may be strongly bending within 500 au from the source , otherwise , these features can be associated with a wide - angle outflow , discussed above . a vlbi position and proper motion of
the extremely red - shifted component will allow us to solve these questions , because such high - velocity component probably reflects an intrinsic motion of the westward jet .
an interferometric imaging of the thermal molecular jet found in @xcite will be also helpful to understand the overall extent and trajectory of the jet .
( 85mm,85mm)fig11.eps ( 145mm,145mm)fig12.eps the recurrent formations of a filamentary maser distribution indicates episodic shock propagations , as suggested in paper i. the short maser lifetime in g353 can simply emphasize one - to - one relation between maser flares and shock propagations .
paper i discussed that such a recurrent shock is possibly caused by an episodic jet - launching or precession .
although our new data do not exclude both cases , no precession signature in the radio jet may support the episodic jet - launching .
the spatial scale in which the maser filament recurrently appeared is @xmath1 100 @xmath3 100 au@xmath4 .
this scale can be resolved by the maximum resolution of atacama large millimeter / submillimeter array ( alma ) in near future .
therefore , there will be a chance to directly determine which case is correct , by monitoring of how shocked gas varies using a thermal tracer .
such a monitoring was already succeeded in the case of the molecular jet associated with a nearby low mass protostar @xcite .
figure [ fig : dynamic ] indicates that the acceleration of the main velocity component also occurs recurrently .
in addition , the acceleration had clearly continued beyond a lifetime of individual maser feature .
these facts suggest that the h@xmath0o maser in g353 traces a region where molecular gas is steadily accelerated .
the spatial distribution of accelerating velocity components is almost consistent with the region where maser features are recurrently formed , i.e. , @xmath1 100 @xmath3 100 au@xmath4 wide around the root of the bipolar jet .
if we adopt the expected inclinations and opening angles , the los extent of such an accelerating region is 180 300 au .
the thermal sio spectrum reported in @xcite has indicated that the maximum los velocity of shocked molecular gas is @xmath1 -120 km s@xmath9 , similar to that of the maser emission .
an important fact is that such highest - velocity maser components have shown no significant acceleration in our monitoring .
it can be interpreted as that the molecular jet is accelerated up to this velocity , and then , turns into free expansion . in order to examine this possibility ,
we must study velocity structures of both pre - shock and post - shock gas along the jet axis in multi - scale .
the most plausible origin of such an acceleration is momentum supply from an optical jet .
a momentum rate of the maser flow in g353 was estimated as @xmath1 10@xmath47 @xmath48 km
s@xmath9 yr@xmath9 in paper i. this order of momentum rate is often found in high mass star formation ( e.g. , @xcite ) .
if we consider a radiation pressure , a bolometric luminosity higher than 10@xmath49 @xmath50 is required to achieve the momentum rate at the radius of 150 au , although such a high luminosity is quite rare and may be unlikely case in g353 .
another possible origin is a magnetic pressure gradient along the jet axis ( e.g. , @xcite ) .
this can be examined by a first - principle mhd simulation , tracing rather longer - term outflow evolution than that commonly reported at present .
@xcite have shown that the number of dbsms without any oh maser ( i.e. , youngest evolutionary stage ) is statistically superior to red - shifted counterparts where maser fluxes dominated by highly red - shifted emissions .
therefore , there must be some intrinsic origin of the blue - shift dominance . @xcite
have proposed that an optically thick disk in a nearly face - on geometry can selectively cause a blue - shift dominance , masking any red - shifted emission .
this scenario is still plausible , because there is no red - shifted component around the center of the bipolar distribution , where the host hmyso is thought to be embedded with a disk .
further test will be given by an exact position of the extremely red - shifted component at + 87 km s@xmath9 .
alternatively , if a circumstellar radiation field is completely dominated by spherically - symmetric fluxes from a central object , only blue - shifted components can receive seed photons for maser amplifications towards the observer .
the two - outflow model may prefer this explanation , allowing a lack of eastward red - shifted masers associated with the wide angle outflow .
this can be applicable in the case of g353 that has no background hii region . in either case
, more statistical samples are required to find out the true origin .
one must compare the geometry of a disk / jet system , optical depth of a disk and circumstellar environment in several dbsms , combining a vlbi array and the highest - resolution connected array such as alma .
we have conducted a follow - up vlbi and single - dish monitoring for the highly variable h@xmath0o masers in g353 .
thanks to the denser monitoring interval than that in paper i , several properties of g353 have been revealed as follows .
( 1 ) : : the h@xmath0o masers are associated with the bipolar jet along the e - w direction .
an inclination angle of the jet axis is expected to be 8@xmath29 17@xmath29 , suggesting a nearly face - on geometry as proposed in cp08 . + ( 2 ) : : intermittent flare activities are caused by recurrent shock propagations .
such shocks are probably explained by episodic jet - launchings .
this will be verified by a direct monitoring of shocked gas with alma , using a thermal tracer .
+ ( 3 ) : : the systematic accelerations were recurrently detected and continued beyond a lifetime of individual maser features .
these accelerating masers probably trace a region , where molecular gas is steadily accelerated up to -120 km s@xmath9 within 100@xmath3100@xmath3300 au@xmath6 box ( @xmath24 , @xmath25 and the los , respectively ) around the root of the bipolar jet . + ( 4 ) : : the measured annual parallax indicates a distance of 1.70 @xmath5 kpc , supporting a commonly - used photometric distance . + ( 5 ) : : the origin of dbsms is still open question .
this can be explained by a masking effect of optically thick face - on disk as proposed in paper i. we require more statistical data about a geometry of a disk / jet system , optical depth of a disk and circumstellar environment towards the known dbsms . the author heavily thanks to all the members in vera and hokkaido univerisity tomakomai radio observatory for their assistance in operations on the very long - term monitoring .
the author also thanks to the anonymous referee for useful suggestions and very encouraging comments .
this work was financially supported by the grant - in - aid for the japan society for the promotion of science fellows ( k.m . ) and the grants - in - aid by the ministry of education , culture , and science of japan 24 - 6525 ( k.m . ) .
cccccccl epoch & day@xmath51 & @xmath19 & pa & @xmath20@xmath52 & @xmath21 & @xmath53 & comments@xmath54 + & & ( mas @xmath3 mas ) & ( @xmath29 ) & ( jy beam@xmath9 ) & ( mas @xmath3 mas ) & ( mas @xmath3 mas ) & + 1 & 183 & 3.16 @xmath3 0.74 & -17.1 & 0.48 & 0.15 @xmath3 0.49 & - & - + 2 & 323 & 3.98 @xmath3 0.82 & -23.2 & 0.30 & 0.92 @xmath3 2.15 & - & @xmath17 @xmath1 1000 k at og + 3 & 403 & 2.63 @xmath3 0.91 & -14.1 & 0.37 & 0.06 @xmath3 0.24 & - & - + 4 & 504 & 2.57 @xmath3 0.88 & -13.2 & 0.30 & 0.04 @xmath3 0.16 & - & - + 5 & 625 & 2.86 @xmath3 0.82 & -12.3 & 0.37 & 0.11 @xmath3 0.48 & - & - + 6 & 675 & 2.79 @xmath3 0.79 & -13.3 & 0.16 & 0.23 @xmath3 0.98 & - & - + 7 & 752 & 3.15 @xmath3 0.85 & -19.5 & 0.21 & 0.09 @xmath3 0.25 & - & - + 8 & 842 & 2.59 @xmath3 0.75 & -14.1 & 0.40 & 0.10 @xmath3 0.41 & - & - + 9 & 976 & 2.84 @xmath3 0.76 & -13.8 & 0.29 & 0.20 @xmath3 0.80 & - & @xmath17 @xmath1 2000 k at og + 10 & 1036 & 2.79 @xmath3 0.77 & -15.0 & 0.19 & 0.21 @xmath3 0.77 & - & @xmath17 @xmath1 1500 k at is + 11 & 1069 & 2.92 @xmath3 0.82 & -14.5 & 0.18 & 0.36 @xmath3 1.40 & 0.03 @xmath3 0.02 & - + 12 & 1107 & 2.67 @xmath3 0.91 & -7.3 & 0.23 & 0.03 @xmath3 0.22 & 0.28 @xmath3 0.00 & - + 13 & 1129 & 2.56 @xmath3 0.83 & -11.7 & 0.14 & 0.16 @xmath3 0.75 & 0.00 @xmath3 0.02 & - + 14 & 1164 & 2.65 @xmath3 0.85 & -10.5 & 0.11 & 0.09 @xmath3 0.50 & 2.44 @xmath3 1.81 & - + 15 & 1210 & 3.03 @xmath3 0.78 & -16.9 & 0.20 & 0.52 @xmath3 1.72 & 0.08 @xmath3 0.11 & - + 16 & 1320 & 2.84 @xmath3 0.78 & -10.0 & 0.48 & 0.15 @xmath3 0.88 & 0.10 @xmath3 0.02 & - + 17 & 1398 & 2.66 @xmath3 0.66 & -15.2 & 0.26 & 0.33 @xmath3 1.19 & 0.21 @xmath3 0.00 & three stations without ir + 18 & 1474 & 2.76 @xmath3 0.85 & -11.1 & 0.16 & 0.07 @xmath3 0.33 & 0.00 @xmath3 0.00 & - + 19 & 1510 & 2.60 @xmath3 0.82 & -12.8 & 0.17 & 0.03 @xmath3 0.13 & 0.00 @xmath3 0.00 & - + 20 & 1547 & 2.54 @xmath3 0.80 & -11.8 & 0.23 & 0.11 @xmath3 0.48 & 1.95 @xmath3 2.65 & - + 21 & 1576 & 2.61 @xmath3 0.84 & -10.0 & 0.22 & 0.10 @xmath3 0.52 & 0.26 @xmath3 1.75 & - + 22 & 1611 & 3.27 @xmath3 0.83 & -15.7 & 0.37 & 0.22 @xmath3 0.78 & 3.72 @xmath3 0.06 & @xmath17 @xmath1 1000 k at og + 23 & 1687 & 2.59 @xmath3 0.76 & -4.7 & 0.58 & 0.10 @xmath3 1.31 & 0.07 @xmath3 0.06 & @xmath17 @xmath1 2000 k at mz and ir + + + + + i d & epoch & @xmath55 & @xmath56@xmath57 & @xmath58 & @xmath25 & @xmath24 - @xmath59@xmath54 & @xmath25 - @xmath60 & @xmath61 + & & & & & ( jy km s@xmath9 ) + i d & epoch & @xmath55 & @xmath56@xmath57 & @xmath62 & @xmath25 & @xmath24 - @xmath59@xmath54 & @xmath25 - @xmath60 & @xmath61 + & & & & & ( jy km s@xmath9 ) + + + + + + + a1 & 3 & -23.72 & 1.26 & -11.90 ( 0.14 ) & 37.09 ( 0.04 ) & -44.96 & -5.80 & 2.63 ( 0.11 ) + & 4 & -23.78 & 1.68 & -12.31 ( 0.18 ) & 35.01 ( 0.13 ) & -45.62 & -6.54 & 7.79 ( 0.21 ) + & 6 & -24.13 & 0.21 & -12.84 ( 0.04 ) & 34.28 ( 0.14 ) & -46.70 & -7.34 & 0.25 ( 0.05 ) + a2 & 4 & -43.83 & 0.42 & 83.35 ( 0.07 ) & 59.54 ( 0.00 ) & 50.03 & 17.99 & 0.93 ( 0.10 ) + & 5 & -43.79 & 0.63 & 83.17 ( 0.08 ) & 58.34 ( 0.09 ) & 50.19 & 17.27 & 1.51 ( 0.19 ) + & 6 & -43.82 & 0.84 & 83.83 ( 0.15 ) & 58.68 ( 0.14 ) & 49.97 & 17.06 & 1.64 ( 0.10 ) + a3 & 2 & -51.44 & 5.69 & 4.46 ( 0.31 ) & -5.32 ( 0.33 ) & -29.05 & -49.20 & 1023.54 ( 9.40 ) + & 3 & -53.08 & 3.79 & 3.24 ( 0.41 ) & -7.18 ( 0.28 ) & -29.82 & -50.06 & 824.36 ( 2.38 ) + & 4 & -53.54 & 1.90 & 2.65 ( 0.09 ) & -9.83 ( 0.06 ) & -30.67 & -51.38 & 16.96 ( 0.63 ) + a4 & 3 & -51.65 & 2.74 & 8.90 ( 0.39 ) & -9.47 ( 0.37 ) & -24.16 & -52.35 & 35.59 ( 1.48 ) + & 4 & -53.22 & 1.26 & 9.91 ( 0.10 ) & -12.67 ( 0.03 ) & -23.41 & -54.22 & 19.55 ( 0.46 ) + & 5 & -56.46 & 1.26 & 9.10 ( 0.23 ) & -14.84 ( 0.16 ) & -23.88 & -55.90 & 3.53 ( 0.31 ) + & 6 & -57.41 & 1.47 & 9.64 ( 0.06 ) & -14.74 ( 0.33 ) & -24.22 & -56.36 & 3.59 ( 0.14 ) + a5 & 2 & -63.77 & 0.63 & 107.57 ( 0.00 ) & 51.48 ( 0.04 ) & 74.06 & 7.60 & 1.87 ( 0.18 ) + & 3 & -64.65 & 2.32 & 107.25 ( 0.14 ) & 50.07 ( 0.24 ) & 74.19 & 7.19 & 6.47 ( 0.17 ) + & 4 & -65.74 & 0.84 & 107.80 ( 0.03 ) & 47.82 ( 0.20 ) & 74.48 & 6.27 & 1.59 ( 0.13 ) + a6 & 2 & -70.00 & 1.69 & 33.79 ( 0.05 ) & 43.07 ( 0.24 ) & 0.28 & -0.81 & 17.90 ( 0.34 ) + & 3 & -72.43 & 2.32 & 32.89 ( 0.37 ) & 42.82 ( 0.16 ) & -0.17 & -0.07 & 28.06 ( 0.15 ) + & 4 & -71.84 & 3.16 & 32.80 ( 0.17 ) & 41.24 ( 0.23 ) & -0.52 & -0.31 & 26.93 ( 0.31 ) + a7 & 2 & -72.15 & 1.05 & 33.51 ( 0.03 ) & 43.88 ( 0.11 ) & 0.00 & 0.00 & 4.44 ( 0.21 ) + & 3 & -74.18 & 2.11 & 33.06 ( 0.12 ) & 42.88 ( 0.08 ) & 0.00 & 0.00 & 13.53 ( 0.13 ) + & 4 & -74.49 & 2.95 & 33.32 ( 0.11 ) & 41.55 ( 0.08 ) & 0.00 & 0.00 & 146.72 ( 0.28 ) + & 5 & -75.41 & 1.90 & 32.98 ( 0.04 ) & 41.06 ( 0.32 ) & 0.00 & 0.00 & 19.77 ( 0.68 ) + & 6 & -75.99 & 2.11 & 33.86 ( 0.04 ) & 41.62 ( 0.04 ) & 0.00 & 0.00 & 24.79 ( 0.23 + & 7 & -77.32 & 0.84 & 33.30 ( 0.05 ) & 42.94 ( 0.12 ) & 0.00 & 0.00 & 1.08 ( 0.11 ) + + b1 & 8 & -46.75 & 1.47 & -6.99 ( 0.04 ) & -2.76 ( 0.37 ) & -80.95 & -71.78 & 10.21 ( 0.44 ) + & 9 & -47.08 & 2.74 & -8.20 ( 0.05 ) & -4.45 ( 0.07 ) & -82.21 & -72.62 & 353.44 ( 0.47 ) + & 10 & -46.86 & 2.74 & -9.24 ( 0.07 ) & -3.74 ( 0.06 ) & -82.50 & -73.20 & 55.49 ( 0.27 ) + & 11 & -47.11 & 2.74 & -8.42 ( 0.17 ) & -5.18 ( 0.17 ) & -82.70 & -73.39 & 7.01 ( 0.21 ) + b2 & 8 & -48.48 & 2.32 & -5.72 ( 0.07 ) & -4.61 ( 0.50 ) & -79.68 & -73.64 & 62.72 ( 0.78 ) + & 9 & -48.54 & 1.69 & -7.39 ( 0.05 ) & -5.58 ( 0.22 ) & -81.39 & -73.76 & 55.04 ( 0.34 ) + & 10 & -48.79 & 1.69 & -8.63 ( 0.13 ) & -4.50 ( 0.25 ) & -81.89 & -73.96 & 13.10 ( 0.21 ) + & 11 & -49.24 & 1.90 & -7.83 ( 0.11 ) & -5.84 ( 0.10 ) & -82.11 & -74.05 & 5.19 ( 0.16 ) + b3 & 9 & -51.50 & 3.37 & -7.00 ( 0.21 ) & -5.46 ( 0.24 ) & -81.01 & -73.64 & 24.30 ( 0.40 ) + & 10 & -51.88 & 2.11 & -8.17 ( 0.06 ) & -4.50 ( 0.28 ) & -81.43 & -73.95 & 12.59 ( 0.28 ) + & 11 & -51.71 & 1.69 & -7.43 ( 0.05 ) & -5.89 ( 0.13 ) & -81.70 & -74.09 & 6.33 ( 0.28 ) + b4 & 8 & -67.03 & 0.63 & 73.96 ( 0.05 ) & 69.02 ( 0.04 ) & 0.00 & 0.00 & 1.76 ( 0.17 ) + & 9 & -66.84 & 1.26 & 74.01 ( 0.11 ) & 68.18 ( 0.09 ) & 0.00 & 0.00 & 2.74 ( 0.19 ) + & 10 & -66.83 & 0.84 & 73.26 ( 0.01 ) & 69.46 ( 0.08 ) & 0.00 & 0.00 & 1.45 ( 0.10 ) + & 11 & -66.86 & 0.84 & 74.28 ( 0.09 ) & 68.21 ( 0.17 ) & 0.00 & 0.00 & 1.54 ( 0.11 ) + + c1 & 13 & -32.50 & 0.63 & 2.36 ( 0.04 ) & 48.24 ( 0.05 ) & 5.91 ) & 10.22 & 0.44 ( 0.05 ) + & 14 & -32.33 & 0.63 & 4.16 ( 0.20 ) & 51.10 ( 0.13 ) & 5.88 ) & 10.13 & 0.54 ( 0.05 ) + & 15 & -32.51 & 1.05 & 1.88 ( 0.20 ) & 49.50 ( 0.12 ) & 6.22 ) & 10.57 & 0.85 ( 0.09 ) + c2 & 12 & -44.45 & 0.42 & -2.80 ( 0.01 ) & 39.62 ( 0.02 ) & 0.00 & 0.00 & 0.55 ( 0.08 ) + & 13 & -44.49 & 1.05 & -3.55 ( 0.21 ) & 38.03 ( 0.24 ) & 0.00 ) & 0.00 & 1.12 ( 0.08 ) + & 14 & -44.56 & 1.05 & -1.72 ( 0.28 ) & 40.97 ( 0.21 ) & 0.00 ) & 0.00 & 1.02 ( 0.07 ) + & 15 & -44.53 & 0.63 & -4.34 ( 0.14 ) & 38.93 ( 0.06 ) & 0.00 ) & 0.00 & 0.72 ( 0.07 ) + c3 & 13 & -45.85 & 1.68 & -3.72 ( 0.06 ) & 22.83 ( 0.10 ) & -0.17 & -15.20 & 1.73 ( 0.09 ) + & 14 & -46.65 & 2.32 & -1.52 ( 0.11 ) & 26.14 ( 0.08 ) & 0.20 ) & -14.83 & 4.10 ( 0.10 ) + & 15 & -47.22 & 1.47 & -3.80 ( 0.04 ) & 24.72 ( 0.08 ) & 0.54 ) & -14.21 & 2.33 ( 0.11 ) + c4 & 12 & -51.63 & 2.32 & -3.44 ( 0.26 ) & 13.21 ( 0.15 ) & -0.64 & -26.41 & 61.82 ( 0.47 ) + & 13 & -51.93 & 3.37 & -4.04 ( 0.32 ) & 12.13 ( 0.15 ) & -0.49 ) & -25.90 & 144.56 ( 0.33 ) + & 14 & -52.26 & 2.95 & -2.14 ( 0.48 ) & 15.34 ( 0.24 ) & -0.42 & -25.63 & 123.73 ( 0.29 ) + & 15 & -52.25 & 3.16 & -4.91 ( 0.15 ) & 13.80 ( 0.21 ) & -0.57 & -25.14 & 520.38 ( 0.67 ) + c5 & 12 & -54.86 & 1.47 & -1.30 ( 0.12 ) & 14.05 ( 0.10 ) & 1.50 & -25.57 & 25.71 ( 0.30 ) + & 13 & -56.18 & 1.90 & -1.68 ( 0.21 ) & 13.18 ( 0.12 ) & 1.87 ) & -24.85 & 16.14 ( 0.16 ) + & 14 & -56.82 & 1.68 & 0.38 ( 0.16 ) & 16.43 ( 0.09 ) & 2.10 ) & -24.54 & 4.63 ( 0.10 ) + c6 & 12 & -56.95 & 2.95 & -0.01 ( 0.52 ) & 14.73 ( 0.22 ) & 2.79 & -24.89 & 38.99 ( 0.32 ) + & 13 & -57.54 & 2.11 & -0.38 ( 0.17 ) & 13.68 ( 0.11 ) & 3.17 ) & -24.35 & 17.58 ( 0.17 ) + & 14 & -58.33 & 1.90 & 1.58 ( 0.11 ) & 16.72 ( 0.14 ) & 3.30 ) & -24.25 & 3.07 ( 0.10 ) + c7 & 12 & -107.60 & 1.05 & 99.21 ( 0.04 ) & 40.58 ( 0.09 ) & 102.01 & 0.96 & 2.21 ( 0.10 ) + & 14 & -107.76 & 1.47 & 101.43 ( 0.02 ) & 42.60 ( 0.04 ) & 103.16 & 1.63 & 8.22 ( 0.08 ) + & 15 & -107.68 & 2.95 & 99.77 ( 0.19 ) & 41.01 ( 0.19 ) & 104.10 & 2.08 & 16.99 ( 0.15 ) + + d1 & 17 & -0.14 & 0.42 & -272.66 ( 0.02 ) & -113.40 ( 0.04 ) & -301.79 & -115.95 & 0.68 ( 0.13 ) + & 18 & -0.08 & 0.63 & -273.74 ( 0.05 ) & -109.01 ( 0.16 ) & -303.30 & -116.29 & 1.11 ( 0.08 ) + & 19 & -0.11 & 0.84 & -273.51 ( 0.09 ) & -109.26 ( 0.08 ) & -303.95 & -116.49 & 1.35 ( 0.14 ) + & 20 & -0.06 & 0.42 & -273.56 ( 0.02 ) & -109.74 ( 0.03 ) & -304.61 & -116.83 & 0.85 ( 0.09 ) + & 21 & -0.01 & 0.42 & -273.67 ( 0.10 ) & -109.43 ( 0.29 ) & -304.97 & -116.97 & 0.39 ( 0.07 ) + & 22 & -0.05 & 0.21 & -273.85 ( 0.08 ) & -109.61 ( 0.25 ) & -305.89 & -117.69 & 0.32 ( 0.11 ) + d2 & 17 & -0.53 & 0.42 & -269.42 ( 0.05 ) & -109.57 ( 0.02 ) & -305.03 & -119.78 & 0.84 ( 0.15 ) + & 18 & -0.36 & 0.84 & -277.09 ( 0.07 ) & -112.55 ( 0.06 ) & -306.65 & -119.82 & 2.77 ( 0.09 ) + & 19 & -0.32 & 1.05 & -276.76 ( 0.15 ) & -112.71 ( 0.11 ) & -307.21 & -119.94 & 2.69 ( 0.14 ) + & 20 & -0.32 & 0.84 & -276.93 ( 0.03 ) & -113.14 ( 0.03 ) & -307.98 & -120.24 & 2.21 ( 0.11 ) + & 21 & -0.31 & 0.63 & -277.15 ( 0.05 ) & -112.99 ( 0.04 ) & -308.45 & -120.54 & 1.30 ( 0.10 ) + d3 & 17 & -53.78 & 1.90 & -0.54 ( 0.31 ) & 9.61 ( 0.23 ) & -32.91 & 3.23 & 11.60 ( 0.37 ) + & 18 & -53.89 & 4.42 & -5.33 ( 0.66 ) & 10.05 ( 0.70 ) & -34.89 & 2.77 & 75.71 ( 0.34 ) + & 19 & -53.96 & 3.16 & -5.22 ( 0.19 ) & 9.98 ( 0.19 ) & -35.67 & 2.75 & 40.02 ( 0.28 ) + & 20 & -54.16 & 2.95 & -5.11 ( 0.19 ) & 10.07 ( 0.22 ) & -36.16 & 2.98 & 25.59 ( 0.34 ) + & 21 & -53.49 & 1.90 & -5.28 ( 0.08 ) & 10.56 ( 0.10 ) & -36.59 & 3.02 & 6.14 ( 0.14 ) + d4 & 18 & -56.53 & 1.6 & -4.01 ( 0.04 ) & 11.49 ( 0.33 ) & -33.57 & 4.22 & 3.53 ( 0.15 ) + & 19 & -56.01 & 2.74 & -3.74 ( 0.19 ) & 11.33 ( 0.12 ) & -34.19 & 4.10 & 27.48 ( 0.27 ) + & 20 & -56.23 & 2.32 & -4.05 ( 0.10 ) & 10.94 ( 0.06 ) & -35.09 & 3.85 & 42.13 ( 0.66 ) + & 21 & -55.98 & 2.11 & -4.08 ( 0.08 ) & 11.73 ( 0.14 ) & -35.38 & 4.18 & 23.03 ( 0.33 ) + & 22 & -55.88 & 4.21 & -3.81 ( 0.67 ) & 12.41 ( 0.49 ) & -35.85 & 4.33 & 147.80 ( 0.53 ) + d5 & 17 & -57.10 & 0.63 & 2.73 ( 0.10 ) & 11.70 ( 0.07 ) & -29.64 & 5.32 & 2.21 ( 0.19 ) + & 18 & -57.75 & 1.69 & -0.96 ( 0.20 ) & 13.13 ( 0.37 ) & -30.52 & 5.85 & 8.73 ( 0.18 ) + & 19 & -58.54 & 2.32 & -0.38 ( 0.21 ) & 13.67 ( 0.53 ) & -30.83 & 6.44 & 23.56 ( 0.40 ) + & 20 & -59.07 & 2.53 & -0.36 ( 0.26 ) & 13.66 ( 0.62 ) & -31.41 & 6.57 & 41.85 ( 0.45 ) + & 21 & -59.27 & 2.32 & -0.34 ( 0.30 ) & 14.22 ( 0.29 ) & -31.65 & 6.68 & 58.96 ( 0.78 ) + & 22 & -59.60 & 2.11 & -0.29 ( 0.25 ) & 14.71 ( 0.18 ) & -32.32 & 6.63 & 35.44 ( 0.79 ) + & 23 & -60.14 & 1.26 & -0.73 ( 0.07 ) & 12.90 ( 0.07 ) & -33.01 & 7.30 & 42.06 ( 0.66 ) + d6 & 17 & -59.11 & 2.74 & 5.41 ( 0.35 ) & 13.70 ( 0.12 ) & -26.96 & 7.32 & 33.64 ( 0.38 ) + & 18 & -60.58 & 3.16 & 1.80 ( 0.40 ) & 14.90 ( 0.20 ) & -27.76 & 7.62 & 133.79 ( 0.29 ) + & 19 & -60.30 & 3.58 & 2.03 ( 0.57 ) & 14.52 ( 0.35 ) & -28.41 & 7.29 & 39.25 ( 0.42 ) + & 20 & -61.21 & 2.32 & 2.81 ( 0.27 ) & 14.48 ( 0.13 ) & -28.40 & 7.29 & 17.95 ( 0.34 ) + & 21 & -61.16 & 2.74 & 2.79 ( 0.12 ) & 15.23 ( 0.14 ) & -28.52 & 7.69 & 84.02 ( 0.29 ) + & 22 & -60.71 & 2.53 & 2.39 ( 0.36 ) & 15.62 ( 0.12 ) & -29.64 & 7.54 & 53.42 ( 0.84 ) + & 23 & -61.38 & 1.05 & 3.35 ( 0.04 ) & 14.04 ( 0.08 ) & -28.93 & 8.44 & 14.58 ( 0.43 ) + d7 & 18 & -63.28 & 1.47 & -6.51 ( 0.05 ) & -5.91 ( 0.11 ) & -36.07 & -13.19 & 2.22 ( 0.12 ) + & 19 & -63.78 & 2.11 & -6.67 ( 0.23 ) & -5.00 ( 0.60 ) & -37.12 & -12.22 & 5.67 ( 0.18 ) + & 20 & -64.63 & 2.11 & -7.12 ( 0.03 ) & -4.41 ( 0.09 ) & -38.17 & -11.50 & 13.95 ( 0.23 ) + & 21 & -64.79 & 2.11 & -7.45 ( 0.08 ) & -3.76 ( 0.15 ) & -38.76 & -11.31 & 11.78 ( 0.15 ) + d8 & 17 & -109.78 & 0.63 & 32.37 ( 0.74 ) & 6.38 ( 0.02 ) & 0.00 & 0.00 & 1.09 ( 0.16 ) + & 18 & -108.43 & 1.69 & 29.56 ( 0.02 ) & 7.28 ( 0.07 ) & 0.00 & 0.00 & 9.03 ( 0.11 ) + & 19 & -108.76 & 2.95 & 30.45 ( 0.06 ) & 7.23 ( 0.18 ) & 0.00 & 0.00 & 16.16 ( 0.16 ) + & 20 & -108.41 & 2.95 & 31.05 ( 0.08 ) & 7.09 ( 0.10 ) & 0.00 & 0.00 & 24.30 ( 0.32 ) + & 21 & -107.93 & 2.95 & 31.30 ( 0.07 ) & 7.62 ( 0.34 ) & 0.00 & 0.00 & 21.77 ( 0.19 ) + & 22 & -107.95 & 2.53 & 32.04 ( 0.06 ) & 8.08 ( 0.07 ) & 0.00 & 0.00 & 42.96 ( 0.38 ) + & 23 & -107.93 & 1.69 & 32.28 ( 0.02 ) & 5.60 ( 0.04 ) & 0.00 & 0.00 & 48.64 ( 0.65 ) + lcc@xmath28/feature & acceleration rate & note + & ( km s@xmath9 yr@xmath9 ) & + & & + + @xmath1 -50 km s@xmath9 & -5.44 @xmath7 0.18 & this work + @xmath1 -50 km s@xmath9 & -5.40 @xmath7 1.20 & paper i + @xmath1 -75 km s@xmath9 & -4.80 @xmath7 0.50 & paper i + average & -5.21 @xmath7 0.44 & + & & + + a5 & -3.97 @xmath7 0.03 & this work + a7 & -3.76 @xmath7 0.50 & this work + a4 & -8.01 @xmath7 0.61 & this work + d5 & -3.89 @xmath7 0.50 & this work + average & -4.91 @xmath7 0.15 & + ccc id@xmath51 & @xmath63 & @xmath64 + & ( mas yr@xmath65 & ( mas yr@xmath9 ) + d1 & -0.6 ( 0.1 ) & -1.1 ( 0.4 ) + d2 & -0.9 ( 0.6 ) & -1.1 ( 0.2 ) + d3 & -0.6 ( 0.2 ) & 1.5 ( 1.3 ) + d4 & -1.1 ( 0.3 ) & 2.6 ( 1.4 ) + d5 & 1.4 ( 0.1 ) & 2.2 ( 1.5 ) + d6 & 4.2 ( 0.3 ) & 0.7 ( 1.3 ) + d7 & -4.4 ( 0.0 ) & 9.0 ( 0.6 ) + d8 & 5.8 ( 0.1 ) & 0.14 ( 1.47 ) + + + c|cc|cc & & + i d & @xmath66 & @xmath67 & @xmath68 & @xmath69 + & & + + a1 & -2.3 ( 0.0 ) & -2.2 ( 0.2 ) & -0.9 ( 0.0 ) & 0.7 ( 0.3 ) + a2 & 0.2 ( 0.4 ) & -2.1 ( 0.1 ) & 1.6 ( 0.4 ) & 0.8 ( 0.2 ) + a3 & -3.2 ( 0.1 ) & -4.5 ( 0.2 ) & -1.8 ( 0.1 ) & -1.6 ( 0.2 ) + a4 & -1.5 ( 0.5 ) & -5.1 ( 0.4 ) & -0.1 ( 0.5 ) & -2.2 ( 0.4 ) + a5 & 0.8 ( 0.0 ) & -2.6 ( 0.3 ) & 2.3 ( 0.0 ) & 0.3 ( 0.3 ) + a6 & -1.6 ( 0.1 ) & 0.8 ( 1.5 ) & -0.2 ( 0.1 ) & 3.7 ( 1.5 ) + a7 & - & - & 1.4 ( 0.0 ) & 2.9 ( 0.1 ) + + b1 & -3.1 ( 0.2 ) & -3.1 ( 0.3 ) & 0.0 ( 0.3 ) & -1.4 ( 0.3 ) + b2 & -4.2 ( 0.4 ) & -0.9 ( 0.2 ) & -1.1 ( 0.4 ) & 0.8 ( 0.2 ) + b3 & -2.9 ( 0.2 ) & -1.8 ( 0.1 ) & 0.2 ( 0.2 ) & -0.1 ( 0.1 ) + b4 & - & - & 3.1 ( 0.1 ) & 1.7 ( 0.1 ) + + c1 & 1.1 ( 0.6 ) & 1.4 ( 0.9 ) & -6.0 ( 0.6 ) & -2.8 ( 0.9 ) + c3 & 3.2 ( 0.2 ) & 4.5 ( 0.3 ) & -3.9 ( 0.2 ) & 0.3 ( 0.4 ) + c4 & 0.1 ( 0.3 ) & 4.3 ( 0.7 ) & -7.0 ( 0.3 ) & 0.2 ( 0.8 ) + c5 & 3.9 ( 0.8 ) & 6.3 ( 1.9 ) & -3.2 ( 0.8 ) & 2.1 ( 1.9 ) + c6 & 1.8 ( 0.9 ) & 2.9 ( 2.4 ) & -5.3 ( 0.9 ) & -1.3 ( 2.4 ) + c7 & 7.3 ( 0.0 ) & 4.2 ( 0.2 ) & 0.3 ( 0.1 ) & 0.0 ( 0.3 ) + c2 & - & - & -7.1 ( 0.0 ) & -4.2 ( 0.2 ) + + d1 & -6.8 ( 0.1 ) & -2.2 ( 0.1 ) & -1.5 ( 0.2 ) & -3.0 ( 0.1 ) + d2 & -7.2 ( 0.1 ) & -1.4 ( 0.3 ) & -1.9 ( 0.1 ) & -2.2 ( 0.3 ) + d3 & -7.0 ( 0.7 ) & -0.2 ( 0.5 ) & -1.7 ( 0.7 ) & -1.1 ( 0.5 ) + d4 & -6.7 ( 0.3 ) & -0.2 ( 1.3 ) & -1.4 ( 0.4 ) & -1.1 ( 1.3 ) + d5 & -4.3 ( 0.1 ) & 2.5 ( 0.1 ) & 1.0 ( 0.1 ) & 1.7 ( 0.1 ) + d6 & -1.8 ( 0.4 ) & 1.4 ( 0.4 ) & 3.6 ( 0.5 ) & 0.5 ( 0.4 ) + d7 & -10.2 ( 0.5 ) & 7.5 ( 1.0 ) & -4.9 ( 0.5 ) & 6.6 ( 1.0 ) + d8 & - & - & 5.3 ( 0.1 ) & -0.9 ( 0.1 ) + + cccccccid & region@xmath51 & 2d motions@xmath52 & @xmath28 - @xmath10@xmath70 & @xmath71 & position angle & inclination@xmath54 + & & ( km s@xmath9 ) & ( km s@xmath9 ) & ( km s@xmath9 ) & ( @xmath29 ) & ( @xmath29 ) + a1 & c & 9.6 ( 1.2 ) & -18.9 & 0.7 & -52.0 @xmath72@xmath73 & 27.0 ( 3.0 ) + a2 & e & 14.7 ( 2.8 ) & -38.8 & 0.4 & 62.8 @xmath74@xmath75 & 20.8 ( 3.6 ) + a3 & c & 19.7 ( 1.1 ) & -47.7 & 2.4 & -131.2 @xmath76@xmath77 & 22.4 ( 1.1 ) + a4 & c & 17.7 ( 3.3 ) & -49.2 & 0.8 & -177.9 @xmath78@xmath79 & 19.8 ( 3.5 ) + a5 & e & 18.3 ( 0.4 ) & -59.7 & 0.8 & 82.7 @xmath80@xmath81 & 17.1 ( 0.3 ) + a6 & e & 29.9 ( 12.0 ) & -66.4 & 1.4 & -3.5 @xmath82@xmath83 & 24.2 ( 9.2 ) + a7 & e & 25.9 ( 0.6 ) & -68.8 & 0.7 & 25.9 @xmath84@xmath85 & 20.7 ( 0.5 ) + b1 & c & 11.6 ( 2.6 ) & -47.0 & 1.7 & 179.2 @xmath86@xmath87 & 13.9 ( 3.0 ) + b2 & c & 11.1 ( 3.0 ) & -48.8 & 1.3 & -52.1 @xmath88@xmath89 & 12.8 ( 3.3 ) + b3 & c & 1.8 ( 1.8 ) & -51.7 & 1.4 & 110.7 @xmath90@xmath91 & 2.0 ( 2.0 ) + b4 & e & 28.5 ( 1.0 ) & -66.9 & 0.6 & 61.2 @xmath92@xmath93 & 23.1 ( 0.7 ) + c1 & c & 53.2 ( 5.6 ) & -27.4 & 0.5 & -115.1 @xmath94@xmath95 & 62.7 ( 2.7 ) + c2 & c & 66.6 ( 0.8 ) & -39.5 & 0.6 & -120.6 @xmath96@xmath97 & 59.3 ( 0.3 ) + c3 & c & 31.9 ( 1.4 ) & -42.6 & 1.1 & -85.3 @xmath98@xmath99 & 37.5 ( 1.3 ) + c4 & c & 56.8 ( 2.7 ) & -47.0 & 2.0 & -88.8 @xmath100@xmath101 & 50.4 ( 1.4 ) + c5 & c & 31.2 ( 10.0 ) & -51.0 & 1.0 & -57.0 @xmath102@xmath103 & 31.5 ( 8.8 ) + c6 & c & 44.1 ( 8.7 ) & -52.6 & 1.4 & -103.6 @xmath104@xmath105 & 40.0 ( 6.0 ) + c7 & e & 2.0 ( 0.4 ) & -102.7 & 1.2 & 95.3 @xmath106@xmath107 & 1.1 ( 0.2 ) + d1 & w & 27.3 ( 0.9 ) & 4.9 & 0.2 & -153.0 @xmath108@xmath109 & 79.8 ( 0.3 ) + d2 & w & 23.3 ( 2.0 ) & 4.6 & 0.4 & -140.3 @xmath110@xmath111 & 78.8 ( 1.0 ) + d3 & c & 16.0 ( 5.0 ) & -48.9 & 1.3 & -122.6 @xmath112@xmath113 & 18.1 ( 5.4 ) + d4 & c & 14.1 ( 6.9 ) & -51.1 & 1.2 & -127.0 @xmath114@xmath115 & 16.3 ( 7.4 ) + d5 & c & 15.7 ( 0.7 ) & -53.8 & 0.7 & 32.0 @xmath116@xmath117 & 16.3 ( 0.7 ) + d6 & c & 29.0 ( 3.6 ) & -55.6 & 1.0 & 81.5 @xmath118@xmath119 & 27.5 ( 3.0 ) + d7 & c & 66.6 ( 6.9 ) & -59.1 & 1.0 & -36.4 @xmath120@xmath121 & 48.4 ( 3.1 ) + d8 & e & 43.4 ( 0.4 ) & -103.4 & 0.8 & 99.1 @xmath122@xmath123 & 22.8 ( 0.2 ) + + + + + + breen , s. l. , caswell j. l. , ellingsen , s. p. , & phillips , c. j. 2010 , mnras , 406 , 1487 caswell , j. l. , phillips , c. j. , 2008 , mnras , 386 , 1521 ( cp08 ) furuya , r. s. , kitamura , y. , wootten , h. a. , claussen , m. j. , saito , m. , marvel , k. b. , & kawabe , r. 2000 , apj , 542 , l135 gibb , a. g. , & hoare , m. g. 2007 , mnras , 380 , 246 greenhill , l. j. , goddi , c. , chandler , c. j. , matthews , l. d. , & humphreys , e. m. l. 2013 , , 770 , l32 jimnez - serra , i. , martn - pintado , j. , winters , j .-
, rodrguez - franco , a. , & caselli , p. 2011
, , 739 , 80 kawaguchi , n. , sasao , t. , & manabe , s. 2000 , , 4015 , 544 kim , j - s . , et al . 2013 , , 767 , 86 honma , m. , et al . , 2003 , pasj , 55 , l57 honma , m. , tamura , y. , & reid , m. j. 2008 , pasj , 60 , 951 machida , m. n. , inutsuka , s .-
i . , & matsumoto , t. 2008 , apj , 676 , 1088 matthews , l. d. , greenhill , l. j. , goddi , c. , chandler , c. j. , humphreys , e. m. l. , kunz , m. w. 2010 , apj , 708 , 80 maud , l. t. , & hoare , m. g. , 2013 , , 779 , l24 moscadelli l. , cesaroni r. , & rioja m. j. 2000 , a&a , 360 , 663 motogi , k. , et al .
2008 , mnras , 390 , 523 motogi , k. , sorai , k. , habe , a. , honma , m. , kobayashi , h. , & sato , k. , 2011 , pasj , 63 , 31 motogi , k. , et al .
2011 , mnras , 417 , 238 ( paper i ) motogi , k. , sorai , k. , niinuma , k. , sugiyama , k. , honma , m. , fujisawa , k. 2013 , mnras , 428 , 349 neckel , t. 1978 , apj , 69 , 51 seifried , d. , pudritz , r. e. , banerjee , r. , duffin , d. , & klessen , r. s. 2012 , , 422 , 347 shepherd , d. s. , borders , t. , claussen , m. , shirley , y. , & kurtz , s. 2004 , apj , 614 , 211 sorai , k. , et al . 2008 , pasj , 60 , 1285 surcis , g. , et al .
2014 , apj , 565 , l8 torrelles , j. m. , et al .
2003 , apj , 598 , l115 torrelles , j. m. , et al .
2011 , mnras , 410 , 627 trinidad , m. a. , torrelles , j. m. , rodrguez , l. f. , & curiel , s. , 2007 , apj , 134 , 1870 wu , y. , wei , y. , zhao , m. , shi , y. , yu , w. , qin , s. , huang , m. 2004 , , 426 , 503
|
we report on a long - term single - dish and vlbi monitoring for intermittent flare activities of a dominant blue - shifted h@xmath0o maser ( dbsm ) associated with a southern high mass young stellar object , g353.273 + 0.641 .
bi - weekly single - dish monitoring using hokkaido university tomakomai 11-m radio telescope has shown that a systematic acceleration continues over four years beyond a lifetime of individual maser features .
this fact suggests that the h@xmath0o maser traces a region where molecular gas is steadily accelerated .
there were five maser flares during five - years monitoring , and maser distributions in four of them were densely monitored by the vlbi exploration of radio astrometry ( vera ) .
the overall distribution of the maser features suggests the presence of a bipolar jet , with the 3d kinematics indicating that it is almost face - on ( inclination angle of @xmath1 8@xmath217@xmath2 from the line - of - sight ) .
most of maser features were recurrently excited within a region of 100@xmath3100 au@xmath4 around the radio continuum peak , while their spatial distributions significantly varied between each flare .
this confirms that episodic propagations of outflow shocks recurrently invoke intermittent flare activities .
we also measured annual parallax , deriving the source distance of 1.70 @xmath5 kpc that is consistent with the commonly - used photometric distance .
|
for over half a century , the crab nebula has played a key role as a laboratory for many seminal discoveries regarding supernovae and their remnants @xcite .
it is also the brightest and best studied example of `` plerionic '' remnants which are powered by a compact central object ( see reviews @xcite ) . despite its many astronomical firsts ,
the detection by @xcite of chains of semi - stellar , optical knots showing unusually strong [ ar iii ] line emission was both remarkable and puzzling .
the discovery came about through a search looking for north - south , bi - polar axial phenomena directly associated with the crab pulsar and related to the remnant s east - west band of strong helium emission filaments @xcite and the n - s hourglass structure seen in polarized light of the crab s synchrotron nebula @xcite . using a fabry - perot imager centered at 5015.3 , @xcite found about a dozen , semi - stellar [ o iii ] @xmath05007 emission knots aligned in arcs , perhaps helical @xcite , from the pulsar s position with seven to the north and four to the south .
the northern arc of knots appeared situated inside a corridor through the remnant s filamentary structure , the eastern edge of which seemed to merge with the western edge of the crab s well - known northern `` jet '' @xcite .
@xcite suggested that this corridor containing the northern knots marked the presence of a directed , north - south collimated wind from the pulsar s vicinity and possibly associated with the pulsar s spin axis @xcite .
subsequent optical spectra of several of these knots revealed remarkably strong [ ar iii ] @xmath07136 and [ s ii ] @xmath1 line emission .
the knots strong [ ar iii ] line emission was deemed more remarkable than that of the [ s ii ] , leading macalpine et al . to refer them as `` argo - knots '' .
their unusual [ ar iii ] line strength was interpreted as being due to either n(ar@xmath2)/n(h@xmath3 ) ratios @xmath4 higher than in typical crab filaments , unusually high knot electron temperatures , or some process(es ) not normally important in nebula astrophysics .
perhaps most remarkable was the knots apparent alignment with , and transverse velocities as high as 900 km s@xmath5 away from , the crab pulsar .
the knots arrangement in north and south arcs in striking alignment with the crab pulsar was highly suggestive of a causal link .
this is because the pulsar has a motion toward the northwest away from the crab s expansion point of @xmath6125 km s@xmath5 ( @xmath7 yr@xmath5 , pa = 290@xmath8 ; @xcite ) .
in addition to the location of the northern knots inside a filamentary corridor associated with the northern jet feature , @xcite also found that the argo - knots north - south alignment to be roughly perpendicular to an apparent east - west torus of high - helium filaments @xcite .
positional measurements for the seven northern and four southern [ ar iii ] knots using ground - based images taken over a two year baseline ( 1991 1993 ) indicated proper motions of @xmath9 yr@xmath5 or @xmath10 900 km s@xmath5 at the crab s 1830 pc distance @xcite .
although their proper motion results had relatively large associated uncertainties ( @xmath11 yr@xmath5 ) , @xcite concluded that `` all measured north or south motions are directed away from the pulsar , as expected '' .
this conclusion , when taken together with the knots remarkable alignment with the pulsar s current position suggested a kinematic link between the pulsar and these so - called `` argo - knots '' .
however , if these knots were dynamically connected with a collimated wind off the pulsar , it would then mean that they comprise a small but previously unrecognized population of line emitting material within the remnant which does not participate in the remnant s radial expansion from the sn 1054 explosion point @xcite .
non - radial motions in the crab nebula and in particular the presence of high - velocity , line - emitting gas moving rapidly away north and south from the pulsar is an unexpected discovery meriting further examination . in this paper , we present a proper motion and morphological study of a few of these knots using 1994 - 1999 _ hst _ wfpc-2 images of the crab nebula .
wfpc-2 images of the center of the crab nebula from 1994 , 1995 and 1999 were obtained from the data archive at the space telescope science institute ( table 1 ) .
figure 1 shows a 1995 _ hst _
wfpc2 image of the west - central region of the crab nebula taken with the [ o iii ] @xmath05007 filter ( f502n ) .
the strong [ ar iii ] emitting knots n1 n6 and s1
s4 described by @xcite are marked .
other nebular knots having similar `` semi - stellar '' [ o iii ] morphologies are shown in the image and are labeled k1 k9 . because all 11 bright [ ar iii ] knots exhibit prominent [ o iii ] @xmath05007 and [ s ii ] @xmath126716,6731 line emissions and positive radial velocities ( + 300 to + 720 km s@xmath5 ; @xcite ) , wfpc-2 images taken using the [ o iii ] f502n and [ s ii ] f673n filters are well suited for studying these knots .
both filter bandpass centers are slightly redshifted ( + 360 km s@xmath5 ) and sufficiently wide ( @xmath13 km s@xmath5 ) to cover the knots observed radial velocity range .
in addition , the [ o iii ] f502n image line center at 5013 is almost identical to the 5015 fabry - perot bandpass center used by @xcite to measure the [ ar iii ] knots kinematic properties .
pairs of wfpc-2 1994 and 1999 images taken in [ o iii ] and [ s ii ] were each combined using the cosmic ray cleaning iraf package `` crrej '' . for the 1995 data ,
two pairs of images were available .
these were also initially combined using `` crrej '' and then one pair was shifted and they were combined again using `` crrej '' .
the combined images for each year were then made into three mosaics representing 1994 , 1995 , and 1999 , and rotated so that all were oriented with north up and east to the left .
pixel image positions of four to six stars present on the 1994 , 1995 , and 1999 images were measured on each of the [ o iii ] or [ s ii ] wfpc-2 images ( @xmath14 pixel@xmath5 ) using the iraf task `` imexamine '' .
image positions for the four southern [ ar iii ] bright nebular knots identified by @xcite were measured both by eye and by using `` imexamine '' , with good agreement between the two methods . since
southern knots s3 and s4 were not imaged on the 1994 images , their proper motion values were determined by comparing only the 1995 and 1999 images . also measured was a similar but somewhat fainter , neighboring feature to knots s1 and s2 which we call `` s0 '' due to its smaller , projected distance from the pulsar than knot s1 .
pixel x and y offsets between the 1994/95 and 1999 images were then obtained and used to predict the 1999 pixel positions of the knots from the 1994 and 1995 images . observed versus predicted knot positions divided by the time interval yielded @xmath15 and @xmath16 ( x and y ) proper motion values .
image pixel x and y differences for the reference stars were each averaged and the standard deviation of each was used as an estimate of the minimum measurement errors .
these were added in quadrature with error estimates based on position differences between the [ o iii ] and [ s ii ] knots . because of the large and irregular morphology of knot s3 , its proper motion determination is significantly more uncertain .
the crab pulsar currently lies some @xmath17 northwest of the remnant s estimated expansion center @xcite . to determine whether , as @xcite concluded , [ ar iii ] bright knots move away from the the crab pulsar `` as expected '' due to a presumed collimated wind off the pulsar or instead move away the remnant s expansion center , we chose to examine the four southern [ ar iii ] bright knots they identified ( s1s4 ) .
we also included one additional likely [ ar iii ] knot ( s0 ) we identified from the wfpc-2 [ o iii ] images .
this latter knot may be the one @xcite referred to as being simply `` to the south '' of the pulsar .
due to locations south_east _ of the pulsar but south_west _ of the crab s center of expansion , the proper motions of the three innermost knots ( s0 , s1 , & s2 ) would be expected to have either a strongly positive or negative right ascension term if moving away from the pulsar or the remnant s center of expansion respectively .
that is , they ought to move toward the southeast if expanding away from the pulsar ( pa @xmath18 ) or toward the southwest if expanding away the crab s expansion center ( pa @xmath19 ) .
in contrast , all seven northern knots would be expected to show proper motions toward the northwest regardless of their origin point , thus making them a less powerful discrimination test without highly accurate positional measurements .
in addition , the innermost northern knot , n1 , lying just @xmath20 north of the pulsar , is a poorly defined feature and lies projected against a relatively bright filamentary background making precise positional measurements more difficult .
table 2 lists the positions of the five southern knots ( relative to the central star just northeast of the pulsar ; star 16 of @xcite ) , our proper motion measurements , and with those cited by @xcite . within measurement
uncertainites both data sets are in agreement .
that is , although our estimated knot motions ( except for knot s1 ) are significantly smaller than those quoted by @xcite , they all lie within the macalpine et al.s relatively large error bar of @xmath11 yr@xmath5 .
however , it is puzzling why @xcite concluded in favor of motions directed away from the pulsar with transverse velocities `` of order '' 900 km s@xmath5 based on their poorly constrained proper motion measurements .
our east - west motions ( @xmath16y values ) of these five southern knots are more than an order of magnitude below the @xcite detection limit .
but they are in line with simple radial motion away from the center of expansion like that experienced by other remnant filamentary features .
a comparison of our knot proper motion estimates relative to the projected displacements of the knots from the center of expansion determine by @xcite is given in table 3 . here the average knot proper motions from the @xmath16x and @xmath16y displacements were calculated assuming these knots experienced the same acceleration as the general nebula leading to a derived explosion date @xmath6 a.d .
1130 @xcite .
as the table shows , our estimated knot proper motions are entirely consistent with the knots projected radial displacements from the crab s center of expansion .
[ note the somewhat better agreement using the @xcite expansion center estimate . ]
table 3 also lists the implied transverse velocities for the knots based on our proper motion estimates , knot radial velocities reported by @xcite , and the resulting space velocities .
the five knots space velocities range from 440 and 740 km s@xmath5 .
this is on the low end of the remnant s @xmath21 km s@xmath5 filamentary expansion range @xcite .
this , in turn , places their location well inside the remnant s filamentary shell and thus relatively close to the x - ray and uv bright synchrotron nebula .
our proper motion estimates for all five knots indicate motions away from the remnant s expansion center and not the pulsar .
specifically , the position angles ( pa ) for knots s0s2 are all @xmath22 180@xmath8 , consistent with expected motions from either trimble s or nugent s estimated expansion centers .
this is in contrast to the expected pa values of around 152@xmath8 if these knots were moving away from the pulsar .
our results are graphically shown in figure 2 where we plot the 100 yr proper motions of the five southern knots based on our measurements . as we found for the knots total proper motion values , our measurements are in somewhat better agreement with the @xcite expansion center than that of @xcite .
as a qualitative test of our results , we registered the 1994/95 images with the 1999 images and then blinked them .
we found the [ ar iii ] knots motions to be visually consistent with radial motion away from the center of expansion as opposed to away from the pulsar , supporting our quantitative results .
we conclude the kinematics of these knots did not appear to be unusual relative to other nebular features .
we believe our derived knot proper motions to be more reliable than those of @xcite for several reasons .
first , they used ground - based [ o iii ] images with a much coarser pixel scale ( @xmath23 ) compared the _ hst _ wfpc-2 images ( @xmath24 ) and with much lower image quality ( @xmath25 seeing vs. wfpc-2 s fwhm of @xmath26 ) .
second , their images covered just a 2.0 yr time span compared to our 5.66 yr ( 19941999 ) and 4.81 yr ( 19951999 ) time coverage .
one criteria @xcite used to help identify [ ar iii ] knots was a semi - stellar and/or isolated appearance on their ground - based images . when examined using the higher resolution wfpc-2 images , however , the knots described by @xcite show a broad range of morphologies and angular dimensions . in figure 3 , we show enlargements of magnified wfpc-2 images for several of these [ ar iii ] knots . while many clearly do have a somewhat stellar appearance with diameters of @xmath6 @xmath27 , others appear fairly diffuse with diameters of up to @xmath28 ( e.g. , knots n1 , n6 & s3 )
. figure 3 also shows enlarged magnifications of nine other nebular knots ( labeled k1
k9 ) with sizes and morphologies similar to the `` argoknots '' .
one of these , k7 , was noted by @xcite as one of two other possible fainter and less distinct gas condensations .
similar knots are found throughout the central part of the remnant ( see fig
. 1 ) and often lie near the north and south arcs of [ ar iii ] knots .
the strong similarities in morphology of the [ ar iii ] knots to these other nebular knots , and the placement of these other knots around the n - s knot arc , suggests that the [ ar iii ] knots are not unusual remnant features .
our study of the proper motions of five [ ar iii ] bright knots using _ hst _ wfpc2 data from 1994 to 1999 show ordinary radial motions away from the nebula s center of expansion with magnitudes consistent with their projected radial displacement .
the measurements indicate the knots are moving away from the remnant s expansion center and not the pulsar .
in addition , the [ ar iii ] knots do not appear to have unusual morphologies , and we have identified several other similarly appearing knots in the remnant .
these results do not support the suggestion by @xcite that these knots might be moving rapidly away from the crab pulsar due to a collimated pulsar - driven wind or that these knots are unusual in appearance .
on the other hand , the spectral data of @xcite clearly show they possess strong [ o iii ] @xmath124959,5007 with unusually strong [ ar iii ] and [ s ii ] line emissions .
macalpine et al .
and @xcite noted a possible spatial association of the northern argo - knots with a filament at a higher radial velocity ( + 900 to + 1300 km s@xmath5 ) .
the [ ar iii ] bright knots were seen mainly at lower - intensity breaks in the filament s emission .
they suggested that this might indicate that the material in the knots is derived from this filament possibly though some instability ( rayleigh - taylor or kelvin - helmholtz ) .
magnetic rayleigh - taylor instabilities at the interface of this filament and the pulsar - generated synchrotron nebula as discussed by @xcite and @xcite might well explain the [ ar iii ] knot morphologies .
the tips of the rayleigh - taylor , finger - like filaments seen edge - on near the outer parts of the remnant ( e.g. , filaments f , g , & h ; @xcite ) show clumps of similar size and shape to the [ ar iii ] knots .
the only difference here may be the viewing angle , where the [ ar iii ] knots are seen face - on .
the observed difference in radial velocity between the coincident filament and the chain of [ ar iii ] knot would be consistent with their more interior remnant positions .
the estimated 440 740 km s@xmath5 space velocities for these knots , at the extreme low end for the crab s filaments , is also consistent with this picture .
this would place them closer to the synchrotron nebula thus exposing them to higher x - ray and uv photoionization fluxes .
this , in turn , might be the underlying cause for their strong [ ar iii ] line emission via higher gas temperatures as suspected by @xcite .
lcccll 1994 feb 23 & u24r0203t204 t & f502n & [ o iii ] & center - nw & 2000 ; 2000 + 1994 mar 9 & u24r0401t402 t & f673n & [ s ii ] & center - nw & 2000 ; 2000 + 1995 jan 2 & u2bx0401t402 t & f673n & [ s ii ] & center - west & 2000 ; 2300 + 1995 jan 2 & u2bx0403t404 t & f673n & [ s ii ] & center - west & 2300 ; 2300 + 1995 jan 5 & u2bx0301t302 t & f502n & [ o iii ] & center - west & 2000 ; 2300 + 1995 jan 5 & u2bx0303t304 t & f502n & [ o iii ] & center - west & 2300 ; 2300 + 1999 oct 24 & u5d10305r306r & f502n & [ o iii ] & center - sw & 2600 ; 2600 + 1999 oct 24 & u5d10307r308 m & f673n & [ s ii ] & center - sw & 1300 ; 1300 + cclcccccc s0 & @xmath29 & @xmath30 & @xmath31 & @xmath32 & 0.013@xmath33 & & & + s1 & @xmath34 & @xmath35 & @xmath36 & @xmath37 & 0.014@xmath38 & 0.00 & 0.00 & @xmath39 + s2 & @xmath40 & @xmath41 & @xmath36 & @xmath42 & 0.018@xmath33 & + 0.05 & @xmath43 & @xmath44 + s3 & @xmath45 & @xmath46 & @xmath47 & @xmath48 & 0.021@xmath49 & 0.00 & @xmath50 & @xmath51 + s4 & @xmath52 & @xmath53 & @xmath54 & @xmath55 & 0.036@xmath56 & 0.00 & @xmath50 & @xmath51 + cccccccccc s0 & 0.011 & 0.013@xmath33 & 120 & & & 215 @xmath57 & 206 & 214 & 152 + s1 & 0.012 & 0.014@xmath38 & 130 & 420 & 440 & 206 @xmath58 & 197 & 205 & 151 + s2 & 0.015 & 0.018@xmath33 & 170 & 720 & 740 & 198 @xmath59 & 187 & 194 & 152 + s3 & 0.022 & 0.021@xmath49 & 200 & 630 & 660 & 169 @xmath60 & 167 & 173 & 147 + s4 & 0.035 & 0.036@xmath56 & 340 & 400 & 525 & 175 @xmath61 & 168 & 172 & 153 + blair , w. p. , davidson , k. , fesen , r. a. , uomoto , a. , macalpine , g. m. , & henry , r. b. c. 1997 , , 109 , 473 davidson , k. , & fesen , r. a. 1985 , , 23 , 119 fesen , r. a. , martin , c. l. , & shull , j. m. 1992 , , 399 , 599 hester , j. j. et al .
1996 , , 456 , 225 lawrence , s. s. , & macalpine , g. m. 1994 , , 26 , 951 lawrence , s. s. , macalpine , g. m. , uomoto , a. , woodgate , b. e. , brown , l. w. , oliversen , r. j. , lowenthal , j. d. , & liu , c. 1995 , , 109 , 2635 kafatos , m. c. , & henry , r. b. c. 1985 , `` the crab nebula and related supernova remnants - an overview '' , ( cambridge , cambridge univ . press ) macalpine , g. m. , mcgaugh , s. s. , mazzarella , j. m. , & uomoto , a. 1989 , , 342 , 364 macalpine , g. m. , lawrence , s. s. , & brown , b. a. 1993 , , 25 , 785 macalpine , g. m. , lawrence , s. s. , uomoto , a. , woodgate , b. e. , brown , l. w. , oliversen , r. j. , lowenthal , j. d. , & liu , c. 1992 , , 24 , 791 macalpine , g. m. , lawrence , s. s. , brown , b. a. , uomoto , a. , woodgate , b. e. , brown , l. w. , oliversen , r. j. , lowenthal , j. d. , & liu , c. 1994 , , 432 , l131 michel , f. c. , scowen , p. a. , dufour , r. j. , & hester , j. j. 1991 , , 368 , 463 nugent , r. l. 1998 , , 110 , 831 trimble , v. 1968 , , 73 , 535 trimble , v. 1985 , in `` the crab nebula and related supernova remnants - an overview '' , ( cambridge , cambridge univ . press ) , p 257 sankrit , r. et al . 1998 , , 504 , 344 uomoto , a. , & macalpine , g. m. 1987 , , 160 l27 van den bergh , s. 1970 , , 160 , l27 wyckoff , s. , & murray , c. a. 1977 , , 180 , 717
|
the kinematic and morphological properties of a string of [ ar iii ] bright knots in the crab nebula are examined using 1994 1999 _ hst _
wfpc-2 images of the remnant .
we find that five southern [ ar iii ] bright knots exhibit ordinary radial motions away from the nebula s center of expansion with magnitudes consistent with their projected radial displacements .
this result does not support the suggestion by @xcite that these knots might be moving rapidly away from the crab pulsar due to a collimated wind .
the _ hst _
images also do not show that the [ ar iii ] knots have unusual morphologies relative to other features in the remnant .
our proper motion results , when combined with radial velocity estimates , suggest these knots have relatively low space velocities implying relatively interior remnant locations thus placing them closer to the ionizing radiation from the crab s synchrotron nebula .
this might lead to higher knot gas temperatures thereby explaining the knots unusual line emission strengths as @xcite suspected .
|
its unusual electronic energy band structure and charge carriers resembling massless dirac fermions have made graphene honeycomb structure an active field of research@xcite .
quasi 1d graphene ribbons have even more interesting electronic and magnetic properties depending on their size and symmetry@xcite .
these are edge states of zigzag ribbons with opposite spin polarization@xcite and band gaps varying with the width of the ribbon@xcite .
theoretical studies have predicted that energetic electrons and ions can induce polymorphic atomic defects , such as vacancies in graphene @xcite . using high - resolution tem the observation of vacancies
have been reported @xcite .
recent studies have shown that vacancies created on two - dimensional ( 2d ) graphene by high - energy electron or ion irradiation can induce magnetism in a system consisting of only @xmath0-electrons@xcite .
it has been argued that stoner magnetism with high @xmath1 originates from the spin - polarized extended states induced by the vacancy defects , while rkky coupling is suppressed .
these effects of defects on 1d semiconducting graphene nanoribbons should be more complex and interesting , because their band gap , magnetic state and symmetry are expected to intervene .
present study based on extensive first - principles as well as empirical tight - binding calculations ( etb ) has shown that the band gap and magnetic state of any armchair or zigzag nanoribbons can be modified by single or multiple vacancies ( holes ) .
the effects of these defects depend on their symmetry , repeating periodicity and positions .
even if the formation of periodic defects of desired symmetry may not easily be achieved , similar effects can be created through the potential difference across a ribbon applied by periodically arranged tips . when combined with various properties of nanoribbons these results
can initiate a number of interesting applications .
we believe that our results are important for further studies , since the graphene ribbons can now be produced with precision having widths sub 10 nm@xcite and nanodevices can be fabricated thereof@xcite .
we have performed first - principles plane wave calculations within density functional theory ( dft ) using paw potentials @xcite .
the exchange correlation potential has been approximated by generalized gradient approximation ( gga ) using pw91 @xcite functional both for spin - polarized and spin - unpolarized cases .
all structures have been treated within supercell geometry using the periodic boundary conditions .
a plane - wave basis set with kinetic energy cutoff of 500 ev has been used . in the self - consistent potential and total energy calculations
the brillouin zone ( bz ) is sampled by ( 1x1x35 ) special * k*-points for ribbons .
this sampling is scaled according to the size of superlattices .
all atomic positions and lattice constants are optimized by using the conjugate gradient method where total energy and atomic forces are minimized .
the convergence for energy is chosen as 10@xmath2 ev between two steps , and the maximum force allowed on each atom is less than 0.02 ev / .
numerical plane wave calculations have been performed by using vasp package @xcite
we start by summarizing the electronic properties of graphene nanoribbons which are relevant for the present study . we will consider hydrogen terminated nanoribbons if it is not stated otherwise .
armchair graphene nanoribbons , agnr(@xmath3 ) ( @xmath3 being the number of carbon atoms in the primitive unit cell ) , are non - magnetic semiconductors .
the band gap@xcite , @xmath4 , of a bare or hydrogen terminated agnr(@xmath3 ) is small for @xmath5 ( @xmath6 being an integer ) , but from @xmath7 to @xmath8 it increases and passing through a maximum it becomes again small at the next minimum corresponding to @xmath9 . as @xmath4 oscillates with @xmath3 its value shall decrease eventually to zero as @xmath10 .
bare and hydrogen terminated zigzag nanoribbons , zgnr(@xmath3 ) , are also semiconductors with @xmath4 decreasing consistently as @xmath3 increases for @xmath118 , but their edge states give rise to antiferromagnetic ( afm ) ground state@xcite .
we will show that these magnetic and electronic properties can be modified by defects originating from vacancies or holes created in those ribbons .
a hole can be created when carbon atoms at the corners of any hexagon of an armchair ribbon are removed and subsequently remaining six two - fold coordinated carbon atoms are terminated with hydrogen atoms .
figure [ fig:1-a ] ( a ) - ( b ) shows that the electronic structure of agnr(@xmath12 ) is strongly modified by such a hole which is placed at the center of the ribbon .
the hole repeats itself at each supercell which comprises six primitive cells corresponding to a repeat period of @xmath13=6 .
this ribbon having a periodic hole ( or defect ) is specified as agnr(@xmath3;@xmath13 ) and has non - magnetic ground state . despite large separation of periodic defect which hinders their direct coupling
such a strong modification of the band gap is somehow unexpected .
however , it is an indirect effect and occurs since the itinerant ( bloch ) states of band edges are modified by the defect . at the end , the direct band gap@xcite at the @xmath14-point has widened from 0.09 ev to 0.40 ev due to a defect situated at the center of the ribbon .
agnr ( 36 ) exhibit the same behavior ; namely its band gap increases when a similar hole repeats itself at each supercell consisting of six primitive cell .
however , in contrast , the band gap of agnr(38 ) , which is normally larger than that of agnr(34 ) , is reduced if the same hole is introduced at its center .
in addition , states localized around the defect have formed flat bands near the edge of valence and conduction bands because of their reduced coupling . moreover , the effect of this defect is strongly dependent on its position relative to the both edges of the nanoribbon as depicted in fig .
[ fig:1-b ] ( a ) . as shown in fig .
[ fig:1-b ] ( b ) , the changes in band gap depend on @xmath3 of agnr ( in the family specified as @xmath15 , @xmath16 being -2 , 0 and + 2 ) as well as on the position of the hole .
we note that the variation of the band gap with the position of the hole relative to the edges of the ribbon show similar trends for both agnr ( 34,6 ) and agnr ( 36;6 ) .
however , agnr ( 38;6 ) displays an opposite trend .
= 34;@xmath13=6 ) with and ( b ) without a hole consisting of six carbon vacancies .
( c ) charge density isosurfaces of selected states .
carbon atoms ( represented by black circles ) , which have coordination number lower than 3 are terminated by hydrogen atoms ( represented by small gray circles).,width=300 ] , @xmath13 ) with the position of the hole specified as 6v .
( d ) variation of band gap with repeat periodicity , @xmath13 .
results for @xmath17 are obtained by first - principles calculations.,width=300 ] etb calculations indicate that a similar behavior is also obtained in agnr(@xmath3;@xmath13 ) s having relatively larger @xmath3 ( i.e. @xmath3=58 , 60 , 62 ) in fig [ fig:1-b ] ( c ) . as expected the effect of the hole on the band gap
depends on the repeat periodicity .
as shown in fig [ fig:1-b ] ( d ) , the effects of defect decrease with increasing repeat periodicity @xmath13 .
we found that larger holes with different geometry and rotation symmetry can result in diverse electronic structure and confined states .
it should be noted that a repeating hole can also modify the mechanical properties .
for example , the stiffness of a ribbon is reduced by the presence of a hole .
the force constant , @xmath18 ( @xmath19 being the lattice constant ) calculated for agnr(34;6 ) with a hole at its center ( @xmath20 6.03 ev / ) is found to be smaller than that without a hole @xmath217.50 ev / ) .
we now show that different types of vacancies in the same armchair nanoribbon gives rise to different changes in electronic and magnetic properties . a divacancy created in agnr(22 )
can cause to a dramatic change in the electronic state of the ribbon when it is repeated with the periodicity of @xmath13=5 .
such an armchair ribbon is specified as agnr(22;5 ) .
the divacancy first relaxes and forms an eight fold ring of carbon atoms which is adjacent to six hexagons and two pentagons . in fig .
[ fig:2 ] ( a ) we see that the non - magnetic and semiconducting agnr(22 ) with band gap of @xmath22=0.18 ev becomes a non - magnetic metal , since a flat band derived from the defect occurs below the top of the valance band edge and causes to a metallic state .
the effect of a single carbon vacancy becomes even more interesting .
a single vacancy created in agnr(22 ) is relaxed and the three - fold rotation symmetry is broken due to jahn - teller distortion . at the end , a nine - sided ring forms adjacent to a pentagon as shown in fig .
[ fig:2 ] ( b ) . owing to the spin - polarization of the @xmath23-dangling bond on the two - fold coordinated carbon atom and adjacent orbitals at the defect site
the system obtains an unbalanced spin .
in fact , the difference of total charge density corresponding to states of different spin states , i.e. @xmath24 is non - zero and exhibits a distribution shown in fig .
[ fig:2 ] ( b ) .
because of unbalanced spin , agnr(22;5 ) gains a net magnetic moment of @xmath25=1 @xmath26 per cell .
this attributes a magnetic state to the nanoribbon , which was nonmagnetic otherwise .
( b ) magnetization of the non - magnetic agnr(22 ) by a defect due to the single carbon atom vacancy with the same repeat periodicity .
isosurfaces around the vacancy corresponding to @xmath27 ; namely the difference of the total charge density of different spin directions.solid ( blue ) and dashed ( red ) lines are for spin - up and spin - down bands ; solid ( black ) lines are nonmagnetic bands.,width=300 ] furthermore , the spin degeneracy of some bands related with this defect is broken and spin - up and spin - down bands are split . the dispersive , non - magnetic band at the edge of the valence band becomes partially emptied , since its electrons are transferred to the flat spin bands below .
eventually , the semiconducting ribbon becomes metallic .
not only armchair , but also zigzag nanoribbons are strongly affected by defects due to single and multiple vacancies . when coupled with the magnetic edge states of the zigzag nanoribbons the vacancy defect brings about additional changes .
the magnetic state and energy band structure of these ribbons depend on the type and geometry of the defects . in fig .
[ fig:3 ] ( a)-(c ) , the effect of a defect generated from the single vacancy with a repeat periodicity of @xmath13=8 is examined in zgnr(14;8 ) for three different positions .
the total energy is 0.53 ev lowered when the defect is situated at the edge rather than at the center of the ribbon .
calculated total energy , @xmath28 ( in ev / cell ) , net magnetic moment , @xmath25 ( in bohr magneton @xmath26/cell ) , band gap between spin - up(down ) conduction and valence bands , @xmath29 ) are shown for each case .
blue and yellow isosurfaces corresponds to the difference of spin - up and spin - down states , width=294 ] zgnr(14;8 ) has a net magnetic moment of @xmath25=1.94 @xmath26 when the defect is situated at the center of the ribbon and hence its antiferromagnetic ground state has changed to ferrimagnetic state through the magnetic moment of the single vacancy .
otherwise , @xmath25 becomes zero , when the position of the defect deviates from the center .
for example , in fig .
[ fig:3 ] ( b ) and ( c ) the sum of the magnetic moments of the edge states is not zero , but the net magnetic moment per unit cell becomes zero only after the spins of the defect is added .
even if the net magnetic moment @xmath25=0 , zgnr(14,8 ) does not have an antiferromagnetic ground state due to the presence of single vacancy .
the edge states , each normally having equal but opposite magnetic moments , become ferrimagnetic when a defect is introduced .
the total magnetic moment of the supercell vanish only after the magnetic moment of defect has been taken into account . since
the spin - degeneracy has been broken , one can define @xmath30 and @xmath31 for majority and minority spin states . not only the magnetic state , but also the band gap of zigzag ribbons in fig .
[ fig:3 ] are affected by the symmetry and the position of the defect relative to edges . in fig .
[ fig:3 ] ( d)-(f ) two defects associated with two separated vacancy and a defect associated with a relaxed divacancy exhibit similar behaviors .
an important issue to be addressed here is the breakdown of lieb s theorem@xcite for those zigzag ribbons . according to lieb s theorem ,
the net magnetic moment per cell is determined with the difference in the number of atoms belonging to different sublattices , and it shall be either @xmath25=1 @xmath26 or 2 @xmath26 for the cases in fig .
[ fig:3 ] .
none of the cases in fig .
[ fig:3 ] is in agreement with lieb s theorem . here
one can consider two features , which may be responsible from this discrepancy .
first is the strong jahn - teller distortion and relaxation of carbon atoms at the close proximity of the defect . as a result
some dangling @xmath23-bonds reconstructed to form new c - c covalent bonds .
the lowering of the total energy , a driving force for such reconstruction , is as high as 0.5 - 0.6 ev / cell .
second is the interaction with the magnetic edge states , which becomes effective for narrow zgnr s . by applying a local bias voltage @xmath32 across the ribbon .
( b ) @xmath22 versus @xmath32 applied at the center of agnr(@xmath3;6 ) for @xmath3=34 , 36 and 38 .
( b ) @xmath22 versus @xmath32 of agnr(34 ) for different tip positions schematically described at the top .
the tip ( or electrodes ) are situated at one of the position 1 - 3 .
the repeat period is @xmath13=20.,width=300 ] finally , we note that introducing periodic vacancies or defects on a given ribbon appear to be difficult with the state of the art technologies .
here we propose a method as described in fig .
[ fig:4 ] to achieve the formation of periodic local defects like holes or vacancies .
the sharp electrodes like stm tips are situated at desired locations , such as one of the cases 1 - 3 in fig .
[ fig:4 ] ( a ) on the graphene with a given repeat periodicity
. a potential difference , @xmath32 common to all electrodes ( tips ) is applied between the tip and underlying insulator through the graphene . this way the electronic potential of graphene atoms just below the tip
is locally lowered or raised depending on the polarity of @xmath32 . here , the effect of locally and periodically applied potential difference has been modelled by etb , where the on - site energies of carbon atoms below the tip have been changed accordingly .
although the present model is crude , it still allows us the realization of monitoring of the properties of nanoribbons . in fig .
[ fig:4 ] ( b ) the variation of @xmath22 with @xmath32 is calculated for agnr(n;20 ) with @xmath3=34 , 36 and 38 by using periodically located tips at the center of the ribbon .
because of the electron - hole symmetry in the ribbons , the band gap variation depends on the magnitude of the bias voltage . in fig
. [ fig:4 ] ( c ) the variation of the band gap of agnr(42;20 ) with @xmath32 and position of the tip is shown . in spite of the fact that the modifications of the band gaps are not the same as in fig .
[ fig:1-b ] , the available parameters , such as @xmath32 , @xmath13 , tip - geometry and its position make the monitoring of the properties possible .
in conclusion , we show that the energy band gaps and magnetic states of graphene nanoribbons can be modified by defects due to single or multiple vacancies . the optimized atomic structure of various vacancy defects have been found to be identical with the tem images reported earlier @xcite .
two different electronic states are distinguished : these are : ( i ) itinerant bloch states perturbed by defects , ( ii ) defect induced states . while the former is dispersive , the latter give rise to flat bands .
the reconstruction and spin - polarization of the orbitals at the close proximity of the defect give rise to net magnetic moments , which , in turn , changes the magnetic ground state of the defect - free ribbon .
numerous studies have been carried out on the electronic structure of armchair and zigzag graphene ribbons .
see for example : v. barone , o. hod and g.e .
scuseria , nano lett .
* 6 * , 2748 ( 2006 ) ; y .- w . son , m.l .
cohen and s.g .
louie , nature * 444 * , ( 2006 ) ; m. y. han , b. ozyilmaz , y. zhang , and p. kim , phys .
* 98 * , 206805 ( 2007 ) . as shown by the recent work based on self - energy corrections ; l. yang , c.h . park , y .- w .
son , m. l. cohen , s. g. louie , phys . rev . lett . * 99 * , 186801 ( 2007 ) ; the band gap , @xmath4 is underestimated by the dft calculations .
since we consider structures which have already a band gap , its actual value does not affect our discussion in any essential manner . using the definition of schottky defect
, we calculated the formation energy of relaxed vacancy of single carbon atom in agnr(34 ) to be 7.79 ev .
this value is smaller than the binding energy of a carbon atom in the graphene , which is calculated to be 9.23 .
lieb , phys .
lett . * 62 * , 1201 ( 1989 ) .
|
using first - principles plane wave calculations we predict that electronic and magnetic properties of graphene nanoribbons can be affected by defect - induced itinerant states .
the band gaps of armchair nanoribbons can be modified by hydrogen saturated holes .
defects due to periodically repeating vacancy or divacancies induce metallization , as well as magnetization in non - magnetic semiconducting nanoribbons due to the spin - polarization of local defect states .
antiferromagnetic ground state of semiconducting zigzag ribbons can change to ferrimagnetic state upon creation of vacancy defects , which reconstruct and interact with edge states .
even more remarkable is that all these effects of vacancy defects are found to depend on their geometry and position relative to edges .
it is shown that these effects can , in fact , be realized without really creating defects
|
in this paper we study numerical integration of twice differentiable functions defined over a triangle @xmath4 . for an integrable function @xmath5 , we denote the true normalized integral of @xmath6 by @xmath7 where @xmath8 denotes the lebesgue measure of @xmath9 . as an approximation of @xmath10
, we consider a linear algorithm of the form @xmath11 for an @xmath12-element point set @xmath13 and a set of real - valued weights @xmath14 .
in particular , a quasi - monte carlo ( qmc ) integration is an equal - weight quadrature rule where the weights sum up to 1 , i.e. , a linear algorithm with the special choice @xmath15 for all @xmath16 .
therefore , @xmath10 is simply approximated by @xmath17 if an infinite sequence of points @xmath18 is given , the first @xmath12 elements of @xmath19 are used as @xmath20 .
we define the norm in @xmath21 by @xmath22 and study the worst - case absolute error over the unit ball of @xmath21 , i.e. , @xmath23 thus an obvious goal in this context is to construct a good point set or sequence in @xmath9 such that the quantity @xmath24 is small either for some @xmath12 or uniformly for all @xmath1 .
the theory of qmc integration has been developed in depth with the particular focus on approximating the integral of functions defined over the unit cube @xmath25^s$ ] , see for instance @xcite .
in fact , not much attention has been paid to qmc integration over non - cubical domains until recently .
we have to point out , however , that many practical problems are not necessarily given by quadrature over the unit cube .
so far , the most standard approach to qmc integration over a non - cubical domain @xmath26 is to find a uniformity - preserving transformation @xmath27^s\to \omega$ ] and then to approximate the normalized integral of @xmath28 by @xmath29 for @xmath30^s$ ] . in the literature , fang and
wang @xcite introduced several transformations from the unit cube to the ball , sphere , and simplex .
pillards and cools @xcite studied 5 different transformations from the unit cube to the simplex .
more recently , basu and owen @xcite gave sufficient conditions on @xmath31 so that @xmath32 is either of bounded variation or satisfies additional smoothness conditions . instead of applying a uniformity - preserving transformation , more direct and explicit constructions of point sets and sequences in a triangular domain @xmath9
have been introduced recently by basu and owen @xcite .
one is based on the van der corput sequence in base 4 in conjunction with a recursive partitioning of @xmath9 .
the other is given by a rotation of an integer lattice through an angle whose tangent is badly approximable .
a discrepancy measure derived in @xcite was employed as a quality criterion of these constructions , and it was shown that the latter one attains a lower discrepancy .
nonetheless , the former one is of practical importance since it is extensible and can be randomized . in this paper , motivated by the first construction of basu and owen
, we study qmc integration for smooth functions in @xmath21 .
in particular , we give an explicit construction of infinite sequences of points including one by basu and owen as a special case , and prove that our quadrature rule achieves the worst - case error of order @xmath0 for any @xmath1 in @xmath21 .
the main result of this paper can be summarized as follows : [ thm : main ] for a triangle @xmath4 , we can explicitly construct an infinite sequence @xmath19 of points in @xmath9 for which there exists a constant @xmath33 such that @xmath34 for all @xmath1 , and in particular , @xmath35 for all @xmath36 .
roughly speaking , our approach for the proof of theorem [ thm : main ] is to exploit the decay of the walsh coefficients for @xmath37 under a suitable recursive partitioning of @xmath9 . by following the essentially same argument using bump functions as in @xcite ,
see also ( * ? ? ?
* section 2.7 ) , we see that a lower bound of order @xmath2 on the worst - case error holds for any linear algorithm in @xmath21 .
namely , there exists a constant @xmath38 such that @xmath39 holds for any choice of @xmath20 and @xmath40 .
thus the upper bound we obtain is best possible apart from the @xmath3 factor .
the rest of this paper is organized as follows . in the next section we present an explicit construction of infinite sequences of points in @xmath9 .
we prove an upper bound on the worst - case error for our quadrature rule in section [ sec : upper ] , whereas the result on the decay of the walsh coefficients for @xmath37 , which is necessary for the proof of an upper bound , is shown later in section [ sec : walsh ] . throughout this paper
we use the following notations .
let @xmath41 be the set of integers , @xmath42 the set of positive integers , and @xmath43 .
we denote the two - element field by @xmath44 , which is identified with the set @xmath45 equipped with addition and multiplication modulo 2 . the addition operation in @xmath44 is denoted by @xmath46 , and in case of vectors or matrices over @xmath44 , @xmath46 is applied componentwise .
further we denote a triangular domain with vertices @xmath47 by @xmath48 and the diameter of a set @xmath49 by @xmath50 . without loss of generality ,
we assume that the center of a triangle @xmath9 is located at the origin in @xmath51 , i.e. , if the center of @xmath9 is not located at the origin , it suffices to shift the whole domain @xmath9 .
in a similar way to ( * ? ? ? * section 3 ) , here we introduce a recursive partitioning of a triangle @xmath52 .
we first partition the triangle into 4 congruent subtriangles , to each of which a pair @xmath53 is assigned with @xmath54 in the center
. then we partition each subtriangle into 4 congruent sub - subtriangles , to each of which a pair @xmath55 is assigned again with @xmath54 in the center .
hence every sub - subtriangle can be now identified with a set of pairs @xmath56 and @xmath57 .
this is illustrated in figure [ fig : partition ] .
it is obvious that this recursive partitioning of the triangle defines the mapping from @xmath58 to @xmath9 , which is surjective but not injective .
moreover , for a matrix @xmath59 with @xmath60 for all @xmath61 , the first @xmath16 rows of @xmath62 determines which of @xmath63 congruent subregions the matrix @xmath62 is mapped within , and from the condition that the pair @xmath54 is always assigned in the center , we see that the matrix @xmath62 is mapped to the center of the corresponding subregion .
.,width=340 ] we now describe our recursive partitioning more precisely .
the subtriangle of @xmath52 for a pair @xmath53 is defined by @xmath64 then the sub - subtriangle for a set of pairs @xmath56 and @xmath57 is defined by @xmath65 in this way , the subregion for a matrix @xmath66 with some @xmath67 is defined recursively by @xmath68 for simplicity of notation , as long as @xmath69 , we write @xmath70 moreover we define the mapping @xmath71 by @xmath72 and , again for simplicity of notation , as long as @xmath73 we write @xmath74 regarding the map @xmath75 we have the following lemma . since the result can be easily proved by induction on @xmath76 , we omit the proof . [
lem : partition ] for a matrix @xmath66 , we define @xmath77 by @xmath78 and @xmath79 for @xmath80 .
let @xmath81 with @xmath82 and @xmath83 .
then the following holds true : 1 . for @xmath69 , we have @xmath84 2 . for @xmath85 ,
define @xmath86 by @xmath87 then we have @xmath88 .
in particular @xmath89 we describe how to generate an infinite sequence of points in a triangle @xmath9 . for this purpose
, we first introduce the definition of digital nets over @xmath44 for the two - dimensional case .
[ def : digital_net ] for @xmath90 with @xmath91 , let @xmath92 . for an integer @xmath93 ,
denote the dyadic expansion of @xmath94 by @xmath95 .
define the matrix @xmath96 by @xmath97 for @xmath98 .
then we call the subset @xmath99 a ( two - dimensional ) digital net over @xmath44 with generating matrices @xmath100 . by using the map @xmath101 defined by @xmath102 for @xmath103 , digital nets over @xmath44 are usually defined as point sets in @xmath104 by @xmath105 , where @xmath106 is applied columnwise . here
we see that the integer @xmath16 denotes the precision of points . in this paper , it is more reasonable to define digital nets over @xmath44 as subsets in @xmath107 instead of point sets in @xmath104 .
the above definition can be extended to digital sequences over @xmath44 .
[ def : digital_seq ] let @xmath108 .
for each @xmath109 , we assume @xmath110 for all sufficiently large @xmath111 . for @xmath112 ,
denote the dyadic expansion of @xmath94 by @xmath113 .
define the matrix @xmath114 by @xmath115 for @xmath98 .
then we call the infinite sequence @xmath116 a ( two - dimensional ) digital sequence over @xmath44 with generating matrices @xmath100 .
[ rem : digital_seq ] it follows from the assumption @xmath110 for all sufficiently large @xmath111 that , for each @xmath112 , there exists a unique @xmath117 such that @xmath118 for both @xmath98 .
furthermore , for @xmath36 , the first @xmath119 elements of @xmath19 can be regarded as a digital net over @xmath44 with generating matrices @xmath120 for some @xmath91 , where we denote by @xmath121 the left upper @xmath122 sub - matrix of @xmath123 .
now we are ready to present how to generate an infinite sequence of points in a triangle @xmath9 .
[ def : digital_seq_t ] let @xmath124 be a digital sequence over @xmath44 with generating matrices @xmath100
. then an infinite sequence of points in @xmath9 is given by @xmath125 where the function @xmath126 is given as in remark [ rem : digital_seq ] .
it is clear from this definition that our infinite sequence of points is determined by generating matrices @xmath100 .
thus we need some quality measure for generating matrices to make an explicit construction of @xmath127 possible , which is discussed in the next subsection .
we first recall the notion of dual net .
[ def : dual ] for @xmath90 with @xmath91 , let @xmath128 a digital net over @xmath44 with generating matrices @xmath92 .
the dual net of @xmath129 is defined by @xmath130 let @xmath124 be a digital sequence over @xmath44 . as mentioned in remark [ rem : digital_seq ] , the first @xmath119 elements of @xmath19 are a digital net over @xmath44 with generating matrices @xmath120 for some @xmath91 .
thus the above definition of the dual net still applies to such an initial finite segment of @xmath19 . the following weight function , introduced in @xcite and @xcite , is well known .
[ def : nrt_weight ] for @xmath131 , where all but only a finite number of @xmath132 are 0 , we define @xmath133 and @xmath134 . in case of a matrix @xmath135 with @xmath136 , where all but only a finite number of elements in @xmath137 are 0 , we define @xmath138 if @xmath111 is an element in @xmath139 for some @xmath67 , by considering an injection @xmath140 we use the same symbol @xmath141 to define the weight function for such @xmath111 .
a similar abuse of notation is also done in case of a matrix @xmath142 for finite @xmath16 . for a digital net @xmath128
, we define the so - called minimum weight by @xmath143 which has been often used as a quality measure of generating matrices for qmc integration over the unit cube .
if a digital net @xmath129 satisfies @xmath144 for some @xmath145 , we call @xmath129 a digital @xmath146-nets over @xmath44 . furthermore , for a digital sequence @xmath124 , if there exists a non - negative integer @xmath147 such that the first @xmath119 elements of @xmath19 are a digital @xmath146-net over @xmath44 for any @xmath148 , we call @xmath19 a digital @xmath149-sequence over @xmath44 . [ rem : precision ] 1 .
any digital net satisfies the above inequality for @xmath150 . in practice , we prefer a larger value of @xmath151 and thus equivalently a smaller value of @xmath147 , and @xmath152 is best possible .
we refer to @xcite for several explicit constructions of digital nets and sequences with small @xmath147-value .
2 . in what follows ,
we restrict ourselves to digital @xmath149-sequences with upper triangular generating matrices @xmath100 which satisfy @xmath110 for @xmath153 .
explicit constructions of digital sequences by sobol @xcite and tezuka @xcite hold this property .
besides , by allowing the situation @xmath154 , the first @xmath119 elements of a digital @xmath149-sequence can be regarded as a digital @xmath146-net for any @xmath36 .
now we introduce a new weight function which suits our purpose .
[ def : new_weight ] let @xmath155 .
for a matrix @xmath156 with @xmath157 , where all but only a finite number of elements in @xmath137 are 0 if @xmath158 , we define @xmath159 we can define the weight function @xmath160 equivalently as follows : for a matrix @xmath161 , where all but only a finite number of @xmath162 are 0 if @xmath158 , define @xmath163 and @xmath164 .
similarly to @xmath151 , we define the minimum weight of a digital net @xmath129 by @xmath165 here we prefer a digital net @xmath129 with a large value of @xmath166 . in the following lemma
, we show that a digital @xmath146-net with small @xmath147 is exactly what we want .
[ lem : nrt - new - weights ] let @xmath128 be a digital @xmath146-net over @xmath44 .
then we have @xmath167 moreover let @xmath116 be a digital @xmath149-sequence over @xmath44 . then for any @xmath148 , we have @xmath168 let @xmath135 with @xmath136 , where all but only a finite number of elements in @xmath137 are 0 . from the definitions of @xmath141 and @xmath160 , we have @xmath169 which gives @xmath170 this proves the first statement .
the second statement directly follows from the definition of a digital @xmath149-sequence .
hence our explicit construction of an infinite sequence of points in @xmath9 is to use a digital @xmath149-sequence over @xmath44 with upper triangular generating matrices which is mapped to @xmath9 according to definition [ def : digital_seq_t ] . in the next section
, we prove that such an infinite sequence of points in @xmath9 achieves the almost optimal order of convergence for smooth functions in @xmath21 . before going into the proof of an error bound , we provide another explicit construction inspired by the first construction due to basu and owen @xcite .
let @xmath108 be given by @xmath171 for @xmath112 with finite dyadic expansion @xmath172 , we have @xmath173 thus for even @xmath174 , it is obvious that the first @xmath119 elements of @xmath19 generated by these matrices are given by @xmath175 considering the image of the map @xmath176 , we can easily check that the point set in @xmath9 obtained in this way is the same as that of basu and owen .
this implies that our construction scheme includes their explicit construction as a special case .
moreover it is easy to show that the first @xmath119 elements of @xmath19 are actually a digital @xmath177-net over @xmath44 .
it can be seen from lemma [ lem : nrt - new - weights ] that the minimum weight for @xmath160 is bounded below by @xmath178 , which can be improved as follows .
since the result follows from direct calculation , we omit the proof .
let @xmath108 be given by ( [ eq : basu - owen - matrix ] ) . for @xmath36
, let @xmath129 be a digital net over @xmath44 with generating matrices @xmath179
. then we have @xmath180
here we prove an upper bound on the worst - case error for our quadrature rule in @xmath21 by using the result later shown in section [ sec : walsh ] .
[ lem : disc_approx ] let @xmath183 and @xmath186 . for any @xmath37
, we have @xmath187 we have @xmath188 and consider the line segment @xmath189 . since @xmath190 is a triangle , and thus is convex , the set @xmath191 is included in @xmath190 .
hence we have @xmath192 which completes the proof . in order to exploit the smoothness of functions in @xmath21
, we shall conduct a discrete walsh - fourier analysis of the discretized function @xmath185 defined on @xmath107 later in section [ sec : walsh ] .
right now we just introduce the definition of walsh functions and briefly review some basic facts so as to make the proof of the main result in the next subsection accessible .
[ thm : main2 ] let @xmath207 be either a digital @xmath149-sequence with upper triangular generating matrices or a digital sequence with generating matrices given by ( [ eq : basu - owen - matrix ] ) , and let @xmath208 be constructed according to definition [ def : digital_seq_t ] . denote the first @xmath12 elements of @xmath127 by @xmath20 . for any @xmath37 , the following holds true :
we now show a bound on @xmath358 . from the first item of lemma [ lem : dyadic - hat ] and the triangle inequality , we have @xmath359 from the obvious fact @xmath360 and the first item of lemma [ lem : shift ] , we have @xmath361 where we use the result in lemma [ lem : dif - deriv - lemma ] with @xmath362 in the second inequality , and then the third item of lemma [ lem : shift ] in the last inequality .
thus we obtain a bound on @xmath358 : @xmath363 next we show a bound on @xmath364 for @xmath325 . from the third item of lemma [ lem : dyadic - hat ] and the triangle inequality
, we have @xmath365 from the second item of lemma [ lem : shift ] , we have @xmath366 in what follows , we continue with further arguments separately for the cases @xmath367 and @xmath368 .
let us consider the case @xmath368 .
in this case we have @xmath369 .
thus we obtain @xmath370 it is easy to see by definition that @xmath371 for @xmath368 .
since @xmath372 implies @xmath373 we have @xmath374 .
further it follows from the third item of lemma [ lem : shift ] that @xmath375 .
thus we obtain @xmath376 and @xmath377 comparing these equalities gives @xmath378 from which we see that @xmath379 form a parallelogram . by using the result in lemma [ lem : dif - deriv - lemma ] with @xmath380 and @xmath381 and then the third item of lemma [ lem : shift ] again
, we have @xmath382 thus we obtain a bound on @xmath364 : @xmath383 let us move onto the case @xmath384 . in this case
we have @xmath385 .
thus we obtain @xmath386 since @xmath384 , we see that @xmath387 thus , from we obtain @xmath388 and @xmath389 comparing these equalities gives @xmath390 from which we see that @xmath391 form a parallelogram .
( note that the points @xmath392 and @xmath393 form not the diagonal but the edge of the parallelogram unlike in the case of @xmath368 . ) by using the result in lemma [ lem : dif - deriv - lemma ] with @xmath394 and @xmath381 and then the third item of lemma [ lem : shift ] again together with the triangle inequality , we have @xmath395 thus we obtain a bound on @xmath364 : @xmath396 99 n. s. bakhvalov , approximate computation of multiple integrals ( in russian ) , vestnik moskov .
ser . mat .
fiz . him . 4 ( 1959 ) , 318 . k. basu , a. b. owen , low discrepancy constructions in the triangle , siam j. numer .
53 ( 2015 ) , 743761 .
k. basu , a. b. owen , transformations and hardy - krause variation , siam j. numer .
54 ( 2016 ) , 19461966 .
l. brandolini , l. colzani , g. gigante , g. travaglini , a koksma hlawka inequality for simplices , in : trends in harmonic analysis , springer , new york , 2013 , pp .
j. dick , a. hinrichs and f. pillichshammer , proof techniques in quasi - monte carlo theory , j. complexity 31 ( 2015 ) , 327371 .
j. dick , f. pillichshammer , _ digital nets and sequences : discrepancy theory and quasi - monte carlo integration _ , cambridge university press , cambridge , 2010 .
fang , y. wang , _ number - theoretic methods in statistics _ , chapman & hall , london , 1994 . h. niederreiter , low - discrepancy point sets , monatsh .
, 102 ( 1986 ) , 155167 . h. niederreiter , _ random number generation and quasi - monte carlo methods _ , cbms - nsf regional conference series in applied mathematics , vol . 63 , siam , philadelphia , 1992 .
t. pillards , r. cools , transforming low - discrepancy sequences from a cube to a simplex , j. comput .
( 2005 ) , 2942 .
rosenbloom , m. a. tsfasman , codes for the @xmath174-metric , probl .
inf . transm . 33 ( 1997 ) , 5563
. i. h. sloan , s. joe , _ lattice methods for multiple integration _ , oxford university press , oxford , 1994 .
i. m. sobol , the distribution of points in a cube and approximate evaluation of integrals , zh .
( 1967 ) , 784802 . s. tezuka , polynomial arithmetic analogue of halton sequences , acm trans . model .
simul . 3 ( 1993 ) , 99107 . t. yoshiki , bounds on walsh coefficients by dyadic difference and a new koksma - hlawka type inequality for quasi - monte carlo integration , arxiv:1504.03175 , 2015 .
|
we study quasi - monte carlo integration for twice differentiable functions defined over a triangle .
we provide an explicit construction of infinite sequences of points including one by basu and owen ( 2015 ) as a special case , which achieves the integration error of order @xmath0 for any @xmath1 .
since a lower bound of order @xmath2 on the integration error holds for any linear quadrature rule , the upper bound we obtain is best possible apart from the @xmath3 factor .
the major ingredient in our proof of the upper bound is the dyadic walsh analysis of twice differentiable functions over a triangle under a suitable recursive partitioning .
keywords : quasi - monte carlo , digital nets and sequences , numerical integration on triangle , dyadic walsh analysis + msc classifications : primary , 42c10 , 65d32 ; secondary , 41a55 , 65c05 , 65d30
|
in this article , the ffs method is used to calculate switching rates for biochemical networks simulated with the gillespie algorithm @xcite .
this algorithm is an application to chemical reactions of the kinetic monte carlo technique @xcite , first introduced by bortz _
et al _ @xcite .
the system is specified by a set of chemical components @xmath4 and a list of allowed reactions , together with their rate constants .
the concentrations @xmath5 of all the components are assumed to be homogeneous in space ; the state of the system at any instant in time is defined by @xmath6 .
the concentrations @xmath6 are propagated stochastically in time , assuming each reaction to be a poisson process .
this time propagation is consistent with the chemical master equation , so that a gillespie simulation is in fact a numerical solution of the master equation .
an important feature of the gillespie algorithm , and of other methods for simulating reaction networks , is that the distribution of states , _
i.e. _ the phase space density , is not known _ a priori _ , but is an output of the simulation .
the phase space density can be obtained by solving the chemical master equation @xcite , but this is generally a demanding task , which is indeed often the motivation for carrying out a gillespie simulation . transition path sampling ( tps ) has been developed to study rare events in condensed - matter systems . in tps ,
paths belonging to the tpe are obtained by importance sampling in trajectory space .
new paths connecting stable states @xmath2 and @xmath3 are generated by making changes to existing paths .
a new path is accepted or rejected according to its weight in the tpe , which depends on the phase space density of its initial point , as well as the transition probability for each subsequent step . without prior knowledge of the stationary distribution of states ,
however , this approach can not conveniently be applied .
the ffs algorithm which we present in this article differs fundamentally from these methods . rather than generating transition paths one at a time ,
many paths are grown simultaneously , in a series of layers , each of which forms the basis for the next one .
prior knowledge of the stationary distribution of states is not required .
ffs is well suited for convenient and efficient calculation of switching rates in biochemical reaction networks .
the ffs method is not limited to reaction networks : although it can not be used for systems with deterministic dynamics , it is applicable to any stochastic dynamical scheme , such as langevin or brownian dynamics . in this context , it could be used to study rare events in soft condensed matter systems such as protein folding and crystal nucleation , or non - equilibrium processes such as dna or rna stretching .
the expression for the rate constant that is used in the ffs algorithm is the same as that described by van erp _
et al _ @xcite .
the transition occurs between two phase space regions @xmath2 and @xmath3 , which must both be `` stable '' in the sense that if the the system is placed outside these regions , it will rapidly evolve in time towards one of them . @xmath2 and @xmath3 are characterized by the functions @xmath7 and @xmath8 ( where @xmath9 denotes all co - ordinates of the phase space : in the case of the gillespie algorithm the concentrations of all the system components ) , such that : @xmath10 we also define the functions @xmath11 and @xmath12 , which depend not only on @xmath13 but also on the history of the system : @xmath14 if the system was more recently in @xmath2 than in @xmath3 , and is zero otherwise , while @xmath15 if the system was more recently in @xmath3 than in @xmath2 , and is zero otherwise .
thus @xmath16 for any point on any path in phase space .
the rate constant @xmath17 for transitions from region @xmath2 to region @xmath3 is given by : @xmath18 here , @xmath19 is the average number of trajectories per unit time entering region @xmath3 , coming directly from @xmath2 ( _ i.e. _ which were in @xmath2 more recently than they were in @xmath3 ) .
here , the overbar denotes an average over all phase space points , with their associated histories .
the flux @xmath19 in equation ( [ eq3 ] ) is difficult to obtain accurately from a simulation because the system makes few , if any , spontaneous crossings from @xmath2 to @xmath3 in a typical run . to alleviate this problem ,
a parameter @xmath0 is chosen , such that the functions @xmath20 and @xmath21 can be written as : @xmath22 an increasing series of values of @xmath0 , @xmath23 , is then chosen , such that @xmath24 and @xmath25 .
these must constitute non - intersecting surfaces in phase space .
it is not necessary for @xmath0 to be the reaction coordinate , merely that @xmath26 and @xmath27 describe the two stable states ( the exact positioning of these surfaces is not critical ) .
moreover , the system should not reach any @xmath28 before it has crossed the preceding surface @xmath29 .
defining @xmath30 as the probability that a trajectory which passes through @xmath31 coming directly from @xmath2 ( _ i.e. _ having been in @xmath2 more recently than it last crossed @xmath29 ) , will subsequently reach the surface @xmath28 before returning to @xmath2 , equation ( [ eq3 ] ) can be written as : @xmath32 expression ( [ eq6 ] ) indicates that the total flux from @xmath2 to @xmath3 is simply the total flux from @xmath2 to @xmath33 , multiplied by the probability that a trajectory reaching @xmath33 from @xmath2 will eventually arrive in @xmath3 , before returning to @xmath2 .
a key point is that @xmath34 can be expressed as the product of the probabilities of reaching each successive interface from the previous one , without returning to @xmath2 : @xmath35 so that @xmath36 expressions ( [ eq3 ] ) and ( [ eq6])-([eq8 ] ) are used in the ffs method to calculate @xmath17 .
the first stage of the ffs algorithm involves the choice of the parameter @xmath0 and values for @xmath26 , @xmath27 and @xmath23 . in any one chemical reaction step ,
the system must be able to cross at most one surface @xmath29 .
of course , some choices of @xmath0 will lead to more efficient path sampling than others , but we shall demonstrate in the next sections that for a typical genetic switch , a rather simple definition of @xmath0 gives very satisfactory results .
it is also convenient to define a series of `` sub - surfaces '' @xmath37 , in between each pair of surfaces @xmath29 and @xmath28 , such that @xmath38 and @xmath39 . a simulation
is then carried out starting from a point in region @xmath2 .
after an equilibration period , the value of @xmath0 is monitored during a run of length @xmath40 .
whenever the trajectory crosses the surface @xmath33 , coming directly from @xmath2 , a counter @xmath41 is incremented .
if @xmath41 is less than a user - defined number @xmath42 , the phase space co - ordinates of the system are also stored .
the run is then continued . after simulation time @xmath40 ,
one is left with a collection of @xmath42 points at or just beyond @xmath33 , as well as a measurement of the flux @xmath43 .
this procedure is illustrated schematically in figure [ fig1 ] : crossings of surface @xmath33 that are labeled with a black circle contribute to @xmath41 and to the collection of points at @xmath33 . during this run
, it is replaced in @xmath2 , re - equilibrated , and the run continued . ]
figure [ fig2 ] illustrates the next stage of the algorithm .
the collection of points at @xmath33 is used to initiate a large number @xmath44 of short simulation trial runs . in each of these trials , a phase space point from the collection at @xmath33 is chosen at random .
this is then used as the starting point for a simulation run , which is continued until the system crosses either @xmath45 or @xmath26 . during this run ,
the maximum value of @xmath0 , @xmath46 , achieved by the system is recorded .
counters @xmath47 for all the sub - surfaces @xmath48 are then incremented by one .
after @xmath44 trials , a good estimate is obtained for @xmath49 , for @xmath50 .
we note that @xmath51 and @xmath52 . during the trial procedure outlined above
, one also makes a new collection of @xmath53 points at or just beyond the surface @xmath45 : these are the final phase space points of those trial runs starting from @xmath33 which make it to @xmath45 .
the number @xmath54 of trials must be large enough to generate @xmath53 points at @xmath45 .
the values of @xmath54 , @xmath53 and @xmath55 and @xmath56 for all the subsequent surfaces @xmath57 are chosen by the user : the @xmath56 should be large enough to allow good sampling of the phase space .
the trial run procedure is repeated for each subsequent surface @xmath29 , starting from the collection of @xmath56 phase space points generated by the successful runs from @xmath58 .
eventually @xmath27 is reached , and one is left with a series of histograms @xmath59 , for @xmath60 and @xmath61 . using equation ( [ eq8 ] ) , these histograms can be fitted together to obtain a smooth curve @xmath62 @xcite , the value of which at @xmath63 is @xmath34 .
the rate constant @xmath17 is obtained on multiplying @xmath34 by the flux @xmath64 calculated in the first stage of the algorithm .
it is important to remark that the ffs algorithm does not assume that the distribution of phase space points at the interfaces @xmath65 is equal to the stationary distribution of states .
for the example which we present in this paper , this turns out to have significant consequences for the transition mechanism .
we have applied the ffs method to a simplified model of a genetic toggle switch @xcite .
this model could be regarded as a minimal representation of the lysis - lysogeny switch in bacteriophage @xmath0 @xcite ; a synthetic switch of this type has also been constructed _ in vivo _
@xcite .
the model switch consists of two proteins @xmath66 and @xmath67 and their corresponding genes @xmath2 and @xmath3 .
@xmath66 and @xmath67 form homodimers @xmath68 and @xmath69 which can bind to the dna strand ( here labeled @xmath70 ) and influence transcription . when the dimer @xmath68 is bound to the dna , gene @xmath3 is not transcribed , while @xmath69 , when bound , correspondingly blocks transcription of gene @xmath2 : thus @xmath66 and @xmath67 mutually repress one another s production .
both proteins are also degraded in the monomer form .
we consider two versions of this switch : the `` general '' switch , in which both dimers can bind simultaneously to the dna , forming the species @xmath71 , and the `` exclusive '' switch , in which only one dimer can be bound at any time . the exclusive switch models the case where the operator regions of genes @xmath2 and @xmath3 are overlapping .
the switch is represented by the set of reactions ( [ eq9 ] ) .
@xmath72 the asterisk indicates that reaction ( [ equationc ] ) happens only for the general switch . here
, we study a symmetrical version of the switch : the rate constants for the reactions on the left and right - hand sides of scheme ( [ eq9 ] ) are identical .
these are all expressed in terms of the protein production rate constant @xmath73 ( for reactions ( [ equationd ] ) and ( [ equatione ] ) ) , so that the unit of time in our calculations is @xmath74 . the rate constants for both the forward and backward dimerization reactions ( [ equationa ] ) are @xmath75 . binding to
the dna occurs with rate constant @xmath75 and dissociation of the complex with rate constant @xmath73 ( reactions ( [ equationb ] ) and ( [ equationc ] ) ) . finally , the rate constant for protein degradation , reactions ( [ equationf ] ) , is @xmath76 .
these parameters are chosen such that in a simulation using the gillespie algorithm , the switch flips between the @xmath66- and @xmath67-rich states at a rate that can be measured by brute - force simulation .
this allows us to test the ffs method .
the model is , of course , highly simplistic : our aim is here to demonstrate the ffs scheme using a simple example .
figure [ fig3 ] shows the results of gillespie simulations of the model genetic switch , in the general ( a ) and exclusive ( b ) cases .
the difference @xmath77 in the total copy numbers of the two proteins , @xmath78 , is plotted as a function of time , where @xmath79 and @xmath80 are defined by : @xmath81 of course , @xmath82 for the exclusive switch . noting the different scales on the time axis , the exclusive switch ( b ) has a much lower flipping rate than the general switch ( a ) , in agreement with previous work @xcite .
the probability @xmath83 of obtaining a particular value of @xmath77 is shown in the insets , demonstrating clearly that both the general and exclusive switches are bistable . using long brute - force gillespie simulations ,
the rate constant for the transition from the @xmath66-rich to the @xmath67-rich state was obtained for each switch .
we define phase space region @xmath2 to be where @xmath84 , and region @xmath3 to be where @xmath85 .
the system flips stochastically between the state where @xmath14 and @xmath86 ( _ i.e. _ it was most recently in @xmath2 ) and that where @xmath87 and @xmath15 ( it was most recently in @xmath3 ) .
the times @xmath88 between flipping events are distributed according to a poisson distribution @xmath89}$ ] , where @xmath90 since the switch is symmetrical in @xmath2 and @xmath3 .
the rate constant @xmath17 can conveniently be measured by fitting the cumulative distribution @xmath91 to the function @xmath92}$ ] .
this procedure resulted in values of @xmath93 for the general switch and @xmath94 for the exclusive switch ( using simulation runs of total length @xmath95 [ 12546 flips observed ] and @xmath96 [ 8808 flips observed ] respectively ) .
we next re - calculated @xmath17 using ffs .
the surfaces @xmath23 , were defined in terms of @xmath77 : _ i.e. _ @xmath97 .
regions @xmath2 and @xmath3 are given by @xmath98 and @xmath99 , respectively . to be sure that the exact values of @xmath26 , @xmath27 and @xmath23 did not affect the result ,
these parameters were varied in a series of separate calculations . in all cases ,
we set @xmath100 , and at each surface @xmath29 , @xmath101 points were stored and @xmath102 shooting trials were made .
results were averaged over @xmath103 independent calculations , to obtain error bars similar to those of the brute - force results .
figure [ fig4 ] shows @xmath62 as a function of @xmath0 , for the general and exclusive switches .
this function can also be obtained from an analysis of the brute - force simulation trajectories : the figure shows excellent agreement between the brute - force results ( shown in black ) , and those of ffs ( shown in red ) .
ccccc + @xmath27 & @xmath104 & @xmath105 & @xmath106 & @xmath107 + @xmath108 & @xmath109 & @xmath110 & @xmath111 & @xmath112 + @xmath113 & @xmath114 & @xmath115 & @xmath116 & @xmath117 + @xmath118 & @xmath119 & @xmath120 & @xmath121 & @xmath122 + @xmath123 & - & @xmath124 & @xmath125 & @xmath126 + & & & & @xmath127 + + @xmath27 & @xmath104 & @xmath105 & @xmath128 & @xmath129 + @xmath108 & @xmath130 & @xmath131 & @xmath132 & @xmath133 + @xmath113 & @xmath114 & @xmath134 & @xmath135 & @xmath133 + @xmath118 & @xmath103 & @xmath136 & @xmath137 & @xmath138 + @xmath123 & - & @xmath139 & @xmath140 & @xmath141 + & & & & @xmath142 + .[tab3 ] relative cpu time required to calculate the values of @xmath17 given in table [ tab2 ] , using parameter values as in the text . for the ffs calculations ,
the calculation of the flux @xmath143 accounted for @xmath144% of the total cpu time . [
cols="^,^,^,^,^ " , ] table [ tab2 ] lists the results for @xmath143 and @xmath34 , as well as @xmath17 , for various choices of @xmath26 and @xmath27 . the number @xmath104 of surfaces used in each calculation is also listed . in all cases ,
the rate constant @xmath17 is in good agreement with the brute - force simulation result .
an additional test is provided by measuring @xmath145 and @xmath62 using the brute - force runs , to provide a second brute - force estimate for @xmath17 .
this value is also listed in table [ tab2 ] and , as expected , is in good agreement both with the value obtained by the fitting to poisson statistics , and with the ffs results .
table [ tab3 ] shows the relative cpu time required for each of the calculations . even for the general switch with a relatively fast flipping rate , estimates of @xmath17
are obtained by the ffs method @xmath146 times faster than using brute - force simulations , with similar accuracy . in the case of the exclusive switch ,
the ffs method is @xmath147 times more efficient .
table [ tab3 ] also demonstrates that the cpu time required for an ffs calculation does not increase as the rate @xmath17 decreases , in contrast to the time for brute - force calculations .
thus ffs should allow the calculation of rates of even very rare flipping events within reasonable computational time .
as an example , we have calculated the rate of flipping for a more stable version of the exclusive switch , in which the rate constant for protein degradation is reduced to @xmath148 . using the ffs method
, we obtain a rate @xmath149 , 500 times slower than the switch considered above .
this result would have been extremely difficult to obtain using brute - force simulation . , @xmath150 and operator occupancy ( color - coded ) are plotted versus time .
the path for the general switch ( a ) happens to be longer than that for the exclusive switch ( b ) , but this is not true of all paths in the ensemble . (
b ) : average values of @xmath151 and average operator occupancies , as functions of the committor @xmath150 , for the general ( i ) and exclusive ( ii ) switches .
( c ) : probability distribution @xmath152 , for selected values of @xmath150 .
@xmath152 is divided into color - coded contributions from the different operator states .
the sum of the areas of the histograms for a particular @xmath150 is unity .
for all plots , @xmath153 test trajectories were used to estimate @xmath150.[fig6 ] , width=321 ] the results presented in table [ tab2 ] show that for the same mean number of protein molecules , the exclusive switch has a flipping rate approximately @xmath154 times slower than that of the general switch , in agreement with previous work @xcite . in order to elucidate the origin of this difference
, we have analysed the flipping mechanism .
the ffs method generates a collection of switching trajectories ( members of the tpe ) .
figure [ fig5 ] shows a sample of five of these transition paths , for the general switch ( a ) and the exclusive switch ( b ) , plotted as a function of @xmath79 and @xmath80 . to obtain these paths
, we begin with the collection of partial trajectories that reach @xmath27 from @xmath155 and trace these back via the intervening surfaces to @xmath26 .
it is clear from figure [ fig5 ] that in the general switch , protein @xmath66 is lost before protein @xmath67 is gained , so that the transition passes through a region of phase space where both @xmath79 and @xmath80 are low .
however , this is not the case for the exclusive switch .
an important quantity associated with transition paths is the committor , @xmath156 .
this is the probability that a new simulation trajectory fired from point @xmath9 will reach region @xmath3 before @xmath2 @xcite .
we have measured @xmath157 for points along the trajectories in the transition path ensembles generated using ffs , for the general and exclusive switches .
figure [ fig6](a ) shows @xmath150 , as well as @xmath158 and the occupancy of the operator sites , as functions of time for typical transition paths .
@xmath151 measures the difference in the number of _ free _ protein molecules : it is similar but not identical to @xmath77 .
a key point is that for the general switch ( figure [ fig6]a(i ) ) , the operator makes two important changes of state , from @xmath159 to @xmath160 early in the transition process , and later from @xmath160 to @xmath161 .
both of these changes influence @xmath150 .
for the exclusive switch ( figure [ fig6]a(ii ) ) , however , the operator is intermittently in states @xmath159 and @xmath161 during the transition
. in figure [ fig6 ] ( b ) , values of @xmath162 and operator occupancies are shown , averaged over the paths in the tpe , as a function of the committor @xmath150 .
results are shown for the general ( i ) and exclusive ( ii ) switches .
figure [ fig6 ] ( c ) analyses the state points along the paths in the tpe which have values of @xmath163 , @xmath164 and @xmath165 . for each of these values of the committor
, points are grouped according to their operator state . for each @xmath150 and operator state , the histograms in figure [ fig6](c ) show the probability distribution function @xmath166 .
clearly , for points on a constant @xmath150 surface , the operator state and the number of free molecules are correlated : when @xmath66 is bound to the dna , on average , a larger @xmath167 is required to obtain a particular value of @xmath150 than when @xmath67 is bound .
this shows that @xmath150 , and hence the reaction co - ordinate , depends not only upon the difference in copy number @xmath167 , but also on the state of the operator .
importantly , the plots of figure [ fig6 ] ( b ) and ( c ) are not symmetric on making the transformations @xmath168 , @xmath169 and @xmath170 .
this demonstrates that the distribution of transition paths does not follow the steady state phase space density , which is symmetric on interchanging @xmath66 and @xmath67 , since the switch is by construction symmetric @xcite .
it also means that the tpe for the reverse transition , from @xmath3 to @xmath2 , would occupy a different region of phase space , as compared to the tpe for the transition from @xmath2 to @xmath3 ( see also supplementary material ) .
the origin of this asymmetry is that the dynamics of our system involves irreversible reactions ( reactions ( [ equationd]-[equationf ] ) ) .
indeed , our system does not satisfy microscopic reversibility .
finally , figure [ fig6 ] clearly demonstrates why the elimination of the operator state @xmath71 for the exclusive switches enhances its stability with respect to that of the general switch . in the general switch ,
as soon as a @xmath69 dimer is produced by some rare fluctuation , it can bind to the dna , switch off the production of @xmath66 and thereby accelerate the flipping of the switch .
for the exclusive switch , however , any @xmath69 dimer that is produced must wait for a second fluctuation by which @xmath68 is released from the dna , before it can bind .
this is the origin of the enhanced stability of the exclusive switch .
this article presents the forward flux sampling ( ffs ) method for the calculation of the rates of rare events in stochastic kinetic simulation schemes such as those used for biochemical reaction networks .
in contrast to previously developed methods for sampling the transition path ensemble @xcite , ffs does not require knowledge of the phase space density .
it is this feature that makes it possible to study rare events in biochemical networks .
our algorithm samples the tpe in a way that is , to our knowledge , new : many paths are grown simultaneously from state @xmath2 to state @xmath3 in a series of layers of partial paths , each layer forming the basis for the next .
the phase space separating stable states @xmath2 and @xmath3 is traversed by the algorithm in a `` ratchet - like '' manner , making the method highly suitable for very rare events , where one - at - a - time path generation tends to be inefficient .
ffs is not applicable to systems whose dynamics is deterministic .
it could be used , however , in combination with any stochastic simulation technique .
this will make it useful for a wide range of problems in soft condensed matter systems , including rare events and non - equilibrium processes . we have demonstrated our method using stochastic simulations of a simple genetic switch consisting of two mutually repressing genes .
following earlier work @xcite , we compare the case where both protein products can bind simultaneously as dimers to the dna ( the general switch ) , to that where each protein dimer excludes the binding of the other ( the exclusive switch ) .
the results obtained using ffs are in good agreement with those of long brute - force simulations for both switches .
the computational time required for the ffs calculations is far less than for the brute - force simulations , and in addition , does not increase as the rate constant decreases . indeed , using ffs
we could simulate a switch that was too stable to be studied using brute - force calculations . by analysing the transition path ensembles we were able to discover the differences between the flipping mechanisms for the general and exclusive switches .
these allow us to understand the origin of the enhanced stability of the exclusive switch .
the ffs method will be easily applicable to many important biochemical switches , for which prediction of the rates and pathways of switching should lead to a better understanding of the design principles underlying their stability .
we thank peter bolhuis for very helpful discussions and daan frenkel , rutger hermsen and harald tepper for their careful reading of the manuscript .
the work is part of the research program of the stichting voor fundamenteel onderzoek der materie ( fom ) " , which is financially supported by the nederlandse organisatie voor wetenschappelijk onderzoek ( nwo ) " .
|
bistable biochemical switches are ubiquitous in gene regulatory networks and signal transduction pathways . their switching dynamics , however , are difficult to study directly in experiments or conventional computer simulations , because switching events are rapid , yet infrequent .
we present a simulation technique that makes it possible to predict the rate and mechanism of flipping of biochemical switches .
the method uses a series of interfaces in phase space between the two stable steady states of the switch to generate transition trajectories in a ratchet - like manner .
we demonstrate its use by calculating the spontaneous flipping rate of a symmetric model of a genetic switch consisting of two mutually repressing genes .
the rate constant can be obtained orders of magnitude more efficiently than using brute - force simulations . for this model switch ,
we show that the switching mechanism , and consequently the switching rate , depends crucially on whether the binding of one regulatory protein to the dna excludes the binding of the other one .
our technique could also be used to study rare events and non - equilibrium processes in soft condensed matter systems .
biochemical switches are essential for the functioning of living cells .
these switches are networks of chemical reactions that exhibit more than one stable steady state ; in the presence of noise , flipping can occur between these states .
well - characterized examples include the lysis - lysogeny switch in bacteriophage @xmath0 @xcite and the lac repressor in _
e. coli _ @xcite .
experimental and theoretical studies have established the presence of bistability in other biochemical networks , including those regulating the cell cycle and developmental fate @xcite .
in addition , synthetic switches have been constructed _ in vivo _
@xcite .
computational modeling has an important role to play in explaining the properties of biochemical switches .
a stochastic approach is required to obtain the mechanism and rate of switching , since these switches are flipped by noise .
examples of such approaches are the chemical langevin technique @xcite , analysis of the chemical master equation @xcite or stochastic simulation techniques .
simulation algorithms that generate trajectories consistent with the chemical master equation include the gillespie algorithm @xcite and stochsim @xcite . where spatial resolution is required ,
methods such as green s function reaction dynamics can be used @xcite .
biochemical switches are often very difficult or impossible to simulate using the above techniques in a brute - force manner .
this is because they can be extremely stable , showing few or no flips during the accessible simulation time .
the average number of spontaneous transitions from the lysogenic to the lytic state for bacteriophage @xmath0 , for example , is about one in @xmath1 bacterial generations @xcite .
new methods are therefore required to model such important but rare events in biochemical networks
. techniques for the simulation of rare events have been developed in the field of soft condensed matter physics @xcite .
recent developments focus on the transition path ensemble ( tpe ) . for a rare transition between stable states @xmath2 and @xmath3
, this is the set of all ` reactive ' trajectories leading from @xmath2 to @xmath3 ( transition paths ) .
analysis of the tpe gives detailed information on the transition mechanism and leads to a prediction of the rate constant .
transition path sampling ( tps ) methods have been developed to generate members of this ensemble in a computationally efficient way @xcite .
tps has been applied to a wide variety of problems , including chemical reactions in solution , conformational transitions in biopolymers and protein folding @xcite .
biochemical switches , however , differ fundamentally from these problems .
as we shall discuss , in simulations of reaction networks the stationary distribution of states is generally not known _ a priori_. as a result , tps methods can not straightforwardly be applied . in this article , we present a new scheme for sampling the tpe and computing the rate constant .
this `` forward flux sampling '' ( ffs ) method is efficient and straightforward . it does not require prior knowledge of the phase - space density and can be applied to simulations of biochemical networks .
the method could also be implemented in any other stochastic dynamics scheme . to our knowledge , ffs constitutes a novel approach to sampling the tpe . rather than generating transition paths one at a time ( as in tps ) ,
a large number of paths are grown simultaneously from state @xmath2 to state @xmath3 in a series of connected layers . as an application of the ffs method
, we have calculated the spontaneous flipping rate of a simple genetic switch , consisting of two mutually repressing genes .
we show , in agreement with previous work @xcite , that the stability of this switch is greatly enhanced when the operator regions for the two genes are mutually exclusive , and that this is due to an important change in the flipping mechanism .
|
since the launch of the chandra x - ray observatory , it has become evident that x - ray emission is a common feature of jets in radio galaxies and quasars ( see , e.g. , @xcite for low - power radio galaxies , @xcite and @xcite for x - ray surveys of powerful radio galaxies and quasars , the overview articles by @xcite and @xcite for emission mechanisms , and the xjet home page for an up - to - date list of x - ray emission associated with extragalactic jets ) .
typically , the x - ray emission from low - power jets ( @xcite [ fr ] class i ) fits on a single synchrotron spectrum with their radio and optical emission and is thus satisfactorily explained as synchrotron emission ( with the interesting problem of having to accelerate x - ray emitting synchrotron electrons _ in situ _ ) .
however , the spectral energy distribution ( sed ) of high - power ( fr ii ) jets usually shows the so - called `` bow - tie '' problem , i.e. , their x - ray spectrum does not fit on an extrapolation of the observed radio / optical synchrotron spectrum .
a representative case in point is the first new x - ray jet detection by chandra , pks 0637 - 752 @xcite , with an observed cutoff to the synchrotron emission in the optical range , but a fairly hard x - ray spectrum . ]
( @xmath0 ) and an x - ray flux per frequency decade ( i.e. , @xmath1 ) exceeding the optical one . as the predicted x - ray intensity from synchrotron self - compton ( ssc ) emission
is usually orders of magnitude below the observed one , it has been suggested that these jets are still highly relativistic at kiloparsec scales , with bulk lorentz factors in the range 530 and beyond , and that the x - rays are inverse - compton ( ic ) scattered cosmic microwave background ( cmb ) photons ( the beamed ic - cmb process ; * ? ? ?
* ; * ? ? ?
the high lorentz factors are necessary in order for the energy density of the cmb photon field to be boosted sufficiently in the jet rest frame ( by a factor @xmath2 , where @xmath3 is the _ bulk _ lorentz factor of the jet fluid ) to account for the observed x - ray : radio flux ratios .
the beamed ic - cmb process is generally invoked to account for the x - ray emission from jets in which a single synchrotron component can not account for the emission in all wavebands .
if correct , the inferred values of @xmath3 would constitute the first firm measurement of the velocity of high - power jets ( for the velocities of low - power jets , see the modeling by * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) with profound implications for the analysis of jet energetics and composition .
since the increase of the cmb energy density with redshift would cancel the surface brightness dimming with redshift , beamed ic - cmb x - ray jets at high redshift could be cosmic beacons that outshine their quasars @xcite .
moreover , there _ are _ jets in fr ii radio galaxies and quasars whose x - ray emission is either well explained as synchrotron emission on an extrapolation of the radio - optical spectrum ( e.g. , in some of the jets in the survey by @xcite , and that in 3c403 [ @xcite ] ) , or because the necessary doppler factors can not be achieved from geometrical considerations , such as in pictor a @xcite where there is x - ray emission from the counter - jet as well as the approaching jet . a detailed check of the beamed ic - cmb model is therefore warranted .
there are two basic testable predictions of the beamed ic - cmb model which arise from the fact that those electrons responsible for the upscattering of cmb photons into the x - ray band have low lorentz factors ( @xmath4 ) and hence emit synchrotron radiation at wavelengths of a few tens or hundreds of megahertz , below the typical gigahertz - range observing frequencies allowing high - resolution studies of radio jets .
essentially , a similar part of the electron energy distribution can be observed in two different wavebands ( the electrons producing ic emission in the chandra band emit synchrotron radiation in the range 40hz1mhz ; * ? ? ?
* ) , leading us to expect a close correspondence both of morphology and of spectral index between a beamed ic - cmb jet s x - ray and radio emission . we have chosen the jet in 3c273 for a detailed test of the beamed ic - cmb model with the chandra x - ray observatory .
this jet has a unique combination of properties that make it worthy of detailed study : it has a large angular size ( bright radio , optical and x - ray emission being observed from 11 out to 22 from the core ) , it is one of the closest high - power jets ( @xmath5 ) so that smaller physical scales and @xmath6 , leading to a scale of @xmath7 per second of arc at 3c273 s redshift of 0.158 .
] can be resolved than in most similar jets , and high - resolution vla and hst data are available ( * ? ? ? * and references therein ) that allow the construction of detailed seds and the study of the relation between the jet s x - rays and the radio - optical synchrotron emission .
3c273 s jet is among those from which x - ray emission had been detected already with the _ einstein _ satellite @xcite .
based on a comparison of rosat hri observations with radio and optical data , @xcite concluded that the x - ray emission could only be due to the same synchrotron component as the jet s radio - optical emission in the first bright feature of the jet ( `` knot a '' ) , but not in the remainder of the jet , where there is a cutoff to the synchrotron component in the near - infrared / optical range ; later , @xcite and @xcite showed that at 03 resolution , already the radio - optical seds of most of the jet , including knot a , require a two - component model to account for the emission .
hence , the x - ray emission there can not be due to a single radio
optical x - ray synchrotron component , either ( but see * ? ? ?
* who suggests that an additional contribution of synchrotron emission from particles moving in fields that are tangled on small scales may explain the observed flattening , at least in parts of the jet ) .
@xcite suggested that the second optical component may be the same component as the jet s x - ray emission , but could not constrain the emission mechanism for the high - energy component further .
regarding the x - ray emission mechanism , @xcite ruled out ssc as well as thermal bremsstrahlung .
the first chandra observations of this jet were presented by @xcite and @xcite .
@xcite presented the first x - ray image of this jet with both high resolution and high signal - to - noise ratio . regarding the seds , they came to similar conclusions as @xcite .
@xcite had analyzed a smaller set of the early chandra data and favored a beamed ic - cmb model for the emission from all parts of the jet .
@xcite compared the optical and x - ray morphology at the chandra resolution of 078 and found that the x - ray and optical emission come from the same parts of the jet and show very similar features . however , there are offsets between the x - ray and optical peaks in one or both of the first two bright knots . in the remainder of the jet ,
the relative variations in x - ray brightness are smaller than those in the radio and optical . such size and
brightness differences require some parameter fine - tuning in the ic - cmb model .
thus , previous observations had left the nature of the x - ray emission mechanism of 3c273 unclear .
here , we present a spectral analysis of our new chandra observations , which we combine with archival data and our vla + hst dataset from @xcite .
observations and data reduction are described in [ s : obs ] , our spectral analysis and its results are presented in [ s : analysis ] , and we discuss in [ s : discussion ] the implications of our results for the beamed ic - cmb model and other emission mechanisms .
we summarize our findings in [ s : conc ] .
we have obtained four observations of 3c273 and its jet in chandra observing cycle 5 , totaling just under 160 ksec observing time .
the observations were split into four exposures of equal length in order to search for variability within the observing cycle .
the quasar and jet were placed on the acis - s3 chip in each of these .
we also use calibration observations of 3c273 from the chandra archive : two acis - s3 data sets from cycle 1 , with 30 ksec exposure time each , and seven acis+hetg calibration observations , totaling just under 200 ksec .
the gratings only transmit about 10% of the soft x - ray photons ( @xmath8kev ) , but about 75% of the hard x - ray photons ( @xmath9kev ) .
hence , we use the grating data to perform consistency checks on the spectral shape of the jet at the high - energy end of the chandra bandpass .
we will not use them for morphological analysis .
this decision is based purely on the fact that their inclusion would have required substantial additional effort without a commensurate gain in the total number of detected photons ; as the calibration data were used by the calibration team to check the effective area , not the pileup or contamination correction , they could be used without causing any circular reasoning .
table [ t : obslog ] gives the observing log .
lrrrr none & 1711 & 28.09 & 2000 - 06 - 14 05:13:19 & 1.1 + & 1712 & 27.80 & 2000 - 06 - 14 13:43:27 & 3.2 + & 4876 & 41.30 & 2003 - 11 - 24 23:08:40 & 0.4 + & 4877 & 38.44 & 2004 - 02 - 10 03:40:39 & 0.4 + & 4878 & 37.59 & 2004 - 04 - 26 20:55:16 & 0.4 + & 4879 & 39.23 & 2004 - 07 - 28 03:35:33 & 0.4 + hetg & 459 & 39.06 & 2000 - 01 - 10 06:46:11 & 2.5 + & 2463 & 27.13 & 2001 - 06 - 13 06:41:21 & 2.1 + & 3456 & 25.00 & 2002 - 06 - 05 10:03:12 & 1.9 + & 3457 & 25.38 & 2002 - 06 - 05 17:19:11 & 2.5 + & 3573 & 30.16 & 2002 - 06 - 06 00:43:51 & 2.5 + & 4430 & 27.60 & 2003 - 07 - 07 12:08:58 & 3.2 + & 5169 & 30.17 & 2004 - 06 - 30 12:39:18 & 2.5 we used the chandra analysis software ciao v. 3.2 or later and calibration database caldb v. 3.0 or later to reprocess all data , taking advantage of updated calibrations , in particular the acis contamination correction @xcite , and to remove the pixel randomization . the spectral analysis
( see following section ) was performed using spectral fits in the sherpa software package @xcite .
in addition , we constructed flux - calibrated brightness maps for the acis - s3 observations .
registering chandra data sets is challenging .
the absolute pointing calibration of chandra of about 04 ( * ? ? ?
* table 5.1 ) is not sufficient for our analysis , since the jet width is only about 1 .
the quasar core is the only bright x - ray point source in the vicinity of the jet , but it is so bright that it is severely piled up ( * ? ? ?
* section 6.14 ) even with a frame time of 0.4s ( chosen in our new observations ) or with the hetg in place .
however , the readout streak in the grating - free observations contains so many counts that it can be used to constrain the location of the quasar core in the perpendicular direction with high precision and accuracy .
the unpiled wings of the psf provide the necessary second constraint .
we further attempt to mitigate the effect of pileup by constructing an `` ev per second '' ( evps ) map , an energy - weighted count rate map .
three of the authors ( sj , deh , hlm ) have performed independent fits of the non - piled up part of the psf and the readout streak , both on the evps maps and on the brightness maps , to obtain the position of the quasar core in each of the six acis - s3 sets . from the rms of the individual determinations
, we estimate that the alignment is better than 01 in each coordinate .
we average the individual measurements to get a final quasar position in `` physical '' acis coordinates .
we then update the world coordinate system ( wcs ) of the individual event files and maps to assign 3c273 s right ascension and declination as determined from radio observations to the quasar s fitted physical coordinates .
for the hetg zeroth - order images , which are used for consistency checks , we fit a gaussian to the wings of the quasar image to fix the astrometry relative to the quasar core .
we perform a separate spectral analysis in each of nine regions , which we have defined considering both the radio / optical and x - ray morphology .
figure [ f : xraymap ] shows the regions on a chandra , hst and vla image .
the nomenclature for the subdivisions of the original regions abcd @xcite is that established by ( * ? ? ?
* their table 4 ) ; later , @xcite and thence @xcite used different subdivisions .
there are no gaps between the extraction regions , and each is about 1 - 2 in size .
while the jet shows similar morphology at all wavelengths , there are differences in some places , which create some ambiguity as to where the boundaries between different jet regions ( or knots ) should be placed .
the differences in morphology at different wavelengths imply spatial variations in the spectral shape .
hence , some of our measurement regions mix emission from jet features with different spectral shapes .
given that the chandra resolution of 078 ( full - width at half - maximum intensity of a gaussian fitted to the readout streak ) is comparable to the typical diameter of the jet features , such mixing of spectral shapes can not be avoided entirely .
however , the morphology differences are mostly minor , the sole exception being region b1 , in which the x - ray and optical emission clearly arch to the north , while the radio emission has an arc to the south ( compare * ? ? ?
this difference may in fact contain a crucial hint to the nature of the emission and particle acceleration mechanism acting in the jet ; however , the two arcs are not fully resolved by chandra , so that we include both in a single region .
still , the need to consider the emission from the arcs in b1 separately should be borne in mind when analyzing the spectral shape there .
for the estimation of the background count rate , we define annuli of similar radial extent as the corresponding jet region that are centered on the quasar , as large - angle scattered photons from the quasar core are the dominant contribution to the background .
the background regions exclude the jet regions themselves as well as the readout streak . considering the locations of those events classified
as `` afterglow events '' by the ciao pipeline suggests that obsids 1711 and 1712 include some piled - up events at the brightness peak of region a. the other acis - s3 observations used a 1/8th subarray , shortening the frame time correspondingly , and therefore do not suffer from pileup .
the hetg observations have lower count rates in the jet and pileup is much less than 1% there . to determine the flux and x - ray spectral index , we extract spectra and response files for each source region using the ` psextract ` ciao script .
we searched for variability in the count rates of each of the regions over all six acis - s3 data sets and found no variations in excess of the expected poisson fluctuations for a constant count rate .
in addition , we performed separate sherpa power - law fits for each knot and each epoch , as detailed below .
we used the @xmath10 test to assess the significance of the fluctations .
the observed fluctuations and their probabilites are given in table [ t : variab ] . in nearly all regions ,
the @xmath10 test returned very high probabilities for obtaining the observed fluctuations in the fit parameters across the six epochs as random fluctuations of an underlying spectrum with both constant flux and spectral index .
only the flux variability for region d2h3 is significant at the 96% confidence level ; however , at 3% , it is also amongst the smallest observed fluctuations , and there is no systematic trend .
thus , there is no detection of significant systematic variability in any of the regions .
lrrrr a & 6 & 0.92 & 4 & 0.90 + b1 & 9 & 0.99 & 8 & 0.88 + b2 & 7 & 1.00 & 2 & 0.08 + b3 & 5 & 0.08 & 11 & 0.49 + c1 & 6 & 0.35 & 9 & 0.58 + c2 & 4 & 0.10 & 7 & 0.40 + d1 & 9 & 0.75 & 9 & 0.51 + d2h3 & 3 & 0.04 & 6 & 0.39 + h2 & 14 & 0.12 & 18 & 0.16 in the remainder of this paper , we therefore concentrate on the question whether the spectral shape of the jet emission is consistent with the beamed inverse - compton model .
the morphology will be discussed in detail in a separate paper . for the analysis in the remainder of the paper , we combine all six epochs of acis - s data . in each region , we perform a joint sherpa spectral fit for data and background , using simple power laws to describe both the source and the background components . as the quasar itself shows both spectral and flux variability in the soft x - ray band @xcite , we allow an independent background model component for every epoch .
we include a ` jdpileup ` model @xcite for the fit of region a in obsids 1711 and 1712 ; the use of this model is appropriate since this region is unresolved by chandra @xcite and it is justified by the fit results which indicate a pileup fraction of 3.6% and 9.4% , respectively .
trial fits including a pileup model for the other 4 data sets and for all other jet regions had results consistent with 0 piled up events , even though scaling the piled up fraction by the frame time would lead us to expect a pileup fraction of about 1% ( see frame times in table [ t : obslog ] ) . as in @xcite
, we use the hydrogen column density of @xmath11@xmath12 given by @xcite .
we perform fits using two methods .
first , we used the likelihood method ( cash statistic ; * ? ? ? * ) for unbinned events , using only the energy range 0.5 kev8.0 kev .
secondly , we grouped the data to a minimum of 15 counts per bin with @xmath10 primini statistics @xcite , in this case excluding all bins below 0.5 kev .
we do separate fits for the data from the hetg zeroth - order images and from the grating - free observations .
the fit results from both data sets ( hetg zeroth - order images and grating - free s3 images ) and both fitting methods ( unbinned / cash and grouped/@xmath10 ) are consistent with each other . we have carried out further spot checks using xspec power - law fits of the grouped data sets , in which background events are subtracted before fitting ( instead of being included in the fit as a separate model component ) , and found excellent agreement with the sherpa results . thus , the following analysis will make use of the fit parameters obtained from the fits of unbinned data with cash statistics , and in the remainder of the paper , we only bin data for display purposes .
lrrrrr a & 10849 & @xmath130.83 & 0.02 & 46.54 & 0.54 + b1 & 2632 & @xmath130.80 & 0.03 & 10.89 & 0.25 + b2 & 3992 & @xmath130.97 & 0.03 & 19.98 & 0.33 + b3 & 768 & @xmath131.13 & 0.07 & 3.41 & 0.14 + c1 & 1115 & @xmath131.07 & 0.06 & 4.85 & 0.16 + c2 & 1467 & @xmath130.96 & 0.05 & 6.25 & 0.18 + d1 & 1182 & @xmath131.02 & 0.05 & 5.16 & 0.17 + d2h3 & 1707 & @xmath131.04 & 0.04 & 7.82 & 0.20 + h2 & 317 & @xmath131.27 & 0.12 & 1.30 & 0.09 table [ t : fits ] gives the fit results . in all cases ,
a simple power law was a good fit to the data , according to the probabilities obtained for the @xmath10 fits .
we also generated x - ray seds in six energy bands from corresponding brightness maps .
the seds for knots a and b2 show some curvature in the range 45kev and a slightly softer spectrum at lower energies than the sherpa fits , which would be consistent with a marginally detected high - energy cutoff .
however , no evidence for a cutoff was found in the fits using the event lists , neither in the pure power - law fits nor in fits with a cut - off power law ( sherpa model ` xscutoffpl ` ) .
the x - ray spectral indices of a and b1 are consistent with each other and both approximately @xmath14 .
the knots in the remainder of the jet have slightly softer x - ray spectra ; all knots from b2 up to and including d2h3 again have spectral indices that are consistent with each other and a value of @xmath15
. the radio hot spot region h2 is significantly detected in x - rays in the combined data set ; in fact , @xcite already noted that the jet was detected out to 22 from the core , with the radio peak in h2 being located 214 from the core .
a comparison of the x - ray flux profile of regions d2h3 and h2 with that of the nearly point - like region a confirms the reality of the detection of h2 .
its spectrum is marginally softer than the rest of the jet .
figure [ f : seds ] shows the spectral energy distributions ( seds ) of the nine jet regions from radio up to x - rays .
it is evident from the seds that the jet emission from radio up to x - rays can not be explained as a single synchrotron component for _ any _ part of the jet . except for region
a , this was already known from the analysis of rosat data @xcite and of the first chandra data sets @xcite .
@xcite showed that a single - component synchrotron spectrum was not in fact viable for knot a , chiefly because the new multi - frequency multi - configuration vla data set used by them , and here , yielded substantially different radio sed shapes .
our spectral index measurement for knot a of @xmath16 is consistent with the 90% confidence interval @xmath17 given by @xcite at the @xmath18 level .
however , @xcite obtained a much harder x - ray spectrum with @xmath19 .
the discrepancy is due to two systematic effects .
first , pileup slightly hardened the data used by @xcite .
secondly , corrections for acis contamination , which absorbs x - ray photons mostly below 1 kev , were released only after publication of their paper ( see * ? ? ?
we can reproduce the @xcite result by turning off the pileup and contamination correction . in fig .
[ f : alpha_alpha ] , we compare the radio and x - ray spectral indices directly .
it clearly shows the basic and important finding of our spectral analysis : except for regions a and b1 , the x - ray spectrum is significantly softer than the radio spectrum .
the aim of the chandra observations presented here was a test of both the synchrotron and the beamed ic - cmb models for the jet s x - ray emission .
the non - detection of significant variability does _ not _ rule out synchrotron as x - ray emission process , as the chandra resolution element is sufficiently large ( the fwhm of 078 corresponds to 2.1kpc ) to allow possible variability to be washed out .
we now discuss the implications of the observed sed shape for both synchrotron and beamed ic - cmb as x - ray emission mechanism .
the models have to account for the following observational facts : 1 .
[ i:2spec ] there is a spectral hardening of the near - infrared / optical / ultraviolet emission in all parts of the jet .
the shape of the second optical / uv emission component can not be constrained to a high degree of certainty based on the present data set , but it may well be the extrapolation of the x - ray power law to lower frequencies @xcite .
[ i : xopt ] the x - ray emission can not be explained by a simple extrapolation of the total optical emission ( the bow - tie problem ; see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
[ i : radioalpha ] the radio synchrotron emission from the jet in 3c273 has a spectral index @xmath20 at all observed frequencies , down to 330mhz ( * ? ? ? * and confirmed by our unpublished vla data ) .
[ i : pol ] radio and optical emission show very similar degrees of linear polarization out to 18 from the core , and the polarization vectors are parallel to each other in all parts of the jet except for the optically quiet radio `` backflow '' or `` cocoon '' to the south of the jet @xcite .
[ i : xrad ] the x - ray emission has a softer spectrum than the radio emission in most parts of the jet ( figures [ f : seds ] and [ f : alpha_alpha ] ) . 6 .
[ i : xmorph ] the same morphological features are seen at all wavelengths , from radio to optical @xcite to x - rays @xcite .
items [ i:2spec ] and [ i : xopt ] individually rule out a one - zone synchrotron model , but would be compatible with a two - zone synchrotron model as well as an ic - cmb model with one or more zones . item [ i : radioalpha ] implies that the jet s total radio emission is dominated by a single synchrotron component down to the lowest currently observed frequencies .
the high linear polarization ( item [ i : pol ] ) lead to the conclusion that the optical emission is synchrotron light @xcite , and the similarity of the radio and optical polarization indicated that the radio and optical synchrotron emission are due to the same electron population @xcite . remarkably , if the x - ray emission could be shown to be of the same origin as _ most _ , and not just _ some _ , of the optical emission ( compare * ? ? ? * ) , the fact that the optical emission is synchrotron would imply that the x - ray emission is synchrotron , too ; this might be expected from the similarity of the optical and x - ray morphology ( item [ i : xmorph ] above ) , and is very strongly supported by the analysis of spitzer mid - infrared observations by @xcite .
the clarification of this issue requires the analysis of further mid - infrared ( spitzer ) and ultraviolet ( hst ) data . for
now , we discuss the viability of the ic - cmb and the two - zone synchrotron models given item [ i : xrad ] . in the beamed ic - cmb model , cmb photons
are upscattered into the chandra band by electrons with lorentz factors of order 50 - 200 . using the formula for the characteristic frequency @xmath21 of synchrotron emission for electrons with lorentz factor @xmath22 in a magnetic field with flux density @xmath23 , @xmath24 ( @xmath25 g )
, we obtain that these same electrons produce synchrotron emission at frequencies 20250mhz@xmath26 i.e. , at lower frequencies than the vla observations presented here ( the minimum - energy field in this jet is of order 1020nt for @xmath27 , and an order of magnitude smaller for @xmath28 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
. the electron energy distribution should be either a power law , or have curvature such that lower - frequency emission has a harder spectrum .
hence , in the beamed ic - cmb model one expects that the x - ray spectrum should have a spectrum that has the same spectral index as , or is harder than , the radio emission from the jet .
as shown by figures [ f : seds ] and [ f : alpha_alpha ] , the x - ray spectrum is significantly softer than the radio spectrum in all parts of the jet except for regions a and b1 .
when the beamed ic - cmb model is put forward to account for x - ray emission from high - power radio jets , or used to infer physical parameters such as the doppler factor and the line - of - sight angle , the usual assumption is a single - zone single - component model , i.e. , one in which the electron population producing synchrotron emission is identical to that upscattering cmb photons into the x - ray band ( see * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* e.g. ) and where identical jet volumes contribute to the synchrotron and inverse - compton emission .
this conventional single - zone beamed ic - cmb model is difficult to reconcile with our new chandra data for those parts of the jet in which the x - ray spectrum is softer than the radio spectrum , i.e. , the entire jet from region b2 outwards , at projected distances greater than 15 from the core .
compared to positing that the x - rays from jets in fr ii sources are due to synchrotron emission , the single - zone beamed ic - cmb model seemed to be more conforming to occam s razor , since it only needed the assumption that the bulk lorentz factors inferred for parsec - scale jets from vlbi observations of blazars persist on the kiloparsec scale . by comparison
, the synchrotron hypothesis would require at least a two - zone model .
in fact , single - zone models are inadequate for many jet observations ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ?
* ) , leading to the invocation of a second spatial zone and/or a second spectral component .
progress in distinguishing between the various models can only be made if they are falsifiable by observations ; x - ray telescopes with spatial resolution comparable to that achieved with radio and optical telescopes would be highly desirable in this context .
[ [ a - two - zone - model - from - velocity - shear ] ] a two - zone model from velocity shear + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the presence of a transverse velocity gradient , i.e. , velocity shear , in high - power jets would naturally create zones in the jet with different inverse - compton emissivities .
based on high - resolution observations and a physical model of the jet flow , the presense of velocity shear has been established in a number of low - power jets : those in the radio galaxies 3c31 @xcite , b2 0326 + 39 and b2 1553 + 24 @xcite and ngc 315 @xcite .
a `` spine - sheath '' shear structure has also been implied for low - power jets based on beaming statistics of fr i and bl lacs in the b2 survey @xcite , as well as from polarimetric observations of the jet in m87 @xcite .
however , both theoretical considerations @xcite and the analogy between accretion in agn and black hole x - ray binaries @xcite may lead us to expect that the velocity structure of high - power jets is different from that of low - power jets ( see the discussion in 7 of * ? ? ?
while the polarization properties of some jets in fr iis show a spine - sheath structure @xcite that may suggest a similar velocity structure as in fr is , @xcite have pointed out that the polarization structure may be caused by the magnetic field structure rather than a velocity gradient .
thus , even though such a structure has not yet been firmly established , it would not be implausible for fr ii jets to have a spine - sheath structure , warranting its consideration here . in a two - zone beamed ic - cmb model for 3c273 , the highly relativistic part of the flow ( the `` spine '' ) would account for the bulk of the x - ray emission , while the slower part of the flow ( the `` sheath '' ) would dominate the synchrotron emission ( fig .
[ f : twozone ] a ) .
thus , the observed radio spectral index of the jet is that of the sheath , with @xmath29 . to match the observed x - ray spectral index ,
the fast spine would need to have a synchrotron spectrum with @xmath30 . to keep the observed radio spectral index unchanged
, the spine would also need to have a much lower radio luminosity .
as the inverse - compton model requires the spine to have a large doppler factor , the radio emission from the spine _ can not _ be `` doppler - hidden '' by having a doppler factor much less than unity ( cf .
the suggestion of such doppler hiding making the jet in 3c353 edge - brightened ; * ? ? ?
instead , the spine has to have a much lower radio brightness than the sheath .
if the spine has an ( observer - frame ) flux density at 1.6ghz of 10% of that of the sheath , its steeper - spectrum contribution would not change the total spectral index between 1.6ghz and 330mhz by more than 0.05 , i.e. , within the spectral index errors .
hence , we consider a 10% contribution at 1.6ghz the maximum plausible flux density of the spine that would keep it undetectable as a separate steep - spectrum component . [ [ estimating - the - bulk - lorentz - factor ] ] estimating the bulk lorentz factor + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we can now estimate the bulk lorentz factors within a 2-zone model from the x - ray : radio ratio .
the x - ray emission has to be produced in the spine via upscattering of cmb photons by an electron distribution that would contribute only about 1% of the jet s total synchrotron luminosity ( where the latter is inferred from the radio flux ) . we just estimated that the spine could contribute at most 10% of the radio flux density at 1.6ghz ; by extrapolating the x - ray flux density to 1.6ghz using the observed spectral index , we obtain that the x - ray : radio ratio for the spine ( @xmath31 in the notation of * ? ? ?
* ) is of order 2000 .
this is much larger than the value @xmath32 inferred from the jet s total radio emission @xcite .
following the method presented by @xcite , we can infer the likely doppler and lorentz factors for the spine if we have an estimate of the magnetic field .
if there is equipartition of energy between the magnetic field and the particles in both the spine and the sheath , we can use the scaling of the minimum - energy field @xmath33 with source luminosity @xmath34 , volume @xmath35 and doppler factor^{-1}$ ] is the relativistic doppler factor for an object moving at speed @xmath36 at angle @xmath37 to the line of sight ] @xmath38 to estimate the relative magnetic flux densities . in this case , @xmath39 ( * ? ? ? * equation a7 ) is the appropriate scaling ; with our estimate that the spine has a total synchrotron luminosity of about 1% of that of the sheath , and assuming that the spine occupies @xmath40 of the jet diameter and hence @xmath41 of the sheath s volume , the magnetic field in the spine would still be @xmath42 20% of the sheath s ( where @xmath43 is the relative relativistic doppler factor of spine and sheath @xcite ) .
hence , following @xcite , we obtain that the bulk @xmath44 would have to be of order 50100 to produce @xmath45 .
[ [ jet - deceleration - in - a - two - zone - ic - cmb - model ] ] jet deceleration in a two - zone ic - cmb model + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the synchrotron emission from the sheath will appear boosted in the frame of the spine , thus providing an additional seed photon field for compton scattering @xcite .
depending on the relative sizes and speeds of spine and sheath , the energy density of the sheath photon field as perceived in the frame of the spine could be comparable to that of the cosmic microwave background .
if this was the case , the boosted cmb would need to contribute only half of the seed - photon energy density .
as the boosted cmb energy density scales like @xmath2 , the required bulk lorentz factor would be reduced only by about @xmath46 . at the same time
, the sheath electrons will upscatter seed photons from the spine , creating an additional ic component . in the case of 3c273
, the dominant contribution to the synchrotron luminosity would arise in the sheath ( layer in the terminology of * ? ? ? * ) rather than the spine as assumed for the blazars considered by @xcite .
this additional `` mutual - compton '' ( mc ) scattering will produce detailed seds that are quite different from those arising in a one - zone synchrotron + beamed ic - cmb model .
as noted by @xcite , the mc scattering could cause a deceleration of the spine ( an `` inverse compton - rocket '' effect ) .
this might provide a more natural explanation for the deceleration that is necessary to account for the changing x - ray : radio ratios along some extragalactic jets , including 3c273 s ( see @xcite and in particular @xcite for the deceleration of this jet implied by a one - zone beamed - ic model ) .
the fr i flow modeling by @xcite shows that these jets decollimate as they decelerate , presumably by entrainment , while no sign of decollimation is observed in 3c273 and other high - power jets .
however , as noted by @xcite , the large number of free parameters of this kind of two - zone model makes it very difficult to obtain definite predictions for the expected shape of the sed .
the most stringent observational constraints will be imposed by verifying whether or not the radio spectrum shows any signs of the steeper - spectrum component assumed to arise in the spine at lower frequencies than observed so far , e.g. , in the frequency range around 100mhz that will soon be accessible with the low - frequency array ( lofar ) . in any case
, the two - zone ic - cmb model requires that the `` spine - sheath '' jet should develop very different low - energy electron populations in the different parts of the flow .
this may seem somewhat artificial if the velocity structure arises from simple velocity shear , unless the shape of the electron energy distribution in jets is governed entirely by a distributed acceleration mechanism , rather than by shock acceleration in the innermost part of the jet ( compare the discussion of shock - like versus jet - like acceleration in * ? ? ?
the different energy distributions might arise more naturally if the two - zone flow existed from the outset ( e.g. , as in the model by * ? ? ? * ) .
a detailed parameter study is necessary to explore the range of plausible total seds ( including synchrotron emission , beamed ic - cmb , and mc scattering ) ; this is beyond the scope of this paper . [ [ constraints - on - two - zone - models - from - the - jet - morphology ] ] constraints on two - zone models from the jet morphology + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we will present a detailed study of the x - ray morphology based on the new data in a separate paper ; the most relevant finding so far is that most of the jet is now clearly resolved transversely even at the chandra resolution .
however , the limit to the width of the brightness peaks in knots a and b2 is smaller than the measured sizes in the optical and radio bands @xcite , where they are clearly resolved @xcite .
this is incompatible with the beamed ic - cmb model even in a decelerating one - zone flow such as that considered by @xcite , the optical emission should be the most concentrated because the `` optical synchrotron '' electrons have the shortest synchrotron loss scales . thus , even though the sed of knot a admits a one - zone ic - cmb model , its morphology does not .
in general , the presence of unresolved knot emission superimposed on resolved more diffuse jet emission compels us to consider two - zone models ( although not necessarily only spine - sheath models ) . in summary ,
if fr ii jets have a spine - sheath structure , and the spine is highly relativistic , there will naturally be two zones in the jet with different ic - cmb emission .
the `` mutual - compton '' scattering of the radiation from the one zone by electrons in the other might provide a natural explanation for the apparent deceleration of 3c273 implied by the changing x - ray : radio ratio @xcite .
however , a two - zone ic - cmb model appears to require unrealistically large bulk lorentz factors for the spine of the jet , of order 50100 .
moreover , the two - zone ic - cmb model does not intuitively explain why the emission in the x - ray band has a softer spectrum than that in the radio band .
as pointed out above , a one - zone synchrotron model is already excluded by the spectral hardening in the near - infrared / optical / uv region .
hence , we only discuss a two - zone synchrotron model here .
the evidence in favor of a spine - sheath structure in fr ii jets summarized in
[ s : disc.ic.twozone ] is of course equally relevant to the discussion of a two - zone synchrotron model . in a two - zone synchrotron model
, there would be the same need to account for these different electron energy distributions , but the difference would now be between the _ low_-energy electron distribution of one zone ( emitting at radio wavelengths ) and the _ high_-energy electron distribution of the other one ( producing x - rays ) .
it is already known that many low - power jets emit synchrotron x - rays @xcite , and possibly _ all _ such jets do @xcite .
the extremely short synchrotron loss timescales of 10s of years for the x - ray emitting electrons require an _ in - situ _ acceleration process to generate these electrons .
it would then be an obvious assumption that high - power jets are also able to accelerate such electrons .
indeed , some high - power jets show evidence for synchrotron x - ray emission @xcite .
moreover , @xcite , @xcite and @xcite have all argued that efficient particle acceleration is possible in regions of velocity shear .
* thus , particle acceleration in a sheared region of the flow might provide a direct physical link between a two - zone velocity structure and a two - component synchrotron spectrum , since an additional acceleration mechanism would be available in the sheared part of the flow .
* we stress that the presence of shear acceleration does not preclude the possibility that there is another distributed acceleration mechanism acting in the other , or both , fluid zones , e.g. , one related to the dissipation of magnetic fields @xcite in the context of such a model , the soft x - ray spectrum in 3c273 with @xmath47 for regions b2 and beyond could be caused either by a soft energy spectrum being generated by the distributed acceleration process , or because it arises from a loss - dominated part of the energy distribution .
the difference in x - ray spectral index between the first bright knot a and the faint bridge b1 on the one hand , and the second bright knot b2 and beyond on the other hand , would require at least some difference of the physical parameters ( either by producing a different electron energy distribution , or by producing stronger cooling ) , and perhaps even a different acceleration mechanism acting there .
this physical difference might have been expected given that a is the first bright region of the optical and x - ray jet , after a gap of 12 with only very faint optical @xcite and x - ray @xcite emission .
the difference between a and b2 may well be related to the reason for the jet lighting up at a. we will revisit the emission from the inner part of the jet in a forthcoming paper on the x - ray morphology . in figure
[ f : colorjet ] , we give a summary of the energetically dominant wavebands and the associated spatial component and emission mechanism , both for the two - zone inverse - compton and the synchrotron model .
we have presented new deep chandra acis - s3 observations of the jet in 3c273 and analyzed them in conjunction with archival data , concentrating on the shape of the radio / optical / x - ray sed in this paper .
fitting power laws to the radio and x - ray spectra , we find that a single - zone beamed ic - cmb model for the x - ray emission is viable only in regions a and b1 ; in the remainder of the jet , the x - ray spectra are softer than the radio spectra .
this is in contrast with the similarly well - studied x - ray jets of pks0637@xmath13752 , 1136@xmath13135 , and 1150@xmath48497 , where the radio and x - ray slopes are compatible with each other in all parts of the jet where both have been measured @xcite .
thus , most of the x - ray emission from the jet in 3c273 must be due to a different emission process than compton scattering of cmb photons by the same electrons producing the bulk of the synchrotron radio emission .
a single - zone synchrotron model extending from radio to x - rays had already been ruled out for all parts of the jet based on previous observations @xcite . a two - zone beamed ic - cmb model does not seem capable of accounting for the x - ray emission from 3c273 s jet with any fewer assumptions than a two - zone synchrotron model .
a two - zone ic - cmb model may still be viable , but would require extreme bulk lorentz factors to produce the observed x - ray emission from much fewer electrons than inferred from the jet s total radio synchrotron emission .
in addition , there is no intuitive explanation for obtaining different electron energy distributions in the spine and the sheath .
by contrast , a two - zone synchrotron model , in which a distributed particle acceleration mechanism related to velocity shear is producing the x - ray emitting particles , provides a causal relationship between the difference in fluid velocities and the different spectral properties .
there would still be velocity and hence beaming differences between the two zones , but they would not have to be as extreme as in the two - zone ic - cmb models , and therefore one might hope to obtain better - constrained parameters for the two zones , and hence falsifiable predictions of this model .
an x - ray polarimeter would enable significant progress to be made in this question ; also , the mid - infrared wavelength region that has now been made accessible by the spitzer space telescope will be of great importance , as shown by the analysis of this jet s mid - infrared emission by @xcite
. it would clearly be desirable to test the consistency of the radio and x - ray spectral indices with the one - zone beamed ic - cmb model for more jets in the future .
we will present a detailed analysis of the x - ray morphology of 3c273 s jet elsewhere ( for first results , see * ? ? ?
this will allow more tests of both emission models for the x - rays .
in particular , understanding how the ( mostly ) wavelength - independent morphology of this and other jets is produced will provide crucial insights into both emission and particle acceleration mechanisms acting in this and other jets .
work at sao was partially supported by nasa contract nas8 - 03060 and grants nasgo3 - 4124a and nasgo4 - 5120c .
hlm was supported from sao contract sv3 - 73016 to mit for support of the chandra x - ray center , operated by sao on behalf of nasa under contract nas8 - 03060 .
sj received support from nasa contract nasgo4 - 5120a , from the u.s .
department of energy under contract no .
de - ac02 - 76ch03000 , and from the max - planck - gesellschaft through an otto hahn fellowship .
he thanks both the cfa and mit for hospitality .
we are grateful to tom maccarone and especially aneta siemiginowska for providing essential advice on spectral fitting , and to the anonymous referee for a constructive report that helped in improving the presentation of our results . , h. l. , tennant , a. , grant , c. e. , hitchcock , a. p. , odell , s. l. , & plucinsky , p.
p. 2004 , in proceedings of the spie , vol .
5165 , x - ray and gamma - ray instrumentation for astronomy xiii , ed .
k. a. flanagan & o. h. w. siegmund , 497508
|
the jet in 3c273 is a high - power quasar jet with radio , optical and x - ray emission whose size and brightness allow a detailed study of the emission processes acting in it .
we present deep chandra observations of this jet and analyse the spectral properties of the jet emission from radio through x - rays .
we find that the x - ray spectra are significantly softer than the radio spectra in all regions of the bright part of the jet except for the first bright `` knot a '' , ruling out a model in which the x - ray emission from the entire jet arises from beamed inverse - compton scattering of cosmic microwave background photons in a single - zone jet flow . within two - zone jet models ,
we find that a synchrotron origin for the jet s x - rays requires fewer additional assumptions than an inverse - compton model , especially if velocity shear leads to efficient particle acceleration in jet flows .
|
the phenomenon of collectively enhanced emission , described by dicke for ensembles of atoms @xcite , can also occur in quantum dots ( qds ) @xcite .
the cooperative radiation effect is due to the collective interaction of the two emitters with quantum radiation field . in the case of non - identical dots and in the absence of interactions
, the collective emission appears if the interband transition energies in various qds differ by no more than the emission line width @xcite , which requires the dots to be nearly identical , beyond the recent technological possibilities .
however , coupling between the dots restores the collective nature of the emission and leads to accelerated or slowed down emission even for dots with different transitions energies @xcite .
two typical couplings that may appear in a system of qds are long - range coulomb interactions and short - range couplings that may result from a combination of carrier tunneling and coulomb correlation effects .
as we have shown , these two couplings show no qualitative difference in either accelerating or slowing down the emission in spite of their essentially different physical origin and properties @xcite .
the simulations of spontaneous emission from ensembles of coupled qds show that the emission rate in such a system can indeed be increased due to collective coupling of the emitters to the electromagnetic field @xcite .
as the positions of the dots are random within the sample plane , it might happen that this enhanced emission effect results from clustering of dots in a self - assembled sample , as a result of which strongly coupled pairs of dots are formed and contribute to the emission .
therefore , in order to fully understand the role of the inter - dot coupling it is interesting to study the role of the planar arrangement of the dots on the observed collective emission .
moreover , so far the collective emission from qd ensembles was discussed only under weak excitation conditions ( one excitation delocalized over the ensemble ) @xcite , where it is manifested only in the form of an enhanced emission rate . in order to observe actual superradiance ,
that is , a non - monotonic evolution of the emission with a delayed outburst of radiation @xcite , one has to study the photon emission under strong excitation conditions . in this paper
, we study the collective spontaneous emission from small ensembles of qds under strong excitation ( full inversion ) conditions for two kinds of spatial arrangements of the qds in the sample plane : regular arrays and randomly scattered dots .
we show that the dynamics of emission very weakly depends on the qd arrangement .
the paper is organized as follows . in sec .
[ sec : model ] , we describe the model of a small ensemble of qds in either random or regular planar arrangement . next , in sec . [
sec : results ] , we present and discuss the simulation results for the photon emission in these two cases .
finally , sec .
[ sec : concl ] concludes the paper .
the planar arrangement of the qds : ( a ) regular array ; ( b ) random ensemble ( sample dimensions in nanometers).,width=321 ] we consider an ensemble of a few ( up to six ) qds placed in the @xmath0 plane at positions @xmath1 , where @xmath2 numbers the dots .
each qd is modeled as a two - level system ( empty dot and one exciton ) with the fundamental transition energy @xmath3 , where @xmath4 is the average transition energy in the ensemble .
we consider two different arrangements of the qds in the sample plane : regular arrays of qds ( double dots , triangular , rectangular aligned structure of the dots , penta- and hexagonal structures , see fig .
[ fig : diagram](a ) ) and ensembles of randomly distributed qds , as schematically depicted in fig .
[ fig : diagram](b ) . in the latter case ,
the ensemble is randomly distributed with a fixed surface density over square mesas with the restriction that the inter - dot distance can not be smaller than 10 nm .
the spectral properties of the dots are modeled by a gaussian distribution of their transition energies @xmath5 with the standard deviation @xmath6 around the mean @xmath4 .
the dots are coupled by an interaction @xmath7 which can be either of dipole dipole character ( long - range _ dispersion force _ ) or short - range ( exponentially decaying with the distance ) .
we introduce the transition operators for the dots : the `` exciton annihilation '' operators @xmath8 which annihilate an exciton in the dot @xmath2 , and the `` exciton creation '' operators @xmath9 which creates an exciton in the dot @xmath2 .
the exciton number operator for the dot @xmath2 is then @xmath10 . using these operators ,
the hamiltonian of the many dots system is written in the frame rotating with the frequency @xmath11 in the form @xmath12 the long - range dipole coupling is described by @xmath13 where @xmath14 , @xmath15 is the spontaneous emission ( radiative recombination ) rate for a single dot , @xmath16 is the vacuum permittivity , @xmath17 is the relative dielectric constant of the semiconductor , @xmath18 , where @xmath19 is the speed of light and @xmath20 is the refractive index of the semiconductor , and @xmath21,\end{aligned}\ ] ] where @xmath22 and @xmath23 , where @xmath24 is the interband matrix element of the dipole moment operator which is assumed identical for both dots . for a heavy hole exciton , @xmath25^{t}$ ] , so that for a vector @xmath26 in the @xmath0 plane one has @xmath27 .
the short - range coupling is described by @xmath28 the effect of the coupling to the radiation field is accounted for by including the dissipative term in the evolution equations , which describes radiative recombination of excitons .
the equation of evolution of the density matrix is then given by @xcite @xmath29 + \sum_{\alpha,\beta=1}^{n}\gamma_{\alpha\beta}\left [ \sigma_{\alpha}\rho\sigma_{\beta}^{\dag } -\frac{1}{2}\left\ { \sigma_{\alpha}^{\dag}\sigma_{\beta},\rho \right\}_{+ } \right],\ ] ] where @xmath30 , with @xmath31,\end{aligned}\ ] ] and @xmath32 denotes the anticommutator .
the diagonal decay rates @xmath33 describe the emission properties from a single dot , while the off - diagonal terms @xmath34 , @xmath35 , account for the interference of emission amplitudes resulting from the interaction with a common reservoir and are responsible for the collective effects in the emission .
the values of the two couplings as well as the interference term of the decay rate @xmath34 are plotted as a function of the distance between the dots in fig .
[ fig : gv ] . in our simulations , we use the parameters for a typical cdse / znse qd system : @xmath36 ns@xmath37 , @xmath38 , the average transition energy of the qd ensemble @xmath39 ev and the qd surface density @xmath40 @xmath41 . for the tunnel coupling
we choose the amplitude @xmath42 mev and the range @xmath43 nm . the interference term of the decay rate @xmath44 and the short- and long - range coupling amplitudes @xmath45 as a function of the inter - dot distance . in ( a ) , the small distance section is shown , while in ( b ) the oscillating tail at larger distances is visible.,width=321 ]
in fig . [ fig : g14 ] we show the results of numerical simulation based on eq . [ evol ] for regular qd arrays . at the initial time , the system is strongly excited to the fully inverted state @xmath46 = @xmath47 and we subsequently determine the photon emission rate as a function of time . on each plot ,
the photon emission in the system is shown as a function of time for identical dots ( @xmath48 ) and for non - identical dots with the realistic value of the fundamental transition energy standard deviation @xmath49 mev the evolution of photon emission for the fully inverted initial state in regular arrays of a few qds.,width=321 ] as can be seen in figs .
[ fig : g14](a ) and ( b ) , the spontaneous emission from a regular array of identical qds develops a maximum , which is a few - emitter counterpart of the delayed outburst of photon emission observed in macroscopic samples @xcite .
this effect appears already for three qds and becomes more and more pronounced as the number of dots increases .
it implies that the true superradiance effect appears in the regularly arranged small ensemble of identical qds with either long- or short - range couplings .
however , if the array is formed by non - identical dots with a realistic inhomogeneity of the fundamental transition energies ( fig .
[ fig : g14](c ) and ( d ) ) , when the value of @xmath6 is greater than both the coupling strengths and the inverse exciton life time ( the emission line width ) , then the maximum is completely washed out . thus , under conditions that led to enhanced emission in the weak excitation regime @xcite , full superradiance effects under strong excitation are not observed . the evolution of photon emission for the fully inverted initial state in an ensemble of a few qds with a random spatial distribution.,width=321 ] in order to find out whether a randomly distributed ensemble of qds behaves similar to a regular array of qds we now study the photon emission from a random qd ensemble in the same strong excitation regime . in each case
, we calculate an average of 100 realizations of the evolution in systems with a given number of dots but differing in their positions within the sample plane . figs . [
fig : g17](a ) and ( b ) show the simulated photon emission for an ensemble of identical dots .
again , a superradiant maximum builds up as the number of qds grows .
similarly to the case of regular qd arrays , the maximum disappears and the decay of luminescence becomes monotonic when the realistic distribution of fundamental transition energies is taken into account ( figs .
[ fig : g17](c ) and ( d ) ) . by comparing fig .
[ fig : g14 ] with fig .
[ fig : g17 ] one can see that there is very little difference in arranging the dots either randomly or regularly under strong excitation conditions .
regular ordering of qds only slightly enhances the peak emission in the hypothetical case of identical dots and slightly suppresses the emission in the realistic case of an inhomogeneously broadened ensemble .
we have shown that the delayed outburst of radiation , typical for the superradiant emission develops in the luminescence from regular arrays or randomly distributed ensembles of quantum dots in the strong excitation regime .
the way the dots are distributed in the sample plane ( regular vs. random ) makes very little difference on the photon emission .
this means that the system response is not dominated by accidental clustering that might appear in the random distribution case and lead to strongly enhanced contribution to the overall emission from pairs of accidentally very closely spaced dots .
moreover , in both cases , the superradiant maximum is completely washed out if the realistic degree of inhomogeneity of the fundamental transition energies is taken into account .
thus , enhanced emission observed experimentally under weak excitation does not imply that true superradiance will be manifested for a fully inverted system .
|
we study the spontaneous emission from a regular lateral array or a randomly distributed ensemble of quantum dots under strong excitation ( full inversion ) conditions .
we focus on the similarities and differences between the cases of random and regular arrangement of the dots and show that there is very little difference between the evolution of luminescence in these two cases , both for identical dots and for a realistically inhomogeneously broadened ensemble .
this means that the enhanced emission or superradiance effect is not due to accidental clustering of pairs of dots .
moreover , we point out that observation of an enhanced emission under weak excitation does not prove that true superradiance will develop in a fully inverted system .
|
our theoretical knowledge of the universe is based on the standard cosmological model , that provides a convenient framework to satisfactorily explain the majority of cosmological observations , like the anisotropy pattern of the cosmic microwave background ( cmb ) @xcite , the large - scale structure of the universe @xcite , the hubble diagram of distant type ia supernovae @xcite and the abundances of light elements @xcite .
the standard cosmological model relies on the assumption that the universe , at least at large scales , is highly homogeneous and isotropic , and its geometry is thus described by the robertson - walker metric .
in fact , the distribution of luminous red galaxies shows that the present universe is homogeneous on scales greater than @xmath4 mpc @xcite , while the isotropy of the cmb itself ( which has a black - body distribution at @xmath5k with temperature fluctuations of order @xmath6 or less ) is an indication of the isotropy of the universe as a whole and a strong evidence for homogeneity at the time of hydrogen recombination ( nearly 400.000 years after the big bang , corresponding to a cosmological redshift @xmath7 ) . on the other hand , below the `` homogeneity scale '' of 100 mpc ,
the distribution of matter is definitely inhomogeneous .
such a dichotomy between the smoothness in the matter - energy distribution at @xmath8 and the clumpiness of the recent universe ( for @xmath9 ) below a certain scale is explained by the mechanism of gravitational instability : the structures we observe today have been formed through the growth of tiny density perturbation seeds that , accordingly to the currently accepted model , were created in the early universe during a phase of inflationary expansion .
the presence of a large scale ( _ i.e. _ , coherent over a hubble length ) , strong magnetic field is forbidden by the observed isotropy , as it would naturally single out a preferred spatial direction .
however , a background and uniform magnetic field could be present at cosmological scales provided that its intensity is small enough .
in particular , upper limits on the present field intensity of order @xmath10 g have been derived from observations of the cmb temperature anisotropies @xcite and of its temperature - polarization correlation @xcite .
smaller scale fields , on the other hand , could be as strong as @xmath11 g. the effects of large - scale magnetic fields on the evolution of cosmological structures have been studied extensively in the literature ( for a complete review , see ref . @xcite and references therein ) , where both newtonian and general relativistic treatments , the latter often using covariant and gauge - invariant techniques , are present .
it is known that , among others , magnetic fields slow down ( and possibly prevent ) the growth of perturbations and can produce vorticities and shape distortions in the density field @xcite . in this letter , our goal is to revisit the issue of the existence of a `` magnetic jeans length '' and of its dependence on the direction along which the perturbation propagates , other than to give a realistic estimate of its value in a matter - dominated universe .
the presence of a magnetic jeans length has been discussed in several papers @xcite both in a newtonian and in general relativistic framework ( for an analysis of the standard jeans mechanism in the presence of dissipative effect see also @xcite ) , but some of the analyses failed to recognize its angular dependence . here , we present a neat derivation of the relevant instability scales bases on the equations of magnetohydrodynamics ( mhd ) on an expanding universe . we discuss a simple generalization of the magnetic jeans length suited for two - fluid systems , and clarify some misunderstandings that are present in the literature .
we also show the results of the numerical integration of the coupled mhd and poisson equations .
finally , we numerically study the distortion introduced in the density field by the anisotropy in the critical length .
the letter is organized as follows . in sec .
2 , we characterize the universe as a plasma . in sec . 3
, we introduce the basic equations , and in sec .
4 we carry out the linearization procedure .
we derive the existence of the magnetic jeans length from analytical considerations in sec . 5 , while in sec .
6 we show some numerical results .
finally , we draw our conclusions in sec .
in this section , we aim at characterizing the plasma features of the cosmological fluid . between the time of @xmath12 annihilation ( _ i.e. _ , for a temperature . ]
@xmath13 , corresponding to a redshift @xmath14 ) and the present ( @xmath15 ) , the matter - energy content of the universe is provided by electrons , protons and neutrons ( the three species being collectively referred to as baryons in the cosmological jargon ) , photons , neutrinos , and two elusive components dubbed dark matter and dark energy , that presently account for more than 99% of the total energy budget of the universe . however , dark energy has been subdominant for most of the past history of the universe and can be safely neglected at redshifts @xmath16 .
dark matter and neutrinos interact only gravitationally with the other components and can be neglected as long as the plasma properties of the fluid are concerned .
the cosmological baryon - to - photon ratio is extremely small and equal to @xmath17 , where @xmath18 and @xmath19 are the photon and baryon number densities , respectively .
both @xmath19 and @xmath18 scale with redshift as @xmath20 ; their present values are @xmath21 and @xmath22 .
most of the baryons in the universe are in the form of @xmath23 nuclei ( i.e. , isolated protons ) so that , for simplicity , in the following we assume @xmath24 ( @xmath25 being the proton number density ) .
furthermore , @xmath25 is also equal to the electron number density @xmath26 , because of the charge neutrality of the universe .
we study separately the properties of the cosmological fluid before and after the time of hydrogen recombination occurring at @xmath27 ( @xmath28 ) . for @xmath29 ,
the protons and electrons are free and thus one deals with a fully ionized plasma . in this regime ,
photons and baryons are tightly coupled due to thomson scattering , and share a common temperature @xmath30 , where the present photon temperature @xmath31 .
after recombination ( @xmath32 ) , most of the electrons and protons exist in the form of neutral hydrogen atoms , and only a small residual ionized fraction @xmath33 survives , making the fluid a weakly ionized plasma . in this regime
, the photon temperature still scales as @xmath34 , while that of baryons evolves in the same way only until @xmath35 , due to residual scatterings that keep them in thermal equilibrium with photons ; after that time , their temperature decreases faster , as @xmath36 .
the baryonic fluid remains neutral until the time of reionization , when the uv radiation produced by the first stars ionizes again the hydrogen present in the cosmological medium .
this is likely to have happened around @xmath37 , however the precise details of the reionization history are still largely unknown , and for this reason we limit our analysis to redshifts @xmath38 . in the following , we will assume the presence of a background homogeneous magnetic field @xmath39 , whose contribution to the total energy density of the universe can be considered negligible .
we recall that the field intensity @xmath40 scales as @xmath41 and , unless otherwise stated , we take its value at the present time to be @xmath42 g. a fundamental quantity characterizing a plasma is the debye length @xmath43 , namely the length over which electrons screen out electric fields in a plasma .
it defines the length scale over which a system can consistently considered to be a plasma .
the debye length of a hydrogen plasma at temperature @xmath44 is @xmath45 where @xmath46 is the proton charge . using @xmath47 and the values given above
, one gets @xmath48 the redshift dependence of @xmath43 implies that the comoving debye length @xmath49 is constant during the cosmological evolution and equal to @xmath50 .
plasma effects can be important in a system when its physical dimension @xmath51 is much larger than the debye length .
for the universe , the relevant length is the hubble radius @xmath52 , where @xmath53 is the hubble parameter .
this length represents the maximum scale at which microphysical processes can operate in order to establish the thermodynamical equilibrium . today , @xmath54 ; during the matter - dominated era @xmath55 , while in the radiation - dominated era @xmath56 . from the analysis of both these scales ,
it is evident that @xmath57 turns out in the period considered here .
moreover , under the same hypotheses , the baryonic matter @xmath58 within a debye sphere is also constant and given by @xmath59 where @xmath60 is the proton mass .
this `` debye mass '' clearly results to be much smaller than any other of cosmological interest and we can conclude that the cosmological fluid can be considered as neutral at all relevant scales .
another meaningful index is the so - called plasma parameter @xmath61 , _
i.e. _ , the number of particles within a debye sphere : @xmath62 the dependence of @xmath43 and @xmath25 on the redshift @xmath63 implies that also @xmath61 is a constant . in particular , since @xmath64 , the cosmological fluid results to be a weakly coupled plasma . in order to provide a complete characterization of the cosmological plasma
, we now turn our attention to the plasma dissipative properties , starting from the plasma resistivity @xmath65 . for an electron - proton plasma , this is given by @xmath66 , where @xmath67 is the electron mass and @xmath68 is the electron - ion collision frequency . for the case under consideration , @xmath68 is well approximated by the electron - electron collision frequency @xmath69 @xcite , _ i.e. _ , @xmath70 where @xmath71 is the coulomb logarithm , introduced to quantify the effects that small - angle - diffusion collisions have in the coulomb scattering . a simply estimate of @xmath72 in a plasma
is given by @xmath73 , so that for the cosmological fluid , the coulomb logarithm is @xmath74 .
substituting eq.([eq : nu_ei ] ) into the expression for the resistivity given above , we get @xmath75 close to recombination , the cosmological plasma has an electric resistivity equal to @xmath76 , _
i.e. _ , a conductivity @xmath77 siemens @xmath78 , a value typical of a semiconductor [ in gaussian units , @xmath79 .
let us now turn our attention to the viscous properties of the plasma .
the shear viscosity coefficient of matter strongly coupled with radiation can be expressed as @xmath80 where @xmath81 is the stefan - boltzmann constant , while @xmath82 denotes the mean collision time between particles and can be estimated as @xmath83 ( here , @xmath84 and we have introduced @xmath85 as the cross section for thomson scattering ) . for a photon gas at equilibrium at temperature @xmath44 , @xmath86 and
then @xmath87 the resistivity and viscosity coefficients enter the mhd equations through the following diffusion coefficients @xmath88 and @xmath89 , where @xmath90 is the density of the fluid .
taking @xmath91 kg m@xmath92 and using @xmath93 , we get @xmath94 the relative magnitude of the viscous and magnetic diffusion rates can be parameterized through the magnetic prandtl number @xmath95 . using the expression above for @xmath96 and @xmath97
, we obtain @xmath98 so that @xmath99 , _
i.e. _ , viscous diffusion is more important than resistive diffusion , at recombination and indeed always at the redshifts under consideration .
having discussed the relative importance of viscosity- and resistivity - driven dissipative effects , we now analyze at which scales these effects are relevant . in a magnetized plasma ,
a useful parameter is the lundquist number @xmath100 , where @xmath51 is a typical length scale and @xmath101 is the alfvn velocity .
the lundquist number is basically the ratio between the resistive diffusion timescale @xmath102 and the alfvn crossing timescale @xmath103 .
assuming @xmath104 and @xmath105 , we obtain @xmath106^{1/2}$ ] and thus @xmath107 we also compare @xmath108 with the viscous diffusion timescale defined as @xmath109 . the ratio @xmath110 can be thought as a viscous analogous to the lundquist number : @xmath111 the baryonic mass ( at the average background density ) contained in a sphere of radius equal to the length where @xmath112 can be calculated to be @xmath113^{-3}$ ] , while the analogous quantity for @xmath114 is @xmath115^{-9/2}$ ] .
both mass scales are well below the values of cosmological relevance at the redshifts of interest .
the discussion above shows how the following hierarchy among the relevant time scales holds for all mass range and redshifts of interest : @xmath116 meaning that viscosity always dominates over resistivity , and that both dissipative effects can indeed be neglected when studying the propagation of alfvn waves . after the time of recombination @xmath117 , the cosmological plasma exists in a weakly ionized state , the neutral and ionized components having densities @xmath118 and @xmath119 , respectively .
we take the residual ionization fraction @xmath120 constant and equal to @xmath121 in the range @xmath122 .
the results of the previous subsection can be generalized to show that the hierarchy ( [ eq : hierdiss ] ) holds also in this regime . in spite of the small value of the ionization fraction
, the magnetic field could still affect the dynamics of the whole system in view of the interactions between neutral and charged particles .
in particular , the magnetic forces acting on the charged particles can be communicated to the neutrals through collisions . however , if the coupling is not tight enough , the neutrals feel the magnetic field but drift with respect to the ions in a process termed _
ambipolar diffusion_. its relevance at a given length scale @xmath51 is quantified by the ambipolar reynolds number @xmath123 @xcite @xmath124 where @xmath125 @xcite is the ion - neutral drag coefficient due to collisions between the two species , @xmath126 is the alfvn velocity in the tightly coupled limit , @xmath127 is the characteristic velocity of the fluid , and @xmath128 is the ambipolar length , i.e. , the scale where @xmath129 . the ambipolar reynolds number is just the ratio between the ambipolar diffusion timescale @xmath130 ( where @xmath131 is the neutral collision timescale ) and some characteristic timescale @xmath132 .
if @xmath133 , the neutrals are uncoupled from the plasma .
on the contrary , when @xmath134 the dynamics of the two components can be described through ordinary single - fluid mhd with an additional dissipative term @xcite .
this term becomes progressively less important as @xmath123 grows and can be neglected in the limit @xmath135 , or @xmath136 .
assuming that the evolution of the fluid is driven by alfvnic phenomena , i.e. , @xmath137 , the ambipolar length is given by @xmath138 and @xmath139 .
thus , @xmath135 if and only if the tight - coupling condition @xmath140 is satisfied . defined in eq.([eq : lamb ] ) as a function of redshift @xmath63 . above line , the condition @xmath141 holds.[fig : lamb ] ] it is straightforward to check that for @xmath142 , it is always @xmath143 at scales larger than a few tens of comoving kiloparsecs , meaning that the following hierarchy holds : @xmath144 so that the ions and neutrals are tightly coupled and ambipolar diffusion can be safely neglected . in order to better illustrate this point , in figure [ fig : lamb ] we plot the mass contained in a sphere of radius @xmath128 at the background baryon density as a function of redshift .
it is evident that , in the redshift range considered , @xmath135 for all scales @xmath145 . in this regime
, we can therefore neglect the dissipative term mentioned above and use single - fluid ideal mhd .
we conclude that ambipolar diffusion does not affect the dynamics of the cosmological plasma after recombination .
in this section , we derive the basic equations describing the linear evolution of instabilities in the cosmological fluid , modeled as a magnetized plasma . in this respect , we underline that the full investigation of the perturbative dynamics of the universe would require a general - relativistic treatment , in order to correlate the matter and geometrical fluctuations . however , as long as one is interested in scales much smaller than the hubble radius , _
i.e. _ , @xmath146 , a newtonian treatment provides a consistent description of the dynamics .
nonetheless , in this scenario the expansion of the universe can be accounted as the bulk background motion of the fluid .
the starting point of our treatment is the eulerian set of equations governing the fluid motion , on which one can develop a perturbative theory by adding small fluctuations to the unperturbed cosmological background solution .
the zeroth - order dynamics is derived by considering a flat homogeneous and isotropic universe whose energy density is dominated by non - relativistic matter , and correctly describes the expansion of the universe .
we assume that a background magnetic field is present , whose contribution to the total energy density of the universe can be considered negligible .
let us now start by briefly recalling the basic equations of non - relativistic , ideal and single fluid mhd , which govern the plasma motion .
the mass conservation and the newtonian gravitational field are described by the continuity and poisson equations , the single - fluid dynamics is described by the euler equation in presence of a magnetic field @xmath147 and , finally , the electromagnetic interaction can be summarized by the frozen - in and the gauss laws .
such equations read [ initial - system ] @xmath148 respectively , where @xmath90 is the mass density , @xmath149 is the velocity field , @xmath150 is the gravitational potential and @xmath151 is newton constant .
this system constitutes the base of our perturbative approach . to derive the zeroth - order dynamics
, we assume the usual robertson - walker metric , _
i.e. _ , @xmath152 , where @xmath153 represents the cosmological scale factor , and a perfect fluid energy - momentum tensor as the matter source of the gravitational field , _ i.e. _ , @xmath154 $ ] , with @xmath155 . in this scheme
, the behavior of the mass density with time is obtained from the energy - momentum conservation law @xmath156 and from the friedmann equation , _
i.e. _ , @xmath157 respectively ( the dot ( @xmath158 ) denotes the total derivative with respect to synchronous time ) . here
@xmath159 is the hubble parameter and @xmath160 is the curvature factor . setting the matter - dominated universe equation of state ( eos ) @xmath161 ( @xmath162 ) in eq.([emt ] )
, the zeroth - order solution of the system ( [ initial - system ] ) turns out to be @xmath163 where @xmath164 and @xmath165 are dimensional constants , @xmath166 ( @xmath167 ) denotes the radial coordinate vector and , of course , @xmath168 satisfies eq.([friedmann ] ) .
we observe how this non - stationary solution characterizing the background dynamics is not affected by the so - called `` jeans swindle '' proper of the static solution .
to obtain now the explicit time dependence of the unperturbed quantities involved in the model , we restrict the analysis to the flat case , _
i.e. _ , @xmath169 . from the friedmann equation ( [ friedmann ] ) and
using the solution for @xmath170 , one readily obtains [ time - parameters ] @xmath171 finally , we recall that the adiabatic sound speed is defined by @xmath172 . for a general specific heat ratio @xmath173
, we assume that the pressure varies as @xmath174 , so that the speed of sound is given by @xmath175
in order to analyze the implications that the physics of an ideal magnetized plasma can have on the structure formation , we will follow the standard perturbation approach . in this respect ,
we consider small perturbations around the zeroth - order cosmological solution derived above , _
i.e. _ , we write @xmath176 ( with @xmath177 ) and similarly for the other quantities @xmath178 , @xmath149 , @xmath150 and @xmath147
. substituting the perturbed quantities in eqs.([initial - system ] ) and keeping only terms up to first order , one gets [ perturbation - system ] @xmath179 where , as already discussed , the pressure and density perturbations have been related through the adiabatic sound speed , _
i.e. _ , @xmath180 .
we are assuming that @xmath181 , where @xmath182 , in order to preserve the isotropy of the background flow . in the following , we replace @xmath183 with the dimensionless magnetic fluctuation @xmath184 .
moreover , the analysis of the system above can be simplified by fourier - transforming the spatial dependence of the involved quantities , _
i.e. _ , using perturbations in the form of plane waves , taking @xmath185 with @xmath186 and @xmath187 is the physical wavenumber scaling as @xmath188 .
it is convenient to consider also the comoving wavenumber @xmath189 , that stays constant during the expansion .
the evolution for a given harmonic can be obtained by the equations in real space with the substitutions @xmath190 , @xmath191 and @xmath192 . in the following , for the sake of simplicity
, we will drop the tilde over the fourier transformed variables . then
, the system ( [ perturbation - system ] ) reduces to ( hats denote unit vectors ) : @xmath193 \dot{{\boldsymbol{v}}}_1+h{{\boldsymbol{v}}_1}+i\left[\frac{v_s^2}{{\rho_0}}-\frac{4\pi g } { k^2}\right]{\rho_1}{{\boldsymbol{k}}}+iv_a^2\,{\hat{{\boldsymbol{b}}}_0}\times({{\boldsymbol{k}}}\times{{\boldsymbol{b}}_1})&=0\;,\\[0.2 cm ] \dot{{\boldsymbol{b}}}_1 + i{\hat{{\boldsymbol{b}}}_0}({{\boldsymbol{k}}}\cdot{{\boldsymbol{v}}_1})-i({\hat{{\boldsymbol{b}}}_0}\cdot{{\boldsymbol{k}}}){{\boldsymbol{v}}_1}&=0\;,\end{aligned}\ ] ] where we have already eliminated @xmath194 by means of the poisson equation in @xmath195space , _ i.e. _ , @xmath196 .
it is understood that the constraint @xmath197 always hold .
decomposing now @xmath198 in its components @xmath199 and @xmath200 parallel and orthogonal to the direction of @xmath201 respectively , _
i.e. _ , @xmath202 ( where @xmath203 ) , and introducing the following scalar variables : @xmath204 we finally get a further simplified system : [ master - system ] @xmath205 where we have defined @xmath206 and , of course , @xmath207 .
we stress that @xmath208 is _ not _ positive definite .
the form ( [ master - system ] ) of the evolution equations has the advantage that it clearly expresses the relationship between the physical quantities involved , other that being very well suited for numerical integration .
some further analytical insight can however be gained by reducing it to an unique higher - order equation for the variable @xmath209 . considering the case of a matter - dominated universe , and using the explicit time dependence of the quantities involved in the model in that case , _
i.e. _ , eqs.([time - parameters ] ) , with some algebra one can derive the following fourth - order differential equation for @xmath209 : @xmath210\,t^2\,\delta^{(2)}+\nonumber\\ & \quad+\left[(12{\lambda_2}+8)+12{\lambda_1}(1 - 3\nu)\,t^{-2\nu}\right]\,t\,\delta^{(1)}+\\ & \qquad+\left[-6{\lambda_2}\mu^2 + 3{\lambda_1}(3{\lambda_2}\mu^2 + 12\nu^2 - 2\nu)\,t^{-2\nu}\right]\,\delta=0\;,\nonumber\end{aligned}\ ] ] where @xmath211 denotes the @xmath212 derivative of @xmath213 with respect to time , and we have defined the following constants : @xmath214 we recall that @xmath173 is the specific - heat ratio ( @xmath215 ) and that @xmath216 , _
i.e. _ , @xmath217 . the most general solution of eq.([master - delta ] ) for @xmath213 is found to be the superposition of four independent solutions @xmath218 @xmath219 , given by : @xmath220\;,\end{aligned}\ ] ] where @xmath221 $ ] denotes the generalized hypergeometric function of argument @xmath63 , the @xmath222 s are arbitrary integration constants and @xmath223 the constant coefficients @xmath224 and @xmath225 depend , in general , on @xmath226 , @xmath227 , @xmath228 and we report their complete expressions in [ hyp_ab ] .
we are now interested in discussing the asymptotic behavior of the hypergeometric functions in the limit of very small or very large argument , _
i.e. _ , @xmath229@xmath230@xmath231 or @xmath232@xmath231 . as in the non - magnetic case discussed in ref . , we restrict the analysis to the range @xmath233 , _
i.e. _ , treating the standard regime @xmath234 . from the asymptotic expansion of the @xmath235 functions in the case of large argument , _
i.e. _ , @xmath236 , the density contrast always shows a damped oscillating behavior with time .
in fact , in this regime it always exists at least one asymptotic solution proportional to positive power of the argument @xmath229 , which results to be the leading term of the solution superposition . in this case
, @xmath213 decreases with time . on the other hand , in the limit @xmath237 ,
the asymptotic expansion of the solutions ( [ delta_sol ] ) , can be written as @xmath238 in order for the gravitational collapse to occur , at least one of the modes has to be growing , _
i.e. _ , @xmath239 .
it is fairly easy to show that @xmath240 , @xmath241 and @xmath242 are always negative , whereas the sign of @xmath243 depends on @xmath228 and @xmath227 . in particular , when @xmath244 , we obtain @xmath245 irregardless of the value of @xmath227 , while when @xmath246 , @xmath243 is positive if @xmath247 .
this means that , on the plane orthogonal to the magnetic field ( @xmath246 ) , a new stability condition arise if the magnetic field is strong enough . the threshold value related to @xmath248 that should discriminates the two regimes of growing and decreasing density contrast can be set as @xmath249 . remembering that @xmath250 , such condition rewrites in terms of the wave number as @xmath251 which is substantially the same as the usual jeans condition for gravitational instability .
in fact , in the non - magnetic case , ( to which our analysis reduces for @xmath252 ) this is the only criterion that separates the growing and the decaying modes . in a similar way , the new threshold @xmath253 yields to the following condition @xmath254 summarizing , we find that the presence of a background magnetic field introduces an anisotropy in the stability criterion . while outside the plane orthogonal to @xmath255 , the stability of the perturbations is dictated only by the standard jeans condition @xmath256 , on that plane the unstable modes are those for which the conditions @xmath257 and @xmath258 both hold is much shorter than both the acoustic and alfvn timescales @xmath259 and @xmath260 . ] . in other words , if @xmath261 ( basically equivalent to @xmath262 ) , there are jeans - unstable modes ( those in the window @xmath263 ) that , in the orthogonal plane , are stabilized by the magnetic pressure .
the window of stable modes gets wider for larger values of the ambient magnetic field , as expected .
we underline that these results are qualitatively the same as those obtained for a static and uniform background @xcite .
a similar analysis was carried on by the authors of ref .
@xcite obtaining a similar results .
however , their derivation contained a mistake when separating the real and imaginary components of the evolution equations .
for this reason they find a second - order differential equation instead than the fourth - order one discussed here .
in the previous section , we have gained an important insight on the effect of a background magnetic field on the evolution of density perturbations .
we now show some results obtained through the direct numerical integration of the differential system ( [ master - system ] ) .
g. the discontinuity at @xmath264 corresponds to the recombination on neutral hydrogen .
bottom panel : magnetic ( red solid line ) and standard ( blue dashed line ) jeans mass as a function of redshift .
the mass contained inside the hubble radius ( black dotted line ) is also shown for comparison .
[ fig : soundspeed],title="fig : " ] g. the discontinuity at @xmath264 corresponds to the recombination on neutral hydrogen .
bottom panel : magnetic ( red solid line ) and standard ( blue dashed line ) jeans mass as a function of redshift .
the mass contained inside the hubble radius ( black dotted line ) is also shown for comparison .
[ fig : soundspeed],title="fig : " ] we will focus on the period of the cosmological evolution that goes from the onset of matter domination ( @xmath265 ) to the time of reionization ( @xmath37 ) .
we start from matter domination because , before that time , the growth of density perturbation was slowed down and practically frozen by the rapid expansion of the universe . in the matter - dominated universe , @xmath266 and @xmath267 .
we can ignore the presence of a dark energy component since this is sub - dominant until very recent times .
the time period that we consider can be divided into two distinct phases , _
i.e. _ , before and after the recombination of hydrogen occurring at @xmath268 .
before recombination , the baryons are completely ionized and they are tightly coupled to photons , at least at scales larger than the comoving photon mean free path @xmath269^{-2}$ ] .
at these scales , the total pressure of the fluid is given by radiation pressure .
after recombination , most of the protons and electrons are in the form of neutral hydrogen atoms , leaving a small ionized fraction @xmath270 . at the scales of interest , the neutral and ionized components are tightly coupled by collisions ( see sec .
[ ssec : post_uni ] ) and can be treated as a single fluid .
however , photons are now free streaming so that the baryon pressure is given just by kinetic pressure , dropping down by several orders of magnitude with respect to its pre - recombination value . in view of this , we take the speed of sound of the cosmological fluid before and after recombination to be @xmath271 respectively , where @xmath272 is the specific entropy .
we recall that @xmath273 for @xmath274 , while afterwards @xmath275 .
the expression above for the sound speed is rigorously valid only for scales corresponding to baryonic masses @xmath276 , that stay above the photon mean free path until the time of recombination . at smaller scales , baryons lose the radiation support before recombination , roughly when the given scale goes below the photon mean free path , and it is at this time that the switch between the two expressions in eq.([vsvs ] ) should take place . in the following
, we shall consider masses nearly as small as @xmath277 , for which the photon decoupling effectively takes place at @xmath265 ( very close to the time of matter - radiation equality ) .
however , we choose to switch between the two expressions for the sound speed at @xmath8 irregardless of scale and shall comment later on how we expect this choice to affect our results . following the discussion in sec .
[ ssec : post_uni ] , the alfvn velocity is taken to be @xmath278 where @xmath279 should always be intended as the total baryon density , both before and after recombination .
a plot of the alfvn and sound speeds as functions of redshift is shown in the upper panel of figure [ fig : soundspeed ] .
the sudden drop in the sound speed at recombination is due to the sharp decrease of baryon pressure after photon decoupling . in a detailed model
, the presence of different uncoupled components making up the matter content of the universe should be taken into account .
in fact , most of the matter ( @xmath280 ) is in the form of cold dark matter ( cdm ) , interacting with the baryon - photon fluid only through the gravitational force .
thus , a proper treatment should rely on a two - fluid description . in the following
, we shall ignore perturbations in the cdm component , however we argue that we can still draw meaningful conclusions about the perturbations in the baryonic component
. in fact , in the pre - recombination era , the large radiation pressure prevents baryons to fall into the potential wells created by cdm ; this is known to be true in the non - magnetic case but we expect it to hold also in the case under consideration since , as seen in the last section , the magnetic field only acts to increase stability . thus in this regime
the cdm and baryon perturbations are effectively decoupled .
after recombination , the baryon density perturbations will take some time to catch up with those in the cdm component and we expect our treatment will rigorously remain valid for some time . before showing the results of the numerical integration , we illustrate in the lower panel figure [ fig : soundspeed ] the evolution of the standard jeans wavenumber ( [ eq : jeans ] ) and of its `` magnetic '' counterpart ( [ eq : jeansm ] ) .
in order to take out the change in @xmath281 and @xmath282 due to the expansion , we follow the convention to express the results in terms of the mass contained inside the corresponding length scales @xmath283 . in particular , we consider the total baryonic mass ( irrespective of the ionization state ) , contained inside a sphere of radius @xmath284 . it can be seen that the window of modes that are made stable by the magnetic field , _ i.e. _ , those between the red dashed and the black solid line , spans , right after recombination , five orders of magnitude in mass . as noted in the introduction
, the existence of a magnetic jeans length has been studied previously and all expressions for the critical wavenumber agree , apart from numerical factors , with expression ( [ eq : jeansm ] ) .
however , the numerical estimates of this and associated quantities that are found in the literature sometimes differ from our results .
the reason seems to be that often the density @xmath285 that appears in eq.([eq : jeansm ] ) is taken to be the present critical density @xmath286 kg m@xmath92 .
this yields at the present time a magnetic jeans length @xmath287 kpc for @xmath288 g and @xmath289 mpc for @xmath290 g .. the reason why a galaxy is much smaller than 1 mpc is that it has detached from the hubble flow and undergone non - linear evolution , so that its density is much larger than the cosmological average .
] using instead the baryon density for @xmath285 will yield values of @xmath291 a factor @xmath292 larger , _
i.e. _ , @xmath293 mpc for @xmath294 g. this amounts to a factor @xmath295 difference in the corresponding mass scale at the numerator of eq.([eq : jeansm ] ) , while keeping the same expression for @xmath296 .
this makes the value of @xmath291 roughly twice smaller than in the case in which only baryons are considered , _
i.e. _ , @xmath297 mpc for @xmath294 g. ] .
we now discuss the results of the direct numerical integration of eqs.([master - system ] ) .
the initial conditions for the integration have been chosen using the fact that power - law solutions for @xmath213 can be found in the limit @xmath298 .
there are four distinct solutions of this kind , but only one corresponds to a growing mode .
we have matched the initial conditions to the asymptotic growing solution at the initial time of integration .
the latter has been chosen so that all the modes of interest were outside the horizon at that time .
even if the initial time falls in what would be the radiation - dominated era , nevertheless we always consider a matter - dominated universe .
all results have been normalized to the initial value of the density contrast . in figure
[ fig : deltak ] , we show the evolution of the density contrast for three different wavenumbers @xmath299 ( normalized at the present time ) , _ i.e. _ , for the following baryonic masses @xmath300 .
these masses roughly correspond to the scale of a dwarf galaxy , of a galaxy and of a galaxy cluster respectively . for each mode
, we show the evolution in both the direction parallel to the background magnetic field ( @xmath301 ) and orthogonally to that direction ( @xmath246 ) . in all cases ,
the perturbation is initially growing but then starts to oscillate once the jeans mass ( that is growing ) becomes larger than the mass of the perturbation .
this happens earlier for smaller scales . in this phase
the magnetic pressure does not play any role , as the much larger radiation pressure is actually providing the force that prevents the collapse .
in fact , there is no difference in the evolution parallel and orthogonal to the field , as the radiation pressure is isotropic .
the situation changes dramatically after recombination , when the baryon pressure drops and only the magnetic pressure can possibly oppose the growth , at least in the plane orthogonal to the field .
thus , perturbations in the direction of the field can grow unhindered , while the perturbations that are orthogonal can be stabilized . as it can be seen from figure
[ fig : deltak ] , this is what happens for perturbations at the dwarf galaxy scale : at @xmath2 , the relative growth of parallel perturbations with respect to orthogonal ones is of order 100 . for perturbations at the galactic scale and larger , instead
, the evolution is basically the same in all directions .
this can be understood by looking at the lower panel of figure ( [ fig : soundspeed ] ) , from which it is clear that the pressure induced by a magnetic field of @xmath3 g can only stabilize perturbations with mass @xmath302 .
we recall that we have neglected the fact that , at the dwarf galaxy scale , the support of radiation pressure is not lost at recombination but some time before ( see previous subsection ) . from the discussion above ,
it is clear that , had we taken into account this fact , the evolution of orthogonal and parallel perturbations would have begun to differentiate earlier .
this goes in the direction of enhancing the anisotropic growth of perturbation and the `` squeezing '' effect studied in the following .
the fact that , after recombination , the evolution of the density contrast in the presence of a magnetic field changes for different directions leads to the reasonable expectation that some degree of anisotropy will be generated even in initially symmetric structures . in order to show this
, we consider a gaussian density fluctuation with standard deviation @xmath303 in coordinate space at recombination : @xmath304 where the @xmath305 are comoving coordinates centered at the maximum of the perturbation .
after fourier - transforming , we separately evolve the different harmonics in momentum space using eqs.([master - system ] ) and we finally transform back to obtain the perturbation in coordinate space at a later time . in figure
[ fig : deltax ] we show , for a background magnetic field directed along the @xmath306 axis and with a present intensity of 1 ng , the evolution of a perturbation with @xmath307 kpc at recombination ( so that the @xmath308 region encloses a mass @xmath309 at the mean baryonic density , _
i.e. _ , roughly the mass of a dwarf galaxy ) .
in particular , we show equal density contours at @xmath310 and @xmath311 .
it is evident from the figure how the perturbation becomes progressively squeezed along the direction orthogonal to the magnetic field . in order to quantify the anisotropy in the perturbation
, we consider the isodensity contour corresponding to half the value at the peak and calculate its eccentricity @xmath312 , where @xmath224 and @xmath225 are the lengths of the semi - major and semi - minor axis of the contour , respectively . in figure
[ fig : ecc ] , we show how the eccentricity changes with redshift ; for the parameters used above , we get @xmath313 at @xmath2 .
we have studied the effect of a background magnetic field on the linear evolution of cosmological density perturbations at scales well below the hubble length , where a newtonian treatment can be used , focusing on the matter - dominated era .
the conditions that allow for the growth of small density perturbations have been clearly stated .
in particular , we have found that in the plane orthogonal to the ambient magnetic field , a new critical length appears , related to the presence of the magnetic pressure , while everywhere else outside that plane the stability is dictated by the standard jeans criterion .
this is also confirmed through a direct numerical integration of the relevant mhd equations during the matter - dominated era , and this effect is shown to be possibly important after recombination , when the magnetic pressure of baryons is much larger than their the kinetic pressure .
finally , it has been shown how the dependence of the critical scale on the angle between the perturbation wavevector and the magnetic field could lead to a sizable anisotropy in the perturbations at sub - galactic scales at the onset of non - linearity .
our analysis has relied on some approximations : in particular , we have ignored the gravitational effects of dark matter perturbations .
we have argued that this approximation limit the validity of our treatment to some time after recombination .
we defer a more detailed and fully general relativistic analysis , also taking into account the different fluid components , to a future work .
* acknowledgment * : n.c . gratefully acknowledges the _ cpt - universit de la mediterrane aix - marseille 2 _ and the financial support from `` sapienza '' university of roma .
m.l . acknowledges support from a joint _ accademia dei lincei / royal society _ fellowship for astronomy .
the work of m.l . has been supported by _ miur - ministero dellistruzione , delluniversit e della ricerca _ through the prin grants `` matter - antimatter asymmetry , dark matter and dark energy in the lhc era '' ( contract number prin 2008nr3ebk-005 ) and `` galactic and extragalactic polarized microwave emission '' ( contract number prin 2009xz54h2 - 002 ) .
in following , we write the complete form of the coefficients of the hypergeometric function of eq.([delta_sol ] ) .
they read :
|
we study how the presence of a background magnetic field , of intensity compatible with current observation constraints , affects the linear evolution of cosmological density perturbations at scales below the hubble radius .
the magnetic field provides an additional pressure that can prevent the growth of a given perturbation ; however , the magnetic pressure is confined only to the plane orthogonal the field . as a result ,
the `` jeans length '' of the system not only depends on the wavelength of the fluctuation but also on its direction , and the perturbative evolution is anisotropic .
we derive this result analytically and back it up with direct numerical integration of the relevant ideal magnetohydrodynamics equations during the matter - dominated era . before recombination
, the kinetic pressure dominates and the perturbations evolve in the standard way , whereas after that time magnetic pressure dominates and we observe the anisotropic evolution .
we quantify this effect by estimating the eccentricity @xmath0 of a gaussian perturbation in the coordinate space that was spherically symmetric at recombination . for a perturbations at the sub - galactic scale
, we find that @xmath1 at @xmath2 taking the background magnetic field of order @xmath3 gauss .
cosmological perturbations , cosmological magnetic fields , structure formation , magnetohydrodynamics .
|
driven transport and coherent control at the nanoscale are well established areas of research .
quantum ratchets,@xcite molecular charge,@xcite spin@xcite and heat pumps,@xcite and nano - plasmonics@xcite are just several examples of areas of recent developments .
advances in optical techniques , in particular near - field optical microscopy , allow single molecule manipulation@xcite and induction of bond specific chemistry.@xcite combined with molecular junction fabrication techniques,@xcite optical spectroscopy methods are becoming an important observation and diagnostic tool in molecular electronics.@xcite experimental developments led to surge of theoretical activity in the field of optically assisted transport@xcite and optical response of molecular junctions.@xcite in particular , ref .
considered molecular junctions composed of molecules with strong charge - transfer transition into their excited state@xcite as a possible constituent for light - induced molecular charge pump , when change of molecular dipole occurs along the junction axis .
consideration was done within a two - level ( homo - lumo ) model with ground and excited ( homo and lumo ) states of the molecule strongly coupled to different contacts . in junction
setup optical excitation brings electron from occupied ground to empty excited state , and asymmetry in coupling to contacts assures appearance of current .
the model was treated within non - equilibrium green function approach , and perturbation theory in coupling to laser field was employed .
later ref .
generalized the consideration of ref .
to strong laser fields .
pumping optical field was treated as a classical driving force , and closed set of eoms for observables ( electronic populations and coherences of the levels and single time exciton correlation function ) was formulated .
one of the most important advances in ref . was consideration of chirped laser pulses , which allowed formulation of charge transfer between ground and excited states in terms of landau - zener problem
. chirped laser pulses enable to produce complete population inversion in molecular systems ( a molecular bridge ) where the well - known @xmath0-pulse excitation@xcite fails . in realistic molecular junctions optical field driving the molecule
is a _ local field _ formed by both incident radiation and scattered response of the system ( mostly plasmonic response of metallic contacts ) .
another feature of molecular junctions is hybridization of states of a molecule with those of contacts .
the latter leads to non - markov effects in response of the junction . in this paper
we generalize studies reported in ref . incorporating the aforementioned effects into consideration .
dynamics of local electromagnetic fields is simulated within the fdtd technique for realistic geometry of a molecular junction similar to our previous publication.@xcite non - markov effects of junction response are introduced within non - equilibrium green functions equation - of - motion ( negf - eom ) approach .
structure of the paper is the following .
after introducing the model in section [ model ] , we describe a junction geometry and numerical approach used in calculations of local electromagnetic fields in section [ optics ] .
section [ cur ] discusses calculation of local field - induced electron flux through the junction , and section [ eom ] introduces set of negf - eoms .
numerical results and discussion are given in section [ res ] .
section [ conclude ] summarizes our findings .
a model junction consists of a molecule coupled to two metallic contacts driven by external radiation field .
the radiation is a time - dependent local electromagnetic field @xmath1 calculated within fdtd technique for bowtie geometry of the contacts ( see section [ optics ] for details ) .
molecule is represented by a two - level system @xmath2 and @xmath3 ( homo and lumo or ground and excited states ) , and is placed in a ` hot spot ' of the local field .
contacts @xmath4 and @xmath5 are assumed to be free charge carrier reservoirs , each at its own equilibrium .
difference in their electrochemical potentials defines bias applied to the junction @xmath6 .
following refs .
we consider two types of coupling between molecule and contacts : charge and energy transfer .
hamiltonian of the junction is @xmath7 here @xmath8 ( @xmath9 ) and @xmath10 ( @xmath11 ) are creation ( annihilation ) operators of electron in level @xmath12 of the molecule and in state @xmath13 in the contact(s ) , respectively , @xmath14 is the operator of electronic population in level @xmath12 , @xmath15 is operator of molecular de - excitation ( @xmath16 ) , and @xmath17 is molecular transition dipole moment .
terms on the right - hand side of ( [ h0 ] ) represent molecular structure ( two - level system ) , contacts , and coupling to the driving field .
right - hand side of eq.([v ] ) introduces electron and energy transfer between molecule and contact(s ) .
( [ h])-([v ] ) introduce the model of ref . with rotating wave approximation relaxed , and
with driving force treated as a local electromagnetic field .
( color online ) a sketch of local field driven molecular charge pump . ] to simulate molecules with strong charge - transfer transition with dipole moment oriented along the junction axis below we assume that ground state ( or homo ) , @xmath2 , is coupled strongly to the left contact @xmath4 , while excited state ( or lumo ) - to the right contact @xmath5 .
such setup works as a local field driven charge pump ( see fig . [ fig1 ] ) .
note that similar selective coupling can also be obtained for the bridge made of a quantum dots as discussed in refs . .
calculations of the local electromagnetic field dynamics are carried out utilizing fdtd technique.@xcite following ref .
we assume that the incident field , @xmath18 , has the form of a linear chirped pulse @xmath19 where @xmath20 is the incident peak amplitude , @xmath21 is the incident frequency , and parameters @xmath22 and @xmath23 describing incident chirped pulse are given by @xmath24 with @xmath25 ( the value of the pulse duration @xmath26 of the corresponding transform - limited pulse used in simulations is @xmath27 fs ) and @xmath28 is the chirp rate in the frequency domain . throughout the simulations
the incident field is taken in the form of ( [ chirp ] ) and is normalized to preserve the total energy of a laser pulse at different chirp rates according to @xmath29 ( color online ) results of fdtd simulations for chirped pulses exciting the bowtie antenna schematically depicted in the inset of panel ( a ) .
panel ( a ) shows the enhancement of the local intensity @xmath30 detected in the gap of the bowtie antenna as a function of the incident frequency .
panel ( b ) presents local field component of electric field along the axis of symmetry of the structure , @xmath31 , as a function of time for two chirp rates at the plasmon resonance , @xmath32 ev : black solid line - @xmath33 fs@xmath34 , red dashed line - @xmath35 fs@xmath34 .
panel ( c ) shows the amplitude of the local field in the frequency domain ( note that it is independent from the phase rate ) .
panel ( d ) represents the phase of the local field in the frequency domain at two chirped rates : black solid line - @xmath33 fs@xmath34 , red dashed line - @xmath35 fs@xmath36 . ]
the geometry considered is depicted in the inset of fig .
[ fig2]a showing the top view of the bowtie antenna . to investigate the influence of chirped incident pulses on plasmon dynamics we choose incident field in the form ( [ chirp ] ) and vary @xmath37 .
below we shall write @xmath38 , having in mind @xmath39 .
we further presume that the incident pulse is @xmath40-polarized and propagates along the @xmath41 axis with the incident frequency at the plasmon resonance ( see the inset of fig .
[ fig2]a ) .
material dispersion of silver is taken in the drude form with other numerical parameters as in ref . .
for a given set of material and geometric parameters the local electric field enhancement exhibits well pronounced plasmon resonance as seen in fig .
[ fig2]a reaching the value of 2800 near @xmath42 ev .
our goal is to take plasmonic effects ( local field enhancement and phase accumulation ) directly into account and investigate how such crafted local fields affect transport properties of molecular junctions placed in the gap of bowtie antennas .
however it is informative first to examine general features of chirped pulses interacting with plasmonic materials .
it has been noted in several papers @xcite that local field enhancement depends sensitively on the sign of chirped excitation pulses .
moreover careful examination of spatiotemporal dependence of local fields on chirp rates@xcite revealed a complex dynamics of plasmon wavepackets that are noticeably influenced by chirped laser pulses - one may find different local points for a given plasmonic system , where positive chirps lead to higher local fields and the other way around . generally speaking , plasmonic materials can be considered as pulse shapers @xcite due to high material dispersion near plasmon resonance , which induces a phase in the frequency domain resulting in shaping of the total electromagnetic field in time domain .
this is illustrated in fig .
[ fig2]b - d , where one can clearly see that the positive chirp leads to the compression of the local field ( fig .
[ fig2]b ) and hence stronger field enhancement . while the field amplitude in the frequency domain is not affected by the chirp sign ( fig .
[ fig2]c ) , obviously the phase of the field is significantly different for positive and negative chirp as shown in fig .
we note that one can not recover data obtained for negative chirp , for instance , by simply flipping the sign of the phase for the positive chirp .
additional phase induced by the plasmonic system , which depends on the sign of the chirp rate , makes this problem time irreversible.@xcite
time - dependent current through the junction under external driving is@xcite @xmath43 \right . \\ + & \left .
\frac{1}{\pi}\mbox{im}\int_{-\infty}^{+\infty}de\,f_k(e)\ , \mbox{tr}\left[\mathbf{\gamma}^k\,\mathbf{g}^r(t , e)\right]\right ) \nonumber\end{aligned}\ ] ] here the trace is taken over molecular subspace , @xmath44 is matrix of electronic decoherence due to coupling to contact k @xmath45 which is energy independent in the wide - band approximation , @xmath46 is fermi - dirac thermal distribution in the contacts , @xmath47 is non - equilibrium reduced density matrix of molecular subsystem , @xmath48 are matrices in molecular subspace of retarded and lesser projections of single - electron green function @xmath49 ( @xmath50 is contour ordering operator ) , and @xmath51 is the right side fourier transform of the retarded green function @xmath52 we are interested mostly in effectiveness of the device as a charge pump , i.e. we will calculate _ excess charge _ transferred through the system during the laser pulse @xmath53 where @xmath54 is defined in eq.([ikt ] ) and @xmath55 is current at bias induced steady - state condition , i.e. in the absence of radiation
markov approximation employed in ref
. comes from consideration of time - local quantities only .
this approach is sufficient when one can neglect broadening of molecular states induced by hybridization with states in the contacts . in realistic molecular junctions
such hybridization is non - negligible , since molecules are usually chemisorbed on at least one of the contacts . here
( in addition to local field formation ) we are going to explore how non - markovian effects influence characteristics of laser pulse induced charge pumping . to keep non - markov effects a time - nonlocal quantity single - particle green function , eq.([g ] ) is at the focus of our consideration .
we employ keldysh contour based eom approach , similar to the one employed in our earlier publication@xcite ( see appendix [ appa ] for derivation ) @xmath57 here @xmath58 is molecular level other than @xmath12 ( e.g. for @xmath59 @xmath60 ) , @xmath61 is a single - particle green function of free electron in the contacts @xmath62 @xmath63 is the self - energy due to coupling to contacts with @xmath64 and @xmath65 is molecular subspace two - particle green function @xmath66 note , in derivation of ( [ eomg1 ] ) we treated the energy transfer term , eq.([v ] ) , at the second order of the perturbation theory .
presence of many - body interaction does not allow to close hierarchy of equations exactly . to make the problem tractable we employ markov approximation in treating energy transfer , last term on the right in ( [ eomg1 ] ) , and in writing eom for two - particle gf ( see below ) .
these approximations are similar to those introduced previously in refs . .
molecule - contact coupling in ( [ eomg1 ] ) is treated exactly , thus introducing non - markov effects into description .
this leads to system of equations ( see appendix [ appa ] for derivation ) @xmath67-\gamma_{mm}n_m(t ) -\gamma_{m\bar m}\mbox{re}\left[p(t)\right ] + 2\mbox{re}\sum_{m_1}\int_{-\infty}^{+\infty}\frac{de}{2\pi}g^r_{mm_1}(t , e)\sigma^{<}_{m_1m}(e ) \\ &
\qquad\qquad -(-1)^m\bigg(b(\varepsilon_{21})n_m(t)-b(\varepsilon_{12})\left[n_1(t)-n_2(t)+n_m(t)\right]\bigg ) \nonumber \\ \label{eomp } & \frac{d}{dt}p(t ) = -i\mu e(t)\bigg(n_2(t)-n_1(t)\bigg)-i\bigg(\varepsilon_2+\varepsilon_1\bigg)p(t ) -\frac{\gamma_{21}}{2}\bigg(n_1(t)+n_2(t)\bigg)-\frac{\gamma_{11}+\gamma_{22}}{2}p(t ) \\ & + \sum_{m_1=1,2}\int_{-\infty}^{+\infty}\frac{de}{2\pi}\bigg ( g^r_{2m_1}(t , e)\sigma_{m_11}^{<}(e)-\sigma_{2m_1}^{<}(e)\overset{*}{g}{}^r_{1m_1}(t , e ) \bigg ) -i\sum_{m_1}b(\varepsilon_{m_1\bar m_1})\mbox{im}\left[p(t)\right ] \nonumber \\ \label{eomnm } & \frac{d}{dt}n_m(t ) = 2\mu e(t)\mbox{im}\left[p(t)\right ] -i\sigma_{22}^{<}(\varepsilon_2)n_1(t ) + i\sigma_{11}^{>}(\varepsilon_1)n_2(t ) -2i\bigg(\sigma_{12}^{>}(\varepsilon_1)+\sigma_{12}^{<}(\varepsilon_2)\bigg)\mbox{re } \left[p(t)\right ] \\ & \qquad\qquad -\bigg(\gamma_{11}+\gamma_{22}+b(\varepsilon_{21})\bigg)n_m(t ) + b(\varepsilon_{12})\bigg(n_1(t)-n_2(t)+n_m(t)\bigg ) \nonumber \end{aligned}\ ] ] here @xmath68 , @xmath69 ( @xmath70 ) are populations of molecular levels , @xmath71 is molecular coherence , @xmath72 is the molecular excitation correlation function , @xmath73 is the matrix of electronic decoherence due to electron transfer between the molecule and contacts , with @xmath74 defined in eq.([gamma ] ) , @xmath75 greater ( lesser ) projections of self - energy due to coupling to contacts with @xmath76\end{aligned}\ ] ] and @xmath77 is the dissipation rate due to energy transfer @xmath78\end{aligned}\ ] ] note that in ( [ eomgre ] ) we omitted term coming from energy transfer , since contribution to the total retarded self - energy @xmath79 from molecule - contacts electron transfer @xmath80 is much bigger than corresponding contribution from energy transfer @xmath81 ( @xmath82 ) for @xmath59 ( @xmath83 ) in a reasonable parameter range.@xcite eoms ( [ eomgre])-([eomnm ] ) form a closed set of time - dependent equations to be solved simultaneously on energy grid starting from a steady - state initial condition corresponding to biased junction before the laser is switched on .
density matrix @xmath84 and retarded gf @xmath51 obtained as the solution are used in ( [ ikt ] ) and ( [ qkt ] ) to calculate time - dependent current and excess charge pumped through the junction , respectively . in the limit of weak molecule - contact coupling @xmath85 , neglecting local field and non - markov effects , disregarding off - diagonal terms in spectral function , and assuming rotating - wave approximation eqs .
( [ eomgre])-([eomnm ] ) are reduced to results of ref .
( see appendix [ appb ] for details ) .
here we present results of numerical simulations for the model ( [ h])-([v ] ) with local field formation and non - markov effects taken into account as described above .
time dependent local electromagnetic field is calculated solving maxwell s equations on a grid ( see section [ optics ] ) for metallic contacts of a bowtie geometry . molecule is placed in a ` hot spot ' situated between the contacts , and local field plays a role of external driving force in electronic calculations ( as described in section [ eom ] ) . unless stated otherwise parameters of the electronic simulations : temperature
is @xmath86k , molecular electronic level positions @xmath87ev and @xmath88ev , elements of electronic decoherence matrix are @xmath89ev , @xmath90ev , @xmath91 , coupling to external field @xmath92ev ( after normalization ( [ norm ] ) for @xmath93 fs@xmath36 ; also below coupling to external field below is given renormalized according to ( [ norm ] ) for particular @xmath38 ) .
fermi energy is taken as origin @xmath94 , bias is applied symmetrically @xmath95 . all calculations except fig .
[ fig6 ] below are done at equilibrium , @xmath96 .
only processes of energy relaxation on the molecule are taken into account with @xmath97 and @xmath98ev .
time grid is taken from the external driving field simulations .
energy grid spans region from @xmath99 to @xmath100ev with step @xmath101ev .
other parameters are introduced separately for each calculation .
( color online ) time dependence of charge pumped through the junction q(t ) , eq.([qkt ] ) for several chirp rates . shown
are results for ( a ) @xmath93 and @xmath102 fs@xmath36 without energy transfer @xmath103 ( solid line , red and dashed line , blue , respectively ) , and taking into account energy transfer term @xmath98 ev ( dash - dotted line , red and dotted line , blue , respectively ) , and ( b ) @xmath104 and @xmath105 fs@xmath36 for @xmath98 ev and @xmath106 ( solid line , red and dashed line , blue , respectively ) and @xmath107 ev ( dash - dotted line , red and dotted line , blue , respectively ) . ]
figure [ fig3 ] demonstrates pumped charge build - up during the laser pulse excitation .
one sees that the local field formation leads to asymmetry in pumped charge for opposite chirp rates .
negatively chirped incoming field creates longer local pulse ( see fig .
[ fig2]b ) , which results in increase in total charge pumped through the junction .
role of electron - hole excitations in the contacts on charge buildup is shown in fig .
[ fig3]a . since processes of escape from lumo into the right contact and energy relaxation on the molecule compete for the excited state population , current ( and consequently pumped charge ) diminish with increase of coupling to electron - hole excitations in the contacts .
[ fig3]b shows effect of intensity of incoming pulse on the transfered charge buildup . for higher intensity
the build - up demonstrates saturation in the middle of the pulse .
the reason for this behavior is the competition between timescales related to rabi oscillation induced by local field between molecular levels and electronic escape rate from molecule into contacts ( @xmath108 ) . on the one hand , both negatively and positively chirped pulses in the middle have frequency approximately at resonance with homo - lumo transition , @xmath109 , which is a prerequisite to effective electron transfer and thus increase in pumped charge . on the other hand , at resonance rabi oscillations@xcite at high enough intensities
compete with electron escape rate , thus effectively blocking current through the junction . depending on parameters this may lead either to most effective charge transfer in the middle of the pulse ( dash - dotted and dotted lines in panel b ) , or to suppression of charge transfer at this point ( solid and dashed line in panel b ) . note that the effect is not related to non - markov relaxation , i.e. this behavior is observed also in the absence of hybridization between molecule and contact(s ) states , and its relation to landau - zener problem@xcite in terms of total charge pumped across the junction was discussed in ref . .
note also , that with positively chirped pulse changing frequency from lower to higher transfered charge buildup is more effective at the start of the pulse ( at lower frequencies ) , while for negatively chirped pulse more effective buildup takes place at the end of the pulse ( compare solid and dashed lines in figs .
[ fig3]b and [ fig5]b ) .
contrary to buildup suppression in the middle of the pulse , this effect is due to molecule - contact hybridization .
the latter leads to broadening of molecular levels , and effectiveness of homo - lumo charge transfer depends ( among other conditions ) on integral of occupied states at homo and empty states at lumo separated by frequency of incident light @xmath110 .
clearly , at frequencies below resonance the latter is greater than at frequencies above it .
( color online ) charge pumped through the junction during pulse vs. chirp rate : ( a ) total charge , i.e. integral of q(t ) , eq.([qkt ] ) , over local field pulse duration , and ( b ) asymmetry in charge transfer between positively and negatively chirped incoming laser pulses , @xmath111 . shown
are results with ( @xmath98ev , dashed line with triangles , blue ) and without ( @xmath103 , solid line with circles , red ) energy transfer . ]
local field asymmetry relative to the sign of the chirp rate in the frequency domain leads to asymmetry in charge pumping contrary to symmetric situation presented in ref .
, as is demonstrated in fig .
one can see that the pumped charge is almost symmetric at high rates with asymmetry confined to the low rate region .
difference between charge pumped through the junction at positive and negative chirp rates is shown in fig .
[ fig4]b . as discussed above duration of local field due to
positively chirped incoming pulse is shorter than the one due to negatively chirped analog .
this compression is the cause of less charge pumped through the system in the former case , which results in decrease in @xmath112 in the region of @xmath39 from @xmath113 to @xmath114 fs@xmath36 . indeed , at the very low rates frequency of the pulse does not change much , so the asymmetry is solely due to difference in pulse length . at a higher rates
an additional factor appears : the most effective contribution to charge transfer takes place at a particular region of frequencies ( at and just below resonance , as is discussed above ) .
this region is passed quicker in the positively chirped local pulse , and in the region up to @xmath114 fs@xmath36 this results in increase of asymmetry , since negatively chirped pulse spends more time in its effective frequencies zone .
further increase of chirp rate leads to decrease and almost disappearance of the asymmetry .
the reason is decrease of ratio of the pulses difference to overall local pulse duration .
coupling to electron - hole excitations not only diminishes pumped charge ( compare solid and dashed lines in panel a ) , but also decreases asymmetry ( panel b ) .
the latter results from the fact that rate for molecular energy relaxation ( lumo @xmath115 homo transition due to coupling to excitations in the contacts ) is proportional to population in the lumo ( see discussion in ref . ) .
so for higher currents also energy relaxation will be more efficient , thus effectively compensating for the difference .
( color online ) dependence of charge pumping on molecule - contact states hybridization at @xmath106ev . shown
are ( a ) charge pumped through the junction during pulse vs. electronic escape rate for chirp rate @xmath104 fs@xmath36 ( solid line , red ) and @xmath105 fs@xmath36 ( dashed line , blue ) , and ( b ) normalized transferred charge build - up ( charge normalized to total charge transferred during the pulse ) vs. time for chirp rate @xmath104 fs@xmath36 and @xmath89ev ( solid line , red ) and @xmath116ev ( dashed line , blue ) .
red line in panel ( b ) is the same as solid red line in fig .
[ fig3]b . ]
importance of non - markov behavior for charge pump is demonstrated in figure [ fig5 ] .
[ fig5]a shows pumped charge as function of level width ( for two opposite choices of chirp rate ) .
increase in total charge pumped through the junction with increase in hybridization saturates at high strength of coupling between molecule and contacts .
such behavior is expected : at low hybridization there is only one frequency corresponding to resonance , where pumping is most effective , so only an instant of chirped pulse contributes to charge transfer .
as molecule - contact coupling grows the condition of resonance transition becomes less and less strict .
eventually any frequency within the chirped pulse has roughly same effectiveness
this is the reason for saturation .
also , stronger coupling means more effective molecule - contact electron transfer , which competes more effectively with intra - molecular rabi oscillation at resonance .
this competition is demonstrated in fig .
[ fig5]b , where middle - of - the - pulse saturation ( see discussion of fig .
[ fig3 ] ) disappears for stronger molecule - contact coupling .
( color online ) dependence of charge pumping on bias .
( a ) total excess charge pumped through the junction during pulse vs. bias for chirp rate @xmath104 fs@xmath36 ( solid line , red ) and @xmath105 fs@xmath36 ( dashed line , blue ) . ( b )
excess charge build - up ( normalized to total excess charge transferred during the pulse ) vs. time at chirp rate @xmath104 fs@xmath36 for @xmath117v ( dashed line , blue ) , @xmath113v ( solid line , red ) , and @xmath116 v ( dotted line , black ) . here
@xmath118 ev .
red line in panel ( b ) is the same as solid red line in fig .
[ fig3]b . ] finally , in figure [ fig6 ] we discuss influence of bias on charge pumping . here
we define optically pumped charge ( excess charge ) as a difference between charge pumped through the junction with and without laser field .
[ fig6]a demonstrates total excess pumped charge during laser pulse for opposite choices of chirp rate as function of bias .
application of bias has 2 effects on the pumping process : 1 .
it depletes ( populates ) the homo ( the lumo ) and 2
. it may block or release channels for electron transfer from lumo to contact @xmath5 .
this leads to a situations when most effective optical pumping does not correspond to zero bias , rather we see shallow peak at @xmath119v .
explanation is related to the fact that broadened molecular levels are essentially a set of scattering channels with different transmission probabilities : high conducting channels are in the center of lorentzian , while channels in the sides of distribution are poor conductors .
optical process takes electron from an occupied ground state and puts it into one of empty excited state .
effectiveness of the charge pump is defined by increase or decrease of current through the junction under optical pulse ( see discussion in ref . ) . in particular ,
negative bias decreases effectiveness of the pump mostly due to blocking part of lumo-@xmath5 escape routes .
positive bias opens additional escape routes at the tail of lumo lorentzian , facilitating increase in pump efficiency . however , additional effect of depleting homo and populating lumo partially blocks optically induced homo - lumo electron transfer , thus reducing overall effectiveness of optical pumping .
competition between the two proceses reveals itself as a shallow peak at @xmath120v .
time resolved charge buildup is presented in fig [ fig6]b .
middle - of - the - pulse saturation observed previously at equilibrium ( solid line , same as in fig .
[ fig3]b ) is enhanced at negative ( dashed line ) and disappears at positive bias ( dotted line ) .
the reason is similar to competition between rabi frequency and escape rate discussed above .
indeed , with negative bias partially blocking fast escape route for the electron from excited state into right contact , rabi oscillation plays an important role at quasi - resonant situation in the middle of the pulse .
positive opening additional routes makes rabi oscillation less effective .
we consider a two - level ( homo - lumo ) model for optically - driven molecular charge pump .
such pump may be realized as a junction formed by a molecule with strong charge - transfer transition between its ground and excited states .
the junction is driven by both applied bias and laser pulse .
the latter is treated as an external classical driving force .
our consideration is the generalization of previous study@xcite which takes into account effects of local field ( ` hot spot ' formation ) and hybridization between the states of molecule and contact(s ) ( non - markov effects ) .
we formulate approximate closed set of eoms for single- and two - particle gfs .
electron transfer in the former is treated exactly . to close set of equations the latter
are considered within markov approximation .
our eoms are reduced to set of equation derived in ref . under several simplifying assumptions :
weak molecule - contact coupling ( neglect of hybridization ) , neglect of non - diagonal terms in molecular spectral function , and within rotating wave approximation .
incoming laser pulse is assumed to be linearly chirped .
local field is calculated within fdtd technique on a grid with bowtie antenna geometry used to represent junction metallic contacts .
we find that contrary to symmetric behavior of the pump relative to sign of the chirp rate , duration of the corresponding local field pulse depends on the sign of incoming chirp , which results in asymmetric operation of the pump .
the asymmetry depends on the incoming pulse chirp rate in a non - monotonic manner .
junction response to optical driving is symmetric to both low and high chirp rates , going through a maximum between the two extremes .
we find that this behavior is caused by the correspondence between pulse duration of the local field and detuning of its frequency at the end of the pulse from energy difference between molecular states ( @xmath121 ) .
we note that at quasi - resonance charge pump becomes ineffective , due to the competition between intra - molecular rabi oscillation induced by the pulse with electron transfer from molecule to contact .
increase of the molecule - contact coupling strength increases electron escape rate , thus reducing ineffectiveness of the pump due to rabi oscillations .
also we study the effect of bias on optically - facilitated charge transfer through the junction .
we find that in the non - markov situation ( i.e. when hybridization between molecule and contacts is non - negligible ) most effective charge pump regime is at finite positive bias , rather than at equilibrium as one may expect from markov consideration of ref . .
the effect comes from optically - assisted charge redistribution between low and high conducting scattering channels in broadened molecular states , as was discussed in our previous publication.@xcite within the model negative bias reduces ( positive increases ) excess charge pumping due to blocking ( facilitating ) outgoing scattering channels in the excited molecular state and thus increasing ( decreasing ) the role of intra - molecular rabi oscillations . finally , direct electron - hole excitation in contacts , heating , and inelastic effects are examples of effects beyond current consideration which may also have a significant impact on the properties of a molecular charge pump .
we acknowledge support by the national science foundation ( mg , che-1057930 ) , the us - israel binational science foundation ( bf and mg , # 2008282 ) , and by the ucsd ( mg , startup funds ) .
eom for ( [ eomg1 ] ) in contour variable @xmath122 starts from writing heisenberg equation for @xmath123 @xmath124 the last two terms on the right come from electron and energy transfer terms in eq.([v ] ) . treating the two within non - crossing approximation allows to find the first exactly within a standard procedure@xcite @xmath125 two particle green function in the second term is treated ( still keeping non - crossing approximation in mind ) within first order perturbation theory in energy transfer @xmath126 substituting ( [ t1 ] ) and ( [ t2 ] ) into ( [ heisenberg ] ) yields ( [ eomg1 ] ) .
eq.([eomgre ] ) is retarded projection of eq.([eomg1 ] ) with omitted energy transfer term .
the approximation is based on an estimate that in usual situation electron escape rate should be much bigger than corresponding energy transfer , @xmath127.@xcite eqs .
( [ eomnm ] ) and ( [ eomp ] ) is lesser projection of ( [ eomg1 ] ) taken at equal times , @xmath128 .
note that ( [ eomp ] ) is exact , while in ( [ eomnm ] ) we employ markov approximation in derivation of the energy transfer term similar to previous publications,@xcite for example @xmath129n_{k_1}\pi\delta\left(\varepsilon_{k_2}-\varepsilon_{k_1 } + \varepsilon_{m_1}-\varepsilon_{\bar m_1}\right ) \nonumber\\ & \qquad\times \left\langle\hat d_m^\dagger(t)\hat d_{\bar m}(t)\hat d_{m_1}^\dagger(t ) \hat d_{\bar m_1}(t)\right\rangle \nonumber \end{aligned}\ ] ] using ( [ markov ] ) and similar expressions for other parts of the keldysh contour deformed in accordance with langreth rules@xcite in the energy transfer term of lesser projection of diagonal element of ( [ eomg1 ] ) , and utilizing ( [ b ] ) leads to ( [ eomnm ] ) .
finally , eq.([eomnm ] ) is treated with markov approximation ( see eq.([markov ] ) above ) applied to both electron and energy transfer terms .
then the derivation goes along the lines presented in ref .
eoms derived in ref .
are markov limit of eqs.([eomnm])-([eomnm ] ) within static quasiparticle approximation assumed for molecular states .
the latter implies disregarding eq.([eomgre ] ) , and assuming instead @xmath130=2\pi\delta_{m , m'}\delta(e-\varepsilon_m)\ ] ] then , disregarding level mixing due to coupling to contacts @xmath131 , eqs .
( [ eomnm ] ) and ( [ eomp ] ) reduce to eqs .
( 33 ) and ( 34 ) ) differs from eq.(34 ) in ref .
, since the latter is written under additional assumption of rotating - wave approximation . ] of ref . .
after omitting non - diagonal elements of self - energy in eq.([eomnm ] ) one gets eq.(35 ) of ref . .
|
we consider a two - level system coupled to contacts as a model for charge pump under external laser pulse .
the model represents a charge - transfer molecule in a junction , and is a generalization of previously published results [ b. d. fainberg , m. jouravlev , and a. nitzan .
phys .
rev .
b * 76 * , 245329 ( 2007 ) ] . effects of local field for realistic junction geometry and non - markov response of the molecule are taken into account within finite - difference time - domain ( fdtd ) and on - the - contour equation - of - motion ( eom ) formulations , respectively .
our numerical simulations are compared to previously published results .
|
in simple models of quantum dots , the conduction electrons ( or holes ) of a semiconductor are confined into a two - dimensional harmonic trap ( for reviews see @xcite ) .
the band structure of the material is taken into account through the effective mass approximation , and screening effects are accounted for by the dielectric constant . the problem
is then reduced to solving the many - particle problem of interacting electrons in a two - dimensional harmonic potential .
the electrons have spin as an internal degree of freedom. neglecting spin - orbit coupling , the spin - up and spin - down electrons can be treated as separate interacting particles , and consequently we can say that the normal electron gas is a two - component gas , the components being the spin - up and spin - down electrons . similarly , we will call the polarised electron gas as a one - component system ( sometimes also called a system of spin - less fermions ) . the simple picture will fail in describing more complex structures where the number of degrees of freedom of the electrons is increased either by several two - dimensional layers or by multiple valleys of the band structure .
for example , in a vertical double - layer quantum dot the electrons confined in the two layers form ( in the vertical direction )
odd and even states @xcite which could be approximated as different components , or as different isospin - states of the electron @xcite .
the isospin together with the spin would make the system a four - component electron gas .
other multicomponent electron systems would be quantum dots in silicon . in this case
the conduction electrons originating from four valleys of the conduction band could be approximated as different , but mutually interacting fermions .
similarly , quantum dots with holes would always have particles belonging to different bands , i.e. heavy holes and light holes @xcite .
density functional theory in the local ( spin ) density approximation ( lsda ) provides a flexible method to study the ground state properties of interacting electrons in quantum dots @xcite . in lsda the exchange and correlation effects of the interacting conduction electrons are locally approximated by the exchange - correlation energy ( @xmath0 ) of the two - dimensional , homogenous gas .
similarly , in a multicomponent electron system the starting point for the local density approximation is the exchange - correlation energy of a two - dimensional multicomponent gas .
recently , we have suggested that the multicomponent @xmath0 can well be approximated by extending the parametrised two - component function of attaccalite _ et al . _
@xcite to a higher number of components @xcite .
this parametrisation was shown to have correct high and low density behaviour and it has rather simple dependence on the densities of different components .
moreover , the dependence of the effective masses of the components can be approximated through a simple scaling of the density parameter by an effective average mass . in this paper
we study the electronic structure of quantum dots where the electron gas has from one up to eight internal degrees of freedom , using the exchange - correlation functional suggested in ref .
the inclusion of more components increases the number of internal degrees of freedom in the system . as a result ,
in the self - consistent scheme the electrons can access more kohn - sham states in addition to the usual two spin states , leading to new features in the electronic structure .
we will discuss the electronic shell structure of four - component quantum dots by studying the addition energy spectrum at high densities . in the low density limit the formation of wigner molecules in an eight - component quantum dot
is investigated .
we found that the seven - electron configuration is particularly stable like the classical point - charge calculation predicts @xcite . in the wigner molecule
limit the fermions of different components are distributed spatially so that antiferromagnetic frustration is avoided .
finally we notice that even a slight increase of mass favors the heavier components as the heavy component states are pushed down in energy .
we write the total density of the multicomponent electron gas as @xmath1 where @xmath2 is the number of components , @xmath3 are the densities of components and @xmath4 dimensionless concentrations of the components . in the multicomponent case ,
the use of concentrations is simpler than the use of total density and polarization @xmath5 , as usually done for the normal electron gas ( in the two - component case , @xmath6 and @xmath7 and @xmath8 ) . following the notations of ref .
@xcite we define numbers @xmath9 which allows us to write the exchange - correlation functional of attaccalite _
et al_. @xcite as @xmath10+\epsilon_x^{(6)}+\alpha_0(r_s ) + \alpha_1(r_s)(2z_2 - 1)+\alpha_2(r_s)(2z_2 - 1)^2~,\ ] ] where @xmath11 .
as shown earlier @xcite this analytic continuation of the originally two - component functional approximates very well all existing results for the exchange - correlation energy of a multicomponent electron gas .
the exchange energy is independent of the particle mass .
if the effective masses of all the particles are the same , the mass dependence of the total energy becomes just a scaling factor .
if the masses are not the same we can use as a first approximation a properly weighed average mass as a scaling factor and write @xcite @xmath12 where @xmath13 is the bare mass of the electron ( or any suitably chosen effective mass ) , @xmath14 is the effective mass of component @xmath15 , and @xmath16 is an average mass defined as @xmath17 note that the density parameter @xmath18 in our formulation always refers to the total number density of all particles : @xmath19 .
the kohn - sham equations have to be solved self - consistently for each component of the electron gas .
the effective potential @xmath20 consists of the external harmonic confinement ( assumed here to be the same for all components ) , of the coulomb repulsion of the electron density distribution ( hartree term ) and of the exchange - correlation potential , which can be directly derived from the multicomponent exchange - correlation energy : @xmath21 where @xmath22 is the strength of the external confinement , @xmath23 the dielectric constant .
note that the exchange - correlation potential depends locally on the concentrations of each component .
the kohn - sham equations to be solved simultaneously for all components are @xmath24 resulting in the densities @xmath3 of all components @xmath15 , @xmath25 where @xmath26 is the number of electrons of component @xmath15 . in the ground state
the lowest single particle levels are filled and the numbers @xmath26 are known only after the ground state is found . in practice
we solve the equations by keeping the numbers @xmath26 fixed and then choose the configuration which gives lowest total energy .
the kohn - sham equations were solved using a plane - wave expansion and the fast fourier transform technique .
we use up to @xmath27 plane waves and , correspondingly a @xmath28 lattice at which the density and potential is derived .
when iterating the kohn - sham equations , a mixing of the new and old potential is necessary to obtain convergence .
the convergence is slow especially at low densities where broken symmetry solutions ( localization of electrons to wigner molecules ) emerge .
we use effective atomic units where energy is given in effective hartree , ha@xmath29=@xmath30 and the unit of distance is the effective bohr radius @xmath31 .
as confined , fermionic quantum systems , quantum dots show shell structure with degeneracies determined by the symmetries of the confining potential .
the independent electron energy spectrum for a 2d harmonic oscillator is @xmath32 , where @xmath33 is the principal quantum number and @xmath34 is the angular momentum ( or its z - projection ) . at high densities
the independent electron scheme describes the qualitative characteristics of the shell filling of the interacting system reasonably well , as known from the earlier studies @xcite of normal quantum dots .
the main effect of the electron - electron interaction is to produce the spin determined by hund s first rule @xcite .
the shell structure effects are usually shown in a form of addition energy spectrum where the second difference of the total energy is plotted as a function of the number of electrons in the dot .
addition energies measure the changes @xmath35 in the chemical potential @xmath36 .
we will first study the addition energy spectrum of an ideal four - component electron system in a quite high electron density , @xmath37 .
a physical realization of such system could be a vertical double dot , where the interlayer distance is very small .
figure 1 shows the addition energy spectrum and a schematic picture of the single particle levels .
since each single particle level can now occupy four electrons the closed shells correspond to total electron numbers 4 , 12 , 24 , etc .
these magic numbers are seen as pronounced maxima in the addition energy spectrum . in between , we see smaller maxima at every even electron number up to 12 electrons and at every third electron number between 12 and 24 . these are manifestations of hund s first rule generalized to the multicomponent case : degenerate states
are filled one component at a time .
this minimizes the total energy , since the exchange energy favors a polarized electron gas .
for example , in the third shell we see maxima at @xmath38 , 18 , and 21 , corresponding to filling the three orbitals of the first , second and third component of the four - component electron gas . at low enough densities the electrons in the parabolic quantum dot
are expected to form wigner molecules as the correlation effects start to dominate the electronic structure . in this limit
the electrons localise in the classical configuration that minimises the electrostatic repulsion @xcite and this charge distribution symmetry is reflected in the internal structure of the many - body wavefunction @xcite . in the multicomponent systems
the localisation is eased up due to the fact that electrons can access more than two internal states . in the low density limit
we study an eight component system in fixed external confinement with @xmath39 effective atomic units .
this choise corresponds to densities that are only slightly higher than those where a polarised ( one - component ) state becomes the ground state for different electron numbers . at these low densities
the multicomponent local density approximation localises the electrons in the classically predicted configurations .
figure 2 shows the total electron densities for seven , eight and nine electrons . in the cases of seven and eight electrons ,
one electron is in the center and the rest form a ring around it , while in the case of nine electrons two electrons form a dimer at the center and seven electrons from a ring around them .
the geometries are in perfect agreement with those of classical electrons @xcite .
figure 3 shows the addition energy spectrum of the eight - component quantum dot at the low density .
the spectrum does not any more show features of shell structure and hund s rule , but shows a small kink at @xmath40 , in agreement with the maximum in the addition energy spectrum of the purely classical system , determined from the results of ref .
@xcite , and also shown in fig .
3 . classically
, the seven electrons can form a perfect hexagon with one electron at the center .
the quantum mechanical solution is the same as shown in fig .
2 . in the density functional theory
the localized electrons are not point charges as in the classical case . as a consequence ,
the spectrum is smoother than that of the classical result .
the evolution of the ground state as a function of density parameter @xmath18 is investigated in an eight electron quantum dot with four components . as discussed
earlier , at high densities @xmath41 the ground state obeys hund s rule giving the configuration ( 3,3,1,1 ) for the ground state . as the density is lowered , the electron structure shows a wigner molecule - like state already at @xmath42 .
six localised electrons are at the outer circumference with two non - localised in the middle , as shown in fig .
the six electrons at the outer radius belong to two components and the two electrons in the middle occupy also two components .
the electrons in the outer ring are distributed spatially so that the two nearest neigbours of each electron belong to other component .
this means that the densities of the outer components are rotated by @xmath43 with respect to one another . in this way
the system will avoid antiferromagnetic frustration .
for @xmath44 the electrons localise into a classically predicted configuration with seven electrons at the outer radius and one in the middle .
the frustration is again avoided by taking the nearest neighbours for each electron from other components , as shown in fig .
the effect of the varying mass was tested in the four - component system with @xmath45 and @xmath46 .
the masses of two components were increased ( @xmath47 ) while the other two masses were kept constant ( @xmath48 ) . for @xmath49 the sd - shell
is filled giving the magic configuration ( 6,6,6,6 ) .
the mass increase shifts the orbitals of the lighter components up in energy relative to heavier components , as shown in fig.5 . as a consequence , at @xmath50
the sd - orbitals of the light components are empty and the fp - orbitals of the heavier components are occupied according to hund s rule leading to configuration ( 10,8,3,3 ) . already at @xmath51 the lighter orbitals have only two electrons and the heavier components obey hund s rule resulting in the configuration ( 12,10,1,1 ) .
we have studied the general features of quantum dots confining a multicomponent electron gas . at high densities
the exchange energy favors polarization of electrons and the degenerate energy levels are filled with one component at a time .
this leads to an addition energy spectrum which besides the peaks at full shells also shows peaks coming from the generalization of hund s rule .
the increased number of internal degrees of freedom of the electrons make it easier for the electrons to localize to wigner molecules .
we have demonstrated this for four- and eight - component systems . in the low density limit
the addition energy spectrum then does not any more show the electronic shell structure but the geometrical shell structure of wigner molecules .
if the different electrons ( or holes ) have different masses , the localization pushes the energy states of the heavy particles down as compared to those of the light particles .
if the mass difference is small the addition energy spectrum is expected to be complicated due to the mixture of light and heavy particle states . however ,
if the mass ratio is at a typical value of heavy and light holes , say five , the light holes do not play any role until the dot has several tens of particles .
t. chakraborty , _ quantum dots : a survey of the properties of artificial atoms _ ( north - holland , amsterdam 1999 ) .
s.m . reimann and m. manninen , rev .
* 74 * , 1283 ( 2002 ) .
b. partoens and f.m .
peeters , phys .
84 * , 4433 ( 2000 ) .
m. pi , a. emperator , m. barranco , f. garcias , k. muraki , s. tarucha , and d.g .
austing , phys .
87 * , 066801 ( 2001 ) .
m. pi , a. emperator , m. barranco , and f. garcias , phys .
b * 63 * , 115316 ( 2001 ) .
p. hawrylak , and j.j .
palacios , phys .
b * 51 * , 1769 ( 1995 )
marder , _ condensed matter physics _ ( john viley & sons , new york 2000 ) . c. attaccalite , s. moroni , p. gori - giorgi and g.b .
bachelet , phys .
lett . , * 88 * , 256601 ( 2002 ) .
k. krkkinen , m. koskinen , s.m .
reimann and m. manninen , phys .
b , * 68 * , 205322 ( 2003 ) .
v.m . bedanov and f.m .
peeters , phys .
b * 49 * , 2667 ( 1994 ) .
m. koskinen , m. manninen , and s.m .
reimann , phys .
lett . * 79 * , 1389 ( 1997 ) .
v.m . bedanov and f.m .
peeters , phys .
b * 49 * , 2667 ( 1994 ) .
reimann , m. koskinen and m. manninen , phys .
b * 62 * , 8108 ( 2000 ) .
|
quantum dots with conduction electrons or holes originating from several bands are considered .
we assume the particles are confined in a harmonic potential and assume the electrons ( or holes ) belonging to different bands to be different types of fermions with isotropic effective masses .
the density functional method with the local density approximation is used .
the increased number of internal ( kohn - sham ) states leads to a generalisation of hund s first rule at high densities . at low densitites
the formation of wigner molecules is favored by the increased internal freedom .
|
figure [ inv1 ] depicts an involution @xmath192 on @xmath136 under which @xmath193 is invariant , has its boundary components interchanged , and @xmath194 .
then @xmath192 extends to an involution of @xmath99 since its restriction to @xmath195 coincides with the restriction to @xmath139 of the standard involution of @xmath137 .
evidently @xmath196 and @xmath197 .
figure [ inv2 ] depicts an involution @xmath198 on @xmath142 under which each of the annuli @xmath199 , and @xmath200 are invariant .
further , it interchanges the components of @xmath201 and as in the previous paragraph , @xmath198 extends to an involution of @xmath100 .
note that @xmath202 for @xmath203 .
next consider the orientation preserving involution @xmath204 on @xmath104 . by construction
we have @xmath205 for @xmath203 , and therefore @xmath206 where @xmath207 is a homeomorphism whose restriction to @xmath208 is isotopic to @xmath209 .
the latter fact implies that @xmath210 is isotopic to a homeomorphism @xmath211 which commutes with @xmath212 .
hence , @xmath213 is isotopic to @xmath212 through orientation preserving involutions whose fixed point sets consist of two points . in particular , @xmath192 and @xmath198 can be pieced together to form an orientation preserving involution @xmath214 . for each slope @xmath43 on @xmath6
, @xmath215 extends to an involution @xmath216 of the associated dehn filling @xmath217 , where @xmath218 is the filling solid torus .
orbifold theorem applies to our situation and implies that @xmath44 has a geometric decomposition . in particular
, @xmath2 is a seifert fibred manifold whose base orbifold is of the form @xmath219 , a @xmath37-sphere with three cone points of orders @xmath220 respectively .
it follows immediately from our constructions that @xmath221 and @xmath222 are @xmath1-balls .
thus @xmath223 and since @xmath224 , it follows that @xmath225 is a @xmath1-ball .
more precisely , @xmath226 is an orbifold @xmath227 , where @xmath228 is a 3-ball , @xmath229 is a properly embedded 1-manifold in @xmath228 that meets @xmath230 in four points , and @xmath0 is the double branched cover of @xmath227 .
we will call @xmath227 a _ tangle _ , and if we choose some identification of @xmath231 with a standard model of ( @xmath232 , _ four points _ ) , then @xmath227 becomes a _ marked _ tangle .
capping off @xmath230 with a 3-ball @xmath233 gives @xmath234 .
then , if @xmath43 is a slope on @xmath6 , we have @xmath235 , where @xmath236 is the rational tangle in @xmath233 corresponding to the slope @xmath43 .
hence @xmath237 where @xmath238 is the link in @xmath239 obtained by capping off @xmath229 with the rational tangle @xmath236 .
we now give a more detailed description of the tangle @xmath227 . for @xmath166 ,
let @xmath240 , @xmath241 , @xmath242 , and @xmath243 .
figure [ quo1 ] gives a detailed description of the branch sets in @xmath244 , @xmath245 , @xmath246 with respect to the corresponding branched covering maps .
note that @xmath228 is the union of @xmath247 , and a product region @xmath248 from @xmath249 to @xmath250 which intersects the branch set @xmath229 of the cover @xmath251 in a @xmath37-braid .
in fact , it is clear from our constructions that we can think of the union @xmath252 as a @xmath52-braid " in @xmath253 with two fat strands " formed by @xmath254 .
see figure [ filling](a ) . by an isotopy of @xmath253 fixing @xmath250 , and which keeps @xmath253 , @xmath249 , and @xmath255 invariant , we may untwist the crossings between the two fat strands in figure [ filling](a ) so that the pair @xmath256 is as depicted in figure [ filling](b ) .
the slope @xmath5 is the boundary slope of the planar surface @xmath51 , and hence the rational tangle @xmath257 appears in figure [ filling](b ) as two short horizontal arcs in @xmath233 lying entirely in @xmath258 .
since @xmath259 , @xmath260 is a tangle of the form shown in figure [ filling6](a ) . recall that @xmath2 is a seifert fibred manifold with base orbifold of type @xmath261 , and is the double branched cover of @xmath262 .
write @xmath263 .
[ type ] @xmath264 is a montesinos link of type @xmath265 . by thurston s
orbifold theorem , the seifert fibering of @xmath2 can be isotoped to be invariant under @xmath266 .
hence the quotient orbifold is seifert fibered in the sense of bonahon - siebenmann , and so either @xmath264 is a montesinos link or @xmath267 is seifert fibred . from figure 6(a ) we see that @xmath264 is a 2-component link with an unknotted component and linking number @xmath268 .
but the only link @xmath264 with this property such that @xmath267 is seifert fibred is the hopf link ( see @xcite ) , whose @xmath37-fold cover is @xmath269
. thus @xmath264 must be a montesinos link .
since the base orbifold of @xmath2 is @xmath219 , @xmath264 has type @xmath270 ( c.f 12.d of @xcite ) .
it s easy to check that any montesinos link @xmath264 of the type described in the lemma [ type ] has two components , one of which , say @xmath271 , is a trivial knot , and the other , @xmath272 , a trefoil knot .
our goal is to use the particular nature of our situation to show that the branch set @xmath264 can not be a montesinos link of type @xmath265 , and thus derive a contradiction . from figure [ filling ]
, we see that @xmath229 has a closed , unknotted component , which must be the component @xmath271 of the montesinos link of type @xmath265 described above .
then @xmath273 , which we denote by @xmath274 .
now delete @xmath271 from @xmath228 and let @xmath275 be the double branched cover of @xmath228 branched over @xmath274 .
then @xmath275 is a compact , connected , orientable @xmath1-manifold with boundary a torus which can be identified with @xmath6 . in particular , if we consider @xmath3 and @xmath5 as slopes on @xmath276 , then both @xmath277 and @xmath278 are the lens space @xmath135 , since they are @xmath37-fold covers of @xmath239 branched over a trefoil knot .
hence the cyclic surgery theorem of @xcite implies that @xmath275 is either a seifert fibred space or a reducible manifold .
@xmath275 is not a seifert fibred space .
suppose @xmath275 is a seifert fibred space , with base surface @xmath40 and @xmath279 exceptional fibres .
if @xmath40 is non - orientable then @xmath275 contains a klein bottle , hence @xmath280 does also .
but since non - orientable surfaces in @xmath135 are non - separating , this implies that @xmath281 , which is clearly false .
thus @xmath40 is orientable
. if @xmath275 is a solid torus then clearly @xmath282 implies @xmath283 ( mod @xmath1 ) , contradicting the fact that @xmath284 .
thus we assume that @xmath275 is not a solid torus , and take @xmath285 to be the slope on @xmath276 of a seifert fibre .
then @xmath286 is reducible @xcite so @xmath287 , and @xmath277 is a seifert fibred space with base surface @xmath40 capped off with a disk , and @xmath288 or @xmath289 exceptional fibres , according as @xmath290 or @xmath291 .
since @xmath277 is a lens space and @xmath275 is nt a solid torus , we must have that @xmath40 is a disk , @xmath292 , and @xmath290 .
similarly @xmath293 .
in particular , without loss of generality @xmath294 in @xmath295 .
the base orbifold of @xmath275 is of the form @xmath296 , with @xmath297 .
then @xmath298 is the abelian group defined by generators @xmath299 and the single relation @xmath300 .
suppose @xmath301 in @xmath298 .
then @xmath302 is presented by the matrix @xmath303 .
similarly , since @xmath304 in @xmath298 , @xmath305 is presented by @xmath306 .
but the determinants of these matrices differ by @xmath307 , so they can not both be @xmath1 in absolute value .
this completes the proof of the lemma .
thus @xmath275 is reducible , say @xmath308 where @xmath309 and @xmath310 is closed .
consideration of @xmath2 and @xmath12 shows that @xmath311 and @xmath312 , and so theorem 2 of @xcite implies that @xmath313 .
it follows that any simple closed curve in @xmath276 which represents either @xmath3 or @xmath5 is isotopic to the core curve of @xmath314 .
let @xmath315 denote the meridional slope of @xmath314 . then @xmath316 is a basis of @xmath295 and up to changing the sign of @xmath3 we have @xmath317 . since @xmath318
, we can find a homeomorphism between the pair @xmath319 and the tangle shown in figure [ nk20](a ) , with the @xmath5 , @xmath3 , and @xmath320 fillings shown in figures [ nk20](b ) , ( c ) and ( d ) respectively .
( we show the case @xmath321 ; the other possibility can be handled similarly . )
recall that in figure [ filling](b ) , the slope @xmath5 corresponds to the rational tangle consisting of two short `` horizontal '' arcs in the filling ball @xmath233 .
it follows that under the homeomorphism from the tangle shown in figure [ nk20](a ) to @xmath319 shown in figure [ filling](b ) , the tangle @xmath260 ( resp .
@xmath322 ) is sent to a rational tangle of the form shown in figure [ filling6](a ) ( resp .
[ filling6](b ) ) . from figure [ nk20](d )
we see that @xmath238 is a link of three components @xmath323 , where @xmath324 is a trivial knot which bounds a disk @xmath325 disjoint from @xmath326 and which intersects @xmath230 in a single arc ; see figure [ filling6](b ) .
push the arc @xmath327 with its two endpoints fixed into @xmath328 along @xmath325 , and let @xmath329 be the resulting knot ( see part ( c ) of figure [ filling6 ] ) .
then there is a disk @xmath330 ( which is a subdisk of @xmath325 ) satisfying the following conditions : * @xmath331 .
* @xmath330 is disjoint from @xmath326 . *
the interior of @xmath330 is disjoint from @xmath233 .
perusal of figure [ filling6 ] ( c ) shows that the following condition is also achievable .
* @xmath332 has a single arc component , and this arc component connects the two boundary components of @xmath250 and is outermost in @xmath330 amongst the components of @xmath333 . among all disks in @xmath239 which satisfy conditions ( 1)(4 )
, we may assume that @xmath330 has been chosen so that * @xmath332 has the minimal number of components .
[ parallel ] suppose that @xmath332 has circle components .
then each such circle separates @xmath334 from @xmath335 in @xmath250 .
let @xmath336 be a circle component of @xmath332 .
then @xmath336 is essential in @xmath337 , for if it bounds a disk @xmath338 in @xmath337 , then an innermost component of @xmath339 will bound a disk @xmath340 .
we can surger @xmath330 using @xmath341 to get a new disk satisfying conditions ( 1)(4 ) above , but with fewer components of intersection with @xmath250 than @xmath330 , contrary to assumption ( 5 ) .
next since the arc component of @xmath332 connects the two boundary components of @xmath250 , @xmath336 can not separate the two boundary components of @xmath250 from each other .
lastly suppose that @xmath336 separates the two points of @xmath342 .
then @xmath336 is isotopic to a meridian curve of @xmath326 in @xmath239 .
but this is impossible since @xmath336 also bounds a disk in @xmath330 and is therefore null - homologous in @xmath343 .
the claim follows .
it follows from claim [ parallel ] that there are disjoint arcs in @xmath250 , one , say @xmath344 , which connects the two points of @xmath345 and is disjoint from @xmath330 , and @xmath346 the other .
hence we obtain a `` 2-bridge link '' of two components one fat , one thin in @xmath239 by capping off the `` @xmath52-braid '' in @xmath253 with @xmath344 and @xmath347 in @xmath348 and with @xmath349 and @xmath350 in the @xmath1-ball @xmath351 ( see figure [ filling7](a ) ) . furthermore , since the disk @xmath330 gives a disk bounded by the `` fat knot '' which is disjoint from the `` thin knot '' , the link is a trivial link .
now it follows from the standard presentation of a @xmath37-bridge link as a @xmath52-plat ( see 12.b of @xcite ) , that there is an isotopy of @xmath253 , fixed on the ends @xmath352 and on the two fat strands , taking the `` 4-braid '' to one of the form shown in figure [ filling7](b ) .
hence @xmath227 has the form shown in figure [ filling8](a ) .
the filling rational tangle @xmath353 is of the form shown in figure [ filling8](b ) .
since the component @xmath354 of @xmath355 has to be a trefoil , there are only two possibilities for the number of twists in @xmath260 ; see figure [ filling8b ] .
the two corresponding possibilities for @xmath264 are shown in figure [ filling9 ] .
but these are montesinos links of the form @xmath356 and @xmath357 , respectively .
this final contradiction completes the proof of theorem [ rf ] under the assumptions of case ( a ) of lemma [ gluing ] .
in this case we choose an involution @xmath192 on @xmath136 as shown in figure [ inv3 ] .
then @xmath358 , @xmath359 ( @xmath360 ) , and the restriction of @xmath192 on @xmath138 extends to an involution of @xmath137 whose fixed point set is a core circle of this solid torus .
thus we obtain an involution @xmath192 on @xmath99 .
the quotient of @xmath137 by @xmath192 is a solid torus @xmath361 whose core circle is the branch set .
further , @xmath362 is a longitudinal annulus of @xmath361 .
the quotient of @xmath136 by @xmath192 is also solid torus @xmath363 in which @xmath364 is a longitudinal annulus .
figure [ inv3 ] depicts @xmath363 and its branch set .
it follows that the pair @xmath365 is identical to the analogous pair in section [ case a ] ( see figure [ quo1 ] ) .
next we take @xmath198 to be the same involution on @xmath100 as that used in section [ case a ] .
an argument similar to the one used in that section shows that @xmath192 and @xmath198 can be pieced together to form an involution @xmath215 on @xmath0 . from the previous paragraph
we see that the quotient @xmath366 and its branch set are the same as those in section [ case a ] .
hence the argument of that section can be used from here on to obtain a contradiction .
this completes the proof of theorem [ rf ] in case ( b ) .
|
we show that the distance between a finite filling slope and a reducible filling slope on the boundary of a hyperbolic knot manifold is at most one .
v c u _ 2sl(2,c ) ps .
psl(2,c ) _ _ _ let @xmath0 be a knot manifold , i.e. a connected , compact , orientable @xmath1-manifold whose boundary is a torus .
a knot manifold is said to be _ hyperbolic _ if its interior admits a complete hyperbolic metric of finite volume .
let @xmath2 denote the manifold obtained by dehn filling @xmath0 with slope @xmath3 and let @xmath4 denote the distance between two slopes @xmath3 and @xmath5 on @xmath6 .
when @xmath0 is hyperbolic but @xmath2 is nt , we call the corresponding filling ( slope ) an _ exceptional _ filling ( slope ) .
perelman s recent proof of thurston s geometrisation conjecture implies that a filling is exceptional if and only if it is either reducible , toroidal , or seifert fibred .
these include all manifolds whose fundamental groups are either cyclic , finite , or very small ( i.e. contain no non - abelian free subgroup ) .
sharp upper bounds on the distance between exceptional filling slopes of various types have been established in many cases , including : @xmath7 : : @xmath8 if both @xmath3 and @xmath5 are reducible filling slopes @xcite @xmath7 : : @xmath8 if both @xmath3 and @xmath5 are cyclic filling slopes @xcite @xmath7 : : @xmath8 if @xmath3 is a cyclic filling slope and @xmath5 is a reducible filling slope @xcite @xmath7 : : @xmath9 if @xmath3 is a cyclic filling slope and @xmath5 is a finite filling slope @xcite @xmath7 : : @xmath9 if @xmath3 is a reducible filling slope and @xmath5 is a very small filling slope @xcite
@xmath7 : : @xmath10 if both @xmath3 and @xmath5 are finite filling slopes @xcite @xmath7 : : @xmath10 if @xmath3 is a reducible filling slope and @xmath5 is a toroidal filling slope
@xcite @xcite @xmath7 : : @xmath11 if both @xmath3 and @xmath5 are toroidal filling slopes @xcite in this paper we give the sharp upper bound on the distance between a reducible filling slope and finite filling slope . [ rf ] let @xmath0 be a hyperbolic knot manifold . if @xmath2 has a finite fundamental group and @xmath12 is a reducible manifold , then @xmath8 .
example 7.8 of @xcite describes a hyperbolic knot manifold @xmath0 and slopes @xmath13 on @xmath6 such that @xmath12 is reducible , @xmath14 is finite cyclic , @xmath15 is finite non - cyclic , and @xmath16 .
in fact there are hyperbolic knot manifolds with reducible and finite fillings for every finite type : cyclic , dihedral , tetrahedral , octahedral and icosahedral , in the terminology of [ bz1 ] ; see [ k ] .
a significant reduction of theorem [ rf ] was obtained in @xcite . before describing this work
, we need to introduce some notation and terminology .
denote the octahedral group by @xmath17 , the binary octahedral group by @xmath18 , and let @xmath19 be the usual surjection .
we say that @xmath3 is an _
@xmath20-type _ filling slope if @xmath21 for some integer @xmath22 coprime to @xmath23 and the image of @xmath24 under the composition @xmath25 is @xmath26 . clearly @xmath27 .
it is shown in 3 of @xcite that @xmath28 is independent of the choice of isomorphism @xmath29 .
a lens space whose fundamental group has order @xmath30 will be denoted by @xmath31 . [ almost ] let @xmath0 be a hyperbolic knot manifold . if @xmath2 has a finite fundamental group and @xmath12 is a reducible manifold , then @xmath9 .
further , if @xmath32 , then @xmath33 , @xmath34 , and @xmath3 is an @xmath20-type filling slope for some @xmath35 .
this is theorem 1.1 of @xcite except that that theorem only claimed that @xmath3 is an @xmath20-type filling slope for some @xmath27 . since @xmath36 contains @xmath37-torsion , the argument of the last paragraph of the proof of theorem 2.3 of @xcite ( see page 1026 of that article ) shows that @xmath38 .
thus , in order to prove theorem [ rf ] , we are reduced to considering the case where @xmath33 , @xmath34 , and @xmath3 is an @xmath20 type filling slope for some @xmath38 .
we do this below .
we also assume that @xmath39 in order to derive a contradiction .
an _ essential surface _ in @xmath0 is a compact , connected , orientable , incompressible , and non - boundary parallel , properly embedded @xmath37-submanifold of @xmath0 .
a slope @xmath5 on @xmath6 is called a _
boundary slope _
if there is an essential surface @xmath40 in @xmath0 with non - empty boundary of the given slope @xmath5 . a boundary slope @xmath5 is called _ strict _ if there is an essential surface @xmath40 in @xmath0 of boundary slope @xmath5 such that @xmath40 is neither a fiber nor a semi - fiber .
when @xmath0 has a closed essential surface @xmath41 , let @xmath42 be the set of slopes @xmath43 on @xmath6 such that @xmath41 is compressible in @xmath44 .
a slope @xmath45 is called a _ singular slope _ for @xmath41
if @xmath46 and @xmath47 for each @xmath48 .
since @xmath49 is finite , the first betti number of @xmath0 is @xmath50 , @xmath2 is irreducible by @xcite , and neither @xmath3 nor @xmath5 is a singular slope by theorem 1.5 of @xcite . as @xmath12 is reducible , @xmath5 is a boundary slope .
further , by proposition 3.3 of @xcite we may assume that up to isotopy , there is a unique essential surface @xmath51 in @xmath0 with boundary slope @xmath5 .
this surface is necessarily planar .
it is also separating as @xmath12 is a rational homology @xmath1-sphere , and so has an even number of boundary components .
this number is at least @xmath52 since @xmath0 is hyperbolic .
[ o2 ] if @xmath39 , then @xmath3 is of type @xmath53 .
according to theorem [ rf ] , we must show that @xmath3 does not have type @xmath20 for @xmath54 .
let @xmath55 be the unique non - trivial curve .
( we refer the reader to 6 of @xcite for notation , background results , and further references on @xmath56 character varieties . )
since @xmath5 is not a singular slope , proposition 4.10 of @xcite implies that the regular function @xmath57 , has a pole at each ideal point of @xmath58 .
( we have identified @xmath59 with its image in @xmath60 under the hurewicz homomorphism . ) in particular , the culler - shalen seminorm @xmath61 is non - zero .
hence there is a non - zero integer @xmath62 such that for all @xmath63 we have @xmath64 where @xmath65 is the algebraic intersection number of the two classes ( c.f .
identity 6.1.2 of @xcite ) .
fix a class @xmath66 satisfying @xmath67 , so in particular @xmath68 .
we can always find such a @xmath69 so that @xmath70 according to proposition 8.1 of @xcite , if @xmath71 is a slope satisfying @xmath72 ( mod @xmath28 ) , then @xmath73 .
hence @xmath74 .
consideration of @xmath75 and @xmath76 then shows that @xmath77 . if @xmath39 , then @xmath51 has exactly four boundary components .
we continue to use the notation developed in the proof of the previous lemma . by case 1 of 8 of @xcite
we have @xmath78 and so @xmath62 is either @xmath37 or @xmath1 .
we claim that @xmath79 . to prove this
, we shall suppose that @xmath80 and derive a contradiction .
it follows from the method of proof of lemma 5.6 of @xcite that @xmath21 has exactly two irreducible characters with values in @xmath56 corresponding to a representation @xmath81 with image @xmath17 and a representation @xmath82 with image @xmath83 ( the dihedral group of order 6 ) .
further , @xmath82 is the composition of @xmath81 with the quotient of @xmath17 by its unique normal subgroup isomorphic to @xmath84 .
it follows from proposition 7.6 of @xcite that if @xmath80 , the characters of @xmath81 and @xmath82 lie on @xmath58 and provide jumps in the multiplicity of zero of @xmath85 over @xmath86 .
lemma 4.1 of @xcite then implies that both @xmath87 and @xmath88 are non - trivial . by the previous lemma , @xmath3 is a slope of type @xmath53 .
thus @xmath87 has order @xmath37 . since @xmath89 and @xmath82 factors through @xmath81
, @xmath88 also has order 2 .
next we claim that @xmath69 lies in the kernel of the composition of @xmath82 with the abelianisation @xmath90 . to see this , note first that @xmath5 is non - zero in @xmath91 since @xmath92 .
thus exactly one of @xmath69 and @xmath93 is zero in @xmath94 .
( recall that duality implies that the image of @xmath95 in @xmath94 is @xmath96 . )
since @xmath5 lies in the kernel of @xmath82 , it follows that @xmath88 is sent to zero in @xmath97 .
but then @xmath88 has order @xmath1 in @xmath83 , contrary to what we deduced in the previous paragraph .
thus @xmath79 .
now apply the argument at the end of the proof of proposition 6.6 of @xcite to see that @xmath98 .
hence @xmath51 has four boundary components .
the four - punctured @xmath37-sphere @xmath51 cuts @xmath0 into two components @xmath99 and @xmath100 .
if @xmath101 denotes the copy of @xmath51 in @xmath102 then @xmath0 is the union of @xmath99 and @xmath100 with @xmath103 and @xmath104 identified by a homeomorphism @xmath105 .
the boundary of @xmath51 cuts @xmath6 into four annuli @xmath106 listed in the order they appear around @xmath6 , where @xmath107 are contained in @xmath99 and @xmath108 are contained in @xmath100 .
the arguments given in the proof of lemma 4.5 of @xcite show that for each @xmath109 , the two annuli @xmath110 and @xmath111 in @xmath112 are unknotted and unlinked .
this means that there is a neighbourhood of @xmath113 in @xmath112 which is homeomorphic to @xmath114 , where @xmath115 is a thrice - punctured @xmath37-sphere and @xmath116 is the interval @xmath117 $ ] , such that @xmath118 , and the exterior of @xmath114 in @xmath112 is a solid torus @xmath119 .
we label the boundary components of @xmath115 as @xmath120 ( @xmath121 ) so that @xmath122 for @xmath123 .
let @xmath124 be the two sphere in @xmath12 obtained from @xmath51 by capping off @xmath125 with four meridian disks from the filling solid torus @xmath126 .
these disks cut @xmath126 into four @xmath37-handles @xmath127 ( @xmath128 ) such that the attaching annulus of @xmath127 is @xmath129 for each @xmath130 .
let @xmath131 be the manifold obtained by attaching @xmath127 to @xmath112 along @xmath129 ( @xmath123 ) .
then @xmath132 is a once - punctured @xmath133 and @xmath134 a once - punctured @xmath135 .
it follows from the description above that @xmath99 is obtained from @xmath136 and @xmath137 by identifying @xmath138 with an annulus @xmath139 in @xmath140 whose core curve is a @xmath141 curve in @xmath140 .
we can assume that @xmath139 is invariant under the standard involution of @xmath137 whose fixed point set is a pair of arcs contained in disjoint meridian disks of @xmath137 .
note that the two boundary components of @xmath139 are interchanged under this map .
similarly , @xmath100 is obtained from @xmath142 and @xmath143 by identifying @xmath144 with an annulus @xmath145 in @xmath146 whose core curve is a @xmath147 curve in @xmath146 .
again we can suppose that @xmath145 is invariant under the standard involution of @xmath143 which interchanges the two boundary components of @xmath145 .
see figures [ inv1 ] and [ inv2 ] .
the map @xmath148 is constrained in several ways by our hypotheses .
for instance , the fact that @xmath6 is connected implies that @xmath149 for some @xmath150 .
other conditions are imposed by the homology of @xmath0 .
[ gluing ] we can assume that either + @xmath151 @xmath152 , and @xmath153 , or + @xmath154 @xmath155 and @xmath156 . without loss of generality
we can suppose that @xmath157 .
hence , as @xmath6 is connected , one of the following four possibilities arises : * @xmath158 and @xmath153 .
* @xmath159 and @xmath156 . * @xmath160 and @xmath161 , or * @xmath162 and @xmath153 .
let @xmath163 be represented , respectively , by @xmath164 , @xmath165 , and a core of @xmath119 , @xmath166 .
then @xmath167 is the abelian group generated by @xmath168 , subject to the relation @xmath169 since @xmath170 , we may orient @xmath171 , @xmath172 so that in @xmath173 we have @xmath174 .
then @xmath173 is the quotient of @xmath175 by this relation together with the additional relations corresponding to the four possible gluings : * @xmath176 , @xmath177 * @xmath178 , @xmath179 * @xmath180 , @xmath176 * @xmath181 , @xmath177 taking @xmath182 coefficients , equation ( i ) allows us to eliminate @xmath183 , while ( ii ) gives @xmath184 .
hence @xmath185 , where the @xmath182 summand is generated by @xmath186 and @xmath187 is defined by generators @xmath188 , and relations @xmath184 plus those listed in ( a ) , ( b ) , ( c ) and ( d ) above .
thus @xmath189 in cases ( a ) and ( b ) , and @xmath190 in cases ( c ) and ( d ) .
since @xmath191 , we conclude that cases ( c ) and ( d ) are impossible .
|
land surface temperature ( lst ) is a vitally important remotely observable tracer of mass and energy exchange across the interface between the atmosphere and the ground .
lst derived from satellite observations is of interest in its own right for local climate studies and climate change monitoring , and as a component of studies of land cover , land cover change , surface moisture , and precision farming , among others @xcite .
unfortunately , few surfaces at the bottom of the atmosphere radiate as blackbodies , and the parameterization of their surface state by means of emissivities complicates the task of accurate temperature determination .
this paper continues the development of bayesian lst estimators that do not require accurate knowledge of surface emissivity , given radiance in multiple bands @xcite .
the approach may be considered as complementary to the widespread use of regression - law based split - window algorithms .
section [ section2 ] reviews the elements of the bayesian lst formalism .
two implementations of the bayesian approach have been investigated .
the first is the iterative contraction mapping algorithm presented in @xcite .
the second is a maximum - a posteriori approach to lst estimation .
section [ section3 ] presents bayesian estimates of lst and band emissivity for modis imagery .
section [ section4 ] discusses results , and a final section [ section5 ] offers conclusions .
appendix a presents a proof of convergence for the iterative algorithm , and estimates of the effect on lst retrieval of uncertainty in aerosol loading .
we begin by recalling some of the results obtained in @xcite , which may be consulted for details .
the probability of sensing radiance @xmath7 at the top of the atmosphere ( toa ) , given @xmath8 and an estimate @xmath9 of the variance of the sensor radiance noise , is obtained by a maximum entropy ( maxent ) argument .
the maxent estimator assumes the existence of a forward model for the toa radiance in terms of @xmath10 and @xmath11 . for the case of a sensor aperture radiance forward model @xmath12 that is _ linear _ in band emissivity @xmath11 ,
the posterior probability @xmath13 is gaussian in the mismatch between @xmath7 and @xmath12 in each band @xmath14 .
we also have @xmath15 as the posterior probability for all bands jointly . given radiance @xmath7 in n bands detected at the top of the atmosphere ( toa ) that originates from a patch on the earth s surface by a sensor with noise radiance @xmath9 , the posterior probability that the surface is at a temperature t is given by bayes theorem as @xmath16 equation ( [ eqn3 ] ) is evaluated with aid of the prior probability for the surface to be at temperature t and emissivity @xmath11 given available knowledge @xmath17 , @xmath18 the result is a posterior probability density .
if one is ignorant of , or unconcerned with , the value of emissivity , it is possible to marginalize on the nuisance variable @xmath11 : @xmath19 and @xmath20 may be expressed in closed form using error functions .
the explicit form is given in @xcite .
we desire to find an estimator for surface temperature using the posterior probability given by bayes theorem ( [ eqn3 ] ) .
two lst algorithms have been developed .
the first , recounted in @xcite , iteratively computes expectation values of lst and band emissivities from their posterior probabilities while contracting the range of _ a - priori _ limits on these parameters .
the second calculational method is search for the maximum _ a - posteriori _ probability ( map ) value of lst that maximises @xmath21 .
the motivation for investigating a map formulation for the bayesian lst estimator is simple .
the original bayesian lst algorithm presented in @xcite obtains estimates of lst and band emissivity by iterative refinement of expectation values for these quantities . even though the posterior probability for surface t is a closed - form expression
, this procedure is computationally intensive .
moreover , extensions to the estimator ( such as , for example , a more careful treatment of the forward model , or incorporation of calibration error effects @xcite ) may deprive us of the comfort of the closed - form solution .
the map criterion for an estimator promises to dramatically reduce the number of cpu cycles expended per lst estimate , and may yet preserve the ability to rely upon closed - form solutions .
map lst estimates are obtained simply and quickly by a golden section search .
it was conjectured in @xcite that the iterative algorithm converges in general .
appendix a presents a proof that the iterative algorithm converges , and , in the limit , converges to the map estimate of lst .
granted the robustness of the golden section search , the existence of a solution for the map estimator may not appear to be a concern .
however , the map algorithm in general includes iterative refinement of the _ a - priori _ limits on surface emissivity .
the existence of a unique map estimate for lst in that case follows from a straighforward adaptation of the method of proof given in the appendix .
a desirable feature of an lst estimator is some means of assessing one s confidence in the accuracy of individual estimates . in the present case lst estimates
are obtained by marginalizing over imperfectly known surface emissivities .
it is naturally of great interest to explore means of judging the effect of an uninformative range for the _ a - priori _
probability for @xmath11 .
direct numerical integration of the posterior probability @xmath21 with respect to @xmath10 was used to calculate emissivity prior confidence intervals for lst retrievals presented in this study .
suppose that lst uncertainties arising from an uninformative emissivity prior are independent of those originating from other sources of error
. then the emissiviy prior confidence level may be combined with an estimate of the net effect of all other sources of error by adding in quadrature to obtain an overall lst uncertainty estimate .
both the map and iterative versions of the bayesian algorithm have been applied to modis imagery .
this effort took advantage of the availability of level 1b calibrated radiances and level 2 land surface temperatures ( used in lieu of ground truth ) for terra modis data granules through the goddard space flight center ( gsfc ) eos / daac website @xcite .
level 2 atmospheric profiles were obtained from the same site .
the level 2 surface temperature product is used in preference to the standard level 3 product because it is in the same 1 km swath format as the level 1 b radiance data .
the gsfc - supplied hdflook display package was used to manipulate the data granules and associated standard products .
four modis data granules were selected for comparison of bayesian lst estimates with the mod11_l2 lst product . to the extent possible , granules were chosen to have large areas with gentle topographic relief and minimal cloud cover effects .
a list of the granules used in this study appears in table 1 .
two granules are daytime , and two nighttime . the nighttime granule for 2006 day - of - year ( doy )
258 includes the alluvial plains of iraq and both the coastal plains and central plateau of saudi arabia .
2006 doy 268 , also nighttime , includes the great plains portion of central canada .
2006 doy 350 is a daytime dataset covering northeast africa , including large plains in sudan .
2006 doy 347 , also daytime , covers much of the plains of central asia .
the mod11_l2 product reported lst values for approximately @xmath22 or more of the pixels in all four granules .
cloud - free pixels for which level 2 modis lst values were reported were selected for analysis .
a number of cuts were made in selecting data from the modis products for surface temperature estimation .
the use of level 1b radiance products combined with level 2 atmospheric profiles to generate lst for comparison with a level 2 modis temperature product introduces sources of error originating in , among others , imperfect georeferencing . thus one may expect errors in surface altitude to affect retrieval accuracy .
these effects may be exacerbated by use of bilinear interpolation of 4 km horizontally reported profiles to the 1 km sampling of the spectral radiances . in order to minimize this source of potential error ,
lst retrieval was limited to sensor zenith angles @xmath0 .
cloudy pixels were masked by rejecting pixels for which the lst quality control mask bits 0,2,3,4 , or 5 were nonzero .
lst retrievals were attempted only for pixels for which the dew point temperature at the base of the water vapor profile exceeded the freezing point of water .
cuts were also made on the basis of nonphysical or otherwise pathological fields in the granule data : pixels for which the modis lst , the modis lst error , the surface temperature field in the profile granule ( otherwise not used in the analysis ) , the surface air temperature at the lowest level of the profile , or the surface pressure were less than zero , were eliminated from processing .
finally , pixels for which surface pressure exceeded 1000 millibars , but for which the surface altitude was less than zero , were omitted from the analysis .
neither ground truth nor surface meterological range ( @xmath23 ) is available as a standard product with modis granules .
however , the presence of lwir bands in the split - window algorithm used to produce the modis lst product@xcite suggests that lst retrievals should not be too sensitive to the exact value assumed for vsby in the forward model .
accordingly , this study replicates modis lst values assuming a nominal , but fairly large , value of @xmath24 for the surface @xmath23 parameter for clear pixels .
appendix b presents a calculation which supports the contention that , for purposes of the present study , this approximation should not introduce large errors for clear air pixels .
a random selection of 500 pixels was made from each granule , subject to data cuts described above .
lst retrievals from both single - pass map and two - pass iterative versions of the algorithm were obtained for all granules . in addition
, @xmath3 confidence intervals confidence levels for lack of emissivity knowledge were chosen to approximate the equivalent of @xmath25 intervals , given that @xmath21 generally displays some skewness .
this should be borne in mind when when comparing with the effect of other error sources described by a standard deviation . ]
were computed for the map retrievals by numerical integration of the posterior probability distribution .
the bandset used in the retrievals appears in table 2 .
the moderate resolution atmospheric transmission ( modtran4 ) code @xcite was used for the forward model calculations , as described in @xcite .
the forward models incorporated mod07_l2 air temperature and water vapor profiles and assumed a @xmath26 mixing ratio of @xmath27 ppm . initially , the prior probability in ( [ eq : prior ] ) was assumed to hold over the range @xmath28 for @xmath10 and @xmath29 in each band .
however , it proved advantageous to restrict the range of the emissivity prior for modis bands 31 and 32 , @xmath30 so as to take advantage of the observation in ( wan and dozier 1996 , wan 1999 ) that most natural land cover types have an emissivity in excess of @xmath31 in these bands . in the simulated lst retrievals presented in @xcite ,
the equivalent noise radiance @xmath9 in each band was parameterized by a signal - to - noise ratio ( snr ) value obtained by approximately inverting the noise - equivalent temperature change ( @xmath32 ) for the specified nominal modis performance .
experimentation with actual modis data revealed that the snr values so chosen greatly underestimated the random variance in radiance measured by the actual terra spacecraft against real terrestrial surfaces , through a real , and sometimes dirty , atmosphere . in consequence ,
the snr parameter was reduced from values characteristic of ideal performance , in which the only source of radiance error in sensor measurements is accounted for by @xmath32 , by a full order of magnitude .
in fact , the label `` snr '' is probably somewhat misleading , since the random variability it describes is likely the concatenation of numerous confounding factors , not all of which need accurately described as `` noise '' .
it is retained because of its use in earlier studies .
the snr values used in all retrievals reported in this study appear in table 2 .
these values were selected on the basis of trial - and - error with pixels from doy 347 that were drawn independently from the selection used for comparison with the modis lst product .
lst estimates are fairly insensitive to the exact values chosen for band snr s , so long as these are fairly large . if the snr values are too large , the noise variance estimate becomes so tight that the retrieval becomes pathologically overconstrained . in the most commonly observed pathology ,
one or more of the individual band posterior probability fails to overlap the others ; _
i. e. _ the support of the joint posterior probability becomes the null set .
if snr is too small , @xmath33 for modis , lst retrievals become badly biased towards underestimates , and confidence intervals become worrisomely large .
tables 3 and 4 present the mean , standard deviation and @xmath34 of the mismatch for each granule for map and iterative retrievals , respectively . in table 3 ,
summary statistics for lst are shown for @xmath35 as well as for @xmath36 .
a small number of pixels required relaxation of the snr parameter or emissivity priors in order to obtain an iterative lst estimate , in a pattern similar to that reported in @xcite .
figures 1 - 8 display plots of the mismatch between map and modis lst values , with @xmath3 confidence intervals , and of mismatch histograms , for 500 modis pixels from each of the four data granules .
overplotted on each mismatch histogram is the histogram for the equivalent gaussian error distribution @xmath37 note that the confidence intervals in the plots ( whose mean and standard deviation for each granule is given in the lst mismatch figure captions ) are for the uncertainty in lst resulting from the assumed uncertainty prior for surface emissivity only .
thus , they do not ( directly ) reflect sensor noise effects or other sources of uncontrollable variance .
agreement between the level 2 modis lst product and map lst estimates obtained in this study is better than @xmath2 in both mean and standard deviation of the mismatch for all granules . on average ,
the mismatch with modis lst in table 3 increases only slightly for the looser emissivity prior .
in addition to the tendency for the bayes lst retrievals to display a slight negative bias with respect to the modis product ( consistent with the observation made in @xcite ) the mismatch plots show evidence of a positive trend in the discrepancy with respect to modis lst .
the trend is readily apparent to the eye for the daytime retrievals , but can also be discerned in figure 3 for ( nighttime ) doy 268 .
map and iterative solutions reproduce modis lst comparably well .
only the doy 347 retrieval shows a notably different mean mismatch compared to modis between iterative and map approaches .
for both algorithms , @xmath34 per degree of freedom ( @xmath38 ) is close to unity .
the best estimate of modis lst accuracy lies in the range @xmath39 to @xmath40 @xcite .
the standard deviation of the map temperature mismatch in table 3 also lies in this range . a t - test shows the difference in means between modis and map lst values to be statistically significant for both nighttime granules and marginally ( at the @xmath41 level ) for doy 347 .
however , the generally small level of the discrepancy indicates that , given better control of error sources , the bayesian map algorithm is capable of replicating the accuracy of the modis mod11_l2 product .
a small number of retrievals for both algorithms have large discrepancies with respect to mod11_l2 lst .
the cosmetic effect of these retrievals in the mismatch plots may give an exaggerated impression of their effect on the accord between modis and bayesian lst .
judging from the mismatch histograms , they appear to be outliers resulting from pixels whose radiometric accuracy has been compromised by one or more of the error sources described earlier , despite imposition of the data cuts . in the case of the granule for doy 2006/350 ,
two pixels out of 500 have a discrepancy in excess of @xmath42 .
these discrepant values appear significantly out - of - family with respect to the largely gaussian appearance and favorable @xmath34 statistic for the retrievals as a whole . in table 3 ,
the summary statistics for doy 350 are replicated with the two worst outlier pixels removed in the rows marked with an asterisk , with minimal change to the results .
error sources arising from georeferencing errors in level 1b radiance products have already been discussed .
an additional possible origin for the discrepancies with modis lst is errors in the forward model . among other causes , these could result from the fixed choice of @xmath43 ( _ vide .
_ appendix b , however ) , or occur as a consequence of using erroneous values of air temperature or water vapor from the mod07_l2 profiles in modtran ( wan et al .
another posible source is imperfect cloud masking in the level 2 modis products that could corrupt pixel radiances .
this last source of error would afflict modis lst as well , and would have to affect modis and bayesian lst differentially in order to contribute to discrepancies . finally , the assumed band emissivity prior range need not always bracket the actual surface emissivity for a pixel .
a noteworthy feature of the lst retrievals reported for this study is the comparatively small uncertainty for _ individual _ lst estimates originating in the assumed lack of surface emissivity knowledge .
the mean value of the @xmath3 interval for individual granules is greatest for doy 350 ( daytime ) at @xmath44 and least for doy 268 ( nighttime ) at @xmath45 .
as noted earlier , the emissivity uncertainty confidence interval depends upon the assumed level of random variance due to noise and other uncontollable factors capable of influencing lst retrieval .
however , insensitivity of bayesian lst to emissivity seems likely to hold generally . in particular ,
confidence intervals for @xmath46 differ little from those for @xmath47 it was remarked earlier that the bayesian approach to lst estimation was complementary to the more common split - window algorithmic family .
the modis generalized split - window algorithm that furnished lst values against which performance of the bayes approach has been tested relies upon regression against a large training dataset , and requires assignment of pixels to emissivity classes . the bayesian approach , while statistical in its treatment of emissivity uncertainty , relies upon radiative transfer theory , rather than regression of brightness temperatures against true lst , to invert radiance to surface temperature .
it is thus a method of remotely determining lst that ( so to speak ) minimizes hostages to the fortunes of emissivity knowledge .
the accurate reproduction in the mean of mod11_l2 lst in this study suggests the possibility of using the bayesian approach for virtual `` ground truth '' to train local or regional split - window lst algorithms .
although the golden section search for the map version runs faster than the iterative solution in protoype idl ( interactive data language ) code , the difference in execution time is not great , and is in any event dominated by the execution time for the forward model calculations in modtran .
however , the map version of the algorithm would seem preferable both on the basis of simplicity and because it readily admits extension to higher - dimensional searches for optimal joint estimates of lst and band emissivity .
the performance of the bayesian land surface temperature retrieval algorithms has been tested in a small - scale study by comparison of bayesian lst estimates against the modis mod11_l2 lst product .
both the original iterative contraction mapping search and a new maximum _ a - posteriori _ algorithm reproduce modis lst values with good accuracy . the mean mismatch with respect to the mod11_l2 lst product for maximum _ a - posteriori _ estimates does not exceed @xmath48 , and its standard deviation is less than @xmath2 , for any granule used in this study . for the iterative algorithm , the corresponding mean mismatch does not exceed @xmath49 , nor its standard deviation , @xmath50 .
the agreement of the map estimates derived from modis level 1b radiance with the mod11_l2 lst is comparable to the claimed accuracy for modis lst .
emissivity prior confidence intervals for individual lst estimates shown in section [ section4 ] show that lst retrieval accuracy with the bayesian approach is insensitive to modest uncertainties in surface emissivity .
we conclude the map lst algorithm is a viable candidate for lst retrieval in circumstances in which band emissivities are poorly known .
it has been shown ( _ vide .
_ appendix a ) that the iterative refinement of expectation values underlying the tes algorithm presented in @xcite converges , in general . in the limit of vanishing support for the lst expectation values , the iterative approach is equivalent to the maximum _ a - posteriori _
estimator for lst .
_ acknowledgement : _ i wish to thank an anonymous referee for @xcite for suggesting the maximum-_a - posteriori _ approach .
the algorithm used in @xcite iteratively refines the calculation of the expectation values ( [ eqn61])-([eqn63 ] ) below of surface temperature and emissivity in multiple bands by systematic contraction of the limits of integration .
it has been found that this procedure , which is described in detail in ( morgan 2005 ) , converges rapidly and reliably in practice .
however , to date there has been no proof that the algorithm actually converges in general .
this appendix supplies such a proof . for reference ,
the expectation values for t given radiance in bands i , obtained from the joint posterior probability for observing radiances @xmath51 are @xmath52 assuming @xmath10 is known to lie between a minimum and a maximum , while a joint estimator for t given radiances in all n bands is @xmath53 an estimator for the emissivity in band @xmath14 is given by @xmath54 both surface temperature and band emissivities are assumed to lie within limits imposed _
a - priori_. the surface temperature is limited to the range @xmath55 and emissivity to the range @xmath56 where the minimum and maximum values may be band - dependent .
it is assumed that the maximum of @xmath21 is always included in the integration . in any workable procedure for iteratively refining surface temperature or emissivity by contracting the support @xmath57 $ ] of the integrations ( [ eqn61]-[eqn63 ] ) which define @xmath58 and @xmath59 , that support must bracket the maximum of @xmath21 as it contracts .
otherwise , the most probable values of @xmath58 and @xmath59 would be excluded from the search , in violation of the maxent assumption that expected values should maximize the likelihood .
_ proposition _ : the iterative refinement of ( [ eqn61])-([eqn62 ] ) converges . for clarity
we limit the discussion to ( [ eqn61 ] ) and ( [ eqn62 ] ) .
the extension to accomodate ( [ eqn63 ] ) is clear .
a total of @xmath60 expectation values for surface temperature @xmath61 ( the index @xmath14 now including the joint estimate ) is obtained at each stage of iteration . the values @xmath61 comprise a vector in @xmath62 .
these are confined within the subset of @xmath62 delimited by ( [ eqn : tlims ] ) that provides nonvanishing support for the integrations in the definitions ( [ eqn61 ] ) and ( [ eqn62 ] ) .
call the vector of expectation values @xmath63 . at the @xmath64-th stage of iteration ,
@xmath63 is calculated over the range @xmath65 $ ] and the mismatch amongst the band lst expectation values @xmath61 is scored by @xmath66 taken over all pairs of estimates .
let the set @xmath67 be the set of all @xmath68-tuplets of the form @xmath69 consistent with the overall limits ( [ eqn : tlims ] ) . here
@xmath70 where the quantity in parentheses is the binomial coefficient .
the set @xmath71 is compact , so all components of any sequence @xmath72 are bounded .
( they are all nonnegative , as well . )
given points @xmath73 define a metric function @xmath74 the pair ( @xmath75 ) can be shown to be a complete metric space @xcite .
define the function @xmath76 by @xmath77 where @xmath78 .
consider a contractive sequence @xmath79 in @xmath71 with @xmath80 for which @xmath81 define the mapping from one iterate of @xmath82 to another in @xmath71 by @xmath83 the function @xmath76 is lower semicontinuous , and @xmath84 is continuous , on @xmath85 . by caristi s fixed - point theorem @xcite the mapping @xmath86 has a fixed point @xmath87 .
the iterative refinement of @xmath58 therefore converges in @xmath71 .
moreover , if @xmath88 then @xmath89 by the definition of @xmath63 .
@xmath90 the fixed point of the iterative mapping drives @xmath91 , returning an @xmath60-tuple of lst expectation values , none of which differ from any of the others by an amount greater than @xmath92 .
it may appear that nothing in the result just obtained says any such estimate necessarily gives an accurate value for lst . in practice , however , it happens that a small number of iterations suffices to yield accurate lst estimates .
the manner in which @xmath93 contracts remains at our disposal .
we are at liberty to force it to tend to zero .
we now show that @xmath6 is equivalent to a contractive mapping chosen with @xmath94 rather than ( [ eqn : contraction ] ) , and that as @xmath95 @xmath61 converges to the maximum _ a - posteriori _
estimate @xmath96 .
the support for the computation of @xmath58 tends to zero as @xmath97 .
there is thus a nested sequence of intervals @xmath98 \subset [ t^{n}_{min},t^{n}_{max } ] $ ] .
cantor s theorem on the intersection of a nested sequence@xcite implies the limit of this process will be a singleton in @xmath85 .
it was assumed at the outset that the support of integrations @xmath65 $ ] in ( [ eqn61]-[eqn62 ] ) brackets @xmath96 ; the singleton thus corresponds to @xmath99 , which is also the value of @xmath100 .
we may see this last by the following argument : in @xmath101 as @xmath102 , we may write @xmath103 with @xmath104 as @xmath6 , it can be shown @xcite that @xmath105 and thus @xmath106 as @xmath97 , @xmath107 and the other expectation values @xmath108 , a unique converged surface temperature estimate .
conversely , if @xmath109 , @xmath110 by applying the preceding argument to each expectation value separately , and @xmath97 .
then a subsequence @xmath111 exists that respects ( [ eqn : recontraction ] ) .
we thus have the _ corollary _ : in the limit @xmath109 , iterative refinement of unbiased expectation values , _
i.e. _ , those for which @xmath112 brackets the maximum of the @xmath21 , is equivalent to the maximum _ a - posteriori _ ( map ) approach , in which one searches directly for the value of @xmath10 which maximizes the joint posterior probability ( [ eqn60 ] ) .
@xmath90 this result is likewise true of the straightforward extension of the proof given here to the case of a contractive iterative search for a fixed point for @xmath113 .
the contribution of aerosol loading of the atmosphere to the forward model radiance calculations was parameterized by specifying the surface meteorological range , or visibility ( @xmath23 ) .
this is the variable used by modtran to scale aerosol extinction .
the meteorological range was assumed to be @xmath24 for all retrievals presented in the main body .
because lst retrievals were only attempted for cloud - free pixels in what were judged to be reasonably clear air conditions , it seems reasonable to suppose that this assumption should not introduce significant error so long as few pixels have @xmath114 .
we now examine the likely validity of this assumption .
we begin by adapting the analysis of @xcite to obtain the prior probabilty for surface meteorogical range @xmath23 and use it to construct posterior probabilities marginalized over a range of @xmath23 .
the starting point is the apparently trivial observation that the meteorological range is a _ length _ : traditionally , and loosely , @xmath23 is the distance at which an observer looking horizontally at a height of two meters just fails to discern the contrast presented by an object against the horizon . in the remote sensing of objects by electromagnetic radiation
, poincar invariance requires that any two observers must be able to relate their description of events by a lorentz transformation .
in particular , the relation between @xmath115 as observed at time @xmath116 in primed and unprimed coordinate systems moving with relative velocity @xmath117 with respect to each other along a stipulated line of sight ( which might as well be taken to be parallel to the ground plane ) is given by @xmath118 @xmath119 where @xmath120 the two observers will assign prior probabilities for the occurence of particular values of the meteorological range @xmath121 in the unprimed system and @xmath122 in the primed one . in order for observers to agree as to the form of the estimator
, the priors must be related by @xmath123 where @xmath124 is the jacobian for the transformation between descriptions in the parameter space . primed and
unprimed observers , if equally cogent , must also agree as to the functional form of the prior probability for @xmath125 , thus @xmath126 leading to a form of schrder s equation for @xmath127 , @xmath128 with solution @xmath129 for the prior probability . we conclude that the @xmath23 parameter , with dimensions of a length , is a scale parameter , for which the prior probability takes the jeffreys form ( [ eqn44 ] ) @xcite .
the prior probability ( [ eqn44 ] ) may be inserted into an estimator for lst which marginalizes on both surface emissivity and @xmath23 in each band i , @xmath130 from which a joint posterior probability may be calculated as a function of lst .
this quantity may be used to compute map estimates of lst and confidence intervals .
the estimator will now be used to examine the sensitivity of lst estimates when @xmath23 was allowed to vary over a range of values by means of an illustrative sample calculation .
lst map retrievals and confidence intervals as a function of @xmath131 for one of the pixels used in the doy 350 granule comparison with modis lst appear in figure 9 .
the @xmath3 confidence intervals shown account for the effects of the @xmath23 and emissivity priors only , and include no estimate of uncertainty originating from noise effects .
the upper limit on the visibility prior was fixed at @xmath132 . the surface t prior for these retrievals was constrained to the range @xmath133 .
the values of lst and associated confidence intervals for this example can not be directly compared with the map value 311.79 k marginalized on surface emissivity alone .
however , note that the range of lst variation in figure 9 is limited , even within the constrained lst prior limits . moreover , the confidence intervals tend to a fixed range ( consistent with the modis lst value ) as @xmath23 increases .
these observations provide a degree of empirical support for the assumption that the lst retrievals presented in the main text should not be too sensitive to the exact value of @xmath23 used to parameterize aerosol loading in the forward model , so long as the actual @xmath23 is great enough .
berk , a. , g. p. andersen , p. k. acharya , j. h. chetwynd , l. s. bernstein , e. p. shettle , m. w. matthew , and s. m. adler - golden , 1999 , _
modtran4 user s manual _ , air force research laboratory , hanscom afb , ma dash , p. , f .- m .
gttsche , f .- s .
olesen , and h. fischer , 2002 , land surface temperature and emissivity estimation from passive sensor data : theory and practice - current trends , _ int .
j. remote sensing , 23 _ , pp .
2563 - 2594 guenther , b. , x. xiong , v. v. salmonson , w. l. barnes , and j. young 2002 , on - orbit perfromance of the earth observing system moderate resolution spectroradiometer ; first year of data , _ rem .
83 _ , pp .
16 - 30 kniezys f. , d. c. robertson , l w abreu , p. acharya , g. p. anderson , l. s. rothman , j. h. chetwynd , j. e. a. selby , e. pl shettle , w. o. ballery , a. berk , s. a. clough , and l. s. bernstein , 1996 , _ the modtran 2/3 report and lowtran 7 model _ , phillips laboratory , hanscom afb , ma savtschenko , a. , d. ouzounov , s. ahmad , j. acker , g. leptoukh , j. kosiana , and d. nickless , 2004 , terra and aqua modis products available from nasa ges daac , _ advances in space research , 34 _ , pp . 710 - 714 + mod021km.a2006258.1900.005.2006260185940.hdf + mod021km.a2006268.0455.005.2006270054543.hdf + mod021km.a2006347.0615.005.2006348125833.hdf + mod021km.a2006350.0830.005.2006352011824.hdf + .................................................................................... + mod11_l2.a2006258.1900.004.2006259185248.hdf + mod11_l2.a2006268.0455.004.2006269162220.hdf + mod11_l2.a2006347.0615.004.2006349135340.hdf + mod11_l2.a2006350.0830.004.2006351140447.hdf + .................................................................................... + mod07_l2.a2006258.1900.005.2006260192510.hdf + mod07_l2.a2006268.0455.005.2006270062431.hdf + mod07_l2.a2006347.0615.005.2006348132045.hdf + mod07_l2.a2006350.0830.005.2006352015143.hdf + table 3 .
a - posteriori _ surface temperature results : mean mismatch , standard deviation , maximum mismatch , and @xmath34 per degree of freedom ( dof ) .
results are given for two choices for the lower limit to the emissivity prior in bands 31 and 32 .
the rows labeled with an asterisk differs from those immediately above it by omission of the pixels with the worst two mismatches compared to modis lst + @xmath36 : + @xmath138@xmath139 @xmath140 @xmath141 + @xmath142 @xmath143 @xmath144 @xmath145 + @xmath146 @xmath147 @xmath148 @xmath149 + @xmath150 @xmath151 @xmath152 @xmath153 + @xmath154 @xmath155 @xmath156 @xmath157 + .......................................................................................... + @xmath35 : + @xmath138@xmath158 @xmath159 @xmath160 + @xmath142 @xmath161 @xmath162 @xmath163 + @xmath146 @xmath164 @xmath165 @xmath166 + @xmath150 @xmath167 @xmath168 @xmath169 + @xmath154 @xmath170 @xmath171 @xmath172 +
|
an approach to land surface temperature ( lst ) estimation that relies upon bayesian inference has been tested against multiband infrared radiometric imagery from the terra modis instrument .
the algorithm employed requires minimal knowledge of surface emissivity , starting from a parsimoniously chosen ( hence , uninformative ) range of prior band emissivity knowledge .
two estimation methods have been tested .
the first is the iterative contraction mapping of joint expectation values for lst and surface emissivity described in a previous paper . in the second method ,
the bayesian algorithm is reformulated as a maximum _ a - posteriori _ ( map ) search for the maximum joint _ a - posteriori _
probability for lst , given observed sensor aperture radiances and _ a - priori _
probabilities for lst and emissivity .
two modis data granules each for daytime and nighttime were used for the comparison .
the granules were chosen to be largely cloud - free , with limited vertical relief in those portions of the granules for which the sensor zenith angle @xmath0 .
level 1b radiances were used to obtain 500 lst estimates per granule for comparison with the level 2 modis lst product .
the bayesian lst estimators accurately reproduce standard modis product lst values . in particular , the mean discrepancy for the map retrievals
is @xmath1 , and its standard deviation does not exceed @xmath2 .
the @xmath3 confidence intervals for individual lst estimates associated with assumed uncertainty in surface emissivity are of order @xmath4 .
the appendix presents a proof of convergence of the iterative contraction mapping algorithm .
the expectation values of surface temperature in multiple bands , and jointly in all bands , converge to a fixed point , within a stipulated convergence criterion . provided the support @xmath5 $ ] for the calculation of the expectation value brackets the maximum in the joint posterior probability for lst , the fixed point coverges to the map lst estimate in the limit @xmath6 .
|
previous to about 1980 , the general paradigm for globular star clusters was that they were simple stellar populations , that is all stars had uniform chemical composition and were all the same age .
however , since that time , numerous spectroscopic studies have shown that many of these clusters exhibit chemical composition variations among their stars likely caused by h - burning via the hot cno cycle ( see gratton et al .
2004 for a nice review ) .
recent imaging observations with the hubble space telescope ( hst ) that produced exquisitely precise photometry has extended this picture to many if not all clusters ( see piotto 2009 for a recent review ) .
however , what is currently lacking is detailed insight into the manner of formation of the various stellar populations in such a cluster .
key input could potentially come from the observation of different dynamics or spatial distributions of the various populations .
sollima et al . (
2007 ) did find radial variations among blue and red main sequence ( ms ) stars in @xmath0 cen , but this observation was not coupled with any proper motion ( pm ) information .
the ms of 47 tuc is broader than observational error alone can explain .
this was first pointed out by anderson et al .
( 2009 ) and examined in some detail by milone et al .
( 2012a ) who clearly demonstrated that the cmd width could be explained as a second - generation population , making up 70% of the total , enriched in both he and n and depleted in c and o. vesperini et al .
( 2013 ) , using n - body simulations , have explored the behavior of first- and second - generation stars in a multi - population cluster . in general
they find that complete mixing of the two populations does not occur until the cluster is well advanced dynamically ; that is when it has lost upwards of 70% of its mass due to two - body relaxation .
giersz and heggie ( 2011 ) performed a monte carlo simulation of 47 tuc and concluded that it has likely lost only about 45% of its initial mass .
we can thus expect that 47 tuc s various populations will not yet be well - mixed and that it might be possible to observe either kinematic and/or spatial differences amongst its various stellar populations .
indeed there already exists a wealth of observations on the radial distributions of various stellar populations going back over 30 years ( e.g. norris and freeman 1979 , milone 2012a ) most of which suggest that stars exhibiting evidence for cn - enhnacement are more centrally concentrated than cn - poor stars .
coupling this with kinematic observations could well provide critical clues to the multiple population scenario . in 2 below
we discuss the observations relevant to the current study and follow this in
3 with an exploration of the pm kinematics of cluster stars yielding strong evidence for differences amongst stars of differing ms colors ( chemical compositions ) . 4
presents a search for radial differences and 5 discusses possible formation scenarios for the various stellar populations based on their currently observed motions and distributions .
our team was awarded 121 hst orbits in cycle 17 to image 47 tuc ( go-11677 ) .
the main science goal was to obtain photometry with the acs f606w and f814w filters that would go deep enough to study the entirety of the white dwarf cooling sequence .
a detailed discussion of the observations can be found in kalirai et al.(2012 ) . we have supplemented these observations with f606w and f814w images in the archive taken between 2002 and 2012 with acs / wfc and wfc3/uvis .
these images were taken at a very wide variety of offsets , orientations , and exposure times .
sources were found in each image independently and measured with a library psf that had been constructed from other data sets ( see anderson and king 2006 , ak06 ) .
positions were corrected for distortion using the bellini et al .
( 2011 ) solution for wfc3/uvis and the ak06 solution for acs , with an adjustment for the fact that the acs solution has changed slightly since sm4 .
we collated these many starlists in a reference frame based on that described in ak06 .
we identified the member stars from the cmd and used their positions in the reference frame and the individual frames to define a linear transformation from each 2048x2048-pixel amplifier into the reference frame . for each star ,
this gives us a reference - frame position for each exposure , and a linear fit can be made to this time series to compute the pms and determine their errors .
we iterated the procedure , which allowed us to do a better job of outlier rejection in the determination of the transformations .
the result of this process is a list of average positions and pms ( and their errors ) for each star in the list .
since the field was observed at a wide variety of pointings and orientations over @xmath1 years , we would expect most minor issues such as errors in the distortion solution or errors in the pixel - based cte correction to average out to zero and introduce only random errrors , which would be reflected in our empirical pm - error estimates .
we have validated this by computing local corrections for the reference - frame position of each star in each exposure based on the 25 closest stars within @xmath2 magnitude of its brightness ( as described in mclaughlin et al .
2006 , and anderson and van der marel 2010 ) .
this strategy should remove any residual distortion or cte errors .
we find that the results of this local test are indistinguishable from those of our more global procedure .
the small magellanic cloud ( smc ) lies in the background of 47 tuc .
fortunately the cluster is moving with respect to the smc by several milliarcseconds ( mas ) per year , so that over the time span of all available images the differential motion of the two systems amounts to nearly a whole acs pixel ( 50 mas ) allowing us to separate out these populations via pms .
figure [ fig : pmall ] displays the pm diagram for all stars in our frames .
the 47 tuc pm distribution was centered around ( 0,0 ) with the distribution centered near ( 4.5,1.5 ) for the smc .
we wanted to have as complete a sample as possible for both 47 tuc and the smc with a minimum of contamination .
for this reason our pm cut was very generous , about 10@xmath3 in each case .
the left panel of figure [ fig : cmd ] presents the cluster cmd using all the stars on the frames detected with s / n @xmath4 30 .
we chose such a restrictive cut as our focus in this reduction is high - precision photometry and astrometry , so that is why the cmd does not penetrate to very faint magnitudes .
the central panel contains the pm - selected sample .
this pm - cleaned cmd allows us to determine the ms ridge line and , in addition , define a color and magnitude selected sample of stars whose magnitudes lie within the black box .
the solid black circle in figure [ fig : pmall ] is the boundary of the pm sample for the smc .
we use both this pm selection and a color cut for the smc sample in a similar manner .
we divide the 47 tuc ms into four groups in color from bluest to reddest .
we first sort the ms stars from brighest to faintest in f814w and divide this sorted sample into thirty non - overlapping groups of 800 stars each . for each group of stars
we determine the median f606w @xmath5 f814w color as well as the standard deviation of the distribution as estimated by the @xmath6 statistic ( rousseoeuw & croux 1991 ) . as in heyl
et al . ( 2012 ) we use this statistic as a robust estimator of the standard deviation of a distribution , and we will denote its value by @xmath7 .
the right panel of figure [ fig : cmd ] displays the four color groups along the straightened ms of 47 tuc as a function of apparent magnitude .
thus we have defined the width through @xmath7 and the center of the ms through the median as a function of f814w . by spline interpolation
we determine the values of the width and median at the value of f814w for each star in the sample and assign each star to one of four color groups . the first group lies greater than one standard deviation and less than two standard deviations blueward of the median .
the second group lies blueward of the median yet within one standard deviation .
the third group lies redward of the median yet within one standard deviation .
the fourth group lies more than one standard deviation and less than two standard deviations redward of the median .
the pms of all of the stars lie within the large circle in figure [ fig : pmall ] , and their magnitudes lie within the box in figure [ fig : cmd ] .
the ridge line of the ms typically lies within the second group .
the fourth group contains some binary stars with nearly equal mass components .
binaries with more unequal mass components lie closer to the ms ridge line .
chemical abundance variations can also affect the color of single stars resulting in a spread in color .
our data are not sufficient to determine whether a particular stellar object is a single star with a peculiar chemical abundance , a binary or an optical double . because the measured binary frequency for mass ratios greater than 0.6 in 47 tuc is only a few percent ( milone et al .
2012b ) , the binaries will not significantly affect our results .
these four color groups , indicated on the 47 tuc cmd in the middle panel of figure [ fig : cmd ] , presumably mostly represent stars of differing chemical composition with a small admixture of binaries .
milone et al .
( 2012a ) found in their analysis of the 47 tuc ms that the various populations were often intertwined and crossed each other in the cmd .
this was generally seen in the most ultraviolet filters which are much more sensitive to chemical composition variations than f606w and f814w used here .
we tested whether he - enhanced ms stars at the metallicity of 47 tuc were always bluer in color than he - normal stars at all luminosities by overlaying a series of models ( dotter et al .
2008 ) with various enhanced he compositions on the cmd . over the entire extent of the ms explored in this paper
, the he - rich ms remained blue - ward of the he - normal sequence .
so we can be reasonably confident that the sequences of various abundances do not cross in the filters used here the populations do not mix . in figure
[ fig : colorpms ] ( top ) we plot the pm dispersions in the radial and tangential directions with respect to the cluster center for the four groups . in calculating these dispersions
we have added an additional cut in the estimated pm error .
we use only the stars whose errors are less than twice the median pm error at their apparent magnitude .
the error bars in the dispersions denote 90% confidence regions as determined by bootstrapping the sample .
if the stars are moving isotropically , the dispersions in these two orthogonal directions should be the same .
clearly they are not .
the anisotropy is largest for the bluest stars and reduces to no discernible anisotropy for the reddest ones .
this signature persists at all magnitudes . in figure 4
, we combine bins 1@xmath82 and bins 3@xmath84 and examine the dispersion of the blue and red stars as a function of magnitude , with 1000 stars in each magnitude bin . at all magnitudes ,
the dispersion of the blue stars in the tangential direction is always significantly smaller than their radial dispersion while for the red stars the dispersions exhibit little difference .
the left - most set of points in this diagram are the median values of the proper motions of the blue and red stars for which there are no discernible trends . as a sanity check
, we carried out a similar analysis on the smc stars .
again we divide the ms of the smc into four color bins .
of course , since the ms of the smc is much bluer than that of 47 tuc , the smc groups have different color ranges compared to those in 47 tuc .
we have restricted our sample here to lie below the turnoff of the smc ( fainter than f814w@xmath9 ) and brighter than 24 in f814w .
we measured the pm dispersions along the same radial and tangential axes of the direction to the 47 tuc center as a check on any potential systematic effects .
clearly these axes have no physical significance for the smc . these dispersions with color group
are illustrated in the bottom panel of figure [ fig : colorpms ] .
the pms and their dispersions here are much smaller than in 47 tuc reflecting mainly the smc s greater distance ; it is twelve times farther from the sun than is 47 tuc .
in contrast with 47 tuc , there is no evidence for any anisotropy in these pms .
in addition to pm effects , we searched for radial differences among the various 47 tuc color groups . since 47 tuc is at best barely relaxed , any discernible signals here could potentially provide important clues to formation scenarios of the populations . from ks tests of significance on the cumulative radial distributions of the various ms color groups , we find that the bluest group is less and less likely to be drawn from the same radial distribution as the other groups as we progress red - ward . a comparison of group 1 s with group 4 s distribution yields a probability of only 2.80@xmath10 that these two distributions were drawn from the same parent sample . in these comparisons ,
the bluest group is always the most centrally concentrated .
if the spread in ms color of 47 tuc was due to a large contribution of binaries , one would expect that the reddest color bin would be the most centrally concentrated due to mass segregation of these heavier stars .
the preceding analysis has shown that the 47 tuc ms stars demonstrate anisotropic pms that are strongly correlated with their colors .
the sense of this result is that the bluest stars possess the most anisotropic motions ( larger radial than tangential dispersions ) while the reddest stars exhibit no measurable anisotropy .
the motions of the smc stars are completely consistent with no anisotropy along orthogonal axes towards and at right angles to the center of 47 tuc . in addition , the bluer 47 tuc ms stars are more centrally concentrated than the redder stars . these differences contradict the notion that the globular cluster formed monolithically .
milone et al .
( 2012a ) demonstrate that the blue cohort exhibits cno processing and therefore likely he enrichment .
this seems to firmly establish these stars as second generation .
what do we expect for the motions and radial distribution of this second generation of stars ?
near the half - mass radius , 47 tuc has gone through only three half - mass relaxation times and our field is near this radius . if the cluster is not yet relaxed at this radius , we would expect this second generation cohort to be more centrally concentrated , which it is .
this is because these stars formed from dissipational gas expelled from massive first generation stars .
their current orbits are more radial as they are in the process of relaxing via two - body interactions to a distribution that is characteristic of their low masses - they are slowing diffusing outward ( radially ) to accomplish this .
the first generation stars exhibit no measurable anisotropy .
this is not entirely easy to understand . whether they underwent violent relaxation
before the second generation formed , or whether they still retain a memory of their initial collapse , they should still be on moderately radial orbits . this can be seen in figure 4 - 21 of binney and tremaine ( 1987 ) where stars outside the central core region of a model cluster that underwent violent relaxation have anisotropic velocities at late times . however , the relaxation of the first generation of stars in 47 tuc may be more complete because the initial state of the cluster may have been quite clumpy achieving more thorough relaxation ( see e.g. heyl et al .
1996 ) and not quasispherical as most simulations assume . these broad - brushstroked descriptions are illustrative only and not unique , but they do demonstrate how dynamical measurements could constrain the birth history of the cluster .
detailed numerical simulations including gas dynamics would be required to get a clearer picture . based on observations with the nasa / esa hubble space telescope , obtained at the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
these observations are associated with proposal go-11677 .
support for program go-11677 was provided by nasa through a grant from the space telescope science institute which is operated by the association of universities for research , inc .
, under nasa contract nas5 - 26555 .
h.b.r . and j.h .
are supported by grants from the natural sciences and engineering research council of canada and by the university of british columbia . j.a .
, j.s.k . ,
r.m.r and m.m.s . were funded by nasa .
the authors benefited from useful conversations with brett gladman .
anderson , j. , & king , i. r. 2003 , , 126 , 772 anderson , j. , & king , i. r. 2006 , instrument science report acs 2006 - 01 , 34 pages , 1 anderson , j. , king , i. r. , richer , h. b. , fahlman , g. g. , hansen , b. m. s. , hurley , j. , kalirai , j. s. , rich , r. m. , & stetson , p. b. 2008 , , 135 , 2114 anderson , j. , & van der marel , r. p. 2010
, , 710 , 1032 bellini , a. , anderson , j. , & bedin , l. r. 2011 , , 123 , 622 binney , j. , & tremaine , s. 1987 , princeton , nj , princeton university press , 1987 , 747 p. , giersz , m. , & heggie , d. c. 2011 , , 410 , 2698 gratton , r. , sneden , c. , & carretta , e. 2004 , , 42 , 385 heyl , j. s. , hernquist , l. , & spergel , d. n. 1996 , , 463 , 69 heyl , j. s. , richer , h. , anderson , j. , et al .
2012 , , 761 , 51 kalirai , j. s. , richer , h. b. , anderson , j. , et al .
2012 , , 143 , 11 mclaughlin , d. e. , anderson , j. , meylan , g. , et al .
2006 , , 166 , 249 milone , a. p. , marino , a. f. , piotto , g. , et al . 2012a , apj , 744 , 58 milone , a. p. , piotto , g. , bedin , l. r. , et al .
2012 , , 540 , a16 norris , j. , & freeman , k. c. 1979 , , 230 , l179 piotto , g. 2009 , iau symposium , 258 , 233 rousseoeuw , p.j . & croux , n.n
. journal of the american statistical association 88 , 20 , 1991 sollima , a. , ferraro , f. r. , bellazzini , m. , et al .
2007 , , 654 , 915 vesperini , e. , mcmillan , s. l. w. , dantona , f. , & dercole , a. 2013 , , 429 , 1913
|
based on the width of its main sequence , and an actual observed split when viewed through particular filters , it is widely accepted that 47 tucanae contains multiple stellar populations . in this contribution
, we divide the main - sequence of 47 tuc into four color groups , which presumably represent stars of various chemical compositions .
the kinematic properties of each of these groups is explored via proper - motions , and a strong signal emerges of differing proper - motion anisotropies with differing main - sequence color ; the bluest main - sequence stars exhibit the largest proper - motion anisotropy which becomes undetectable for the reddest stars .
in addition , the bluest stars are also the most centrally concentrated . a simiilar analysis for smc stars , which are located in the background of 47 tuc on our frames , yields none of the anisotropy exhibited by the 47 tuc stars .
we discuss implications of these results for possible formation scenarios of the various populations .
|
intuitively , a set is a collection of elements ( e.g. , numbers or symbols ) that is completely determined by them . the elements of a set are pairwise different .
if we relax this restriction and allow repeated occurrences of any element , then we end up with a mathematical structure that is known as _ multiset _ ( see @xcite for a historical account of the development of the multiset theory ; also , see @xcite for a recent account of the mathematical theory of multisets ) .
multisets are really useful structures and they have found numerous applications in mathematics and computer science .
for example , the prime factorization of an integer @xmath1 is a multiset @xmath2 whose elements are primes .
also , every monic polynomial @xmath3 over the complex numbers corresponds in a natural way to the multiset @xmath4 of its roots .
in addition , multisets have been used in concurrency theory @xcite .
a rather interesting recent development in the theory of multisets is the discovery that the logic of multisets is the @xmath5-fragment of intuitionistic linear logic ( see @xcite for details ) . if we allow elements of a multiset to occur an integral number of times ( and that includes a _ negative _ number of times ) , we end up with a structure that has been dubbed _
hybrid set_. these structures have been introduced by loeb @xcite .
initially , one may wonder whether hybrid sets are of any use .
however , loeb has shown that they are indeed very useful structures ( see @xcite ) .
for example , one can use a hybrid set to describe the roots and the poles of a rational function .
in particular , if @xmath3 is a monic rational function , then @xmath3 can be written in terms of its roots @xmath6 , and its poles @xmath7 , as follows : @xmath8 from this we can directly form a hybrid set , where elements that occur a positive number of times correspond to the roots of the function and elements that occur a negative number of times corresponds to the poles of the function . in a seminal paper , yager @xcite introduced fuzzy multisets , that is fuzzy subsets where an element may occur more than one time ( see @xcite for an up - to - date presentation of the theory of fuzzy multisets , which , however , does not differ significantly from @xcite ) .
yager defined fuzzy multisets as follows @xcite : assume @xmath9 is a set of elements .
then a fuzzy bag @xmath10 drawn from @xmath9 can be characterized by a function @xmath11 such that @xmath12 where @xmath13 is the set of all crisp bags drawn from the unit interval . in `` modern parlance ''
fuzzy multisets can be characterized by a high - order function . in particular , a fuzzy multiset @xmath10 can be characterized by a function @xmath14 where @xmath15 $ ] and @xmath16 is the set of natural numbers including zero .
it is not difficult to see that any fuzzy multiset @xmath10 is actually characterized by a function @xmath17 which is obtained from the former function by _ uncurrying _ it .
however , it is more natural to demand that for each element @xmath18 there is only one membership degree and one multiplicity . in other words ,
a `` fuzzy multiset '' @xmath10 should be characterized by a function @xmath19 . to distinguish these structures from fuzzy multisets
, we will call them _ multi - fuzzy _ sets @xcite . given a multi - fuzzy set , @xmath10 ,
the expression @xmath20 denotes that there are @xmath21 copies of @xmath18 that belong to @xmath10 with degree that is equal to @xmath22 .
apart from their applicability to mathematics , multisets are really useful structures as interesting models of computations are built upon them .
for instance , the chemical abstract machine of berry and boudol @xcite is an abstract machine that is well - suited to model concurrent computation and manipulates _ solutions _ , which are finite multisets of _ molecules _ where a molecule is simply a term of an algebra .
membrane computing is a model of computation that is built around the notion of multiset rewriting rules .
more specifically , membrane computing is a computational paradigm that was inspired by the way cells live and function ( see @xcite for an overview of the field of membrane computing ) . roughly speaking
, a cell consists of a membrane that separates the cell from its environment .
in addition , this membrane consists of compartments surrounded by porous membranes , which , in turn , may contain other compartments , and so on . at any moment , matter flows from one compartment to any neighboring one .
in addition , the cell interacts with its environment in various ways ( e.g. , by dumping matter to its environment ) .
obviously , at any moment a number of processes occur in parallel ( e.g. , matter moves into a compartment , while energy is consumed in another compartment , etc . ) .
a p system is a conceptual computational device whose functionality is based on an abstraction of the cell .
thus , a p system consists of porous membranes that are populated with multisets of objects , which are usually materialized as strings of symbols .
in addition , there are rules that are used to change the configuration of the system .
a p system behaves more or less like a parser , which is clearly hard - wired to a particular grammar .
thus , a p system stops when no rule can be applied to the system .
the result of the computation is always equal to the cardinality of the multiset that is contained in a designated compartment .
now , since rigid mathematical models employed in life sciences are not completely adequate for the interpretation of biological information , there have been various proposals to use fuzzy sets in the modeling of biological systems ( e.g. , see @xcite ) .
thus , it is quite reasonable to attempt the use of the theory of fuzzy sets in p systems .
indeed , such an attempt has been described in @xcite that is generalized to a certain degree in this paper .
[ [ structure - of - the - paper ] ] structure of the paper + + + + + + + + + + + + + + + + + + + + + + in what follows , i will define @xmath0-multi - fuzzy sets and @xmath0-fuzzy hybrid sets .
next , i will define the basic operations between such structures ( e.g. , union , sum , etc . ) .
also , i will give the definition of certain standard fuzzy - theoretic operators . by replacing multisets with either @xmath0-multi - fuzzy sets or @xmath0-hybrid sets in the definition of both p systems and the chemical abstract machine
, we end up with fuzzy versions of these notational computing devices .
we formally define these devices and briefly investigate their properties .
the paper ends with the customary concluding remarks .
one may say that multisets form an abstraction of the _ token - type _ distinction , which is the basis of the `` token - token identity theory '' @xcite , while ( ordinary ) sets are an abstraction of the denial of the token - type distinction . to make clear the essence of the token - type distinction
, i will borrow an example from @xcite .
if one writes the word `` dog '' three times ( i.e. , `` dog dog dog '' ) , then she has written three instances , or tokens , of the one type of word .
this observation necessitates a distinction between types ( i.e , abstract general entities ) and tokens ( i.e. , concrete particular objects and events ) .
`` a token of a type is a particular concrete exemplification of that abstract general type '' .
@xcite .
there is no question that three instances of the word `` dog '' are tokens of the `` dog '' type .
however , there are many instances where one can not make such a definitive statement regarding the type of some tokens .
in particular , there are many cases where some token @xmath23 is of type @xmath24 to a certain degree . for example , consider the following glyphs : each of them depicts an `` a , '' however , each one is reminiscent of the `` standard a '' to some degree .
for instance , the rightmost is less reminiscent of the `` standard a '' , while the leftmost glyph looks like an ordinary `` a. '' a text is clearly a multiset of letters , but if we are free to use any of these `` as '' to typeset a document , then we need a fuzzy multiset to _ describe _ the text .
however , this is not a common practice and so we need a more restricted structure that better models real life situations .
as was noted above these structures will be called multi - fuzzy sets . clearly , it is too restrictive to demand that tokens are of some type to a degree , which is expressed by some number that belongs to the unit interval .
more generally , we can assume that the likelihood degree is drawn from some frame @xmath0 .
indeed , it is possible to define many other partially ordered sets that are frames .
for example , vickers @xcite shows that finite observations on bit streams have the properties of a frame .
thus , the similarity degree would express the idea that an element resembles , in some way , an element of such a frame .
typically , a frame is defined as follows @xcite : a poset @xmath10 is a _ frame _ iff 1 .
every subset has a join 2 .
every finite subset has a meet 3 .
binary meets distribute over joins : @xmath25 * @xmath0 is a nonempty set ; * @xmath26 and @xmath27 are binary operations on @xmath0 ; * both @xmath26 and @xmath27 are * * idempotent ( i.e. , @xmath28 and @xmath29 ) ; * * commutative ( i.e.,@xmath30 and @xmath31 ) ; * * associative ( i.e. , @xmath32 and @xmath33 ) ; * * distributive ( i.e. , @xmath34 and @xmath35 ) ; * * and they satisfy the absorption law ( i.e. , @xmath36 ) .
note that a frame is clearly a distributive lattice .
so , @xmath0-multi - fuzzy sets are an extension of multi - fuzzy sets just like @xmath0-fuzzy sets @xcite are an extension of fuzzy sets .
kostrikin in his comments in the entry for the conecpt of duality in the _ encyclopaedia of mathematics _ notes that `` [ d]uality is a very pervasive and important concept in ( modern ) mathematics . ''
one could argue that hybrid sets , and , therefore , fuzzy hybrid sets , extend multisets and their fuzzy counterparts to describe and/or to model dualities .
let us now proceed with the formal definition of @xmath0-fuzzy hybrid sets : [ fuzzy : hybrid : set ] an @xmath0-fuzzy hybrid set @xmath37 is a mathematical structure that is characterized by a function @xmath38 , where @xmath0 is a frame , and it is associated with a @xmath0-fuzzy set @xmath39 . more specifically , the equality @xmath40 means that @xmath37 contains exactly @xmath21 copies of @xmath18 , where @xmath41 .
if we substitute @xmath42 with @xmath16 in the previous definition , then the resulting structures will be called _ @xmath0-multi - fuzzy sets_. assuming that @xmath37 is an @xmath0-fuzzy hybrid set , then one can define the following two functions : the _ multiplicity _ function @xmath43 and the _ membership _ function @xmath44 . clearly , if @xmath40 , then @xmath45 and @xmath46 .
notice that it is equally easy to define the corresponding functions for an @xmath0-multi - fuzzy set .
i believe this is a good point to briefly express my prejudices and my intentions regarding the present work . clearly , it is not my intention to develop an axiomatic set theory of @xmath0-fuzzy hybrid sets and @xmath0-multi - fuzzy sets , in the sense of the zermelo - frnkel set theory , but rather a `` nave '' set theory in the sense that i will not present a precise axiomatization .
therefore , i plan to introduce only the basic set - theoretic operations and the basic properties of these sets . to begin with ,
let me now define the cardinality of an @xmath0-fuzzy hybrid set : assume that @xmath37 is an @xmath0-fuzzy hybrid set that draws elements from a universe @xmath9 .
then its cardinality is defined as follows : @xmath47 where @xmath48 is a binary multiplication operator that is used to compute the product of @xmath49 times @xmath50 . if @xmath51 ( i.e , when extenting `` intuitionistic '' fuzzy sets , see @xcite ) , then @xmath52 . when @xmath0 is the unit interval , then @xmath53 is the usual multiplication operator .
the cardinality of a set is equal to the number of elements the set contains .
clearly , the previous definition is not in spirit with this assumption .
however , hybrid sets may contain elements that occur a negative number of times .
thus , one may think that we should take this fact under consideration when computing the cardinality of a hybrid set and , more generally , the cardinality of an @xmath0-fuzzy hybrid set .
so , it makes sense to introduce the notion of a _
strong _ cardinality defined as follows : assume that @xmath37 is an @xmath0-fuzzy hybrid set that draws elements from a universe @xmath9
. then its _ _ s__trong cardinality is defined as follows : @xmath54 . for reasons of completeness
i give below the definition of the cardinality of @xmath0-multi - fuzzy sets : assume that @xmath37 is an @xmath0-multi - fuzzy set that draws elements from a universe @xmath9 .
then its cardinality is defined as follows : @xmath47 where @xmath55 is some binary operator that maps @xmath49 and @xmath56 to some positive real number ( since @xmath57 ) . in order to complete the presentation of the basic properties of fuzzy hybrid sets ,
it is necessary to define the notion of subsethood .
before , going on with this definition , i will introduce the ( new ) partial order @xmath58 over @xmath42 . in particular ,
if @xmath59 , then @xmath60 note that here @xmath26 and @xmath27 denote the classical logical conjunction and disjunction operators , respectively .
in addition , the symbols @xmath61 and @xmath62 are the well - known ordering operators , and @xmath63 is the absolute value of @xmath21
. from the previous definition it should be obvious that @xmath64 , for all @xmath50 .
also , @xmath65 , @xmath66 , and @xmath67 .
but what kind of structure is the pair @xmath68 ?
the answer is easy with the help of the following result : the relation @xmath58 is a partial order .
i have to prove that the relation @xmath58 is reflexive , antisymmetric and transitive : reflexivity : : assume that @xmath69 .
then if @xmath70 , @xmath71 from the first part of the disjunction .
if @xmath72 , then @xmath71 from the fourth part of the disjunction and if @xmath73 , then @xmath71 from the second part of the disjunction .
antisymmetry : : assume that @xmath74 , @xmath75 , and @xmath76 .
then if @xmath70 this implies that @xmath77 and so @xmath78 . if @xmath72 , then it follows that @xmath79 , @xmath80 , and @xmath81 , which implies that @xmath78 .
similarly , if @xmath73 , then it follows that @xmath82 , @xmath83 , and @xmath84 , which implies that @xmath78 .
transitivity : : assume that @xmath85 , @xmath75 , and @xmath86
. then if @xmath70 , then clearly @xmath87 .
if @xmath72 and @xmath79 , then either @xmath88 or @xmath89 , but since @xmath90 , this implies that @xmath87 .
if @xmath73 and @xmath82 , then @xmath89 and since @xmath91 this implies that @xmath87 .
if @xmath72 and @xmath82 , then since @xmath86 , this implies that @xmath89 , which means that @xmath87 .
note that @xmath92 is an alternative form of @xmath76 , which will be used in the rest of this paper .
let us now proceed with the definition of the notion of subsethood for @xmath0-fuzzy hybrid sets : assume that @xmath93 are two @xmath0-fuzzy hybrid sets .
then @xmath94 if and only if @xmath95 and @xmath96 for all @xmath97 .
remark that for all @xmath98 , @xmath99 if @xmath100 is `` less than or equal '' to @xmath101 in the sense of the partial order defined over @xmath0 .
the definition of subsethood for @xmath0-multi - fuzzy sets is more straightforward : assume that @xmath102 are two @xmath0-multi - fuzzy sets .
then @xmath94 if and only if @xmath95 and @xmath103 for all @xmath97 .
the basic operations between sets are their union and their intersection . a third operation , viz .
set sum , is meaningful only for multisets .
also , since both @xmath0-multi - fuzzy sets and @xmath0-fuzzy hybrid sets are actually generalizations of fuzzy sets , one should be able to define the @xmath104-cuts of such sets .
i will start by defining the basic set operations between @xmath0-multi - fuzzy sets .
let me first present the definitions of union and intersection of @xmath0-multi - fuzzy sets : assuming that @xmath105 are two @xmath0-multi - fuzzy sets , then their union , denoted @xmath106 , is defined as follows : @xmath107 where @xmath108 is the join of @xmath109 .
assuming that @xmath105 are two @xmath0-multi - fuzzy sets , then their intersection , denoted @xmath110 , is defined as follows : @xmath111 where @xmath112 is the meet of @xmath109 .
i will now define the sum of two @xmath0-multi - fuzzy sets : suppose that @xmath102 are two @xmath0-multi - fuzzy sets .
then their sum , denoted @xmath113 , is defined as follows : @xmath114 although it is crystal clear , it is necessary to say that @xmath115 and @xmath116 are operators that are part of the definition of the frame @xmath0 . and
as such they have a number of properties ( e.g. , they are idempotent , etc . , see @xcite for details ) that , naturally , affect the properties of the operations defined so far .
indeed , these operations have the following properties : for any three @xmath0-multi - fuzzy sets @xmath117 the following equalities hold : 1 .
commutativity : @xmath118 2 .
associativity : @xmath119 3 .
idempotency : @xmath120 4 .
distributivity : @xmath121 5 .
distributivity of sum : @xmath122 1 .
although this is easy , i will prove all cases : @xmath123 \bigl(\mathscr{a}\cap\mathscr{b}\bigr)(z)&=&\biggl ( \mathscr{a}_{\mu}(z)\sqcap\mathscr{b}_{\mu}(z ) , \min\bigl\{\mathscr{a}_{m}(z ) , \mathscr{b}_{m}(z)\bigr\}\biggr)\\ & = & \biggl ( \mathscr{b}_{\mu}(z)\sqcap\mathscr{a}_{\mu}(z ) , \min\bigl\{\mathscr{b}_{m}(z ) , \mathscr{a}_{m}(z)\bigr\}\biggr)\\ & = & \bigl(\mathscr{b}\cap\mathscr{a}\bigr)(z)\\[5pt ] \bigl(\mathscr{a}\uplus\mathscr{b}\bigr)(z)&=&\biggl ( \mathscr{a}_{\mu}(z)\sqcup\mathscr{b}_{\mu}(z ) , \mathscr{a}_{m}(z)+\mathscr{b}_{m}(z)\biggr)\\ & = & \biggl ( \mathscr{b}_{\mu}(z)\sqcup\mathscr{a}_{\mu}(z ) , \mathscr{b}_{m}(z)+ \mathscr{a}_{m}(z)\biggr)\\ & = & \bigl(\mathscr{b}\uplus\mathscr{a}\bigr)(z)\end{aligned}\ ] ] 2 .
i will prove only the first case as the others can be proved similarly : @xmath124 3 .
as in the previous case , i will prove only the first case as the other can be proved similarly : @xmath125 4 .
the proof of this case follows from the fact that the following equalities are true for the any three elements of a frame : @xmath126 5 .
as with the previous case the proof for this case follows from the fact that for any @xmath127 the following equalities hold : @xmath128 the @xmath104-cut of a fuzzy subset is just a crisp set .
similarly , the @xmath104-cut of an @xmath0-multi - fuzzy set has to be a multiset . indeed ,
if @xmath129_{n}$ ] denotes a multiset that consists of only @xmath21 copies of @xmath18 , the following definition is in spirit with the general theory of fuzzy sets : suppose that @xmath37 is an @xmath0-multi - fuzzy set with universe the set @xmath9 , and that @xmath130 .
then the @xmath104-cut of @xmath37 , denoted by @xmath131 , is the multiset @xmath132_{\mathscr{a}_{m}(x)}.\ ] ] not so surprisingly , the properties of the @xmath104-cut of @xmath0-multi - fuzzy sets are similar to those of plain fuzzy sets .
these properties are summarized below : [ alpha - proof ] assume that @xmath37 and @xmath133 are two @xmath0-multi - fuzzy sets with universe the set @xmath9 .
then the following properties hold : 1 . if @xmath134 , then @xmath135 and 2 .
@xmath136 , @xmath137 , and @xmath138 . 1 .
let @xmath97 and @xmath134 . if @xmath139 , then @xmath140 . if @xmath141 , then @xmath142 . if @xmath143 , then @xmath144 .
thus , for all possible cases @xmath145 , which means that @xmath146 .
2 . assume that @xmath147 .
then this means that @xmath148 also , it implies that @xmath149 and hence @xmath150 . from this
, one can immediately deduce that @xmath151 and @xmath152 .
suppose now that @xmath153 and @xmath154 .
then this means that @xmath155 and @xmath156 and so @xmath157 loeb has shown that the set of all subsets of a given hybrid set with the subsethood relation do not form a lattice .
this means that if @xmath158 and @xmath159 are two hybrid sets , then if they have lower bounds , they do not necessarily have a greatest lower bound .
similarly , if @xmath158 and @xmath159 have upper bounds , then they do not necessarily have a lowest upper bound .
practically , this means that given two hybrid sets @xmath158 and @xmath159 , one can not define their union and their intersection .
fortunately , the sum of hybrid sets is a well - defined operation .
thus , we can easily extend this definition as follows : assume that @xmath93 are two @xmath0-fuzzy hybrid sets .
then their sum , denoted @xmath113 , is defined as follows : @xmath114 let @xmath160 denote a finite collection of hybrid sets with a common universe @xmath9 , where each of these sets contains repeated occurrence of only one element @xmath161 .
in addition , let us insist that no two @xmath162 and @xmath163 will have common elements .
also , let us denote with @xmath164 the unique hybrid set that is the sum of all @xmath162 . with these preliminary definitions ,
the road for the following definition has been paved : suppose that @xmath37 is an @xmath0-fuzzy hybrid set with universe the set @xmath9 and that @xmath130 .
then the @xmath104-cut of @xmath37 , denoted by @xmath131 , is the hybrid set @xmath164 , where @xmath165 iff @xmath166 , for all @xmath161 .
the @xmath104-cut of @xmath0-fuzzy hybrid sets has the following properties : assume that @xmath37 and @xmath133 are two @xmath0-fuzzy hybrid sets with universe the set @xmath9 . then the following properties hold : 1 .
if @xmath134 , then @xmath167 2 . @xmath138 .
the proof is similar to the proof of theorem [ alpha - proof ] and is omitted .
in @xcite the author has proposed fuzzified versions of p systems .
the basic idea behind this particular attempt to fuzzify p systems is the substitution of one or all ingredients of a p system with their fuzzy counterparts . from a purely computational point of view , it turns out that only p systems that process multi - fuzzy sets are interesting .
the reason being the fact that these systems are capable of computing ( positive ) real numbers . by replacing the multi - fuzzy sets employed in the author s previous work with @xmath0-multi - fuzzy sets
, the computational power of the resulting p systems will not be any `` greater , '' nevertheless , these systems may be quite useful in modeling living organisms .
but , things may get really interesting if we consider p systems with @xmath0-fuzzy hybrid sets , in general .
let us begin with the definition of these systems : [ f : data ] a general fuzzy p system is a construction @xmath168 where : 1 .
@xmath169 is an alphabet ( i.e. , a set of distinct entities ) whose elements are called _ objects _ ; 2 .
@xmath170 is the membrane structure of degree @xmath171 ; membranes are injectivelly labeled with succeeding natural numbers starting with one ; 3 .
@xmath172 , @xmath173 , are @xmath0-fuzzy hybrid sets over @xmath169 that are associated with each region @xmath22 ; 4 .
@xmath174 , @xmath173 , are finite sets of multiset rewriting rules ( called _ evolution rules _ ) over @xmath169 .
an evolution rule is of the form @xmath175 , @xmath176 and @xmath177 , where @xmath178 , @xmath179 .
the effect of each rule is the removal of the elements of the left - hand side of each rule from the `` current '' compartment and the introduction of the elements of right - hand side to the designated compartments ; 5 .
@xmath180 is the label of an elementary membrane ( i.e. , a membrane that does not contain any other membrane ) , called the _ output _ membrane .
the really interesting thing with the systems described in @xcite is that i havent managed to find any limits on what can be actually computed .
remember , that a real number @xmath181 is called _ computable _ if there is a computable sequence @xmath182 of rational numbers which converges to @xmath18 effectively , that is for all @xmath56 , @xmath183 ( see @xcite for details ) .
in other words , this means that not all real numbers are _
computable_. however , one should not forget that the definition of _ computability _ is hard - wired to the computational capabilities of the universal turing machine and the so called church - turing thesis , which dictates what can be and what can not be computed .
now , the crucial question is whether there are any limits that prohibit the computation of certain numbers with fuzzy p systems ?
it seems that these system go beyond the church - turing barrier because their set of input values is drastically larger than that of the turing machine .
however , it is an open problem the determination of the exact computational power of these systems .
in this paper i have introduced @xmath0-multi - fuzzy sets and @xmath0-fuzzy hybrid sets as well as their basic operations .
in addition , general fuzzy p systems have been introduced , which can be used to compute real numbers .
i do not believe that this is something really new
it is just another indication that the current theory of computation is simply inadequate to describe all computational phenomena .
after all , this has been elegantly demonstrated by stein in his thought provoking paper @xcite .
in addition , i believe that we need a paradigm shift in computer science so to encompass new `` phenomena '' and practices . .
transactions on rough sets ii : rough sets and fuzzy sets _ , j. f. peters , a. skowron , d. dubois , j. grzymala - busse , and m. inuiguchi , eds .
3135 in lecture notes in computer science .
springer - verlag , berlin , 2004 , pp . 189200 .
|
an orthogonal approach to the fuzzification of both multisets and hybrid sets is presented . in particular , we introduce @xmath0-multi - fuzzy and @xmath0-fuzzy hybrid sets , which are general enough and in spirit with the basic concepts of fuzzy set theory .
in addition , we study the properties of these structures .
also , the usefulness of these structures is examined in the framework of mechanical multiset processing .
more specifically , we introduce a variant of fuzzy p systems and , since simple fuzzy membrane systems have been introduced elsewhere , we simply extend previously stated results and ideas .
|
recently , cassan et al .
( 2012 ) conducted a fairly comprehensive census of planets in the milky way using the microlensing technique . for planets at orbital radii of 0.5 - 10
au they reported that @xmath8 of stars host planets with masses in the range @xmath9 , @xmath10 host cool neptunes with masses @xmath11 , and @xmath12 host super - earths with masses @xmath13 .
the results are encouraging for a discipline so young , but the uncertainties are large for the lower mass planets , and also the measurements do not extend down to planets of earth - mass . in this paper
we attempt to address these problems by proposing a modified strategy for detecting low - mass planets by microlensing .
our strategy is based on a proposal originally made in 1998 by griest & safizadeh for detecting planets in microlensing events of high magnification .
high magnification occurs when the source star passes almost directly behind the lens star , so that @xmath14 , or @xmath15 , @xmath16 and @xmath17 denote the einstein radius , crossing time and time of closest approach between the lens and source stars , and @xmath18 , @xmath19 and @xmath20 the einstein radius , the impact parameter between the lens and source stars in angular coordinates , and the angular radius of the source star respectively .
we denote by @xmath21 the projected impact parameter in the lens plane between the lens and source stars in units of @xmath22 .
thus @xmath23 . finally , for planetary events , we denote by @xmath24 , @xmath25 and @xmath26 the planet : star mass ratio , projected separation in units of @xmath22 and axis relative to the source star track respectively .
the fundamental parameter of microlensing , the einstein radius @xmath22 , is defined in rattenbury et al .
( 2002 ) . ] .
these very well aligned events are rarer than normal microlensing events that have magnifications of only a few , but they are easily recognizable because the peak magnification is approximately @xmath27 and this is @xmath28 . moreover , as griest and safizadeh showed , if a planet orbits the lens star , it almost certainly produces a perturbation to the light curve close to the peak of the light curve .
it thus appeared possible to search for planets with high efficiency merely by monitoring the peaks of events that had high magnification .
this was noted by several groups ( gaudi et al .
1998 ; rhie et al .
2000 ; bond et al . 2002 ; rattenbury et al .
2002 ) and subsequently demonstrated to be correct . since 2002 a steadily growing number of high magnification events was discovered and alerted each year by the moa and ogle collaborations .
several of these events were monitored by ` follow - up ' networksmicrofun + planet : planet.iap.fr + robonet - ii : robonet.lcogt.net + mindstep : www.mindstep-science.org ] and this led to the discoveries of several planets ( udalski et al .
2005 ; gould et al . 2006 ; gaudi et al .
2008 ; bennett et al . 2008 ; dong et al . 2009 ; janczek et al .
2010 ; miyake et al . 2011 ; bachelet et al .
2012 ; yee et al .
2012 ; han et al .
2013 ) , some of the first measurements of planetary abundances ( gould et al .
2006 , 2010 ) , and also to a measurement of limb darkening ( fouque et al , 2010 ) . a new class of events was also discovered by the moa and ogle microlensing collaborations in their quests for high magnifications events .
both collaborations found events having very high magnifications , of order 1,000 ( e.g. , abe at al .
2004 ; dong et al . 2006 ) . when these were first discovered it was natural to assume they would provide greater sensitivity to planets .
this assumption is the subject of the present paper .
our conclusion is that low - mass planets can be efficiently searched for in events with moderately high magnifications of order 50 - 200 , and that the greater frequency of these events in comparison to those with higher magnifications lends advantage to monitoring them .
however , to take full advantage of their discovery potential , telescopes with apertures in the range 1 - 2 m would be needed to photometer them with sufficient precision to detect low - mass planets .
using such telescopes the planetary mass function of cassan et al .
( 2012 ) may be able to be extended down to earth mass . because such planets have special importance in the search for the proverbial ` earth twin ' , and because their abundance will constrain theories of planetary formation
, it is important to optimise strategies to encompass these planets in observing programmes .
recently han ( 2009 ) and han and kim ( 2009 ) examined the sensitivity of the high magnification technique for detecting planets .
however , they did not explicitly examine the sensitivity as a function of magnification , nor did they focus on the lower magnification events considered in this paper .
our results are therefore complementary to theirs .
liebes ( 1964 ) first suggested that planets might be detected by gravitational microlensing .
his analysis was based upon the use of einstein arcs .
1 shows the well - known geometry of the pair of moving einstein arcs that are formed in a normal microlensing event ( paczynski 1996 ) . for clarity
, the radius of the source star ( projected to the lens plane ) has been greatly exaggerated .
typically it is of order @xmath29 and the einstein arcs are of comparable width . and the source star radius is scaled to its projected value in the lens plane .
the magnitude of @xmath22 is approximately 2 au in typical galactic microlensing events .
it is therefore reasonable to suppose the presence of a planet close to the ring as shown in the figure .
this figure ( without the planet ) is adapted from paczynski ( 1996).,width=377 ] fig .
1 may be used to understand qualitatively the main features of planet detection in high magnification events .
straightforward geometrical considerations show that the lengths of the einstein arcs in the lens plane near the time of maximum magnification are approximately @xmath30 .
if the lens star hosts a planet , and if the planet is situated fairly close to either the upper or the lower portions of the einstein ring as shown in the figure , then one of the einstein arcs will slide by it during the event , and be perturbed by it . in this way a planet betrays its presence .
the effect is large because the image of the source star is magnified by the lens star , and the magnified image is perturbed by the planet .
if the planet is closer to the einstein ring than the arc length , as shown in the figure , the duration of the planetary perturbation is given by the time required for the arc to slide by the planet .
this is easily shown to be approximately @xmath31 where @xmath32 in @xmath33 .
their speeds are therefore @xmath34 .
this equals @xmath35 ( rattenbury et al .
2002 ) . dividing the speed into the above expression for an arc length we obtain @xmath31 for the slide - by time . ] .
the result is independent of @xmath21 , and therefore of the peak magnification @xmath36 , as @xmath37 .
higher magnifications produce longer arcs , but they rotate faster , and the slide - by time is unchanged .
the magnitudes of the perturbations are surprisingly large , as discussed below .
they are positive if the source star trajectory threads the lens star : planet system ( @xmath38 ) and vice versa .
as expected , they increase ( approximately proportionally ) with the mass of the planet , and as the planet approaches the einstein ring .
we note that these scaling laws differ qualitatively from those found for perturbations caused by planetary caustic crossings ( horne et al .
2009 ) .
several examples of computed planetary perturbations may be found in the literature that confirm the above expectations ( rhie et al . 2000 ; rattenbury et al . 2002 ; bennett et al .
2008 ) . for typical values of @xmath39 and @xmath40 days planetary pertubations of duration
about 1.4 hr are predicted . for typical ranges of @xmath41 and @xmath16 ( see 6 below ) we expect durations ranging over @xmath4 hr . such perturbations may be conveniently observed from a single observing site .
there are two exceptions to the above .
if a planet lies very close to the einstein ring , the difference from the planetless light curve can become large . in this case
the above perturbative picture fails , and a quantitative analysis ( as carried out below ) is required .
second , for events with very small impact parameters in which the source star passes behind the lens star , the geometry of the moving arcs is different from that depicted in fig .
1 . in these events a pair of arcs is formed fore and aft of the lens , instead of above and below it ( liebes 1964 ) . a planet perturbs both arcs , one after the other , leading to the formation of two very brief spikes of opposing polarities on the light curve .
examples of such spikes have been observed and modelled ( dong et al .
2009 ; janczek et al . 2010 ; han & kim 2009 ) .
further examples are displayed below .
wambsganss ( 1997 ) originally used magnfication maps to demonstrate the detectability of planets by gravitational microlensing . in this method rays
are shot from the observer , which is treated as a point source , through the lens in the thin - lens approximation to the source plane to form a ` magnification map ' that is sampled by the moving source star . and
@xmath42 to the right .
the deviations were computed pixel - by - pixel as the difference in magnification produced by a star with and without a planet , normalised by the magnification without a planet .
the coordinates shown are fractions of @xmath18 . also shown
are typical tracks of a solar - like source star at @xmath43 .
moire - like artefacts caused by the use of rectangular pixels are present .
they are alternately positive and negative , and have small net effects when integrated over the source star.,width=302 ] fig .
2 above shows a magnification map for a typical high magnification event .
the lens consists of a star at the origin and a planet on the positive @xmath44-axis at a distance @xmath25 of @xmath45 .
the planet : star mass ratio @xmath24 is @xmath46 .
this corresponds to a super - earth of about three earth masses if the lens star is an m or k dwarf .
the map shows the pixel - by - pixel fractional perturbation caused by the planet .
it is seen to be approximately @xmath47 for a large region extending to @xmath48 towards the planet , and approximately @xmath49 for a similarly large region extending to @xmath50 away from the planet . also shown are three possible tracks of a source star at a typical inclination to the star : planet axis .
the leftmost track is for a solar - like source star with impact parameter @xmath51 or @xmath52 . for the central track the impact parameter @xmath53 and @xmath54 by liebes s theorem ( 1964 ) .
the righthand track corresponds to @xmath55 and @xmath56 . integrating ` by eye ' over
the area of the source star indicates that the left and right trajectories with @xmath57 of 100 and 200 yield planetary perturbations approaching @xmath49 and @xmath47 respectively in magnitude .
perturbations to light curves of order 10% are readily detectable when the magnification is @xmath58 , yet perturbations of the above type have not been extensively sought in planet searches to date .
it should therefore be possible to increase the current detection rate of planets by making use of the large extent of the regions shown on the left and right sides of fig .
2 . at the centre of the map lies the ` stellar caustic ' .
this is an approximately triangular closed curve of formally infinite magnification extending from the lens star towards the planet with length about @xmath59 , i.e. approximately the diameter of the source star shown in the figure .
the true length of the caustic is @xmath60 ( han 2009 ) but this is not discernable on the map due to the finite size of the pixels .
the caustic provides a means for detecting planets , but source star tracks that traverse it are subject to cancellations from neighbouring areas of opposing polarities . on the other hand , the areas to the left and right of the caustic do not suffer from cancellations to the same extent .
in addition , they are longer , and therefore offer the prospect of more numerous traversals in any random sample of events . and separation @xmath25 are @xmath46 and @xmath45 , as in fig . 2 , and the angle between the planet : star axis and the track of the source star is @xmath61 .
the impact parameter @xmath21 is 0 , 0.002 , 0.005 , 0.01 or 0.02 respectively.,width=264 ] magnification maps similar to that shown in fig .
2 for heavier or lighter planets , at greater or smaller distances from the einstein ring , yield similar results .
in order to further assess the detectability of low - mass planets , light curves were computed for a range of planet masses and separations .
the light curves were generated by subdividing the source plane into three regions .
a central region was used to construct the peaks of light curves ( approximately @xmath62 the fwhm ) at high resolution .
a larger region was used to extend the light curves out to approximately @xmath63 at moderate resolution , and an outer region was used to further extend them to approximately @xmath64 at low resolution . the pixel size in each region , and the number of rays per pixel , were adjusted to keep numerical noise below the precision of typical data .
the wings of the light curves beyond the outermost of the above regions were treated in the single lens approximation .
limb darkening was included in the linear approximation with a limb darkening coefficient of 0.5 .
the procedures were checked by comparison with published analyses of several events . in order for a planetary deviation to be clearly identified on a light curve ,
it is helpful if both the deviation and also a significant portion of the unperturbed light curve is observed .
if we imagine an ideal search being executed with robotically controlled telescopes programmed to monitor the fwhms of selected events from sites of excellent seeing , than this requirement will favour events for which @xmath65 . as remarked in @xmath66 ,
the duration of a typical planetary deviation is @xmath67 whereas @xmath33 is @xmath68 ( rattenbury et al .
2002 ) . to sample both the deviation and the unperturbed light curve equally precisely , the deviation should occupy less than half the fwhm .
this translates to @xmath69 or @xmath70 for a typical value of @xmath71 we note however that , in actual practice , the descending branch of the fwhm will tend to be more fully monitored than the ascending branch , and that coverage on the descending branch may usefully extend beyond the fwhm .
we note also that some previously made planetary discoveries have suffered from less than complete coverage of the fwhm .
3 shows typical planetary deviations .
the planet : star mass ratio and separation are the same as those used in fig .
fractional deviations are plotted for @xmath57 equal to 2,000 , 500 , 200 , 100 and 50 respectively .
the first exhibits the double - spiked structure discussed above , and the others exhibit perturbations of the width and height predicted above . for the last one at @xmath72 the requirement that the planet be closer to the einstein ring than the length of the einstein arcs begins to break down , and the width of the perturbation is greater than predicted . a number of plots similar to those in fig . 3 were constructed and examined .
4 shows the dependence of planetary perturbations on impact parameter @xmath21 .
the parameter @xmath21 was chosen as the independent variable in this plot , rather than the equivalent parameter @xmath57 which approximately equals @xmath73 , because @xmath21 is uniformly populated in any unbiased sample of events , whereas @xmath57 is not .
it is apparent that from fig .
4 that , for all combinations of the other parameters , there is a drop in the fractional deviations for the smallest values of @xmath21 , i.e. for the largest magnifications .
this is caused by the onset of the ` double - spike ' phenomenon discussed above .
4 also shows the dependence of a typical planetary perturbation on the angle @xmath26 between the planet : star axis and the track of the source star .
the dependence is essentially constant for angles larger than @xmath74 . and @xmath75 showing enhanced sensitivity at negative values of @xmath21 and moderate magnifications.,width=291 ] the dependence on the source - size parameter @xmath32 is also shown in fig .
4 . for typical events with solar - like source stars in the galactic bulge , and main sequence m - k lens stars lying in the bulge or in the galactic disc
, @xmath41 takes on values between about 0.0005 and 0.002 .
it is seen that planetary deviations are greater for smaller values of @xmath41 , and they are quite low when @xmath41 is as large as 0.005 .
the latter value is about the minimum possible value of @xmath41 if the source star is a giant .
the dependence of typical planetary perturbations on the planet mass parameter @xmath24 is displayed in fig .
@xmath24 was set equal to @xmath76 , @xmath46 or @xmath77 . for each value of @xmath24
the value of @xmath25 was adjusted ( in increments of 0.05 ) to yield a fractional deviation of approximately 4% when a solar - like source star with @xmath78 was magnified 100 times ( @xmath79 ) .
this procedure yielded approximate ranges of the variables @xmath80 and @xmath41 where planets are likely to be detectable .
the value of 4% was chosen as approximately the minimum detectable fractional deviation currently being achieved with 0.3 - 1.3 m class follow - up telescopes .
the above procedure yielded minimum detectable values of @xmath25 of approximately 0.95 , 0.85 and 0.7 for @xmath24 equal to @xmath76 , @xmath46 and @xmath77 respectively . for values of @xmath25 lying outside the einstein ring , the well - known @xmath25 to @xmath81 symmetry ( griest & safizadeh 1998 ) yields maximum detectable values of approximately 1.05 , 1.2 and 1.4 respectively .
these limiting values of @xmath25 agree with those reported by han ( 2009 ) who assumed a minimum detectable fractional deviation of 5% .
an increase in the breadth of the plots in fig .
5 as @xmath24 decreases ( right to left ) is evident .
this implies , irrespective of any details , that very low mass planets with @xmath82 and @xmath83 will be detectable in events distributed over a wider range of magnifications than heavier planets with @xmath84 and @xmath85 . in hunting for the lowest mass planets it will therefore be advantageous to monitor events with the broadest range of magnifications possible .
we note that the detection efficiences at the above limiting values of @xmath24 and @xmath25 are approximately 67% if only the fwhm s of light curves are monitored intensively .
this follows because only those planets with planet : star axes at angles @xmath86 to the track of the source star are probed during the fwhm ( rattenbury et al .
2002 ) .
we note also that the left panel of fig .
5 exhibits a bias towards negative values of @xmath21 .
this is caused by a co - operative effect between the stellar caustic at the origin and a pair of planetary caustics above and below the negative x - axis for a planet located just inside the einstein ring , as shown in fig .
a similar effect occurs for planets located just outside the einstein ring and positive values of @xmath21 .
these cooperative effects already aided discoveries of planets , in moa-2009-blg-266 ( muraki et al . 2011 ) and ogle-2012-blg-0461 ( c. han , private communication ) for example , and they could boost the detection rate of low mass planets in future events with magnifications of order a few to several tens . however , the lower magnifications of these events may need to be compensated by brighter source stars .
the previous discussion can be extended by allowing for the relative precisions with which planetary deviations can be measured at different magnifications . as a first approximation we assume poisson statistics for the detected photons in any exposure , and multiply fractional deviations by the square root of amplification to approximate relative detectabilities .
the results for the three @xmath87 combinations shown in fig .
5 are reproduced below in figs .
these include perturbations for negative impact parameters as well as the positive ones shown in fig .
3 . they are almost mirror images of one another ( rattenbury et al . 2002 ) . at various magnifications .
the lines at @xmath88 are the estimated limits of detectability with telescopes currently in use as discussed in the text.,width=264 ] the inclusion of the poisson factor tends to reverse the effect of the diminished deviations shown in fig .
3 at large magnifications . in fig .
7 the largest perturbation occurs at @xmath89 .
although the peak signals at @xmath90 are larger in magnitude , their contributions to @xmath91 in a full analysis would be limited by the narrowness of the spikes . the estimate made in 4 of a minimum detectable fractional deviation of 0.04 at @xmath92 with telescopes currently in use moves to 0.4 on figs .
the calculated perturbations at @xmath52 are seen to be almost at the threshold , as expected .
the results shown in figs . 8 and 9 for lighter planets closer to the einstein ring show the same trend that was found in fig .
5 . the importance of lower magnification events grows until at ( @xmath93 events with magnifications from 50 - 500 are seen to provide approximately equal sensitivities on an event - by - event basis , with a slight preference for negative impact parameters over positive ones .
when one allows for the greater frequency of events with lower magnifications , they are seen to provide the greatest discovery potential . at various magnifications .
estimated limits of detectability with current telescopes are shown at @xmath88.,width=264 ] at various magnifications .
the enhancement at negative impact parameters over positive ones results from the cooperative effect between stellar and planetary caustics discussed in 4.,width=264 ] for planets even closer to the ring the advantage of working at lower magnification becomes more evident .
10 shows results for @xmath94 .
here it is seen that the sensitivity to sub - earths is significant at @xmath57 as low as 50 . at various magnfications .
the large enhancement at negative impact parameters results from the cooperative effect between stellar and planetary caustics described in 4.,width=264 ] fig .
10 exhibits the best of all worlds .
the source star is highly magnified , the planet perturbs the highly magnified image , and the magnification is further enhanced by planetary caustics .
moreover , these properties are predictable .
every event with @xmath95 will enjoy these advantages and offer the prospect of detecting sub - earths close to the einstein ring if they are sufficiently abundant .
to further quantify the detectability of low - mass planets in events of moderately high magnification a sample of simulated events was generated and fully analysed using standard procedures .
the simulation was carried out assuming follow - up telescopes with apertures larger than those presently in use become available in the future .
an expected distribution of events was generated by extrapolating the planetary mass function of cassan et al .
( 2002 ) to planets with masses in the range @xmath96 and separations in the range 1.25 - 5.0 au .
these ranges were subdivided into a @xmath97 array as shown in table 1 . [ cols="^,^,^,^,^,^ " , ] [ t2 ]
there are a large number of variables that differentiate one gravitational microlensing event from another .
it is difficult to allow for every possible contingency even when discussing just one particular characteristic such as planet detection efficiency . in the preceding sections we considered those variables whose values are most likely to affect the detectability of low - mass planets in microlensing events of high magnification .
we focused in particular on the peak value of the magnification @xmath57 , but also considered results for various values of the source - size parameter @xmath41 and the einstein radius crossing time @xmath16 .
these variables are the only ones whose magnitudes are likely to be known with any degree of confidence prior to an event reaching peak magnification . on the basis of their values decisions
may be made on whether or not to observe an event intensively .
we have found that events with @xmath98 , @xmath99 and @xmath100 days offer a sensitive hunting ground for low - mass planets provided their fwhm s are monitored continuously with 1 - 2 m class telescopes .
the frequency of such events should enable selections to be made to avoid troublesome events such as those with variable or potentially spotty source stars , peak data taken partly in poor weather or under moonlight , or those with nearby bright stars .
the planetary perturbations shown in the preceding sections are relatively featureless .
it might therefore appear difficult to distinguish a perturbation caused by a light planet close to the einstein ring from that of a heavy planet further away . to quantify this effect we examined the accuracy with which planets could be recovered from their simulated light curves .
figure 18 shows @xmath101 as a function of @xmath24 and @xmath25 for the light curve shown in fig . 12 in which @xmath102 was 5,206 . in this case the uncertainties in @xmath24 and @xmath25 are seen to be @xmath10310% and @xmath1033% at the @xmath104 level respectively .
19 shows that the uncertainties grow to @xmath10350% and @xmath10310% for the borderline detection in fig .
15 with @xmath102 was 304 .
these analyses also yielded uncertainties of @xmath10320% and @xmath10330% for the above events at the @xmath104 level for the recovered value of the source size parameter @xmath41 .
it is apparent from the above that the richest harvest of planets will occur in events where very low - mass planets are situated close to the einstein ring .
this confirms the expectation given in section @xmath105 that was based on the breadth of the left panel of fig .
it is also evident that definitive detections such as those shown in figs .
10 and 11 require the use of 1 - 2 m class telescopes .
as in other planetary detections by microlensing , measurements of orbital and/or terrestrial parallax can assist to extract absolute values of planetary masses and separations from the microlensing parameters @xmath24 and @xmath25 . in this regard
we note that orbital parallax is unlikely to be measurable in the events we have modelled with @xmath106 days , but that terrestrial parallax may be measurable in some cases .
the rise and fall times of the planetary perturbations shown in fig .
3 are relatively steep , of order 20 minutes .
if these perturbations were intensively monitored from different continents and/or different hemispheres it might be possible to measure terrestrial parallax in some cases .
however , we note that terrestrial parallax has been measured previously in events of very high magnification only ( gould et al . 2009 ; yee et al .
2009 ) . if two or more planetary deviations are present on a light curve , as the planetary mass function of cassan et al .
( 2012 ) suggests could be the case when extrapolated to lower masses , it would be easier to identify the deviations if they are well separated .
two conditions must be met to achieve the above .
first , the projected position angles of the planets should differ by @xmath107 or more in order for the individual deviations not be superimposed on one another , and secondly @xmath33 must be several times the duration of an individual planetary deviation .
this translates to @xmath108 approximately , as shown in fig .
we note that multiplanet events have been reported previously ( gaudi et al , 2008 ; han et al .
2013 ) in events with magnifications 100 and 290 respectively .
however , the detected planets in these cases were all giants .
in recent years the microlensing community has focussed attention on events with the highest observed magnifications , in particular on events with magnification @xmath109 ( gould et al .
as mentioned in 1 this resulted in the discoveries of several planets by telescopes with apertures ranging from 0.3 m to 2.0 m .
the detected planets have separations from their host stars of order a few au and correspondingly cool temperatures .
their measured masses range from a few earth masses to a few jupiter masses .
they provide a representative sample of the distribution of planets between us and the centre of the galaxy that orbit a relatively unbiased sample of host stars .
the discoveries are important because other planetary detection techniques are insensitive to such planets . on the basis of these discoveries the first estimates of the abundances of cool planets in the milky way were made . in an effort to build on these successes
we have proposed a modified procedure for extending the measurements down to earth mass .
the modified strategy makes use of the surprisingly good sensitivity of microlensing events with magnifications @xmath110 to low - mass planets situated close to the einstein ring .
it was found that high quality detections of low - mass planets such as those depicted in figs .
11 and 12 could may be made by densely monitoring the peaks of these events with 1 - 2 m class telescopes . if an extrapolation of the planetary mass function of cassan et al .
( 2012 ) to earth mass is approximately valid , a statistically significant sample of detections could be made by monitoring some tens of events . in recent years
@xmath111 events with magnification @xmath109 have been detected by the moa and ogle collaborations annually .
the inclusion of events with @xmath57 down to 50 should quadruple the event rate , and thus permit some tens of these events with moderately high magnifications to be monitored in good conditions relatively rapidly . in this way
the planetary mass function could be measured down to earth - mass .
in addition , sub - earths and planetary systems with more than one low - mass planet might be found .
results obtained following the above procedure would provide an independent check of measurements made by monitoring events of low magnification ( kim et al . 2010 ) and of entirely independent results obtained from radial velocity ( howard et al .
2010 ; mayor and queloz , 2012 ) and transit measurements ( borucki et al .
support by the marsden fund of new zealand , and comments by andrew gould , gus hazel , yasushi muraki , ken prendini , ian ramsay , rachel street and a referee , are gratefully acknowledged .
99 abe , f. , bennett , d.p . , bond , i.a . ,
eguchi , s. et al . , 2004 ,
science 305 , 1264 bachelet , e. , shin , i - g .
, han , c. , fouque , p. , et al . , 2012 , apj 754 , 73 batista , v. , gould , a. , dieters , s. , dong , s. et al . , 2011 , a&a 529 , a102 bennett , d.p . , bond , i.a .
, udalski , a. , sumi , t. , et al . , 2008 , apj , 684 , 663 bond , i.a . ,
abe , f. , dodd , r.j . ,
hearnshaw , j.b .
et al . , 2002 ,
mnras 331 , l19 borucki , w. , koch , d. , basri , g. , batalha , n. et al .
, 2011 , apj 736 , 19 brown , t. et al . , 2012 , in preparation cassan , a. , kubas , d. , beaulieu , j .-
p . , dominik , m. et al . , 2012 ,
nature 481 , 167 dong , s. , depoy , d.l . ,
gaudi , b.s . ,
gould , a. et al . , 2006 ,
apj 642 , 842 dong , s. , bond , i.a . ,
gould , a. , kozlowski , s. , et al . , 2009 ,
apj 698 , 1826 fouqu@xmath112 , p. , heyrovski , d. , dong , s. , gould , a. , et al . ,
2010 , a&a 518 , a51 gaudi , b.s . ,
naber , m.n .
, sackett , p.c .
, 1998 , apj 502 , l33 gaudi , b.s . , bennett , d.p . ,
udalski , a. , gould , a. et al . , 2008 ,
science 319 , 927 gould , a. , udalski , a. , an , d. , bennett , d.p .
, et al . , 2006 ,
apjl 644 , l37 gould , a. , udalski , a. , monard , b. , horne , k. , et al . , 2009 ,
apjl 698 , l147 gould , a. , dong , s. , gaudi , b.s . ,
udalski , a. , et al . , 2010 , apj 720 , 1073 griest , k. , safizadeh , n. , 1998,apj , 500 , 37 han , c. , 2009 , apj 691 , 452 han , c. , kim , d. , 2009 , apj 693 , 1835 han , c. , udalski , a. , choi , j .-
y , yee , j. , 2013 , apj 762 , l28 horne , k. , snodgrass , c. , tsapras , y. , 2009 , mnras 396 , 2087 howard , a. , marcy , g. , johnson , j. , fischer , d. et al .
, 2010 , science 330 , 653 janczek , j. , fukui , a. , dong , s. , monard , b. , et al .
, 2010 , apj 711 , 731 kim , s - l . , park , b - g . , lee , c - u .
, yuk , i - s .
, et al . , 2010 , proc .
spie7733 , 77333f liebes , s. , 1964 , phys .
133 , 835 mayor , m. , queloz , d. , 2012 , new astronomy reviews 56 , 19 miyake , n. , sumi , t. , dong , s. , street , r. , et al . , 2011 , apj 728 , 120 muraki , y. , han , c. , bennett , d.p . ,
suzuki , d. , et al .
, 2011 , apj 741 , 22 paczynski , b. , 1996 , ann .
astron . astrophys .
34 , 419 rattenbury , n.j . , bond , i.a . ,
skuljan , j. , yock , p.c.m . , 2002 ,
mnras 335 , 159 rhie , s.h . ,
bennett , d.p , becker , a.c . ,
peterson , b.a .
et al . , 2000 ,
apj 533 , 378 sumi , t. , kamiya , k. , udalski , a. , bennett , d.p . , et al . , 2011 ,
nature 473 , 349 udalski , a. , jaroszynski , m. , paczynski , b. , kubiak , m. et al . , 2005 , apjl 628 , l109 wambsganss , j. , 1997 , mnras 284 , 172 yee , j.c . , udalski , a. , sumi , t. , dong , s. , et al .
, 2009 , apj 703 , 2082 yee , j.c .
, shvartzvald , y. , gal - yam , a. , bond , i.a .
, 2012 , apj 755 , 102 yee , j.c .
, hung , l .- w . , bond , i.a . , allen , w. et al . ,
2013 , submitted
|
a measurement by microlensing of the planetary mass function of planets with masses ranging from @xmath0 to @xmath1 and orbital radii from 0.5 to 10 au was reported recently . a strategy for extending the mass range down to @xmath2
is proposed here .
this entails monitoring the peaks of a few tens of microlensing events with moderately high magnifications with 1 - 2 m class telescopes .
planets of a few earth masses are found to produce deviations of @xmath3 to the peaks of microlensing light curves with durations @xmath4 hr in events with magnification @xmath5 if the projected separation of the planet lies in the annular region @xmath6 .
similar deviations are produced by earth mass planets in the annular region @xmath7 .
it is possible that sub - earths could be detected very close to the einstein ring if they are sufficiently abundant , and also planetary systems with more than one low mass planet .
[ firstpage ] gravitational lensing : micro stars : planetary systems
|
organic liquid scintillators are used in large quantities for rare event detection in particle astrophysics .
the main objective in these experiments is the real time spectroscopy of neutrinos from steady - state sources such as the sun , nuclear reactors and from beta decays in the crust and mantle of the earth , as well as from transient sources such as supernovae . despite the large target mass of several hundreds of tons ,
the signal rates of the steady - state sources are typically in the range of a few events per day down to a few events per year at mev or sub - mev energies .
thus , background signal rates created by radioactivity and cosmic ray interactions need to be extremely low .
low backgrounds can be achieved by locating the detectors deep underground to suppress the cosmic ray muon flux , shielding the scintillator target against the ambient radioactivity from the surrounding rocks , and suppressing and removing radioactive impurities present in trace amounts in the detector and ancillary systems as well as in the liquid scintillator itself .
this concept of background reduction has been pioneered by the borexino collaboration @xcite in the counting test facility ( ctf ) @xcite and is implemented in the borexino detector , and similarly , in the kamland experiment @xcite .
this paper summarizes the study of 1,2dimethyl4(1phenylethyl)benzene ( phenyl o xylylethane , pxe ) , a new scintillator solvent the key characteristics of which are its high density ( @xmath54 ) and high flash point ( 145@xmath55c ) .
this scintillator solvent has been investigated as a ` back - up solution ' for the borexino experiment .
the design of choice is based on 1,2,4-trimethylbenzene ( pseudocumene , pc ) , both as buffer liquid and as neutrino target .
since the density of pxe is close to that of water , even large scintillator targets can be submerged in water - serving as a shield against ambient radiation - while creating only modest buoyancy forces on the scintillator containment vessel . moreover , a detector with water as shield and pxe as target material provides a substantial higher fiducial target mass because of the improved shielding performance against external radiation compared to a detector with identical dimensions , but with both shield and target of organic liquids with standard densities ( @xmath56 g/@xmath57 ) . finally , a pxe - water configuration reduces the overall inventory of organic liquid in the detector systems , e.g. to about one fourth in the case of borexino .
this , depending on national regulations , may have an impact on the legal classification of the detector systems and therefore on the safety and operational aspects of the experiment .
a further asset of pxe is its high flash point simplifying safety systems relevant for transportation , handling and storage . according to regulations by the united nations ( un )
, pxe is legally non - hazardous for transportation purposes and no special united nation number code applies @xcite . in a paper by majewski et al .
@xcite pxe has been described as a relatively safe solvent with very low toxicity compared to standard liquid scintillators .
the double chooz reactor neutrino experiment @xcite , aiming at a measurement of the mixing angle @xmath58 , will use as target a scintillator mixture based on pxe .
the study reported here of pxe as a solvent for a low - background scintillator for solar neutrino spectroscopy was carried out within the borexino project .
research on pxe scintillators started in 1995 with laboratory measurements focusing on optical properties and radio purification techniques with solid columns .
after completion of the laboratory scale study , about five tons of pxe solvent for testing on prototype scale with the counting test facility ( ctf ) of borexino were acquired in 1996 .
fluors were added on site in hall c of the laboratori nazionali del gran sasso ( lngs ) and the final scintillator was purified with module-0 , a solid column purification and liquid handling system @xcite .
the ctf was loaded with pxe scintillator in october 1996 and first data were acquired until january 1997 . the quality of data was limited because only a small fraction of the photomultipliers in the ctf were operational .
the pxe scintillator was unloaded from the ctf after the shut down of the detector in july 1997 , and moved back into the storage tanks of module-0 .
further batch purification operations were carried out during the period october until december 1997 .
samples for neutron activation analysis were taken to monitor the achieved radio purity after each operation .
after reconstruction of the ctf during 1999 , the pxe scintillator was reloaded into the ctf and measured from june to september 2000 .
the main objective during this period was the analysis of radio purity and optical properties in a large volume detector . beyond this scope ,
physics limits on electron instability @xcite , nucleon instability @xcite , neutrino magnetic properties @xcite , and on violations of the pauli exclusion principle @xcite could be derived from measurements with the pxe scintillator .
the paper is structured as follows : section 2 summarizes the physical and chemical properties of the pxe solvent .
section 3 describes the optical properties of the solvent as well as the mixed scintillator .
section 4 is dedicated to the large scale test of the pxe scintillator with the ctf including the scintillator preparation , purification and analysis of trace impurities , and conclusions are given in section 5 .
pxe is a clear , colorless liquid with an aromatic odor .
it is an industrial product with different applications , as for example : insulating oil in high voltage transformers and capacitors and as oil for pressure sensitive paper .
pxe is produced by reacting styrene and xylene using an acidic catalyst .
it is then washed with water and distilled to improve the purity .
its final industrial purification step uses a solid column .
pxe has the molecular formula @xmath59 with a weight of 210.2 g / mol .
its molecular structure is shown in figure[f : pxemol ] .
chemical and physical details which are of relevance for detector design , safety and operational aspects are listed in table[t : pxe - chem - phys ] .
the key features are its high density of 0.988 g/@xmath57 , its low vapor pressure and high flash point of 145 @xmath55c .
it is therefore classified as a non - hazardous liquid . for our test , the standard production scheme at koch chemical company , corpus christi , texas
, usa was modified omitting the final column purification at the company , since it was expected that the clay column material might leach off radio active impurities , such as uranium and thorium . instead a column purification system was built and operated using silica gel in the underground laboratories of the lngs .
details are discussed below .
the optical and scintillation properties of the pure pxe solvent , of selected fluors , and of pxe - fluor mixtures have been investigated by uv / vis spectrometry , by fluorimetry and by excitation with ionizing radiation .
the objective was to optimize the scintillator performance by maximizing its light yield and attenuation length and minimizing its scintillation decay time .
methods to remove optical impurities were studied , since impurities can potentially reduce light yield and attenuation length .
all optical properties presented in this section are derived from laboratory size samples up to a few hundred ml s .
attenuation length measurements typically are done in ` one - dimension ' only .
scattered light , elastic or inelastic , is undetectable in these measurements whereas in large volume applications with 4@xmath60 geometry , the scattered photons are not necessarily lost .
the performance of the pxe scintillator in a large volume detector , taking scattering into account , has been studied in ctf and is discussed in sec .
pxe diluted in cyclohexane shows an absorption maximum at 267 nm . the emission spectrum after excitation at this wavelength peaks at about 290 nm as displayed in figure [ f : pxe - emi - abs ] .
the fluorescence lifetime after excitation at @xmath61 measured in diluted solutions of cyclohexane shows an exponential decay with a lifetime @xmath62 of @xmath63ns .
preliminary samples obtained from koch had various optical impurities with absorption bands around 300 , 325 , 360 and 380 nm .
they could be observed both by uv / vis and fluorimetric measurements . passing the solvent through a column with acidic alumina reduced the absorption peaks . the band at 380 nm was reduced most efficiently .
the pxe used for the 5 t test , as described section 4 , still had various optical impurities , but at much reduced levels .
the attenuation length at 430 nm ( @xmath64 ) , defined as @xmath65 , with the initial intensity @xmath66 and attenuated intensity @xmath67 after the optical path length @xmath68 , was @xmath69 in the pure pxe solvent .
attenuation lengths were measured with a varian cary 400 uv / vis photospectrometer in 10 cm cuvettes .
a tertiary scintillator system was chosen in order to shift the emission wavelength to about 430 nm , well above the absorption bands of the residual optical impurities .
this is achieved by using 1,4-diphenylbenzene ( para - terphenyl , p - tp ) as primary solute and 1,4-bis(2-methylstyryl)benzene ( bis - msb ) as secondary solute .
absorption and emission spectra of pxe in cyclohexane and those of the fluors in pxe are shown in figures [ f : pxe - emi - abs ] and [ f : pxe - fluors ] .
the latter shows that the emission spectrum of p - tp is satisfactorily matched by the absorption spectrum of the wavelength shifter bis - msb , and therefore an efficient energy transfer is expected in this solvent .
the scintillator properties were tested with p - tp concentrations at 2 and 3 g / l ( close to the solubility limit ) and with bis - msb at 20 mg / l . a scintillation yield of 88 ( 93)% for pxe with an addition of p - tp ( 2.0 ( 3.0 ) g / l)/bis - msb(20
mg / l ) with respect to a scintillator based on 1,2,4-trimethylbenzene ( pc ) and 1.5 g /
l of 2,5-diphenyloxazole ( ppo ) has been found ( uncorrected for the pmt sensitivity ) .
attenuation lengths of @xmath70 m to 3.2 m of the scintillator mixture have been measured depending on the sample treatment ( cf .
section 4.2 ) .
the fluorescence decay time of the p - tp(2.0 g / l)/bis - msb(20 mg
/ l ) scintillator mixture measured by fluorimetry after excitation at 267 nm shows a fast component @xmath71 of 3.7 ns . at 3
g / l p - tp , the decay constant is shortened to 3.2 ns .
the photon emission probability density function ( pdf ) was further studied with ionizing radiation in order to investigate the contribution of long lived triplet states and the possibility to use the emission time for discrimination of alpha versus beta particles ( pulse shape discrimination ) .
the response to gamma radiation has been measured for a concentration of 2.0
g / l and 3.0 g / l p - tp with a @xmath72cs source , and to alpha radiation at a concentration of 3.0
g / l using a @xmath73po alpha source @xcite .
the pdf can be parametrized by the weighted sum of four exponentials @xmath74 with the parameters given in table [ t : em ] .
the emission time distribution is shown in fig .
[ fpxecs137a ] .
in addition to an increased slow component which can be used for pulse shape discrimination , alpha particles emit less light because of the high ionization density compared to electron or gamma radiation .
the quenching factor is mainly solvent dependent and has been measured for different alpha decays in the @xmath1u chain , using @xmath75rn loaded scintillator .
the results are given in table[t : quench ] @xcite .
light attenuation of the standard mixture ( pxe/(2.0 g / l)p - tp/(20 mg / l)bis - msb ) which has been used in the ctf ( cf .
4 ) has been measured at various steps during the preparation of the scintillator in 1996 and after filling the ctf .
after several purification steps of the mixed scintillator through a silica gel column in module-0 in order to remove radio - impurities ( cf .
sections below ) , attenuation lengths in the range between 2.6 and 3.0 m at 430 nm were measured .
after completion of the pxe measurements with ctf and subsequent batch purification operations , the scintillator has been stored in barrels under nitrogen atmosphere . in 2003 , this scintillator was used in the frame of the lens project @xcite and optical properties were remeasured to check for degradation with time .
the attenuation length as well as the light yield were unchanged with respect to the measurements in 1996 .
passing this scintillator mixture through a weak acidic alumina column increased the attenuation lengths to 12 m as displayed in figure [ fpxeatt ] while retaining the scintillation yield @xcite .
the main objectives of the large scale test in the borexino prototype detector ( counting test facility , ctf @xcite ) were the study of 1 ) optical properties on a large scale , 2 ) achievable levels of radioactive trace contaminations of the pxe based scintillator , 3 ) the performance of scintillator purification with a silica gel column , and 4 ) of the liquid handling system , module-0 .
the purity levels required for detection of low - energy solar neutrinos , in particular neutrinos from the @xmath76be electron capture , are at the @xmath77bq / m@xmath78 level corresponding to concentration of @xmath1u and @xmath79th of @xmath80 g / g . for detailed specifications of solar neutrino rates and background requirements ,
the reader is referred to refs .
as the @xmath1u progenies @xmath75rn , @xmath73pb , @xmath73bi and @xmath73po are not necessarily in equilibrium with the progenitor activity , we studied the contamination levels of @xmath1u ( as well as @xmath79th and several other isotopes ) _ directly _ with neutron activation analysis ( naa ) at levels of @xmath81g / g and below .
the short lived progenies are not accessible with naa .
their contamination levels were studied in the ctf together with other backgrounds , as for example the radiogenic produced @xmath51c .
new techniques to remove radioactive isotopes by solid column purification were tested for the first time in a ton scale experiment .
furthermore , the preparation , handling and purification of several tons of liquid scintillator with module-0 was the first test of a subsystem of the liquid handling system to be used in the scintillator operations in borexino .
about five tons of pxe solvent was procured from koch special chemical company , texas , usa . to ensure a controlled procedure , the solvent was loaded by us at the company site in three specially modified and cleaned stainless steel transport containers and shipped to the gran sasso underground laboratories .
the pxe solvent was transferred from the transport containers into module-0 , a liquid handling and purification system , specially built for the pxe test .
it can be used for volumetric loading and unloading of liquid scintillator to / from the ctf inner vessel ( iv ) , for fluor mixing , purification of liquid scintillator with a silica gel column , rn degassing of liquid scintillator by nitrogen sparging , and spray degassing as well as for water extraction .
module-0 consists of a high and low pressure manifold system which are connected by pumps to build up the pressure difference of typically 2 atm .
the manifolds are connected to tanks , columns , filters etc . to allow a variety of flow paths and operations .
the system includes two 7 m@xmath78 electropolished pressure tanks ( ep1/2 ) , two 1 m@xmath78 process tanks ( bt1/2 ) equipped with nitrogen spargers at the bottom and spray nozzles at the top ( for a turbulent injection of liquid together with nitrogen gas ) , pumps , flow meters , millipore filters ( 0.5 , 0.1 and 0.05micron ) and one 70l high pressure column purification unit .
all tanks are connected to a high purity nitrogen gas system to provide a nitrogen blanket at typically 30 mbar overpressure .
the complete system has been designed according to ultra - high - vacuum standards to avoid contaminations from the environmental air , in particular of @xmath75rn and @xmath82kr .
only stainless steel and teflon are in contact with the liquids .
metal surfaces are electro polished and welds carried out with thorium free welding rods .
module-0 has been constructed in a class 1000 clean room .
after completion , all surfaces exposed to scintillator were cleaned following a detailed procedure to remove surface contaminations as radon progenies @xmath73pb , @xmath73bi and @xmath73po .
details about the up - graded system are given in @xcite .
the fluors ( p - terphenyl and bis - msb , both from sigma - aldrich co. , scintillation grade ) were sieved and then added without further purification to ep1 through a glove box connected to the top inlet flange which was flushed with nitrogen while the pxe solvent was agitated with a nitrogen flux from the bottom inlet to facilitate dissolving p - tp . to accelerate the solvation of p - tp ,
the ep1 tank was heated to 38 @xmath55c for four days under continuous nitrogen agitation .
the final scintillator mixture had a concentration of 2.0g / l p - tp and 20mg / l bis - msb .
one objective of the ctf test was to study the performance of scintillator purification with a solid silica gel column .
preceding laboratory tests of the column purification with pxe / p - tp with and without the addition of radio tracer showed a clear reduction of metal impurities @xcite . the silica gel used during the ctf test
has been radio assayed with hp germanium spectrometry and by direct measurements of the emanated radon with proportional counters .
the results are listed in table[t : sigel - radon ] . despite the high bulk impurities of the merck silica gel
, we did not observe any measurable carry over into the scintillator apart from @xmath75rn .
for further use in borexino we have found silica gel material with improved radio purity , in particular a radon emanation rate of @xmath83 mbq / kg .
the scintillator components had been mixed as received from the supplier ; no special pretreatment had been given to the fluor and the wavelength shifter .
the ready mixed pxe scintillator was then passed through a column filled with silica gel ( merck , silica gel 60 , 25 - 70 mesh astm ) at a flow rate of typically 100 l / hour .
radon was removed by purging the scintillator after passing the column as displayed schematically in figure[f : column ] .
samples for neutron activation analysis were taken prior and after all major steps of the scintillator preparation and purification : * sample ( 1 ) * was collected from module-0 after addition of p - tp and bis - msb to the pxe solvent ( without purification ) . * sample ( 2 ) * was collected during the filling of the iv .
the complete scintillator had passed once a column of about 80 cm height ( 15 kg ) which had been exchanged against fresh silica gel after the first two tons had run through the column , and once with a height of about 50 cm , i.e. in total two times .
* sample ( 3 ) * has been taken after the scintillator was circulated from the iv through the column ( re - filled to a height of 80 cm ) and back into the iv .
the flow rate was adjusted such that one cycle ( 5000 l ) took two days .
the scintillator was circulated for four days , i.e. two cycles .
subsequently the scintillator was unloaded from the iv into the ep1-tank .
* sample ( 4 ) * was taken after completion of a water extraction in the ep1 tank . *
sample ( 5 ) * was collected in module-0 after circulating the scintillator from ep1 through the column back to ep1 .
the column packing had not been exchanged ( same as in ( e ) ) .
this final operation lasted for eight days with a flow rate of 3600 l / day , i.e. about 6 cycles . combining neutron activation and low level counting methods
, we developed a novel analytical method with sensitivities of @xmath81g / g and below for @xmath1u and @xmath79th in liquid scintillators .
for this purpose about 250 g samples of liquid scintillator were irradiated at a neutron flux of @xmath84 up to 100 hours at the research reactor in garching .
the long lived primordial radio nuclides are transformed into short - lived radio nuclides ( e.g. @xmath85 , @xmath79th@xmath86pa ) , thus providing a higher specific activity with respect to their progenitors .
after irradiation , liquid - liquid and ion exchange techniques were applied to separate the radio nuclides of interest from interfering activities .
coincidence counting methods @xcite , e.g. @xmath87-@xmath88-conversion electron for @xmath89np and @xmath87-@xmath88 for @xmath90pa , further increase the sensitivity .
details can be found in refs .
@xcite .
the main results of the neutron activation analysis of the scintillator samples ( 1 ) to ( 5 ) are summarized in table [ t : naapxe1 ] @xcite .
we report the concentrations of @xmath1u , @xmath79th and other long - lived isotopes which contribute to the background in borexino . in order to illustrate the performance of the silica gel column , also the concentrations for several non - radioactive metal isotopes are given .
an overall reduction between 1 and 3 orders of magnitude has been reached for all elements where a positive value ( above the detection limit ) could be measured before the purification . except for potassium , where the purity needed for borexino
is below the detection limit of the naa , the requirements for borexino are met for all of the long lived radio nuclides .
new purity records in organic liquid scintillators have been achieved for uranium at @xmath91 , and for thorium at @xmath92 .
the borexino prototype detector ctf is a highly sensitive instrument for the study of backgrounds in liquid scintillators at energies between a few tens of kev and a few mev .
it consists of a transparent nylon balloon with 2 m diameter containing 4.2m@xmath78 of liquid scintillator ( inner vesssel , iv ) .
100 pmts with light concentrators mounted on a 7 m diameter support structure detect the scintillation signals ( optical coverage 20% ) .
the whole system is placed inside a cylindrical steel tank ( 11 m in diameter , 10 m height ) that contains 1000tons of ultra - pure water in order to shield against external @xmath93 rays from the pmts and other construction material as well as neutrons from the surrounding rock . in the upgraded version of the ctf detector ( after 1999 ) , 16 additional pmts mounted on the floor of the water tank detect the cherenkov light created by muons transversing the water buffer and are used as a muon veto .
details of the ctf detector and results of first measurements with pseudocumene scintillator are given in refs .
@xcite .
direct counting after first loading of the pxe scintillator into the ctf in 1997 did not provide conclusive results because of photo - tube and electronics problems of the detector system that hindered data evaluation .
after an upgrade of the ctf detector and the liquid handling system during 1998 and 1999 , the same scintillator was filled again into the ctf in summer 2000 , after passing once through a 40l silica gel column .
the loading of the scintillator into the inner vessel was done in four batches of 1 ton each separated by short periods of data taking .
undisturbed data taking with 4.2 tons of pxe was going on from july 16 until september 5 , 2000 , in total 52 days .
afterwards , a series of calibration measurements with a @xmath75rn point source were carried out where the source was moved inside the inner vessel to map the detector response and tune the position reconstruction software .
the pulse height - energy relation can be derived from the @xmath94po alpha peak ( cf .
figure [ f : popeak ] ) together with the measured quenching factors as given in table[t : quench ] . due to its short half life of 164@xmath77s ,
the @xmath94po decay can easily be tagged .
the @xmath94po stems mainly from @xmath75rn introduced during scintillator loading and therefore is homogeneously distributed in the scintillator volume .
the alpha particle has an energy of 7.69mev which in the scintillator is quenched to an equivalent beta energy of @xmath95kev .
the measured peak position in the ctf corresponds to @xmath96photo electrons leading to a yield of @xmath97 photo electrons per mev .
the pulse height
energy relation is also a parameter of the @xmath51c fit ( see below ) , which is done in the energy range from 70 to 150kev , resulting in a somewhat higher value of 340 photo electrons per mev .
a more detailed analysis of the photo electron yield including a model with ionization quenching , has been presented in ref .
a value of @xmath98 photo electrons per mev has been derived for the asymptotic yield , ie .
for scintillation light created by electrons with energies large compared to their rest mass .
the energy resolution derived from the width of the @xmath94po alpha peak corresponds to @xmath99 , or @xmath100 .
it is expected to scale with @xmath101 .
a similar value of @xmath102 is obtained from the spectral analysis of the @xmath51c spectrum .
the spatial resolution was studied with a localized @xmath75rn source at various positions inside the inner vessel @xcite .
values of @xmath103 , @xmath104 at 600kev were obtained at the detector s center ( cf .
figure [ 793resol ] ) .
the spatial resolution degrades up to 15% close to the surface of the inner vessel .
the photon arrival time distribution was studied with the source located at the center of the detector .
figure [ f : beta ] displays the time distribution for scintillation photons from the @xmath94bi beta decay .
the data are compared with a monte carlo simulation which includes the scintillation decay time from laboratory measurements given in table [ t : em ] , elastic and inelastic scattering and the pmt time jitter .
monte carlo and data show good agreement for arrival times up to 25ns .
the data show a larger slow component than the mc simulation , which could be due to reflected photons and/or pmt late pulses which are not fully accounted for in the simulation . to distinguish alpha from beta particles by pulse shape discrimination , the analog sum of the charge signal from all pmts is split into three identical signals and fed into a charge sensitive adc with different time delays .
the total charge is derived from integration over the entire charge signal 0 - 500ns , while the slow component is derived from the time windows 32 - 532ns and 48 - 548ns respectively .
the tail - to - total ratio r@xmath105 or r@xmath106 can then be used as discrimination parameter @xcite as displayed in figure [ f : alphabeta ] .
the discrimination efficiency can be derived from a clean data sample of @xmath107 and @xmath108 events in the corresponding energy range . only the @xmath94bi(@xmath109)-@xmath94po(@xmath107 ) decay sequence can unambiguously be identified as pure @xmath107 or @xmath108 events due to the short half life of @xmath94po .
only @xmath94bi events in the same energy range as the @xmath94po events ( 0.6 - 1.3mev ) were considered . fixing
the @xmath109 acceptance efficiency at 98 % leads to an @xmath107 identification efficiency of 84.6% ; applying a radial cut on the @xmath94bi and @xmath94po events of @xmath110 cm leads to an increased @xmath107 identification efficiency of 92.4% .
this improvement is due to the fact that for events in the outer regions of the inner vessel the difference in the light path to different pmts is higher , and therefore the fraction of light registered at late times is higher .
also other methods for @xmath111 discrimination like the gatti optimum filter method have been studied with the ctf @xcite , and will be implemented in borexino .
the counting rate below 200kev is dominated by the @xmath51c beta decay with an endpoint of 156kev .
the @xmath51c concentration in the scintillator was determined by fitting a convolution of the theoretical beta spectrum and the energy dependent detector resolution function , plus a background contribution to the measured spectrum in the energy range from 70 to 150kev @xcite . the low energy spectrum together with the fit is displayed in figure [ f : c14-ctf2 ] .
at energies below 70kev there is a background contribution from cherenkov events produced by @xmath93 s in the shielding water . from the fit we derive the @xmath51c activity which translates to a ratio of @xmath112 .
the systematic error is dominated by the uncertainty of the total scintillator mass .
the @xmath1u decay chain is assayed via the delayed coincidence of the @xmath75rn progenies @xmath94bi ( @xmath113 ) and @xmath94po ( @xmath107 ) with a half life of 164@xmath77s . assuming secular equilibrium of the decay chain which is likely to be broken due to the long - lived @xmath114ra and the gaseous @xmath75rn the measured @xmath94bi-@xmath94po activity can be expressed as uranium - equivalent .
after loading the ctf an initial @xmath94bi-@xmath94po rate of about 90 counts per day was observed which was clearly related to @xmath75rn introduced during the filling process , as it decayed away with the @xmath75rn half life of 3.8 days .
the last two 1-week - periods of data taking with a total counting time of 7.32 days ( period i ) and 6.98days ( period ii ) were used for the analysis , when the contribution from the initial @xmath75rn contamination was below 1 count per day . the detector efficiency ( due to the dead time connected with
the read out of each event ) was 91% during that period .
the @xmath94bi-@xmath94po events were selected by applying the following cuts : no muon veto trigger ( efficiency @xmath115 99% ) ; energy of the first event @xmath115 300kev ( efficiency 95% ) ; energy of the following event between 0.6 and 1.25mev ( 99% ) ; coincidence time between 5@xmath77s and 800@xmath77s ( 94% ) .
224 ( 199 ) candidate events survived the cuts for period
i ( ii ) , corresponding to a @xmath75rn activity of 102@xmath77bq / m@xmath78 ( 99@xmath77bq / m@xmath78 ) .
this activity however was not homogeneously distributed within the scintillator volume , but most of the events were localized along the vertical symmetry axis of the detector .
an artefact due to false reconstruction could be excluded after calibration with point like sources which were located both on and off axis .
though the origin of these events could not be fully resolved , it is excluded that they are related to the intrinsic scintillator impurities . to derive a number for the residual @xmath75rn concentration homogeneously distributed in the scintillator ,
a cylindrical cut around the vertical axis was applied .
for @xmath116 we get a @xmath75rn activity of @xmath117 in period i and @xmath118 in period ii , or a @xmath1u equivalent concentration of @xmath119 and @xmath120 . due to
the above stated @xmath75rn problem this value must be considered as an upper limit of the internal @xmath1u contamination .
further radio - isotopes from the uranium decay chain of concern are the @xmath73pb progenies @xmath73bi and @xmath73po .
secular equilibrium typically is disturbed even within this sub - chain , as well as with respect to the progenitor @xmath114ra because of the characteristic lifetimes and the different chemistries involved . from a spectral analysis , about 100 to 200 decays per day with alpha - like pulse shapes and with energy depositions quenched to approximately 0.5 mev are attributed to @xmath73po decays in the scintillator @xcite .
a limit for the intrinsic @xmath79th contamination can be derived via the delayed @xmath109-@xmath107 coincidence of its progenies @xmath121bi-@xmath121po with a half life of 299ns .
it can be distinguished from the @xmath94bi-@xmath94po coincidence by the higher energy of the @xmath107 decay and the shorter coincidence time , though the latter has to be considered as a background .
secular equilibrium in the @xmath79th chain is usually observed for @xmath122th and its progenies .
the following cuts were applied to select the @xmath121bi-@xmath121po events : no muon veto trigger ; energy of the first event @xmath115 300kev ( efficiency 85% ) ; energy of the second event @xmath115 800kev ( 100% ) ; coincidence time between 50 and 1500ns ( 86% ) ; due to a contamination at the bottom of the vessel only decays in the upper hemisphere of the detector were considered ( 50% ) .
the overall efficiency of the cuts for the selection of the @xmath121bi-@xmath121po events is 37% . during the whole period of undisturbed data taking , in total 25.8days of live time ,
12 candidates were detected , leading to a @xmath79th equivalent concentration of @xmath123 .
an estimation of the remaining background for the neutrino detection in borexino is provided by the subtraction of all known contributions from the background spectrum measured in the ctf . in order to reduce the contribution from the external background ,
a radial cut ( @xmath124 m ) has been applied .
also , all correlated events have been tagged and discriminated . in the neutrino window ( nw , @xmath125 ) ,
the counting rate then is 120 events per day and ton . as next step ,
all muon induced events have been subtracted ( 74 events per day and ton ) . in this inner region of the detector , alphas can be discriminated via their pulse shape with an efficiency of @xmath126% , leading to 48 events per day and ton .
the resulting background spectrum is shown in figure [ internalback ] .
clearly visible are the contributions from @xmath51c the remaining alphas from the @xmath1u and @xmath79th chain , and a @xmath127k @xmath93 peak , which could be identified as external background originating from the vectran strings that hold the inner vessel in place ( a dedicated measurement of the strings showed a concentration of 45 ppm k ) .
the large scale test of a scintillator based on pxe in the ctf showed that its optical properties and the detector performance are comparable to a pseudocumene based scintillator .
the purity levels required for the detection of low energy solar neutrinos can be achieved by purification of the scintillator with a silica gel column . though the very low @xmath1u and @xmath79th concentrations measured by naa ( @xmath128g / g and @xmath129g / g ) could not be confirmed by the ctf measurements , the remaining activity levels were close to the detector s sensitivity limit and largely influenced by systematics e.g. due to external background and surface contamination .
pxe based scintillator has been investigated for application in borexino as detector medium for low energy solar neutrino spectroscopy .
key questions under study were the optical properties , radio purity and new techniques of purification by solid column extraction .
both laboratory measurements and the large scale operations in the ctf showed that pxe scintillator as target and water as buffer medium is a viable solution for borexino and it was rated as a backup solution .
the default configuration in the borexino design is based on pseudocumene as scintillator as well as buffer medium . because of its superior safety characteristics and the better self - shielding due to its higher density , future applications of pxe in underground experiments are conceivable , as for example the search for non - vanishing value of @xmath58 with reactor neutrinos , the search for neutrinoless double beta decay with metal loaded scintillators , the study of neutrinos from the interior of the earth and from supernovae , as well as the study of proton decay via the k - meson decay channel .
f.x . hartmann would like to thank the extra - ordinary services of the koch speciality chemical company and the alcoa company as regards the solid column design .
|
we report on the study of a new liquid scintillator target for neutrino interactions in the framework of the research and development program of the borexino solar neutrino experiment .
the scintillator consists of 1,2dimethyl4(1phenylethyl)benzene ( phenyl o xylylethane , pxe ) as solvent and 1,4-diphenylbenzene ( para - terphenyl , p - tp ) as primary and 1,4-bis(2-methylstyryl)benzene ( bis - msb ) as secondary solute .
the density close to that of water and the high flash point makes it an attractive option for large scintillation detectors in general .
the study focused on optical properties , radioactive trace impurities and novel purification techniques of the scintillator .
attenuation lengths of the scintillator mixture of 12 m at 430 nm were achieved after purification with an alumina column . a radio carbon isotopic ratio of @xmath0 has been measured in the scintillator .
initial trace impurities , e.g. @xmath1u at @xmath2 g / g could be purified to levels below @xmath3 g / g by silica gel solid column purification . borexino collaboration h.o . back@xmath4 , m. balata@xmath5 , a. de bari@xmath6 , t. beau@xmath7 , a. de bellefon@xmath7 , g. bellini@xmath8 , j. benziger@xmath9 , s. bonetti@xmath8 , a. brigatti@xmath8 , c. buck@xmath10 , b. caccianiga@xmath8 , l. cadonati@xmath11 , f. calaprice@xmath9 , g. cecchet@xmath6 , m. chen@xmath12 , a. di credico@xmath5 , o. dadoun@xmath13 , d. dangelo@xmath14 , a. derbin@xmath15 , m. deutsch@xmath16 , f. elisei@xmath17 , a. etenko@xmath18 , f. von feilitzsch@xmath19 , r. fernholz@xmath9 , r. ford@xmath20 , d. franco@xmath8 , b. freudiger@xmath21 , c. galbiati@xmath9 , f. gatti@xmath22 , s. gazzana@xmath5 , m.g .
giammarchi@xmath8 , d. giugni@xmath8 , m. gger - neff@xmath23 , a. goretti@xmath5 , c. grieb@xmath19 , e. de haas@xmath9 , c. hagner@xmath24 , w. hampel@xmath10 , e. harding@xmath25 , f.x .
hartmann@xmath10 , t. hertrich @xmath19 , h. hess@xmath19 , g. heusser@xmath10 , a. ianni@xmath5 , a.m. ianni@xmath9 , h. de kerret@xmath7 , j. kiko@xmath10 , t. kirsten@xmath10 , g. korga@xmath26 , g. korschinek@xmath19 , y. kozlov@xmath18 , d. kryn@xmath7 , m. laubenstein@xmath5 , c. lendvai@xmath19 , f. loeser@xmath9 , p. lombardi@xmath8 , s. malvezzi@xmath8 , j. maneira@xmath27 , i. manno@xmath28 , d. manuzio@xmath22 , g. manuzio@xmath22 , f. masetti@xmath17 , a. martemianov@xmath29 , u. mazzucato@xmath17 , k. mccarty@xmath9 , e. meroni@xmath8 , l. miramonti@xmath8 , m.e .
monzani@xmath8 , p. musico@xmath22 , l. niedermeier@xmath30 , l. oberauer@xmath19 , m. obolensky@xmath7 , f. ortica@xmath17 , m. pallavicini@xmath22 , l. papp@xmath26 , s. parmeggiano@xmath8 , l. perasso@xmath8 , a. pocar@xmath9 , r.s .
raghavan@xmath31 , g. ranucci@xmath8 , w. rau@xmath32 , a. razeto@xmath22 , e. resconi@xmath33 , a. sabelnikov@xmath8 , c. salvo@xmath22 , r. scardaoni@xmath8 , d. schimizzi@xmath9 , s. schnert@xmath34 , k.h schuhbeck@xmath35 , e. seitz@xmath19 , h. simgen@xmath10 , t. shutt@xmath9 , m. skorokhvatov@xmath18 , o. smirnov@xmath36 , a. sonnenschein@xmath37 , a. sotnikov@xmath36 , s. sukhotin@xmath18 , v. tarasenkov@xmath18 , r. tartaglia@xmath5 , g. testera@xmath22 , d. vignaud@xmath7 , r.b .
vogelaar@xmath4 , v. vyrodov@xmath18 , m. wojcik@xmath38 , o. zaimidoroga@xmath36 , g. zuzel@xmath38 now at technische universitt mnchen , james franck strae , d-85747 garching , germany + @xmath39 now at sudbury neutrino observatory , inco creighton mine , p.o.box 159 lively , ontario , canada , p3y 1m3 + @xmath40 now at universitt hamburg , luruper chaussee 149 , d-22761 hamburg , germany + @xmath41 now at dept . of physics , queen s university stirling hall , kingston , ontario k7l 3n6 , canada + @xmath42 now at max - planck - institut fuer kernphysik , heidelberg , germany + @xmath43 now at max - planck - institut fuer plasmaphysik , boltzmannstr.2 d-85748 garching , germany + @xmath44 now at center for cosmological physics , university of chicago , 933 e.56@xmath45st . , chicago , il 60637 + @xmath46 marie curie fellowship at lngs + @xmath47 on leave of absence from st .
petersburg nuclear physics inst .
- gatchina , russia + @xmath48 on leave of absence from institute for nuclear research , msp 03680 , kiev , ukraine + @xmath49 on leave of absence from kfki - rmki , konkoly thege ut 29 - 33 h-1121 budapest , hungary + @xmath50 now at lockhead martin corporation , sunnyvale ca + @xmath51 now at physics department , virginia polytechnic institute and state university , robeson hall , blacksburg , va 24061 - 0435 , usa + @xmath52
deceased + @xmath53 + stefan schnert , email : [email protected] , + marianne gger - neff , email : [email protected] .
+ ] phenyl - o - xylylethane , pxe , organic liquid scintillator , solar neutrino spectroscopy , low - background counting , borexino 14.60.pq , 23.40.-s , 26.65.+t , 29.40.mc , 91.35.lj
|
strong couplings involving three mesons are complicated objects posing a great challenge for their theoretical study .
the @xmath6 coupling , for which most theoretical analyses predicted values sizably smaller than the one later measured by cleo @xcite , illustrates this statement very well . in this letter
, we address the strong three - meson couplings involving @xmath4 and @xmath5 states .
these quantities can not be measured directly in strong @xmath4 and @xmath5 decays , but they are important for our understanding of the @xmath4 and @xmath5 properties in a hadronic medium @xcite . most results for charmonium couplings arose from rather detailed qcd sum - rule calculations @xcite . in the past , however , the application of qcd sum rules to three - meson couplings faced a great problem : qcd sum rules strongly underestimated the @xmath6 coupling ( see , e.g. , @xcite ) and the origin of this discrepancy has not been fully clarified .
we thus present an alternative analysis of the family of @xmath4 and @xmath5 couplings using the relativistic dispersion approach @xcite , one of the approaches which managed to predict correctly the @xmath6 coupling @xcite before the cleo measurement .
the strong couplings in the focus of our interest , @xmath7 and @xmath8 are defined by @xmath9 with momentum transfer @xmath10 .
accordingly , @xmath11 is dimensionless whereas @xmath7 has inverse mass dimension .
these strong couplings are related to the residues of the poles in the transition form factors at time - like momentum transfer arising from contributions of intermediate meson states in the transition amplitudes @xmath12 channel .
we study the form factors @xmath13 , @xmath14 , and @xmath15 related to the transition amplitudes induced by vector quark currents @xmath16 or axial - vector quark currents @xmath17 : @xmath18 where dots stand for other lorentz structures .
the poles in the above form factors are of the form @xmath19 in these relations , @xmath20 and @xmath21 label pseudoscalar and vector resonances with appropriate quantum numbers ; @xmath22 and @xmath23 are the leptonic decay constants of the pseudoscalar and vector mesons , respectively , defined in terms of the amplitude of the meson - to - vacuum transition induced by the axial - vector or vector quark currents according to @xmath24
relativistic constituent quark models @xcite proved to constitute an efficient tool for the study of hadron properties , in particular of meson decay constants and transition form factors .
an essential feature of the constituent quark picture is the appropriate matching of the quark currents in qcd ( @xmath25 , @xmath26 , etc . ) and the associated currents formulated in terms of constituent quarks ( @xmath27 , @xmath28 , etc . ) . for light quarks , for instance , partial conservation of the axial - vector current requires the appearance of the pseudoscalar structure in the axial - vector current of the constituent quarks , similar to the case of the axial - vector current of the nucleon @xcite . for the currents containing heavy quarks , the matching conditions are simpler : @xmath29 where the dots indicate contributions of other possible lorentz structures @xcite .
constituent quarks @xmath30 and @xmath31 have masses @xmath32 and @xmath33 , respectively . in general , the form factors @xmath34 and @xmath35 depend on the momentum transfer .
vector current conservation requires @xmath36 at zero momentum transfer for the elastic current and at zero recoil for the heavy - to - heavy quark transition .
the specific values of the form factors @xmath34 and @xmath35 and their momentum dependences belong to the parameters of the model , as well as the quark masses and the wave functions of mesons regarded as relativistic quark antiquark bound states . a relativistic treatment of two - particle contributions to the bound - state structure may be consistently formulated within a relativistic dispersion approach which takes into account only two - particle intermediate quark - antiquark states in feynman diagrams @xcite .
such a formulation is explicitly relativistic - invariant : hadron observables like form factors or decay constants are given by spectral representations over the invariant masses of the quark - antiquark intermediate states .
application of the dispersion formulation of the constituent quark picture to heavy - to - light meson form factors has convincingly demonstrated the reliability of this approach @xcite . within the dispersion formulation of the constituent quark model , the decay constants @xmath22 and @xmath23 of pseudoscalar and vector mesons
are expressed in the form of relativistic spectral representations , over the invariant masses of the intermediate quark
antiquark states , of the spectral densities involving the nonperturbative meson wave functions @xmath37 and @xmath38 respectively @xcite : @xmath39 with @xmath40 . the wave functions @xmath41 , @xmath42 can be written as @xmath43 with @xmath44 normalized according to @xmath45 notice that eqs .
( [ fpv ] ) may be rewritten as the fourier transform of the meson relativistic wave function at the origin .
similarly , the @xmath46 transition form factors induced by the constituent - quark transition current @xmath47 in the kinematical region @xmath48 is given by the double spectral representation @xmath49 the function @xmath50 is the double spectral density of the relevant feynman diagrams with constituent quarks in the loop ( fig .
[ plot:1 ] ) .
it contains the @xmath51-functions corresponding to the quark - antiquark thresholds and a specific constraint coming form the triangle feynman diagram .
the explicit expressions for @xmath50 are given in sect .
3.2 of @xcite and will not be reproduced here .
we point out that at @xmath52 , the form factors obtained within the dispersion formulation are equal to the form factors of the light - front relativistic constituent quark model ( ref . 3 in @xcite ) .
correspondingly , the double spectral representation ( [ ff ] ) at @xmath52 may be rewritten as the convolution of the light - cone wave functions of the initial and the final hadrons ( see eq .
( 2.86 ) of @xcite ) , or , equivalently , as the fourier transform in the transverse variables of the overlap of these wave functions .
the merit of having explicitly relativistic - invariant spectral representations compared to other formulations is the possibility to obtain the form factors in the decay region @xmath53 by the analytic continuation in @xmath12 which was shown to lead to the appearance of the anomalous cut @xcite .
noteworthy , both the normal and the anomalous contributions involve the @xmath54 and @xmath55 integrations over the corresponding two - particle cuts , i.e. for @xmath56 and @xmath57 with @xmath58 given by ( [ k2 ] ) . as the result , the form factors in a broad kinematical region @xmath48 are expressed in terms of the relativistic wave functions of the participating mesons @xmath59 and @xmath60 with @xmath56 and @xmath57 . in this region , the precise form of the wave functions @xmath61 is not crucial ; essential is only that the confinement effects have been taken into account .
that is why , as shown in the applications to meson transition form factors @xcite , a simple gaussian parameterization can be adopted : @xmath62 the spectral representation ( [ ff ] ) is based on constituent - quark degrees of freedom and we apply it to calculate the form factors in the region @xmath63 .
we then numerically interpolate the results of our calculations and use the obtained parameterizations to study the form factors at @xmath64 , where one expects the appearance of meson resonance at @xmath65 .
the use of the dispersion formulation of the constituent quark model allows us to reveal the intimate connection between different decay modes and to perform the calculations in the broad range of @xmath12 which includes the scattering region @xmath52 and the physical region of the quark weak decay @xmath66 .
in fact , quark models are the only approach that leads to relations between the decays of different mesons through the meson wave functions and provides the form factors in the @xmath12-range indicated above .
it is important to emphasize that the form factors ( [ ff ] ) reproduce correctly the structure of the heavy - quark expansion in qcd for heavy - to - heavy and heavy - to - light transitions if the radial wave functions @xmath44 are localized in a region of the order of the confinement scale @xmath67 i.e. , @xmath68 @xcite .
we also investigated the @xmath69-breaking effects in the strong couplings .
the replacement of the light quark by the strange quark leads to the increase of the considered transition form factors and of the corresponding residues . at the same time , however , also the leptonic decay constants of the charmed strange mesons considerably exceed those of their non - strange counterparts . the three - meson strong couplings are derived as ratios of the form - factor residues and the leptonic decay constants , and , eventually , the replacement of the non - strange quark by the strange quark leads to a reduction of the three - meson couplings at the level of some @xmath70 in contrast , qcd sum rules observe an enhancement of the three - meson couplings when the light quark is replaced by the strange one . 30 s. ahmed _ et al . _
( cleo collaboration ) , phys .
* 87 * , 251801 ( 2001 ) .
z. lin and c. m. ko , phys .
c * 62 * , 034903 ( 2000 ) .
m. e. bracco , m. chiapparini , f. s. navarra , and m. nielsen , phys .
b * 605 * , 326 ( 2005 ) .
r. mattheus _ et al . _ ,
int . j. mod .
e * 14 * , 555 ( 2005 ) .
b. osorio rodriguez , m. e. bracco , and m. chiapparini , nucl .
a * 929 * , 143 ( 2014 ) .
b. osorio rodriguez _
et al . _ ,
j. a * 51 * , 28 ( 2015 ) .
v. m. belyaev , v. m. braun , a. khodjamirian , and r. rckl , phys .
d * 51 * , 6177 ( 1995 ) .
d. melikhov , eur .
j. direct * c4 * , 2 ( 2002 ) [ arxiv : hep - ph/0110087 ] .
d. melikhov and m. beyer , phys .
b * 452 * , 121 ( 1999 ) .
d. melikhov and b. stech , phys .
d * 62 * , 014006 ( 2000 ) .
a. le yaouanc , l. oliver , s. ono , o. pene , and j .- c .
raynal , phys .
d * 31 * , 137 ( 1985 ) ; phys . rev .
lett . * 54 * , 506 ( 1985 ) ; w. lucha , f. schberl , and d. gromes , phys . rep . *
200 * , 127 ( 1991 ) ; f. cardarelli , e. pace , g. salm , and s. simula , phys .
b * 357 * , 267 ( 1995 ) ; r. n. faustov and v. o. galkin , z. phys .
c * 66 * , 119 ( 1995 ) ; d. ebert , r. n. faustov , and v. o. galkin , phys .
b * 635 * , 93 ( 2006 ) .
d. melikhov and b. stech , phys .
d * 74 * , 034022 ( 2006 ) ; w. lucha , d. melikhov , and s. simula , phys .
d * 74 * , 054004 ( 2006 ) .
v. v. anisovich , d. i. melikhov , and v. a. nikonov , phys . rev .
d * 55 * , 2918 ( 1997 ) ; a. f. krutov and v. e. troitsky , jhep * 9910 * , 028 ( 1999 ) .
d. melikhov , phys .
d * 53 * , 2460 ( 1996 ) ; * 56 * , 7089 ( 1997 ) .
m. s. bhagwat , m. a. pichowsky , c. d. roberts , and p. c. tandy , phys .
c * 68 * , 015203 ( 2003 ) ; g. eichmann , r. alkofer , i. c. clet , a. krassnigg , and c. d. roberts , phys .
c * 77 * , 042202 ( 2008 ) ; d. melikhov and s. simula , eur .
j. c * 37 * , 437 ( 2004 ) .
k. a. olive _ et al .
_ ( particle data group ) , chin .
c * 38 * , 090001 ( 2014 ) .
w. lucha , d. melikhov , and s. simula , phys .
b * 701 * , 82 ( 2011 ) .
w. lucha , d. melikhov , and s. simula , phys .
b * 735 * , 12 ( 2014 ) .
d. beirevi _ et al .
_ , jhep * 1202 * , 042 ( 2012 ) .
c. t. h. davies , c. mcneile , e. follana , g. p. lepage , h. na , and j. shigemitsu , phys . rev .
d * 82 * , 114504 ( 2010 ) .
g. c. donald , c. t. h. davies , r. j. dowdall , e. follana , k. hornbostel , j. koponen , g. p. lepage , c. mcneile , phys .
d * 86 * , 094501 ( 2012 ) .
r. e. mitchell _ et al .
_ ( cleo collaboration ) , phys . rev .
lett . * 102 * , 011801 ( 2009 ) .
d. beirevi and f. sanfilippo , jhep * 1301 * , 028 ( 2013 ) . j. p.
lees _ et al . _
( babar collaboration ) , phys .
111 * , 111801 ( 2013 ) .
d. beirevi and f. sanfilippo , phys .
b * 721 * , 94 ( 2013 ) ; j. m. flynn _ et al .
_ , arxiv:1506.06413 [ hep - lat ] .
|
we discuss the strong couplings @xmath0 and @xmath1 for vector ( @xmath2 ) and pseudoscalar ( @xmath3 ) mesons , at least one of which is a charmonium state @xmath4 or @xmath5 .
the strong couplings are obtained as residues at the poles of suitable form factors , calculated in a broad range of momentum transfers using a dispersion formulation of the relativistic constituent quark model .
the form factors obtained in this approach satisfy all constraints known for these quantities in the heavy - quark limit .
our results suggest sizably higher values for the strong meson couplings than those reported in the literature from qcd sum rules .
|
this work was supported in part by the eu under grant ist - topqip , `` topological quantum information processing '' ( contract ist-2001 - 39215 ) .
a.c . acknowledges support from marie curie rtn project conquest . v.v . acknowledges also support from epsrc and the british council in austria .
g. m. palma , k .- a .
suominen , and a. k. ekert .
lond . , a * 452 * 567 , ( 1996 ) ; l .- m . duan and g .- c .
lett . , * 79 * 1953 , ( 1997 ) ; p. zanardi and m. rasetti .
* 79 * 3306 , ( 1997 ) ; d. a. lidar , i. l. chuang , and k. b. whaley .
* 81 * 2594 , ( 1998 ) ; a. barenco et al . , * 26 * 1541 , ( 1997 ) .
|
we propose a new way to generate an observable geometric phase by means of a completely incoherent phenomenon .
we show how to imprint a geometric phase to a system by `` adiabatically '' manipulating the environment with which it interacts . as a specific scheme we analyse a multilevel atom interacting with a broad - band squeezed vacuum bosonic bath .
as the squeezing parameters are smoothly changed in time along a closed loop , the ground state of the system acquires a geometric phase .
we propose also a scheme to measure such geometric phase by means of a suitable polarization detection .
whenever a pure quantum state undergoes a parallel transport along a closed path , it gathers information on the geometric structure of the hilbert space in which it lies . in this letter
we will show that a a possible way to generate such a parallel transport is by way of an irreversible quantum evolution .
in several models of interaction with the environment there are some `` protected '' subspaces , like the decoherence free subspaces ( dfs ) , which are left unaffected @xcite .
states lying in these subspaces are _ stationary _ , i.e. they do not evolve in time .
a typical example is the ground state of an atomic system , which , trivially , remain unaffected by the coupling with the electromagnetic field .
however , there are situations in which the interaction between a system and an engineered environment can generate non - trivial ground states @xcite . for instance , when a group of atoms collectively interacts with a broad band squeezed vacuum , the highly non - classical correlations which are present in the field are transferred to the atomic system , which relaxes in a complex pure equilibrium state . in such a scenario ,
the control over the engineered reservoir allows an indirect control on the state of the system to which it is coupled @xcite .
of particular interest is the possibility to change in time the reservoir parameters in such a way that the `` protected '' system subspace evolves in a controlled fashion .
here we show that if this change in time is made slowly enough , a state lying in such a subspace evolves coherently and acquires information about the geometry of the space explored . as an explicit example , we consider a suitable multilevel atomic system interacting with a broad band squeezed vacuum . to be more specific let us consider first a three level atom whose interaction with an electromagnetic field in the rotating wave approximation is described by the following hamiltonian : @xmath0 d\omega \text{,}\nonumber\ ] ] where @xmath1 is the free atomic hamiltonian , @xmath2 is the atomic operator describing the absorption of an excitation and @xmath3 is the annihilation operator of the mode with frequency @xmath4 ( @xmath5 . the field , which we treat as a reservoir , is assumed to be in a broad band squeezed vacuum state . in mathematical terms , this is obtained from the ordinary field vacuum state by means of the unitary operator @xmath6 @xmath7 where @xmath8 d\omega\right\}\ ] ] is a multimode squeezing transformation @xcite , which correlates symmetrical pairs of modes around the carrier frequency @xmath9 and @xmath10 is the squeezing parameter , whose polar coordinates @xmath11 and @xmath12 are called phase and amplitude of the squeezing , respectively .
the use of the born markov approximation , justified by the broadband nature of the field , leads to the following master equation for the atomic degrees of freedom @xcite : @xmath13 where @xmath14 , and @xmath15 from ( [ rmatrix ] ) follows that the state @xmath16 with @xmath17 and @xmath18 , satisfies @xmath19 .
in other words , this state is unaffected by the environment , i.e. it is decoherence free .
moreover @xmath20 represents the new ground state , as all the other states of the atomic system relax to it .
as anticipated , the key idea is to smoothly change the squeezing parameter of the field in order to `` adiabatically '' drag a state initially prepared in @xmath21 into @xmath22 , where @xmath23 is the time dependent squeezing parameter .
we will show the existence of an `` adiabatic '' limit such that the transition probability of @xmath24 to the orthogonal subspace vanishes as the rate of change of @xmath25 becomes sufficiently small .
furthermore , we will show that after a cyclic evolution of @xmath25 , the state @xmath26 acquires a geometric phase .
it is worth stressing that this procedure , although reminiscent of the usual adiabatic evolution , is a different physical phenomenon .
the usual adiabatic approximation refers to a coherent evolution , obtained by tuning the parameters of the system hamiltonian , while the `` steering process '' discussed here is achieved manipulating the environment .
the essential difference is that in the latter case the system state can be adiabatically controlled entirely by means of an incoherent phenomenon and no hamiltonian term contributes to its time evolution . to show how this incoherent adiabatic steering process can take place ,
consider the time dependent version of equation ( [ mastereq ] ) where @xmath27 is explicitly dependent on time through @xmath23 .
it is useful to express the equation of motion in the reference frame where @xmath28 is time independent . to this end , consider the following unitary transformation @xmath29 from the basis @xmath30 , @xmath31 , @xmath32 to the time dependent basis @xmath33 , @xmath34 , @xmath35 , where @xmath26 coincides with @xmath35 under this change of frame , the equation of motion becomes @xmath36\text{,}\ ] ] where , in this new frame , @xmath37 , @xmath38 , @xmath39 and @xmath40 is a hamiltonian term arising from the change of picture . moreover , in this frame the lindbladian term , @xmath41 , assumes a simple diagonal form : @xmath42 the main advantage of this transformation is that it allows to formulate clearly the adiabatic condition , since the rate of change of the environment parameters are contained in the operator @xmath43 .
the limit that we are interested in is the one in which the dominant contribution in equation ( [ eq : newmastereq ] ) comes from the incoherent terms , i.e. @xmath44 . an interesting case is the one in which the squeezing amplitude is kept constant while its phase is slowly changed from @xmath45 to @xmath46 .
this adiabatic evolution can be easily achieved by tuning , for example , the carrier frequency , @xmath47 , of the squeezed state slightly off resonance from the two photon transition @xmath48 . by introducing this detuning , @xmath49 , ( assuming @xmath50 )
, the master equation obtained has the form of eq .
( [ mastereq ] ) and ( [ rmatrix ] ) where @xmath51 is replaced by @xmath52 .
hence , a sufficiently small value of @xmath49 determines the required adiabatic evolution . under this condition
the operator @xmath43 assumes the form @xmath53 where @xmath54 and @xmath55 .
we show that , when @xmath56 is small enough , the state @xmath57 is adiabatically decoupled from its orthogonal subspace and a cyclic evolution in @xmath51 results in a geometric phase acquired by @xmath26 depending only , in this case , on the amount of squeezing @xmath58 . note however that , since the steering process is essentially incoherent , any phase information acquired by a superposition of @xmath28 and a state belonging to the orthogonal subspace is inevitably lost , as the latter is subject to decoherence .
the only way to retain such information is to consider an auxiliary level @xmath59 , unaffected by the noise , playing the role of a reference state for an interferometric measurement . for simplicity
assume that @xmath59 is unaffected by the environment during the whole evolution , and , hence , is time independent . as a consequence ,
the action of the unitary transformation @xmath60 on @xmath59 is trivial , and equation ( [ eq : newmastereq ] ) remains essentially unchanged .
the whole information about the geometrical phase and the coherence retained by the system during its evolution is then recorded in the phase and amplitude of the density matrix term @xmath61 , whose evolution is described by the following set of coupled differential equations @xmath62 where @xmath63 .
assume that initially the excited states @xmath30 and @xmath31 of the system are not populated , hence @xmath64 and the coherence @xmath65 evolves as @xmath66 \text{,}\end{aligned}\ ] ] where @xmath67 in the limit @xmath68 we obtain for the coherence @xmath65 @xmath69 where @xmath70 .
we are interested in a cyclic evolution , corresponding to @xmath71 . by retaining only the leading terms in @xmath72 the total evolution at time @xmath73
is given by @xmath74 and @xmath31 and between @xmath31 and @xmath30 is @xmath9 .
the transitions between these level are coupled to the modes @xmath3 of the reservoir .
the reference state @xmath59 is decoupled from the reservoir .
+ ( b ) five - level system , transitions @xmath75 and @xmath76 are coupled to modes @xmath77 and @xmath78 and @xmath79 are coupled to modes @xmath80 of the reservoir . ] where we have substituted @xmath81 with @xmath82 .
finally , going back to the original frame by means of @xmath83 ) , the corresponding coherence @xmath84 is given by : @xmath85 for example a state initially prepared in @xmath86 , after closing the loop , evolves into @xmath87 where @xmath88 .
it is clear from this expression , that in the limit @xmath89 the dominant contribution to the time evolution is just a phase factor @xmath90 , with @xmath91 .
this proves that in the `` adiabatic '' approximation , the system preserves its coherence .
in fact , according to equation ( [ eq : cohe ] ) , the amplitude damping of @xmath92 occurs only when we take into account the first order contribution in @xmath93 , which shows an exponential decay rate of the order of @xmath94 .
this proves that for small @xmath56 the system admits an adiabatic limit , in which the subspace @xmath95 spanned by @xmath96 and @xmath59 is adiabatically decoupled from its orthogonal subspace @xmath97 .
for this reason , @xmath95 is decoupled from the effects of the decoherence , which only affect states lying in its orthogonal subspace . within this approximation , then
, a state prepared in the space @xmath98 is adiabatically transported rigidly inside the evolving subspace @xmath95 .
as a result of this adiabatic steering , when the system is brought back to its initial configuration , the coherence @xmath92 acquires a phase @xmath91 .
this phase can be interpreted as the geometric phase accumulated by the state @xmath96 . by using the canonical formula for the berry phase , it easy to see that the geometric phase of @xmath96 is given by @xmath99
as expected the value of @xmath100 depends only on the squeezing , and vanishes as the squeezing tends to zero .
moreover , notice that the phase @xmath100 is purely geometrical , i.e. there is no dynamical contribution arising from an existing hamiltonian , since , in absence of any steering process , the states inside @xmath101 have a trivial dynamics .
this makes the measurement of this phase a relatively easy task .
usual procedures to measure geometric phases make use of suitably designed techniques to eliminate dynamical phase contributions , such as spin - echo @xcite or parallel transport conditions @xcite . in this setup ,
the geometric phase is the only contribution to the phase accumulated by @xmath26 , and hence , it is straightforward to measure by a suitable interferometric setup .
a simple scheme to measure the geometric phase obtained by such a steering process can be realized with a simple variation of our system .
let us consider the five - level atomic system shown in picture [ fig : sys4l](b ) .
it essentially consists of two replicas of the three - level system discussed above , with the level @xmath31 in common .
the important ingredient is that transitions @xmath102 and @xmath103 are coupled with modes of the reservoir which are different from those coupled to the transitions @xmath104 and @xmath105 .
a simple way to achieve this , is to choose , for example , polarisation selective transitions , say , left - circularly polarised modes for the former transitions and right - circularly polarised for the latter ones .
the complete hamiltonian of such system is : @xmath106 d\omega \text{,}\nonumber\end{aligned}\ ] ] where @xmath107 , and @xmath108 and @xmath109 , and @xmath110 is the annihilation operator of the mode with the energy @xmath4 and polarization @xmath111 .
assume broadband squeezed vacuum states for the set of modes @xmath77 and modes @xmath80 with different squeezing parameters @xmath112 and @xmath113 : @xmath114 where @xmath115 are the analogous of the operator ( [ eq : squeezer ] ) acting on the set of modes @xmath116 .
under the same assumptions which lead to equation ( [ mastereq ] ) we obtain the master equation : @xmath117 where @xmath118 , and @xmath119 .
this system admits a two - dimensional decoherent - free subspace , spanned by states @xmath120 and @xmath121 whose definition is the analogous of state @xmath26 of equation ( [ eq : dfstate ] ) .
we assume again time dependent squeezing parameters @xmath122 , and again we examine the time dependence of the system in a rotating frame , i.e. a frame where the state @xmath123 appear stationary .
this leads to the following master equation for the five - level system in the rotating frame : @xmath124\text{,}\ ] ] where @xmath125 , @xmath126 being the unitary transformation producing the change of frame .
assume again , for simplicity , that the parameters @xmath127 and @xmath128 are kept constant and that @xmath129 . under this assumption , the master equation can be exactly solved .
the solution is analogous to the one obtained for system previously analyzed .
suppose that the system is initially prepared in a coherent superposition of state @xmath130 and @xmath131 , for example : @xmath132 .
at a later time one has @xmath133 with @xmath134 and @xmath135 .
when the parameter @xmath51 closes a loop , at @xmath136 , the coherence has gained a phase @xmath137 which is the difference between the geometric phases @xmath138 acquired by the states @xmath139 , respectively . as in the previous scheme ,
the visibility is reduced by a factor which is linear in the `` adiabatic parameters '' @xmath140 , which guarantees the existence of the adiabatic limit .
the advantage of this modified scheme is that the value of the geometric phases can be readily measured from the polarisation of the light emitted when the system relaxes .
infact , if the value of the squeezing parameters @xmath141 is suddenly switched to zero , the states @xmath139 are no longer decoherece free , and decay to a superposition of the ground states @xmath32 and @xmath142 .
this dissipation process is accompanied by two photon emissions into the reservoir . due to the structure of the interaction ( [ hamiltonian2 ] ) with the reservoir , the photon emitted due to the transitions @xmath143 and @xmath144 , is polarised according to the geometric phase accumulated between @xmath120 and @xmath121 .
for example , if @xmath77 and @xmath80 are right and left circularly polarised modes , respectively , the first dissipation process produces the linearly polarised photon : @xmath145 the detection of the polarisation of the emitted photon makes possible a direct measurement of of the geometric phase .
we have presented a scheme to generate a geometric phase via a completely incoherent control procedure .
this scheme is conceptually different from the usual coherent adiabatic control .
the latter is realized through a smooth evolution of suitable hamiltonians , whereas here , the adiabatic steering is the effect of an externally controlled environment .
the phase generated is purely geometrical , and , therefore , experimentally detectable without resorting to techniques for the elimination of dynamical contributions . due to its very nature , this scheme is immune from unwanted environmental effects .
moreover , like any geometric effects , it presents an inherent degree of robustness against uncertainties in the control parameters .
|
tidal disturbances have played an important role in reshaping galaxies and triggering star formation over cosmic time . this is confirmed by h@xmath0 , far - infrared , and mid - infrared studies showing that the mass - normalized star formation rates of pre - merger optically - selected interacting galaxies are enhanced by a factor of two on average compared to normal spirals @xcite . with the advent of the galaxy evolution explorer ( galex ) , a new window on star formation in galaxies
is now available .
the addition of uv helps to break the age@xmath1extinction degeneracy in population synthesis modeling ( e.g. , @xcite ) .
furthermore , since the uv traces somewhat older and lower mass stars ( @xmath2400 myrs ; o to early - b stars ) than h@xmath0 ( @xmath210 myrs ; early- to mid - o stars ) , it provides a measure of star formation over a longer timescale than h@xmath0 studies .
galex imaging has shown that some tidal features in interacting galaxies are quite bright in the uv ( e.g. , @xcite ) . in some cases ,
tidal features previously thought to be purely gaseous have been detected by galex ( e.g. , @xcite ) .
in other systems , galex images have been used to identify new tidal features ( e.g. , @xcite ) . to address these issues ,
we have used the galex telescope to image more than three dozen strongly interacting galaxies in the uv ( the ` spirals , bridges , and tails ' ( sb&t ) sample ) .
these galaxies were selected from the arp ( 1966 ) atlas using the following criteria : 1 ) they are relatively isolated binary systems ; we eliminated merger remnants , close triples , and multiple systems in which the galaxies have similar optical brightnesses .
2 ) they are tidally disturbed .
3 ) they have radial velocities less than 10,350 km / s .
4 ) their total angular size is @xmath3 3@xmath4 , to allow for good resolution with galex .
each galaxy was imaged for @xmath51500 seconds in the fuv and nuv broadband filters of galex , which have effective bandpasses of 1350 @xmath1 1705 and 1750 @xmath1 2800 , respectively .
some of the galaxies that fit our selection criteria were previously observed by guaranteed time projects . for these galaxies
, we used the archival galex images .
the circular galex field of view has a diameter of 1.2 degrees .
the pixel size is 15 , and the spatial resolution @xmath6 5@xmath7 . about 2/3rds of our galaxies
have broadband optical images available from the sloan digitized sky survey ( sdss ) , while 3/4 have broadband spitzer infrared images available @xcite .
about half have published 21 cm hi maps .
the sb&t galaxies have a large range of collisional morphologies , including m51-like systems , wide pairs with long tails and/or bridges , wide pairs with short tails , close pairs with long tails , and close pairs with short tails . in the current paper
, we discuss unusual tidal morphologies in a subset of the galaxies . in @xcite
, we present an atlas of uv images of additional sb&t galaxies .
the full survey is described in detail in @xcite . for four of the galaxies in the sb&t sample
, we have already published the galex images as part of detailed studies of the individual galaxies , and compared with numerical simulations of the interaction @xcite .
there is a large variety of star formation morphologies within the tidal features in this sample . in many cases ,
the tidal features are quite bright in the uv .
this is illustrated by arp 72 ( figure 1a ) , whose eastern tail is very prominent in the galex images , and has very blue uv / optical colors .
arp 72 is also a good example of the so - called ` beads on a string ' morphology , in which regularly - spaced clumps of star formation are seen along spiral arms and tidal features .
these clumps are generally spaced about 1 kpc apart , the characteristic scale for gravitational collapse of molecular clouds @xcite .
such beads are seen in many other systems in our sample , including the northern tail of the western galaxy in arp 65 ( figure 1b ) , arp 82 @xcite , and arp 285 @xcite . in a few systems , we see luminous star forming regions at the base of a tidal feature .
we call these features ` hinge clumps ' @xcite .
these lie near the intersection of the spiral density wave in the inner disk and the material wave in the tail .
these may form when dense material in the inner disk gets pulled out into a tail .
this lowers the shear , which may allow more massive clouds to gravitationally collapse .
hinge clumps are visible at the eastern end of the arp 72 bridge ( figure 1a ) and the base of the northern tail of arp 65 ( figure 1b ) .
hinge clumps are also seen in arp 82 @xcite and arp 305 @xcite .
our sample also includes some candidate ` tidal dwarf galaxies ' ( tdgs ) , massive concentrations of young stars near the tips of tidal features .
the prototypical tdgs in arp 244 and arp 245 @xcite are included in the sb&t sample , along with the bridge tdg in arp 305 @xcite .
another possible tdg is seen in arp 202 ( figure 2 ) , an interaction between an edge - on disk galaxy and a smaller irregularly - shaped galaxy to the south .
a long clumpy tail is visible to the west of the southern galaxy .
the tip of this tail is particularly prominent in the galex images , and has very blue uv / optical colors .
our optical spectrum shows that this clump is at the same redshift as arp 202 .
this source was not detected in our spitzer 8 @xmath8 m map @xcite or in our sara h@xmath0 map , suggesting that it is in a post - starburst stage .
another sb&t system that may have tdgs is arp 181 ( figure 3 ) .
a clump is visible near the end of the western tail in the galex and sdss images , with very blue optical / uv colors .
however , no optical spectrum is available , thus it is unclear whether it is at the same redshift as arp 181 .
further west , another galaxy is visible , without any obvious link to the tail .
our optical spectrum shows that it is at the same redshift as arp 181 . in the sdss image
it looks like a spiral galaxy or a disturbed disk with short tidal tails .
it is extremely blue in nuv @xmath1 g , and is detected at 8 @xmath8 m @xcite
. this may be either a pre - existing dwarf galaxy or a recently detached tdg .
the sb&t sample also contains numerous examples of accretion from one galaxy to another .
one of the best - studied examples is the northern tail of arp 285 , which was likely produced from material accreted from the southern galaxy @xcite . according to our numerical simulations ,
the material in this tail fell into the gravitational potential of the northern galaxy , overshot that potential , and is now gravitationally collapsing and forming stars @xcite .
we call such features ` accretion tails ' , to distinguish them from classical tidal features
. the inner western tail of arp 284 was likely produced by the same process @xcite .
another system which may have an accretion tail is arp 105 ( figure 4 ) .
the spiral in this system has a long tail extending to the north , previously classified as a tdg @xcite .
the spiral is connected by a bridge to an elliptical galaxy to the south .
south of the elliptical is a bright star formation knot @xcite .
both the northern tdg candidate and the southern knot of star formation are luminous in hi maps @xcite . in the galex images , the spiral and the northern tdg
are quite bright in the uv , but the highest uv surface brightness is found in the knot of star formation south of the elliptical .
we suggest , based on analogy to arp 285 @xcite and proximity to the elliptical , that the southern star forming region in arp 105 is an accretion tail , rather than simply a classical tidal tail coincidently seen in projection behind the elliptical .
another system with possible mass transfer is arp 269 ( figure 5a ) , an unequal - mass pair of galaxies with a bridge . in both the galex and sdss images ,
an off - center group of blue star forming regions is visible in the smaller galaxy ngc 4485 , as well as along the bridge .
as noted by @xcite , several of these knots of star formation lie in a tail - like structure southwest of ngc 4485 .
@xcite suggest that ngc 4485 passed through the disk of the larger galaxy ngc 4490 , and ram pressure caused an offset in the location of the interstellar gas in ngc 4485 , and thus the observed offset in star formation .
we suggest an alternative possibility , that the star formation was triggered by gas flow from ngc 4490 along the bridge .
thus this may be an example of accretion from one galaxy to another .
our sample contains only a few tidal features with high hi column densities that are not detected in our galex maps or published optical maps .
one of these systems is arp 269 . in the @xcite hi maps ,
two large plume - like features extend 10@xmath4 ( @xmath620 kpc ) to the north and south of the pair . neither of these plumes is strongly detected in the galex or sdss images , although smoothed sdss images show a possible hint of the southern plume .
this is surprising in light of the relatively high n@xmath9 in the inner 5@xmath4 ( 10 kpc ) sections of these plumes of 4 @xmath10 10@xmath11 @xmath12 .
it was suggested by @xcite that these hi features were produced by sn - driven outflow from the main galaxy ngc 4490 .
instead , we suggest that the hi plumes are simply gas - rich tidal features . deeper optical and uv imaging is needed to search for a stellar component to these features .
another system with a gas - rich tidal feature without a galex counterpart is arp 84 ( ` the heron ' ) , a pair of unequal mass spiral galaxies connected by a bridge ( figure 5b ) . in hi maps , a large gaseous plume extends to the south of the main galaxy @xcite . in spite of its high n@xmath9 @xmath6 4 @xmath10 10@xmath11 @xmath12 and narrow hi line width of @xmath6100 km s@xmath13 ,
no diffuse uv emission is seen in this feature , although a few uv - bright clumps are present .
no redshifts are available at present for these clumps .
arp 84 also exhibits ` beads on a string ' along the inner edge of the northern tail of the smaller galaxy .
the most northern knot along this ` string ' is bright in the galex image , as well as at 8 @xmath8 m @xcite and h@xmath0 @xcite .
in this paper , we present galex uv images of a subset of the sb&t interacting galaxy pairs . we identify examples of ` beads on a string ' , accretion from one galaxy to another , and candidate tidal dwarf galaxies
. there are only a few tidal features that are bright in hi maps that are not detected in galex .
this research was supported by galex grant + galexgi04 - 0000 - 0026 , nasa ltsa grant nag5 - 13079 , and spitzer grant rsa 1353814 .
we thank the lick observatory staff for their support of our optical spectroscopy campaign .
|
we have used the galex ultraviolet telescope to study stellar populations and star formation morphology in a well - defined sample of more than three dozen nearby optically - selected pre - merger interacting galaxy pairs .
we have combined the galex nuv and fuv images with broadband optical maps from the sloan digitized sky survey to investigate the ages and extinctions of the tidal features and the disks .
we have identified a few new candidate tidal dwarf galaxies in this sample , as well as other interesting morphologies such as accretion tails , ` beads on a string ' , and ` hinge clumps ' . in only a few cases are strong tidal features seen in hi maps but not in galex .
|
tropical geometry is a new and active field of research .
roughly , its main idea is to replace algebraic varieties by piece - wise linear objects called tropical varieties .
these objects may be easier to deal with and new methods from combinatorics can be used to handle them .
a lot of work has been done to `` translate '' terms and definitions to the tropical world .
sometimes a translation is justified by the appropriate use of the new tropical object rather than by an argument why this is the correct tropicalization .
this is the case for the @xmath0-invariant of an elliptic curve .
many results predict that the `` tropical @xmath0-invariant '' of a tropical curve of genus one is its cycle length .
the @xmath0-invariant is an invariant which coincides for two smooth elliptic curves over an algebraically closed fields if and only if they are isomorphic . in @xcite ,
isomorphisms ( `` equivalences '' ) between abstract tropical curves are defined , and two elliptic abstract tropical curves are equivalent if and only if they have the same cycle length .
thus the cycle length plays the same role in the tropical setting as the @xmath0-invariant does in the algebraic setting .
also , the possibility to define a group law on the cycle of a tropical curve ( see @xcite ) using distances indicates the importance of the cycle and its length .
furthermore , the numbers of tropical curves with fixed cycle length are in correspondence to the numbers of curves with fixed @xmath0-invariant ( see @xcite ) .
but there is also a result which suggests that the cycle length actually might be the correct tropicalization of the @xmath0-invariant ( as introduced on p. ) .
this is a byproduct of the proof of ( * ? ? ?
* theorem 5.4.1 ) .
it says that given a tropical curve @xmath1 with cycle length @xmath2 and a puiseux series @xmath0 of valuation @xmath3 , we can embed the elliptic curve with @xmath0-invariant @xmath0 such that its tropicalization is equal to @xmath1 .
all these results indicate that the `` tropical @xmath0-invariant '' should be defined as the cycle length ( respectively , its negative ) .
the aim of this paper is to show that for a plane cubic the @xmath0-invariant really tropicalizes to the negative of the cycle length .
more precisely , we define plane cubic curves over the field of puiseux series @xmath4 and use the valuation map to tropicalize them ( see p. ) . the @xmath0-invariant of an elliptic curve over the puiseux series is a puiseux series itself .
our main theorem ( see theorem [ thm - main ] ) is that the negative of the cycle length of the tropicalization of a smooth cubic curve in @xmath5 ( assuming it has a cycle see definition [ def - cycle ] ) is always equal to the generic valuation of the @xmath0-invariant ( see definition [ def - genericvaluation ] ) and it is actually equal to the valuation of the @xmath0-invariant itself if no terms in the @xmath0-invariant cancel which generically is the case .
a corollary ( see corollary [ cor - nocycle ] ) of this theorem is that if an elliptic curve has a @xmath0-invariant with a positive valuation , then its tropicalization does not have a cycle .
there is an intriguing similarity to bad reduction of elliptic curves over discrete valuation rings .
firstly , only elliptic curves whose @xmath0-invariant has a negative valuation have bad reduction , and secondly the negative of this valuation is then the `` cycle length '' of the special fiber in the nron model of the curve in the sense that it is the number of projective lines forming the cycle .
we are working on better understanding of the connection of these two results .
moreover , in a forthcoming paper we will show that the main result of this paper can be generalized to smooth elliptic curves on toric surfaces other than the projective plane .
this paper is organized as follows . in section [ sec - jinv ]
we recall the definition of the @xmath0-invariant of a plane cubic as a rational function in the cubic s coefficients .
its denominator is the discriminant of the cubic .
moreover , we observe that the generic valuation ( see definition [ def - genericvaluation ] ) of the @xmath0-invariant is a piece - wise linear function . in section [ sec - trop ]
we recall basic definitions of tropical geometry and show that the function `` cycle length '' is piece - wise linear as well .
the main theorem is stated in section [ sec - thm ] . as we know
already that the two functions we want to compare are piece - wise linear the proof consists of two main steps : first we compare certain domains of linearity , then we compare the two linear functions on each domain .
we present two proofs since we believe that each of them is interesting on its own .
for the first proof , we choose as domains of linearity the union of @xmath6-equivalent cones of the secondary fan of @xmath7 ( i.e. cones of the grbner fan of the discriminant ) .
the comparison of the two linear functions `` generic valuation of the @xmath0-invariant '' and `` cycle length '' on each such cone is done in section [ sec - delta ] . in the second proof ,
we choose smaller domains of linearity cones of the secondary fan of @xmath8 .
to compare the two linear functions , we have to classify the rays of the secondary fan of @xmath8 and compare them on each ray .
section [ sec - rays ] is concerned with this classification of the rays of the secondary fan of @xmath8 . in section [ sec - discriminant ]
we give an alternative proof of the fact that for an arbitrary convex lattice polytope the grbner fan of the discriminant is a coarsening of the secondary fan , and in section [ sec - num ] we study the numerator of the @xmath0-invariant .
these two sections are important to understand the domains of linearity of the function `` generic valuation of the @xmath0-invariant ''
. parts of our proofs and many examples rely on computations performed using ` polymake ` @xcite , topcom @xcite and singular @xcite .
the singular code that we used for this is contained in the library ` jinvariant.lib ` ( see @xcite ) and it is available via the url http://www.mathematik.uni-kl.de/~keilen/en/jinvariant.html .
more detailed explanations on how to use the code can be found there .
the authors would like to acknowledge vladimir berkovich , jordan ellenberg , bjorn poonen , david speyer , charles staats , bernd sturmfels and john voigt for valuable discussions .
since every smooth elliptic curve can be embedded into the projective plane as a cubic it makes sense to start the investigation of smooth elliptic curves and their tropicalizations by studying cubics in the plane . and since the tropicalization of a curve highly depends on its embedding we want to consider all such planar embeddings of a given curve at the same time . that is , we start with a non - zero homogeneous polynomial @xmath9 of degree @xmath10 as input data . here
the coefficients @xmath11 are thought of as elements of the field @xmath12\big]\big ) = \left\{\sum_{\nu = m}^\infty c_\nu\cdot t^\frac{\nu}{n}\;\big|\ ; c_\nu\in\c , n\in\z_{>0 } , m\in\z\right\}\ ] ] of puiseux series over @xmath13 , and we consider the algebraic curve @xmath14 defined by @xmath15 .
since the tropicalization of @xmath16 only depends on the points of @xmath16 in the torus @xmath17 ( see section [ sec - trop ] ) we may as well replace @xmath15 by its affine equation @xmath18 and we will do so for the remaining part of the paper always keeping in mind that via homogenization @xmath15 defines a cubic in the projective plane . in our investigation
the newton polytope of @xmath15 , the triangle @xmath19 with endpoints @xmath20 , @xmath21 and @xmath22 , plays an important role .
we denote by @xmath23 its integer points .
we can write the equation for @xmath15 as @xmath24 . cm 0.4 0.05 ( 1 0 ) 0.8 ( 0 0 ) ( 3 0 ) ( 0 3 ) ( 0 0 ) f:0.8 ( 0 0 ) f:0 r:0.1 ( 0 1 ) f:0 r:0.1 ( 0 2 ) f:0 r:0.1 ( 0 3 ) f:0 r:0.1 ( 1 0 ) f:0 r:0.1 ( 1 1 ) f:0 r:0.1 ( 1 2 ) f:0 r:0.1 ( 2 0 ) f:0 r:0.1 ( 2 1 ) f:0 r:0.1 ( 3 0 ) f:0 r:0.1 if the curve @xmath16 is smooth it is fixed up to isomorphism by a single invariant @xmath25 , the _ @xmath0-invariant _ , which can be computed as a rational function , say @xmath26 in the coefficients @xmath11 of @xmath15 now thought of as indeterminates where @xmath27 $ ] are homogeneous polynomials of degree @xmath28 .
since we quite frequently need to refer to polynomials in the @xmath11 we introduce the convention @xmath29 and if @xmath30 is a multi index then @xmath31 .
the denominator @xmath6 is actually the _ discriminant _ of @xmath15 ( see @xcite ) .
the field @xmath32 of puiseux series comes in a natural way with a _ valuation _ , namely @xmath33 and we may extend the valuation to @xmath32 by @xmath34 . if @xmath35 with @xmath36 then we call @xmath37 the _ leading coefficient _ of the formal power series @xmath38 .
we sometimes call @xmath39 the _ tropicalization _ of @xmath38.[p - trop ] throughout the paper we will treat polynomials in the variables @xmath40 as well as in the variables @xmath29 , and many results will be derived for both cases at the same time . we therefore want to introduce a unifying notation which we use whenever we work with polynomials with an arbitrary set of variables .
[ not - x ] we set @xmath41 where @xmath42 is some finite index set . if @xmath43 then we use the usual multi index @xmath44 and if @xmath45 is finite and @xmath46 for @xmath47 then @xmath48\ ] ] is a polynomial over @xmath32 .
later on we will sometimes need laurent polynomials instead of polynomials and we then allow negative exponents .
it is our aim to study the valuation of the @xmath0-invariant of the curve @xmath16 given by fixing values for the @xmath11
. it would be nice if it only depended on their valuations , but this is only true if the leading coefficients of the @xmath11 are sufficiently general .
we therefore introduce the notion of _ generic valuation _ for a polynomial like @xmath49 or @xmath6 . in order to do
so we need @xmath50-initial forms , a concept which is needed in a broader context further down .
given a polynomial @xmath51 $ ] and a point @xmath52 , we define the _ @xmath53-weight _ @xmath54 of @xmath55 , where @xmath56 , and the _ @xmath50-initial form _ @xmath57 of @xmath55 with respect to @xmath53 . [ def - genericvaluation ]
if @xmath58\subset\k[\underline{x}]$ ] has _ constant rational coefficients _ and @xmath59 , then @xmath60 is called the _ generic valuation _ of @xmath55 at @xmath53 . if @xmath61 $ ] , then all @xmath62 such that the @xmath50-initial forms @xmath63 coincide with each other form a cone and the collection of these cones forms the _ grbner fan _ of @xmath55 ( see e.g. ( * ? ? ? * chap .
top - dimensional cones correspond to @xmath50-initial forms which are monomials .
[ lem - genvallinear ] if @xmath64 $ ] , then the function @xmath65 is piece - wise linear , and it is linear on a top - dimensional cone of the grbner fan of @xmath55 . moreover ,
if @xmath62 is in the interior of a top - dimensional cone of the grbner fan of @xmath55 , then @xmath66 for any @xmath67 with @xmath68 . obviously @xmath69 is piece - wise linear .
if @xmath53 is in the interior of a top - dimensional cone of the grbner fan of @xmath55 , @xmath70 is a monomial .
therefore the minimum @xmath71 in the definition of @xmath72 is attained for only one term , namely for @xmath73 , and @xmath74 is linear .
furthermore , if @xmath68 then the terms of @xmath75 of lowest order in @xmath50 come from those terms of @xmath55 for which @xmath71 is minimal .
thus @xmath76 .
note that if @xmath77 is not in the interior of a top - dimensional cone of the grbner fan of @xmath55 , then the terms of lowest order in @xmath50 in @xmath75 might cancel and we can not predict the valuation of @xmath75 . the generic valuation is , however , the valuation of @xmath75 under the assumption that the terms of lowest order in @xmath50 do not cancel . with the above notation
we define the _ generic valuation of the @xmath0-invariant at @xmath78 _ as @xmath79 . from lemma [ lem - genvallinear ]
it follows that @xmath80 is a piece - wise linear function which is linear on intersections @xmath81 of a top - dimensional cone @xmath82 of the grbner fan of @xmath49 and a top - dimensional cone @xmath83 of the grbner fan of @xmath6 . for @xmath84 in the open interior of @xmath81 , @xmath85 for any @xmath86 with @xmath87
in this section we will study the tropicalization of plane cubics as well as the tropicalization of the varieties defined by @xmath49 or @xmath6 in @xmath88 . we therefore start again using the general notation [ not - x ] .
if @xmath89 $ ] then the _ tropicalization of _ @xmath90 , @xmath91 is the closure of @xmath92 with respect to the euclidean topology in @xmath93 , where by abuse of notation @xmath94 denotes the cartesian product of the valuation map from section [ sec - jinv ] .
this definition is not too helpful when it comes down to actually computing tropical varieties .
there it is better to consider tropical polynomials . for a _ tropical polynomial _
@xmath95 , with @xmath45 finite and @xmath96 , ( see e.g.
@xcite ) we define the _ tropical hypersurface _ associated to @xmath97 to be the locus of non - differentiability of the function @xmath98 ( i.e. the locus where the minimum is attained by at least two terms ) .
we call the convex hull of @xmath99 the _ newton polytope _ of @xmath97 respectively of the tropical hypersurface defined by @xmath97 . if @xmath100 then we call a tropical hypersurface simply a
_ tropical curve_. given @xmath101 $ ] we define the _ tropicalization of _
@xmath55 to be the tropical polynomial @xmath102 then by kapranov s theorem ( see ( * ? ? ?
* theorem 2.1.1 ) ) , @xmath103 is equal to the tropical hypersurface associated to @xmath104 .
moreover , this is obviously the collection of @xmath53 for which @xmath63 is not a monomial . in particular , if @xmath105 $ ] has constant rational coefficients then it is the codimension - one skeleton of the grbner fan . we will use this definition mainly in two different settings : 1 . for a plane cubic @xmath24 we will consider the corresponding _ plane tropical cubic _ @xmath106 given by the tropical polynomial @xmath107 where @xmath87 .
2 . for the numerator and denominator of the @xmath0-invariant @xmath49 and @xmath6 which are polynomials in @xmath108 $ ] we will consider their tropicalizations @xmath109 and @xmath110 in @xmath111 .
the latter was recently studied in @xcite .
as @xmath49 and @xmath6 have constant rational coefficients these two tropical hypersurfaces are equal to the codimension - one skeletons of the grbner fans of @xmath49 respectively @xmath6 .
tropical hypersurfaces as defined above are dual to certain marked subdivisions of @xmath99 .
let us recall the necessary facts from @xcite , still using notation [ not - x ] .
a _ marked polytope _ is a pair @xmath112 where @xmath113 is a convex lattice polytope and @xmath114 contains at least the vertices of @xmath115 .
the set @xmath99 is said to be the set of _ marked lattice points_. the newton polytope @xmath116 as shown in figure [ fig - q3 ] is a marked polytope .
let @xmath112 be a marked polytope in @xmath93 with @xmath117 .
a _ marked subdivision _ of @xmath112 is a finite family of marked polytopes @xmath118 such that 1 .
@xmath119 is a marked polytope with @xmath120 for @xmath121 , 2 .
@xmath122 is a subdivision of @xmath115 , i.e. @xmath123 is a face ( possibly empty ) of @xmath124 and of @xmath125 for all @xmath126 , 3 .
@xmath127 for @xmath121 , and 4 .
@xmath128 for all @xmath126 .
we do not mandate that @xmath129 .
example [ ex - markedsubdivision ] shows an example of a marked subdivision of @xmath116 .
let @xmath130 and @xmath131 be subdivisions of @xmath112 .
we say that @xmath132 _ refines _
@xmath133 is for all @xmath134 , the collection of @xmath119 so that @xmath135 is a marked subdivision of @xmath136 .
figure [ fig - gcl ] shows two marked subdivisions of @xmath116 where the right one is a refinement of the left one .
let us note that the _ coarsest subdivision _ of @xmath112 is @xmath137 .
more interesting examples of subdivisions are the so called _ regular _ or _ coherent subdivisions_. given @xmath138 , we can associate a subdivision as follows .
let the _ upper hull _
@xmath139 of @xmath140 be the convex hull of the subset of @xmath141 given by @xmath142 the lower faces of @xmath139 project to @xmath115 giving a subdivision of @xmath115 . define the _ lower convexity _ @xmath143 of @xmath140 to be the function @xmath144 whose graph is the lower faces of @xmath139 .
this is a convex function .
the faces of the subdivision are the maximal domains of linearity of @xmath143 . for a face @xmath124 , define the marked set @xmath145 the set of all points of @xmath99 in @xmath124 that lie on the lower faces of the upper hull .
marked subdivisions that arise in this fashion are said to be _
regular_. the set of all @xmath146 that induce the same regular marked subdivision form an open polyhedral cone in @xmath147 .
all such cones form the _ secondary fan _ of @xmath99 .
the secondary fan is the normal fan to a polytope , the _
secondary polytope_. the poset of cones in the secondary fan is isomorphic to the poset of regular marked subdivisions under refinement .
_ top dimensional cones of the secondary fan _ of @xmath99 therefore correspond to the finest subdivisions where each @xmath124 is a @xmath148-dimensional simplex and @xmath149 is the set of vertices of @xmath124 .
( see ( * ? ? ? * chap .
[ linearityspace ] note that the minimal cone of the secondary fan is a @xmath150-dimensional space @xmath151 and is given by all @xmath138 of the form @xmath152 with @xmath153 and @xmath59 .
all these functions induce the _
coarsest subdivision_. we call @xmath151 the _ linearity space _ of the secondary fan of @xmath8 .
every cone of the secondary fan of @xmath8 contains its linearity space .
therefore , we may consider the secondary fan as living in @xmath154 .
when we speak of _ rays _ in the secondary fan , we mean @xmath155-dimensional cones in the fan containing @xmath151 .
the secondary fan of @xmath8 is an important object because we will see that the `` cycle length '' as a function is linear on each top - dimensional cone of the secondary fan .
since we have already seen that the valuation of the @xmath0-invariant is linear on each cone of the common refinement of the grbner fans of @xmath49 and @xmath6 our strategy will be to compare the secondary fan with these two grbner fans .
[ rem - direction ] the _ duality _ of tropical hypersurfaces in @xmath93 with newton polytope @xmath156 and regular marked subdivisions of @xmath157 with @xmath158 is set up as follows .
given a tropical polynomial @xmath159 we associate to @xmath160 the regular marked subdivision induced by @xmath161 this marked subdivision is then dual to the tropical hypersurface defined by @xmath160 in the sense of ( * ? ? ?
3.11 ) , which in its full generality is rather technical . however , for the cases we are interested in it can easily be described .
if @xmath160 defines a plane tropical curve @xmath162 in @xmath163 then each marked polytope of the subdivision of @xmath157 is dual to a vertex of @xmath162 , and each facet of a marked polytope is dual to an edge of @xmath162 . moreover ,
if the facet , say @xmath97 , has end points @xmath164 and @xmath165 then the _ direction vector _
@xmath166 of the dual edge @xmath167 in @xmath162 is defined ( up to sign ) as @xmath168 and points in the direction of @xmath167 . in particular , the edge @xmath167 is orthogonal to its dual facet @xmath97 . finally , the edge @xmath167 is unbounded if and only if its dual facet @xmath97 is contained in a facet of @xmath169 .
cm 0.4 0.05 ( 1 0 ) 0.8 ( 0 0 ) ( 3 0 ) ( 0 3 ) ( 0 0 ) ( 0 0 ) f:0 r:0.1 ( 0 3 ) f:0 r:0.1 ( 1 0 ) f:0 r:0.1 ( 3 0 ) f:0 r:0.1 0 ( 0 0 ) ( 0 3 ) ( 3 0 ) ( 0 0 ) ( 0 3 ) ( 1 0 ) ( 0.9 0.4)@xmath97 cm 0.3 0.1 ( 1 0 ) 0 ( 0 0 ) ( 3 1 ) ( 0 2 ) ( 3 1 ) ( 2 0 ) ( 0 0 ) ( 0 2 ) ( 0 0 ) ( -1.5 -1.5 ) ( 1 -0.5 ) @xmath167 the second case we are interested in are the tropical hypersurfaces defined by @xmath170 and @xmath171 , i.e. the case when the tropical hypersurface is actually the codimension - one skeleton of a fan . in this case
the corresponding fan is just the normal fan of the newton polytope of the the defining polynomial ( see ( * ? ? ?
* chap . 5 ) ) .
[ ex - markedsubdivision ] the marked subdivision below is for example induced by the tropical polynomial @xmath172 .
let us now restrict our attention to plane tropical cubics , that is , curves defined by tropical polynomials whose newton polytope is _ contained _ in the triangle @xmath19 .
note that this triangle has only one interior point , so also all possible marked subdivisions we consider have at most one interior point .
[ def - cycle ] we say that a plane tropical cubic @xmath1 _ has a cycle _ if the interior point @xmath173 is visible as the vertex of a marked polytope in its dual marked subdivision . if this is the case , the _ cycle _ of @xmath1 is the union of those bounded edges of @xmath1 which are dual to the facets of marked polytopes in the marked subdivision which emanate from @xmath173 , and we say that these edges _ form the cycle_. in the picture below , the left plane tropical cubic has a cycle while the right one does not , since @xmath173 is visible but it is not a vertex of one of the marked polytopes in the subdivision .
[ def - cyclelength ] for a bounded edge @xmath167 of a plane tropical curve with direction vector @xmath166 , defined as in remark [ rem - direction ] ( i.e. @xmath166 is orthogonal to its dual facet in the marked subdivision and of the same euclidean length as this facet ) we define the _ lattice length _
@xmath174 to be the euclidean length of @xmath167 divided by the euclidean length of @xmath166 . for a plane tropical cubic with cycle
, we define its _ cycle length _ to be the sum of the lattice lengths of the edges which form the cycle .
if the plane tropical cubic has no cycle we say its length is zero .
this defines a function `` cycle length '' @xmath175 associating to every plane tropical cubic polynomial @xmath176 the cycle length of the corresponding plane tropical cubic @xmath162 .
the following picture shows a plane tropical cubic with cycle length @xmath177 .
assume a plane tropical cubic has no cycle , i.e. @xmath173 is not the vertex of a marked polytope in the corresponding marked subdivision , but it is contained in a facet of ( necessarily ) two such marked polytopes , say @xmath178 and @xmath179 .
they are dual to two vertices @xmath180 and @xmath181 of the tropical curve .
we define the _ generalized cycle length _ of such a cubic to be four times the lattice length of the edge connecting @xmath180 and @xmath181 .
this definition is necessary because these tropical curves arise as limits of curves with a cycle , and the cycle length tends to the generalized cycle length for such a limit : one factor of @xmath182 is necessary because two edges tend to the same edge in the picture .
the other factor of @xmath182 appears because the direction vector of @xmath167 is of twice the euclidean length than the direction vector of @xmath183 and @xmath184 .
let us generalize definition [ def - cycle ] to plane tropical curves other than cubics .
let @xmath1 be a plane tropical curve with newton polytope @xmath115 and with dual marked subdivision @xmath185 .
suppose that @xmath186 is in the interior of @xmath115 and that the @xmath119 are ordered such that @xmath187 is a vertex of @xmath124 for @xmath121 and it is not contained in @xmath124 for @xmath188 ( see figure [ fig - cycle ] ) .
cm 0.4 0.05 ( 1 0 ) 0 ( 0 0 ) ( 2 3 ) ( 3 0 ) ( 0.1 0.4 ) ( 1.5 -1.5 ) ( 1 0 ) ( 0 0 ) ( 1 -3 ) ( 4 0 ) ( 1 1 ) ( 0.1 0.4 ) ( 1.3 1.3 ) ( 1 0 ) ( 3 0 ) ( 0 0 ) f:0 r:0.1 ( -1.5 -0.2 ) @xmath189 ( 3 0 ) ( 0.5 -1.5 ) f:0 r:0.1 ( -4.2 -2 ) @xmath190 ( 3 0 ) ( 3.5 -3 ) f:0 r:0.1 ( 2.5 -5.7 ) @xmath191\\\omega_0}$ ] ( 3 0 ) ( 5 -3 ) f:0 r:0.1 ( 5.5 -4 ) @xmath192 ( 3 0 ) ( 2 3 ) f:0 r:0.1 ( 2 3.3 ) @xmath193 ( 3 0 ) ( 5 3 ) f:0 r:0.1 ( 5 3.3 ) @xmath194 ( 3 0 ) f:0 r:0.1 ( 3.7 -0.2 ) @xmath187 ( 1.5 -2.5 ) @xmath195 ( 0.5 -1 ) @xmath196 ( 1.2 0.8 ) @xmath197 ( 3 1.8 ) @xmath19 ( 0 0 ) ( -1 3 ) ( 2 3 ) ( -0.7 2 ) @xmath198 ( -1 3 ) ( 3 3 ) ( 3 -3 ) ( 1.5 4.2 ) @xmath199 ( 2 6 ) ( 2 0 ) ( 0.1 0.4 ) ( 2 0 ) ( -1 3 ) ( 1 0 ) ( -1 -1 ) ( 0.1 0.4 ) ( -1.5 -1.5 ) ( 1 -3 ) ( 1 0 ) ( -0.5 -1 ) ( 0.1 0.4 ) ( -0.5 -1 ) ( 5 -3 ) ( 1 0 ) ( 0 -1 ) ( 0.1 0.4 ) ( 0 -1 ) we then say that @xmath187 _ determines a cycle _ of @xmath1 , namely the union of the edges of @xmath1 dual to the facets emanating from @xmath187 , and we say that these edges _ form the cycle _ determined by @xmath187 . the length of this cycle is defined as in definition [ def - cyclelength ] . [ lem - cyclelengthlinear ]
let @xmath200 be a marked polytope in @xmath163 with a regular marked subdivision @xmath185 and suppose that @xmath186 in the interior of @xmath115 is a vertex of @xmath124 for @xmath121 and it is not contained in @xmath124 for @xmath188 .
if @xmath201 is a function defining the subdivision , then @xmath187 determines a cycle in the plane tropical curve @xmath1 given by the tropical polynomial @xmath202 and , using the notation in figure [ fig - cycle ] , its length is @xmath203 where @xmath204 with @xmath205 and @xmath206 . by definition @xmath187 determines a cycle .
it remains to prove the statement on its length . for this
we consider the convex polytope @xmath125 having @xmath207 , @xmath187 and @xmath208 as neighboring vertices : cm 0.2 0.05 ( 1 0 ) 0 t : v ( 0 0 ) ( -4 6 ) f:0 r:0.3 ( 0 0 ) ( 4 6 ) f:0 r:0.3 ( 0 0 ) f:0 r:0.3 ( 1 -1 ) @xmath187 ( -6 6 ) @xmath209 ( 5 6 ) @xmath207 ( -4.3 2 ) @xmath210 ( 4 2 ) @xmath211 ( -0.5 3 ) @xmath125 ( 0.1 0.4 ) ( -4 6 ) ( 2 1 ) ( 4 6 ) ( -2 1 ) the vertex @xmath212 of @xmath1 dual to @xmath125 is given by the system of linear equations @xmath213 where @xmath214 , @xmath215 and @xmath216 . this system can be rewritten as @xmath217 since @xmath207 , @xmath187 and @xmath209 are neighboring vertices of the polytope @xmath125 the vectors @xmath210 and @xmath211 are linearly independent and we may apply cramer s rule to find @xmath218 the lattice length of the edge from @xmath219 to @xmath220 is the real number @xmath221 such that @xmath222 , where where @xmath223 is perpendicular to @xmath210 . thus @xmath224 in order to understand the right hand side of this equation better we need the following observation .
the last row of the matrix @xmath225 is a linear combination of the first two , and thus the determinant of @xmath226 is zero . developing the determinant by the last row we get @xmath227 or equivalently @xmath228 expanding the right hand side of using and plugging in this last equality we get @xmath229 the lattice length of the cycle of @xmath1
is then given by adding the @xmath230 , i.e. it is @xmath231 [ rem - cyclelengthlinear ] an immediate consequence of lemma [ lem - cyclelengthlinear ] is that the function `` cycle length '' , @xmath232 , from definition [ def - cyclelength ] is linear on each cone of the secondary fan of @xmath8 .
[ thm - main ] let @xmath1 be a plane tropical cubic given by the tropical polynomial @xmath233 and assume that @xmath1 has a cycle .
then the negative of the generic valuation of the @xmath0-invariant at @xmath234 is equal to the cycle length of @xmath1 , i.e. @xmath235 furthermore , if the marked subdivision dual to @xmath1 corresponds to a top - dimensional cone of the secondary fan of @xmath8 ( that is , if it is a triangulation ) , then @xmath236 where @xmath24 is any elliptic curve over @xmath32 with coefficients @xmath11 satisfying @xmath87 .
there are two main parts of the proof : the first part is to compare certain `` domains of linearity '' in @xmath111 of the two piece - wise linear functions `` cycle length '' , @xmath232 , and `` generic valuation of @xmath0 '' , @xmath237 , and the second part is to compare the two linear functions on each domain .
the proof uses many results that will be proved in the following sections .
note that our claim only involves curves @xmath1 which have a cycle or , equivalently , where in the dual subdivision the point @xmath173 is a vertex of a marked polytope .
therefore we may replace @xmath111 as domain of definition of @xmath232 and @xmath238 by the union @xmath239 of those cones of the secondary fan of @xmath8 where the corresponding marked subdivision contains @xmath173 as a vertex of a marked polytope .
the coordinates on @xmath239 are given by @xmath240 and the canonical basis vector @xmath241 has a one in position @xmath242 and zeros elsewhere . from lemma [ lem - num ] we know that @xmath239 is contained in a single cone of the grbner fan of @xmath49 , namely the one dual to the vertex @xmath243 of the newton polytope of @xmath49 .
hence the generic valuation of @xmath49 is linear on @xmath239 , namely @xmath244 thus , if we want to divide @xmath239 into cones on which @xmath237 is linear , it suffices to consider @xmath245 , and we know already that the latter is linear on cones of the grbner fan of @xmath6 by lemma [ lem - genvallinear ] .
thus so is @xmath237 restricted to @xmath239 , and by lemma [ lem - cyclinondelta ] and remark [ rem - deltaequivalence ] the function @xmath232 is so as well
. moreover , remark [ rem - deltaequivalence ] tells us that @xmath239 is indeed a union of cones of the grbner fan of @xmath6 , and each such cone is a union of certain @xmath6-equivalent cones of the secondary fan of @xmath8 . hence to prove that the two functions @xmath237 and @xmath232 coincide
it is enough to compare the linear functions on each cone of the grbner fan of @xmath6 contained in @xmath239 . to do this ,
we use theorem 11.3.2 of @xcite which enables us to compute the assignment rule for the linear function @xmath245 on each such cone , say @xmath82 , given a ( top - dimensional ) marked subdivision @xmath246 whose corresponding cone in the secondary fan of @xmath8 is contained in @xmath82 . in fact , it provides us with a formula to compute the coefficient of @xmath247 for each @xmath248 .
since we already know that the two functions @xmath245 and @xmath232 are linear on @xmath82 , we can for our comparison assume that @xmath246 is the representative of its class with as few edges as possible .
the coefficient of @xmath247 in the linear function @xmath232 for the marked subdivision is given by lemma [ lem - cyclelengthlinear ] .
to compare the two coefficients , there are some cases to distinguish , which is done by lemma [ lem - comparelin ] .
this proves the first part of the theorem . finally , for any point @xmath84 in the interior of a cone of the grbner fan of @xmath6 , @xmath236 for any polynomial @xmath249 with @xmath87 by lemma [ lem - genvallinear ] .
as a point @xmath84 in the interior of a top - dimensional cone of the secondary fan of @xmath8 is in the interior of a cone of the grbner fan of @xmath6 , the last statement follows as well .
we would like to give an alternative proof of the statement whose methods we believe to be interesting on their own . here , we will consider smaller domains of linearity , namely the cones of the secondary fan of @xmath8 contained in @xmath239 ( using the notation from the above proof ) . from
10 , thm .
1.2 ) or alternatively from lemma [ discriminant ] we conclude that the codimension @xmath250 cones of the grbner fan of @xmath6 do no meet the interior of any top - dimensional cone of the secondary fan of @xmath8 .
thus an open top - dimensional cone of the secondary fan of @xmath8 is completely contained in some top - dimensional cone of the grbner fan of @xmath6 , and using lemma [ lem - genvallinear ] we conclude that @xmath245 is linear on each top - dimensional cone of the secondary fan of @xmath8 . using lemma [ lem - num ]
we can see that @xmath251 is linear on each cone of the secondary fan of @xmath8 corresponding to a subdivision for which the interior point is visible . by [ lem - cyclelengthlinear ]
we know that the function @xmath232 is linear on a cone of the secondary fan of @xmath8 , too . to show that the two functions agree , we only have to show that they agree on the rays of each cone of the secondary fan of @xmath8 in question . in proposition [ prop - rays ]
we classify the rays .
then a computation for each ray shows that the two functions agree .
we computed this using the procedure ` raysc ` in the library ` jinvariant.lib ` available via the url http://www.mathematik.uni-kl.de/~keilen/en/jinvariant.html .
note that we have to compare with the generalized cycle length , because a point on a ray is a limit of points corresponding to curves with a cycle .
since the rays of the secondary fan of @xmath8 in question are not necessarily contained in the interior of a top - dimensional cone of the grbner fan of @xmath49 and @xmath6 , we have to use the generic valuation .
however , for a point @xmath84 in the interior of a top - dimensional cone of the secondary fan of @xmath8 we know that @xmath236 for any polynomial @xmath252 with @xmath87 by lemma [ lem - genvallinear ] .
[ cor - nocycle ] let @xmath252 define a smooth elliptic curve over @xmath32 such that the valuation of its @xmath0-invariant is positive .
then its tropicalization does not have a cycle .
let @xmath16 be a smooth elliptic curve over @xmath32 .
it is obvious that the tropicalization of @xmath16 depends on the embedding into the projective plane that we choose .
one might ask if the cycle length does not depend on the embedding though , as it should take the role of a tropical @xmath0-invariant .
this is not true however .
for example , each curve can be put into weierstrass - form without @xmath253-term and its tropicalization does not have a cycle , because the interior point is not part of the marked subdivision .
also , an embedding might be such that the valuations of the coefficients lie in a cone of the grbner fan of @xmath6 of higher codimension , and the coefficients might be such that the lowest order terms of @xmath254 cancel .
but as the cycle length is equal to the generic valuation of @xmath0 and the generic valuation of @xmath0 is smaller than the valuation of the @xmath0-invariant in this case , the cycle length would not reflect the @xmath0-invariant .
as an example , we consider the following family of curves over @xmath32 with the same @xmath0-invariant ( hence you could also say : a family of different embeddings for one curve ) .
we choose a given curve and apply the coordinate change ( in affine coordinates ) @xmath255 , where @xmath256 .
since this is an isomorphism , the curves over @xmath32 have the same @xmath0-invariant for any choice of @xmath38 .
in particular , the valuation of the @xmath0-invariant is the same for any @xmath38 .
one might hope that at least for a general choice of @xmath38 the cycle length of the tropicalization is equal to the valuation of the @xmath0-invariant .
but even this is not true , as the following example demonstrates .
let us take a subset of the family given by the coordinate changes @xmath257 , where @xmath258 . for our example we will see that infinitely many choices of @xmath259 lead to the expected cycle length , but also infinitely many lead to the `` wrong '' cycle length .
this shows that it is not true that a general choice of @xmath38 ( in the sense of zariski topology ) leads to the expected cycle length . as example
we choose @xmath260 where the @xmath261 are general . by general
we mean that after applying the coordinate change every coefficient has the expected valuation and nothing cancels .
the valuation of the @xmath0-invariant is @xmath262 .
the cycle length of the tropicalization is @xmath262 , too , as expected .
cm 0.2 t : v 0.2 ( 1 0 ) 2.5 ( -1 0 ) ( 0 2 ) ( -1 0 ) ( 1 0 ) ( -1 0 ) ( -0.93 -0.93 ) ( 0 2 ) ( 1 1 ) ( 0 2 ) ( 0 0.93 ) ( 2 -3 ) ( 2 -1 ) ( 2 -3 ) ( 0.93 0 ) ( 2 -3 ) ( -0.93 -0.93 ) ( 1 0 ) ( 2 -1 ) ( 1 0 ) ( 1 1 ) ( 2 -1 ) ( 0.93 0 ) ( 1 1 ) ( 0.93 0 ) ( -1 -3 ) f:0 r:0.1 ( -1 -2 ) f:0 r:0.1 ( -1 -1 ) f:0 r:0.1 ( -1 0 ) f:0 r:0.1 ( -1 1 ) f:0 r:0.1 ( -1 2 ) f:0 r:0.1 ( 0 -3 ) f:0 r:0.1 ( 0 -2 ) f:0 r:0.1 ( 0 -1 ) f:0 r:0.1 ( 0 0 ) f:0 r:0.1 ( 0 1 ) f:0 r:0.1 ( 0 2 ) f:0 r:0.1 ( 1 -3 ) f:0 r:0.1 ( 1 -2 ) f:0 r:0.1 ( 1 -1 ) f:0 r:0.1 ( 1 0 ) f:0 r:0.1 ( 1 1 ) f:0 r:0.1 ( 1 2 ) f:0 r:0.1 ( 2 -3 ) f:0 r:0.1 ( 2 -2 ) f:0 r:0.1 ( 2 -1 ) f:0 r:0.1 ( 2 0 ) f:0 r:0.1 ( 2 1 ) f:0 r:0.1 ( 2 2 ) f:0 r:0.1 cm 0.6 0.05 ( 3 0 ) ( 0 0 ) ( 0 0 ) ( 0 3 ) ( 0 3 ) ( 3 0 ) ( 3 0 ) ( 1 1 ) ( 1 1 ) ( 1 2 ) ( 0 0 ) ( 1 1 ) ( 1 2 ) ( 0 2 ) ( 0 1 ) ( 1 2 ) ( 1 1 ) ( 0 1 ) ( 0 0 ) f:0 r:0.1 ( 0 1 ) f:0 r:0.1 ( 0 2 ) f:0 r:0.1 ( 0 3 ) f:0 r:0.1 ( 1 0 ) f:0 r:0.1 ( 1 1 ) f:0 r:0.1 ( 1 2 ) f:0 r:0.1 ( 2 0 ) f:0 r:0.1 ( 2 1 ) f:0 r:0.1 ( 3 0 ) f:0 r:0.1 ( 0 -1.2 ) f:1 r:0.1 let us check what happens to the valuations @xmath247 of the coefficients @xmath263 when we apply the coordinate change . in general ( i.e. if no cancellation happens ) the valuations are
as follows : if we choose @xmath259 very small , then all points but the point @xmath21 lie on the same plane and thus the subdivision is as follows : in fact , it corresponds to a ray of the secondary fan of @xmath8 . for our special choice of @xmath15 and the coefficients @xmath247 as above , this marked subdivision is only reached when @xmath264 .
starting from the other end we see that the tropical curve stays unchanged as long as @xmath265 .
in particular , the cycle length is as expected for all those @xmath259 . for @xmath259 in the interval @xmath266 the dual marked subdivision is not changed , but the position of two vertices of the tropical curve changes .
for example for @xmath267 , the tropical curve is : cm 0.2 t : v 0.2 ( 1 0 ) 2.5 ( -1 0 ) ( 0 2 ) ( -1 0 ) ( 1 0 ) ( -1 0 ) ( -1.03 -1.03 ) ( 0 2 ) ( 1 1 ) ( 0 2 ) ( 0 1.03 ) ( 1.5 -3.5 ) ( 1.5 -0.5 ) ( 1.5 -3.5 ) ( 1.03 0 ) ( 1.5 -3.5 ) ( -1.03 -1.03 ) ( 1 0 ) ( 1.5 -0.5 ) ( 1 0 ) ( 1 1 ) ( 1.5 -0.5 ) ( 1.03 0 ) ( 1 1 ) ( 1.03 0 ) ( -2 -4 ) f:0 r:0.1 ( -2 -3 ) f:0 r:0.1 ( -2 -2 ) f:0 r:0.1 ( -2 -1 ) f:0 r:0.1 ( -2 0 ) f:0 r:0.1 ( -2 1 ) f:0 r:0.1 ( -2 2 ) f:0 r:0.1 ( -2 3 ) f:0 r:0.1 ( -1 -4 ) f:0 r:0.1 ( -1 -3 ) f:0 r:0.1 ( -1 -2 ) f:0 r:0.1 ( -1 -1 ) f:0 r:0.1 ( -1 0 ) f:0 r:0.1 ( -1 1 ) f:0 r:0.1 ( -1 2 ) f:0 r:0.1 ( -1 3 ) f:0 r:0.1 ( 0 -4 ) f:0 r:0.1 ( 0 -3 ) f:0 r:0.1 ( 0 -2 ) f:0 r:0.1 ( 0 -1 ) f:0 r:0.1 ( 0 0 ) f:0 r:0.1 ( 0 1 ) f:0 r:0.1 ( 0 2 ) f:0 r:0.1 ( 0 3 ) f:0 r:0.1 ( 1 -4 ) f:0 r:0.1 ( 1 -3 ) f:0 r:0.1 ( 1 -2 ) f:0 r:0.1 ( 1 -1 ) f:0 r:0.1 ( 1 0 ) f:0 r:0.1 ( 1 1 ) f:0 r:0.1 ( 1 2 ) f:0 r:0.1 ( 1 3 ) f:0 r:0.1 ( 2 -4 ) f:0 r:0.1 ( 2 -3 ) f:0 r:0.1 ( 2 -2 ) f:0 r:0.1 ( 2 -1 ) f:0 r:0.1 ( 2 0 ) f:0 r:0.1 ( 2 1 ) f:0 r:0.1 ( 2 2 ) f:0 r:0.1 ( 2 3 ) f:0 r:0.1 for @xmath268 the marked subdivision changes to cm 0.6 0.05 ( 3 0 ) ( 0 0 ) ( 0 0 ) ( 0 3 ) ( 0 3 ) ( 3 0 ) ( 3 0 ) ( 1 1 ) ( 1 1 ) ( 1 2 ) ( 0 0 ) ( 1 1 ) ( 1 2 ) ( 0 2 ) ( 1 1 ) ( 0 1 ) ( 0 0 ) f:0 r:0.08 ( 0 1 ) f:0 r:0.08 ( 0 2 ) f:0 r:0.08 ( 0 3 ) f:0 r:0.08 ( 1 0 ) f:0 r:0.08 ( 1 1 ) f:0 r:0.08 ( 1 2 ) f:0 r:0.08 ( 2 0 ) f:0 r:0.08 ( 2 1 ) f:0 r:0.08 ( 3 0 ) f:0 r:0.08 and it remains like this for @xmath269 , while the cycle length decreases ; e.g. for @xmath270 respectively @xmath271 we get : cm 0.2 t : v 0.2 ( 1 0 ) 2.5 ( -1 1 ) ( 0 3 ) ( -1 1 ) ( 0.66 1 ) ( -1 1 ) ( -1.18 -1.18 ) ( 0 3 ) ( 0.66 2.33 ) ( 0 3 ) ( 0 1.18 ) ( 0.66 -3.33 ) ( 0.66 1 ) ( 0.66 -3.33 ) ( 1.18 0 ) ( 0.66 -3.33 ) ( -1.18 -1.18 ) ( 0.66 1 ) ( 0.66 2.33 ) ( 0.66 1 ) ( 1.18 0 ) ( 0.66 2.33 ) ( 1.18 0 ) ( -2 -4 ) f:0 r:0.1 ( -2 -3 ) f:0 r:0.1 ( -2 -2 ) f:0 r:0.1 ( -2 -1 ) f:0 r:0.1 ( -2 0 ) f:0 r:0.1 ( -2 1 ) f:0 r:0.1 ( -2 2 ) f:0 r:0.1 ( -2 3 ) f:0 r:0.1 ( -2 4 ) f:0 r:0.1 ( -1 -4 ) f:0 r:0.1 ( -1 -3 ) f:0 r:0.1 ( -1 -2 ) f:0 r:0.1 ( -1 -1 ) f:0 r:0.1 ( -1 0 ) f:0 r:0.1 ( -1 1 ) f:0 r:0.1 ( -1 2 ) f:0 r:0.1 ( -1 3 ) f:0 r:0.1 ( -1 4 ) f:0 r:0.1 ( 0 -4 ) f:0 r:0.1 ( 0 -3 ) f:0 r:0.1 ( 0 -2 ) f:0 r:0.1 ( 0 -1 ) f:0 r:0.1 ( 0 0 ) f:0 r:0.1 ( 0 1 ) f:0 r:0.1 ( 0 2 ) f:0 r:0.1 ( 0 3 ) f:0 r:0.1 ( 0 4 ) f:0 r:0.1 ( 1 -4 ) f:0 r:0.1 ( 1 -3 ) f:0 r:0.1 ( 1 -2 ) f:0 r:0.1 ( 1 -1 ) f:0 r:0.1 ( 1 0 ) f:0 r:0.1 ( 1 1 ) f:0 r:0.1 ( 1 2 ) f:0 r:0.1 ( 1 3 ) f:0 r:0.1 ( 1 4 ) f:0 r:0.1 cm 0.2 t : v 0.2 ( 1 0 ) 2.5 ( -1 1 ) ( 0 3 ) ( -1 1 ) ( 0.33 1 ) ( -1 1 ) ( -1.25 -1.25 ) ( 0 3 ) ( 0.33 2.66 ) ( 0 3 ) ( 0 1.25 ) ( 0.33 -3.66 ) ( 0.33 1 ) ( 0.33 -3.66 ) ( 1.25 0 ) ( 0.33 -3.66 ) ( -1.25 -1.25 ) ( 0.33 1 ) ( 0.33 2.66 ) ( 0.33 1 ) ( 1.25 0 ) ( 0.33 2.66 ) ( 1.25 0 ) ( -2 -4 ) f:0 r:0.1 ( -2 -3 ) f:0 r:0.1 ( -2 -2 ) f:0 r:0.1 ( -2 -1 ) f:0 r:0.1 ( -2 0 ) f:0 r:0.1 ( -2 1 ) f:0 r:0.1 ( -2 2 ) f:0 r:0.1 ( -2 3 ) f:0 r:0.1 ( -2 4 ) f:0 r:0.1 ( -1 -4 ) f:0 r:0.1 ( -1 -3 ) f:0 r:0.1 ( -1 -2 ) f:0 r:0.1 ( -1 -1 ) f:0 r:0.1 ( -1 0 ) f:0 r:0.1 ( -1 1 ) f:0 r:0.1 ( -1 2 ) f:0 r:0.1 ( -1 3 ) f:0 r:0.1 ( -1 4 ) f:0 r:0.1 ( 0 -4 ) f:0 r:0.1 ( 0 -3 ) f:0 r:0.1 ( 0 -2 ) f:0 r:0.1 ( 0 -1 ) f:0 r:0.1 ( 0 0 ) f:0 r:0.1 ( 0 1 ) f:0 r:0.1 ( 0 2 ) f:0 r:0.1 ( 0 3 ) f:0 r:0.1 ( 0 4 ) f:0 r:0.1 ( 1 -4 ) f:0 r:0.1 ( 1 -3 ) f:0 r:0.1 ( 1 -2 ) f:0 r:0.1 ( 1 -1 ) f:0 r:0.1 ( 1 0 ) f:0 r:0.1 ( 1 1 ) f:0 r:0.1 ( 1 2 ) f:0 r:0.1 ( 1 3 ) f:0 r:0.1 ( 1 4 ) f:0 r:0.1 this happens because the cycle length is equal to the generic valuation of @xmath0 . the latter is not equal to the valuation of the @xmath0-invariant if and only if the @xmath50-initial form @xmath272 cancels when plugging in the leading terms of the puiseux series coefficients . the @xmath50-initial form of @xmath6 corresponding to this marked subdivision is @xmath273 . the leading term of the coefficient for @xmath274 of @xmath77 is @xmath275 , the leading term for @xmath276 is @xmath277 , the one for @xmath253 is @xmath278 , and the one for @xmath279 is @xmath280 . plugging those into @xmath281 yields @xmath282 .
thus the generic valuation of @xmath0 ( whose negative is equal to the decreasing cycle length ) is not equal to the valuation of the @xmath0-invariant for values of @xmath259 in a whole interval .
we computed this example ( as well as many other examples ) using the procedure ` drawtropicalcurve ` from the singular library ` tropical.lib ` ( see @xcite ) which can be obtained via the url http://www.mathematik.uni-kl.de/~keilen/en/tropical.html .
this library contains also a procedure ` tropicaljinvariant ` which computes the cycle length of a tropical curve as defined in definition [ def - cyclelength ] .
in this section we want to show that the function `` cycle length '' , @xmath232 , is linear on the union of cones of the secondary fan of @xmath8 which are @xmath6-equivalent . also , we provide the classification of the different cases we need to consider in order to compare the two linear functions @xmath238 and @xmath232 on such a union .
this is part of our first proof of theorem [ thm - main ] .
[ rem - deltaequivalence ] the prime factorization theorem , ( * ? ? ?
10 , thm .
1.2 ) , or alternatively lemma [ discriminant ] tells us that the codimension @xmath250 cones of the grbner fan of @xmath6 do not meet the interior of any top - dimensional cone of the secondary fan of @xmath8 .
thus the grbner fan of @xmath6 is a coarsening of the secondary fan of @xmath8 .
two cones of the secondary fan of @xmath8 are called _
@xmath6-equivalent _ if they are contained in the same cone of the grbner fan of @xmath6 .
it has been studied how two top - dimensional marked subdivisions whose cones belong to the same @xmath6-equivalence class can differ . by (
* chap . 11 , prop .
3.8 ) they can be obtained from each other by a sequence of modifications along a circuit ( see ( * ? ? ?
* chap . 7 , sect .
2c ) ) such that each intermediate ( top - dimensional ) marked subdivision belongs to the same equivalence class .
since our point configuration ( i.e. @xmath8 , the integer points of the triangle @xmath19 ) is in the plane , we can use ( * ? ? ?
* chap . 11 , prop .
3.9 ) to see that if a marked subdivision can be obtained from another equivalent one by a modification along a circuit , then this circuit consists of three collinear points on the boundary of @xmath19 . an example is shown in the following picture , the three points are @xmath20 , @xmath283 and @xmath22 .
[ lem - cyclinondelta ] the function @xmath232 ( see definition [ def - cyclelength ] ) is linear on a union of cones of the secondary fan of @xmath8 which are @xmath6-equivalent ( i.e. on a cone of the grbner fan of @xmath6 )
. given two @xmath6-equivalent marked subdivisions @xmath246 and @xmath284 of the secondary fan of @xmath8 , we can use lemma [ lem - cyclelengthlinear ] to determine the function @xmath232 on the cone corresponding to each of them . recall from remark [ rem - cyclelengthlinear ] that the function is linear on each cone of the secondary fan of @xmath8 .
we want to show that the assignment rules of these two linear functions coincide . without restriction
we can assume that @xmath246 can be obtained from @xmath284 by a modification along a circuit , and this circuit consists then of three collinear points on a facet of @xmath19 ( see remark [ rem - deltaequivalence ] ) .
recall from lemma [ lem - cyclelengthlinear ] that the coefficients of the linear function @xmath232 can be determined using the determinants @xmath285 , where @xmath205 .
one easily sees that for @xmath246 and @xmath284 the following two equations hold : @xmath286 \label{eq2 } & d_{i , i+1}=\lambda\cdot d_{i-1,i } \mbox { for } \lambda \mbox { satisfying } \lambda\cdot ( w_{i-1}-w_i)=w_{i}-w_{i+1}. \end{aligned}\ ] ] to show that the two assignment rules of @xmath232 on the cones for @xmath246 respectively @xmath284 coincide we have to show that for @xmath246 the summand for @xmath287 equals @xmath288 and the summand for @xmath289 equals the summand for @xmath289 for @xmath284 . the first statement follows immediately from equation above . to show the second statement
, we subtract the two summands from each other : @xmath290 multiplied with @xmath291 this difference is equal to : @xmath292 where the first equality follows from equation above and the last from .
let us fix a cone @xmath293 of the secondary fan of @xmath8 corresponding to the marked subdivision @xmath246 .
we then denote by @xmath294 the coefficient of @xmath247 in the assignment rule of the linear function @xmath295 on @xmath293 , and by @xmath296 we denote the coefficient of @xmath247 in the assignment rule of the linear function @xmath232 restricted to @xmath293 .
[ rem - eta ] note that by lemma [ lem - genvallinear ] and remark [ rem - deltaequivalence ] @xmath297 for all @xmath298 whenever @xmath246 and @xmath284 belong to @xmath6-equivalent cones of the secondary fan of @xmath8 , and by lemma [ lem - cyclinondelta ] also @xmath299 for all @xmath298 in this situation .
[ lem - comparelin ]
let @xmath246 be a marked subdivision of @xmath116 corresponding to a top - dimensional cone in the secondary fan of @xmath8 ( i.e. a triangulation ) such that @xmath173 is a vertex of some marked polytope in @xmath246 ( i.e. all dual plane tropical curves have a cycle )
. then @xmath300 and @xmath301 for all @xmath302 .
due to remark [ rem - eta ] we may for the proof assume that @xmath303 is the representative of its @xmath6-equivalence class with as few edges as possible .
moreover , if two triangulations @xmath246 and @xmath284 can be transformed into each other by an integral unimodular linear isomorphism , i.e. by linear coordinate change of the projective coordinates @xmath304 with a matrix in @xmath305 , and the claim holds for @xmath246 then it obviously also holds for @xmath284 . in this situation
we say that @xmath246 and @xmath284 are _ symmetric _ to each other .
we therefore only have to prove the claim _ up to symmetry_. we want to use ( * ? ? ?
* chap . 11 , thm .
3.2 ) which explains how @xmath294 can be computed . for each @xmath298
we have to consider all @xmath306 such that @xmath307 .
note that since @xmath246 by assumption is a triangulation then @xmath307 implies necessarily that @xmath308 is a vertex of @xmath309 .
we have to distinguish four cases , where in the formulas @xmath310 denotes he generalized lattice volume ( i.e. twice the euclidean area of @xmath309 ) : * if @xmath308 is a vertex of @xmath19 , then @xmath311 where @xmath312 and @xmath313 denote the lattice lengths of those facets of some @xmath309 adjacent to @xmath308 which are contained in facets of @xmath19 . e.g. if @xmath314 in the following triangulation @xmath246 , then @xmath315 .
+ cm 1 0.03 0.9 f:0.9 0 ( 0 0 ) ( 3 0 ) ( 0 3 ) ( 0 0 ) ( 1 1 ) ( 0 3 ) ( 2 0 ) ( 2 1 ) ( 1 1 ) ( 2 0 ) ( 0 0 ) f:0 r:0.06 ( 0 3 ) f:0 r:0.06 ( 3 0 ) f:0 r:0.06 ( 1 1 ) f:0 r:0.06 ( 2 0 ) f:0 r:0.06 ( 2 1 ) f:0 r:0.06 ( 0.2 1)@xmath316 ( 1 1.2)@xmath317 ( 1.4 0.6)@xmath318 ( 2.1 0.1)@xmath319 ( 1 2.2)@xmath320 ( -1.3 1.2)@xmath321 ( 2.7 0.5)@xmath322 * if @xmath308 lies on a facet of @xmath19 , is not a vertex of @xmath19 , but is a vertex of some @xmath323 , then @xmath324 where again @xmath312 and @xmath313 denote the lattice lengths of those facets of some @xmath309 adjacent to @xmath308 which are contained in facets of @xmath19 , e.g. if in the previous example @xmath325 then @xmath326 . *
if @xmath308 lies on a facet of @xmath19 , is not a vertex of any @xmath309 , then @xmath327 . * and finally @xmath328 .
let @xmath115 be the union of all those @xmath309 which contain @xmath173 , and endow the marked polytope @xmath329 with the subdivision , say @xmath330 , induced by @xmath246 .
we say that @xmath115 meets a facet of @xmath19 if the intersection of @xmath115 with this facet is @xmath250-dimensional ( and not only a vertex )
. moreover , we say that a facet of @xmath115 is multiple if it contains more than two lattice points .
we first want to show that @xmath294 and @xmath296 are as claimed whenever @xmath331 . up to symmetry
, we have to distinguish the following cases for @xmath115 and @xmath330 : * assume @xmath115 meets all three facets of @xmath19 , and assume that for all three facets the intersection with @xmath115 is multiple
. then @xmath115 looks ( up to symmetry ) like one of the following two pictures : + in the second case , @xmath332 . using lemma [ lem - cyclelengthlinear ] we can compute @xmath333 .
it is a sum with a summand for each vertex of @xmath115 .
the summand for @xmath20 is @xmath334 computing the other @xmath10 summands analogously we get @xmath335 . in the first case , @xmath336 and @xmath337 . *
assume @xmath115 meets two facets of @xmath19 multiply and one facet non - multiply .
+ in both cases , @xmath338 and @xmath339 . *
assume @xmath115 meets two facets of @xmath19 multiply and the third facet not at all .
+ in both cases , @xmath340 and @xmath341 . * assume @xmath115 meets only one facet of @xmath19 multiply ( and the two remaining facets non - multiply , or only one of them and that one non - multiply , or none of them at all ) .
+ in the first case , @xmath340 and @xmath341 , in the second and third case , @xmath342 and @xmath343 , and in the last case , @xmath344 and @xmath345 . *
assume @xmath115 meets 3 facets of @xmath19 , but none of them multiply .
+ in the first case , @xmath342 and @xmath343 , and in the second case , @xmath340 and @xmath341 . * assume @xmath115 meets only two facets of @xmath19 , and none of them multiply .
+ in the first case , @xmath342 and @xmath343 , and in the second case , @xmath344 and @xmath345 .
* assume @xmath115 meets only one facet of @xmath19 and it does so non - multiply .
+ in the first and second case , @xmath344 and @xmath345 , and in the third case , @xmath346 and @xmath347 . *
assume @xmath115 meets no facet of @xmath19 at all .
+ finally , in this case , @xmath346 and @xmath347 .
thus the claim for @xmath173 is shown .
now assume @xmath348 is not a vertex of @xmath19 . if @xmath308 is also not a vertex of any @xmath309 then there is no edge in the subdivision from @xmath173 to @xmath308 and thus @xmath308 does not contribute to the formula for the cycle length , i.e. @xmath349 .
however , the same holds true also for @xmath294 .
we may thus assume that @xmath308 is a vertex of some @xmath309 , and we may without restriction assume @xmath350 .
the classification of cases we have to consider is very similar to the above , and we will not give the details in particular , leaving the computation of @xmath296 and @xmath294 to the reader .
we do not have to consider the whole of @xmath115 , but only the triangles which are adjacent to @xmath308 .
if @xmath348 is a vertex of @xmath19 ( without restriction @xmath351 ) , the following cases have to be considered : finally , we have to consider the case were @xmath308 is not part of @xmath115 . obviously , @xmath349 in this case and we have to show the same for @xmath294 .
assume first that @xmath308 is a vertex of @xmath19 , without restriction we can assume @xmath351 .
there must be a facet of @xmath115 such that @xmath20 is on one side of it and @xmath173 is on the other .
then ( up to symmetry ) there are @xmath10 possibilities for that facet .
since we assumed that @xmath246 is the representative with as few edges as possible , the triangle formed by that facet of @xmath115 and @xmath20 can not be additionally subdivided in the second and third picture . in any case
, @xmath20 is a vertex of only one triangle , which has one facet of integer length @xmath250 and one facet of integer length @xmath2 where @xmath352 .
thus @xmath353 . now assume that @xmath308 is not a vertex of @xmath19 , without restriction @xmath354 .
again there must be a facet of @xmath115 such that @xmath283 is on one side and @xmath173 is on the other .
up to symmetry this can only be one of the line segments in the two right pictures above .
we assumed that @xmath246 is the representative of its @xmath6-equivalence class with as few edges as possible .
but that means there is no edge through @xmath283 and @xmath283 is not a vertex of a triangle in the subdivision .
thus @xmath355 . in the proof above the computation that shows that @xmath356 is different in each of the considered cases .
in particular , in the computation of @xmath294 the part of @xmath19 which is not part of the cycle is involved while this is not the case for @xmath296 .
therefore it is most unfortunately not possible to replace the consideration of several cases by an argument which holds for all of them at the same time .
however , using ` polymake ` and singular one can compute the vertices of the newton polytope of @xmath6 and for each vertex one can compute the dual cone in the grbner fan of @xmath6 and the triangulation of @xmath116 with as few edges as possible corresponding to this cone .
that way one can verify the above computations for @xmath357 and @xmath358 , since the values for @xmath358 can be read off immediately from the exponents of the vertex of the newton polytope , while the @xmath357 can be computed with the formula in lemma [ lem - cyclelengthlinear ] .
these computations have been made using the procedure ` displayfan ` and the result can be obtained via the url http://www.mathematik.uni-kl.de/~keilen/en/jinvariant.html .
the advantage is that the file ` discriminant_fan_of_cubic.ps ` available via this url shows the cases not only up to symmetry , but it shows actually all possible cases .
the aim of this section is to show that the grbner fan of @xmath6 is a coarsening of the secondary fan .
this follows from the prime factorization theorem of gelfand , kapranov and zelevinsky ( ( * ? ? ?
10 , sect .
2 ) , see also ( * ? ? ?
* conj . 5.2 ) ) .
we present our own proof because we hope that the techniques will be useful .
they are similar to those of sturmfels ( see @xcite ) .
our reference for toric varieties and the @xmath99-discriminant is @xcite , and we will use notation [ not - x ] . let @xmath359 be a finite set of lattice points .
it defines a projective toric variety @xmath360 over @xmath361 , where @xmath361 is any field with non - archemedian valuation @xmath362 whose value group is dense in @xmath363 with respect to the euclidean topology . our special case is @xmath364 . in this case ,
@xmath365 embedded in @xmath366 by the @xmath10-uple embedding .
we review some facts about toric varieties .
a laurent polynomial with support @xmath99 , @xmath367 can be thought of as defining a hypersurface in the toric variety @xmath368 ( see @xcite ) , and since this hypersurface coincides with the hyperplane section defined by the coefficients of @xmath15 under the embedding @xmath369 we identify a polynomial @xmath370 with the point @xmath371_{\omega\in{{\mathcal{a}}}}$ ] in the dual space @xmath372 of @xmath373 . the _ dual variety _
@xmath374 is the zariski closure of the set of all hyperplanes @xmath160 such that @xmath375 is singular where @xmath376 is the open torus in @xmath368 .
if @xmath377 is of codimension @xmath250 , then its defining polynomial is called the _ discriminant _ @xmath378 , otherwise we say that the @xmath368 has a _ degenerate dual variety _ and we set @xmath379 . for our special case @xmath8 , @xmath380 is the denominator of the @xmath0-invariant .
[ lem - smoothdelta ] suppose @xmath368 is smooth and @xmath381 , then the variety @xmath382 , considered as a hypersurface in @xmath368 , is singular .
consider the universal hypersurface @xmath383 and the flat subfamily defined by the vanishing of @xmath378 : @xmath384\ar[dr ] & v(\delta_{{\mathcal{a}}})\times x_{{\mathcal{a}}}\ar[d]\\ & v(\delta_{{\mathcal{a } } } ) } \ ] ] by definition the general fiber of this family is singular , and thus so must be each fiber .
the fact , however , that this singular point need not be in the open torus is a problem when proving that the grbner fan of the discriminant @xmath378 is a coarsening of the secondary fan of @xmath99 . for this we will have to reduce to the restriction of @xmath15 to some face @xmath385 which has a singular point in the torus orbit @xmath386 ( see lemma [ lem - singularorbit ] ) .
let @xmath387 and suppose @xmath388 .
for each face @xmath385 of @xmath389 of dimension @xmath38 , we have a parameterization of the open torus orbit @xmath390 given by @xmath391 we may consider the restriction @xmath392 of @xmath15 to @xmath390 as a function @xmath393 picking coordinates @xmath394 on @xmath395 , we define the subset @xmath396 of @xmath397 by @xmath398 @xmath396 can be viewed as the subset of @xmath397 consisting of pairs @xmath399 of functions @xmath15 and points @xmath400 such that @xmath401 is singular on @xmath390 at @xmath400 . note that if @xmath385 is just a point , then @xmath390 is a closed point of @xmath368 , and @xmath15 is singular on @xmath390 if and only if it is zero on @xmath390 .
[ singularclosure ] for two faces @xmath402 and @xmath385 of @xmath403 with @xmath404 there is the following inclusion @xmath405 where @xmath406 denotes the zariski closure of @xmath396 in @xmath397 .
we may restrict to the toric variety @xmath407 and therefore , we may suppose that @xmath408 . by applying a monomial change - of - variables ( i.e. a change of variables of the form @xmath409 for @xmath410 where @xmath411 is an integer matrix of determinant @xmath412 )
, we may suppose that @xmath402 lies in a coordinate subspace of @xmath93 given by @xmath413 for @xmath414 with @xmath415 . by construction , there is an embedding @xmath416 of @xmath368 into projective space .
let us pick homogeneous coordinates @xmath417_{\omega\in{{\mathcal{a}}}}$ ] so that on the open torus @xmath418 the map @xmath419 is given by @xmath420 now consider the following morphism @xmath421 which is the identity on the first factor and @xmath419 on the second factor .
we denote the coordinates on @xmath372 again by @xmath371_{\omega\in{{\mathcal{a}}}}$ ] .
there is a universal hypersurface @xmath422 cut out by @xmath423 observe that @xmath424 is the universal hypersurface on @xmath397 defined by the vanishing of @xmath425 .
we may also define a universal singular locus @xmath426 by the @xmath150 equations @xmath427 it is straight forward from the definitions that @xmath428 therefore , @xmath429 .
we may define @xmath430 by @xmath431 again we get immediately from the definition that @xmath432 where @xmath433 thus by taking inverse images by @xmath0 , we get @xmath434 [ lem - singularorbit ] if @xmath381 , then there is some face @xmath385 of @xmath403 so that @xmath401 is singular on @xmath390 .
if @xmath435 denotes the set of all faces of @xmath389 including @xmath389 itself , we have to show that @xmath436 where @xmath437 denotes the projection onto the first factor . since by definition @xmath438 is the zariski closure of @xmath439 it suffices to show that the right hand side in is zariski closed , or equivalently that @xmath440 is so .
this , however , follows once we know that @xmath441 for all @xmath442 , where @xmath406 denotes the zariski closure of @xmath396 .
since the zariski closure of @xmath386 is @xmath443 and @xmath444 , this fact follows from the lemma [ singularclosure ] .
[ lem - singularsimplex ] let @xmath445 $ ] be a polynomial with @xmath446 terms whose newton polytope @xmath447 is a @xmath38-dimensional simplex .
then the hypersurface @xmath448 has no singular points in @xmath449 .
we may divide @xmath450 by a monomial without changing @xmath451 .
therefore , we may suppose that one vertex of @xmath447 is at the origin . by applying a monomial change - of - variables , i.e. a coordinate change of the form @xmath452 with @xmath453 and @xmath454
, we may suppose that the edges from @xmath288 to the other @xmath38 vertices of @xmath447 lie along the axes .
therefore , @xmath455 for @xmath456 . in that case @xmath457 is a monomial and has no root in @xmath458 .
[ discriminant ] let @xmath459 be such that @xmath377 has codimension one and the discriminant @xmath378 exists .
then the tropicalization @xmath460 of @xmath438 is supported on the codimension one cones of the secondary fan of @xmath99 in @xmath147 .
we assume that @xmath460 intersects the interior of a top dimensional cone of the secondary fan of @xmath99 in a point @xmath461 and derive a contradiction . since @xmath462 by the lifting lemma for hypersurfaces
( see e.g. ( * ? ? ?
* thm . 2.13 ) ) we can lift @xmath84 to a point @xmath463_{\omega\in{{\mathcal{a}}}}\in v(\delta_{{\mathcal{a}}})$ ] such that @xmath464 for all @xmath47 in particular , @xmath465 by lemma [ lem - singularorbit ] there exists then a , say @xmath38-dimensional , face @xmath385 of @xmath387 such that @xmath466 has a singular point @xmath467 , and @xmath468 due to . we define @xmath469 to be the valuation of @xmath470 , and we then claim that @xmath471 is a singular point of @xmath472 . in order to see that @xmath473 is a singular point of @xmath472 it suffices to note that @xmath474 and that for any polynomial @xmath475 $ ] with @xmath476 we necessarily have @xmath477 .
since @xmath84 is in the interior of a full - dimensional cone of the secondary fan of @xmath99 the newton subdivision , say @xmath478 , of @xmath15 is a triangulation . by definition
the newton polytope of the @xmath50-initial form @xmath472 is a face of some @xmath309 and is thus a simplex .
but then by lemma [ lem - singularsimplex ] @xmath472 has no singular point in the torus @xmath449 in contradiction to the existence of @xmath473 .
unfortunately , for the numerator @xmath49 of the @xmath0-invariant it is not true that the grbner fan of @xmath49 is a coarsening of the secondary fan , as follows from example [ ex - notcoarse ] .
[ ex - notcoarse ] we provide an example which shows that the grbner fan of @xmath49 is not a coarsening of the secondary fan in the case of curves of a particular form .
the case of the full cubic is more complicated but analogous .
it can easily be proved by a computation using ` polymake ` this can be done using the procedure ` nonrefinementc ` in the library ` jinvariant.lib ` ( see @xcite ) .
let us consider curves in weierstrass form @xmath479 this corresponds to taking @xmath480 .
the fixing the constant coefficient and the coefficients of @xmath279 and @xmath481 has the effect of fixing an isomorphism @xmath482 in light of remark [ linearityspace ] . by the usual formulas for the @xmath0-invariant , we have @xmath483 so that @xmath484 the following picture shows the tropicalization of the numerator @xmath49 , the tropicalization of the denominator @xmath6 , and the secondary fan in @xmath485 .
cm 0.35 0.05 ( 1 0 ) 0.6 t : v ( -4 0 ) ( 4 0 ) ( 0 -4 ) ( 0 4 ) ( -2 0 ) ( 0 -0.2 ) ( -1 0 ) ( 0 -0.2 ) ( 1 0 ) ( 0 -0.2 ) ( 2 0 ) ( 0 -0.2 ) ( 0 -2 ) ( -0.2 0 ) ( 0 -1 ) ( -0.2 0 ) ( 0 1 ) ( -0.2 0 ) ( 0 2 ) ( -0.2 0 ) ( 4.5 0 ) @xmath486 ( -1 4.5 ) @xmath487 0 0.07 ( -1.75 -3.5 ) ( 1.75 3.5 ) cm 0.35 0.05 ( 1 0 ) 0.8 t : v ( -4 0 ) ( 4 0 ) ( 0 -4 ) ( 0 4 ) ( -2 0 ) ( 0 -0.2 ) ( -1 0 ) ( 0 -0.2 ) ( 1 0 ) ( 0 -0.2 ) ( 2 0 ) ( 0 -0.2 ) ( 0 -2 ) ( -0.2 0 ) ( 0 -1 ) ( -0.2 0 ) ( 0 1 ) ( -0.2 0 ) ( 0 2 ) ( -0.2 0 ) 0 0.07 ( -1.75 -3.5 ) ( 0 0 ) ( 0 3.5 ) ( 0 0 ) ( 3.5 0 ) cm 0.35 0.05 ( 1 0 ) 0.8 t : v ( -4 0 ) ( 4 0 ) ( 0 -4 ) ( 0 4 ) ( -2 0 ) ( 0 -0.2 ) ( -1 0 ) ( 0 -0.2 ) ( 1 0 ) ( 0 -0.2 ) ( 2 0 ) ( 0 -0.2 ) ( 0 -2 ) ( -0.2 0 ) ( 0 -1 ) ( -0.2 0 ) ( 0 1 ) ( -0.2 0 ) ( 0 2 ) ( -0.2 0 ) 0 0.07 ( -1.75 -3.5 ) ( 0 0 ) ( 0 3.5 ) ( -3.5 0 ) ( 3.5 0 ) 0.05 1.5 ( -2.5 1 ) ( 1.5 0 ) ( -1.5 1 ) ( 0 -1 ) ( 0.5 0.5 ) ( 1 -0.5 ) ( -2.5 2 ) ( 0.5 -0.5 ) ( -2.5 1 ) f:0 r:0.07 ( -1 1 ) f:0 r:0.07 ( -2 1.5 ) f:0 r:0.07 ( -2.5 2 ) f:0 r:0.07 ( -3.5 -2 ) ( 1.5 0 ) ( -1.5 1 ) ( 0 -1 ) ( 0.5 0.5 ) ( 1 -0.5 ) ( -3.5 -1 ) ( 0.5 -0.5 ) ( 0.5 -0.5 ) ( -3.5 -2 ) f:0 r:0.07 ( -3.5 -1 ) f:0 r:0.07 ( -2 -2 ) f:0 r:0.07 ( -3 -1.5 ) f:0 r:0.07 ( -2.5 -2 ) f:0 r:0.07 ( 1 1 ) ( 1.5 0 ) ( -1.5 1 ) ( 0 -1 ) ( 1 1 ) f:0 r:0.07 ( 2.5 1 ) f:0 r:0.07 ( 1 2 ) f:0 r:0.07 ( 1 -2 ) ( 1.5 0 ) ( -1.5 1 ) ( 0 -1 ) ( 1 -1 ) ( 1 -1 ) ( 1 -2 ) f:0 r:0.07 ( 1 -1 ) f:0 r:0.07 ( 2.5 -2 ) f:0 r:0.07 ( 2 -2 ) f:0 r:0.07 observe that the tropicalization of the denominator is supported on the codimension one skeleton of the secondary fan while that of the numerator intersects a top - dimensional cone of the secondary fan . however , we are only interested in plane tropical cubics which have a cycle , that is , which are dual to marked subdivisions for which the interior point can be seen .
all these cones of the secondary fan are completely contained in one cone of the grbner fan of @xmath49 .
we verified this computationally using ` polymake ` ( see @xcite ) .
as usual we use the coordinates @xmath247 with @xmath298 on @xmath111 and we denote by @xmath488 the canonical basis vector in @xmath111 having a one in position @xmath242 and zeros elsewhere .
lem - num ] let @xmath239 be the union of all cones of the secondary fan of @xmath8 corresponding to marked subdivisions @xmath303 of @xmath116 for which @xmath173 is a vertex of some @xmath309 .
then @xmath239 is contained in a single cone of the grbner fan of the @xmath49 , namely in the cone dual to the vertex @xmath489 of the newton polytope of @xmath49 . as input for
` polymake ` we use all exponents of the polynomial @xmath490 $ ] .
the convex hull of the set of all exponents is the newton polytope , say @xmath491 , of @xmath49 and its vertices are the output of ` polymake ` .
the newton polytope has @xmath492 vertices .
dual to each vertex is a top - dimensional cone of the grbner fan @xmath493 of @xmath49 , because the grbner fan is dual to the newton polytope ( see ( * ? ? ?
2.5 and prop . 2.8 ) ) .
the inequalities describing the cone @xmath16 dual to the vertex @xmath494 are given by the hyperplanes orthogonal to the edge vectors connecting @xmath494 with its neighboring vertices in @xmath491 .
we compute the neighboring vertices for each vertex using ` polymake ` and deduce thus inequalities for each of the top - dimensional cones of the grbner fan of @xmath49 .
we do these computations identifying @xmath495 with @xmath496 via the following ordering of the variables : @xmath497 in order for a marked subdivision @xmath498 of @xmath499 given by @xmath500 to have the point @xmath173 as vertex of some @xmath309 it is obviously necessary that the @xmath247 satisfy the following inequalities : @xmath501 these inequalities determine thus a cone in @xmath111 which contains @xmath239 .
a simple computation with ` polymake ` allows to compute the extreme rays of this cone and to check that they actually satisfy the inequalities of the single cone of the grbner fan of @xmath49 which is dual to the vertex @xmath502 in @xmath491 .
this proves the claim .
the computations were done with the procedure ` testinteriorinequalities ` in the library ` jinvariant.lib ` ( see @xcite ) .
the inequalities in actually determine precisely the cone @xmath239 , which is less obvious than that they are necessary , but this can again be easily tested using ` polymake ` .
in this section , we classify the rays of the secondary fan of @xmath8 .
this is part of our second proof of the main theorem .
let @xmath503 be a lattice point that is not a vertex .
the _ lift _ associated to @xmath504 is the ray in @xmath111 consisting of all functions @xmath505 of the form @xmath506 for some @xmath153 , @xmath507 , and @xmath508 .
the marked subdivision of @xmath509 associated to the lift is @xmath510 .
let @xmath511 and @xmath512 be such that @xmath513 is a line @xmath2 through @xmath514 that intersects the boundary of @xmath514 in lattice points .
the _ fold through @xmath2 _ is the cone consisting of all functions @xmath505 of the form @xmath515 for some @xmath153 , @xmath507 , and @xmath508 .
the subdivision associated to the fold is @xmath516 where @xmath517 and @xmath518 , @xmath519 .
consider three lattice points @xmath520 , @xmath521 and @xmath522 on the boundary of @xmath19 ( ordered counterclockwise ) and the corresponding vectors @xmath523 pointing from the interior point @xmath524 to @xmath525 .
the half lines starting in @xmath526 and passing through @xmath525 divide @xmath163 into three cones , say @xmath527 , @xmath528 and @xmath529 where @xmath530 contains the points @xmath525 and @xmath531 in its boundary .
note that the angle between @xmath523 and @xmath532 is less than @xmath533 degrees .
define the function @xmath534 by @xmath535 for @xmath507 .
in other words @xmath536 is @xmath288 on @xmath537 and is linear on @xmath538 .
the _ pinwheel _ through @xmath520 , @xmath521 and @xmath522 is the cone consisting of all functions @xmath505 of the form @xmath539 for some @xmath153 , @xmath507 and @xmath508 .
the subdivision associated to the pinwheel is subdivision @xmath540 with @xmath541 and @xmath542 . if @xmath545 , let @xmath546 .
the marked subdivision associated to the lift of @xmath547 is a coarsening of @xmath132 and the lift of @xmath547 is a ray of the secondary fan of @xmath543 .
therefore the marked subdivision @xmath132 itself must determine a lift .
suppose @xmath548 .
there must be at least @xmath182 polygons in the subdivision .
there must therefore be an edge @xmath167 of a @xmath309 which is not an edge of @xmath514 .
if the edge has both end - points on the boundary of @xmath514 then the fold through @xmath167 is a coarsening of @xmath132 .
therefore @xmath132 must determine a fold .
otherwise , one end - point of @xmath167 must be the interior lattice point , @xmath526 .
there can be at most @xmath10 edges of the subdivision meeting at @xmath526 .
if there were more , we could remove all but @xmath10 of them and still have less than @xmath533 degrees counterclockwise between any two edges .
therefore , we have a pinwheel .
there is a natural action of @xmath549 on @xmath550 .
this induces an action of @xmath551 on the secondary fan of @xmath8 .
we list the folds and pinwheels that are rays of the secondary fan of @xmath8 up to @xmath551 action .
jrg rambau . . in arjeh
m. cohen , xiao - shan gao , and nobuki takayama , editors , _ mathematical software - icms 2002 _ , pages 330340 .
world scientific , 2002 .
available at http://www.uni - bayreuth.de / departments / wirtschaftsmathematik / rambau / topcom/.
|
several results in tropical geometry have related the @xmath0-invariant of an algebraic plane curve of genus one to the cycle length of a tropical curve of genus one . in this paper
, we prove that for a plane cubic over the field of puiseux series the negative of the generic valuation of the @xmath0-invariant is equal to the cycle length of the tropicalization of the curve , if there is a cycle at all .
|
nowadays , people of various professions use different cad systems .
there are many ways to represent curves and surfaces , therefore , the exchange of geometric data between those systems often requires approximate conversion .
as it was stated in @xcite , there are two main operations that should be considered : degree reduction and merging . in the past @xmath1 years , both problems have been extensively investigated . in this paper , we focus on the constrained merging of segments of a composite bzier curve , i.e. , we look for a single bzier curve that approximates multiple adjacent bzier curves and satisfies certain conditions .
we propose the so - called _ box constraints _ , which appear for the first time in the context of the merging problem .
a conventional problem of merging is to approximate multiple adjacent bzier curves with a single bzier curve which minimizes a selected error function and satisfies some continuity constraints at the endpoints .
most of the papers deal with merging of only two bzier curves ( see @xcite ) .
obviously , to merge more than two curves , one could use those algorithms repeatedly .
however , such an approach increases the error of the approximation as well as the computational cost ( see @xcite ) .
there are three methods that specialize in merging of more than two bzier curves at the same time ( see @xcite ) .
regardless of how many curves are merged , the most frequently used strategy is to solve a system of normal equations ( see , e.g. , @xcite ) . in @xcite
, one can observe a different approach which is based on the properties of the so - called constrained dual bernstein basis polynomials ( to our knowledge , this method is the fastest one available ) .
the parametric ( see , e.g. , @xcite ) or geometric ( see , e.g. , @xcite ) continuity at the endpoints is preserved . in @xcite
, one of us proposed a new approach to the problem of degree reduction of bzier curves .
the author noticed that as a result of the conventional degree reduction , the computed control points can be located far away from the plot of the curve .
he also explained why this is a serious defect .
next , to eliminate this issue , he solved the degree reduction problem with constraints of a new type . in this paper , we show that the same observations may apply to the control points of the merged curve .
therefore , the main goal of this paper is to formulate a new problem of merging of bzier curves . as in @xcite
, the new approach requires completely different methods than in the case of the conventional one .
the outline of the paper is as follows .
further on in this section , we give necessary definitions and notation . in section [ sec : pre ] , we formulate the problem of merging of bzier curves with box constraints .
the example motivating the restricted area approach is given in section [ sec : motiv ] .
section [ sec : mer ] brings a solution of the new problem .
some illustrative examples are presented in section [ sec : ex ] . for a brief summary of the paper ,
see section [ sec : conc ] .
let @xmath2 denote the space of all parametric polynomials in @xmath3 of degree at most @xmath4 ; @xmath5 .
further on in the paper , we use @xmath6 $ ] , where @xmath7 to denote the vector of _ bernstein polynomial basis _ in @xmath8 .
we recall the well - known _ gramian matrix _ @xmath9 \in \mathbb{r}^{(m+1 ) \times ( n+1)}$ ] of the bernstein basis with the elements given by @xmath10 _ forward difference operator _
is defined by @xmath11 let @xmath12 be a matrix , and let @xmath13 $ ] , @xmath14 $ ] be the sets of natural numbers sorted in ascending order .
notation @xmath15 defines a matrix formed by rows @xmath16 and columns @xmath17 of the matrix @xmath18 .
similarly , we use @xmath19 , where @xmath20 is a vector in @xmath21 .
in this section , we formulate the following new problem of merging of bzier curves . [ ] [ p : problem ] + let @xmath22 be a partition of the interval @xmath23 $ ] .
let there be given a _
composite bzier curve _ @xmath24 ( @xmath25 $ ] ) in @xmath26 , which in the interval @xmath27 $ ] ( @xmath28 ) is exactly represented as a _ bzier curve _ @xmath29 , @xmath30 where @xmath31 , and @xmath32^t$ ] with @xmath33 .
+ find a bzier curve @xmath34 , @xmath35 where @xmath36^t$ ] with @xmath37 , satisfying the following conditions : * value of the squared _
@xmath38 where @xmath39 denotes the _ euclidean vector norm _ in @xmath3 , is minimized in the space @xmath2 ; * _ parametric continuity constraints at the endpoints _ are satisfied , i.e. , @xmath40 r^{(j)}(1)=p^{(j)}(1 ) \qquad ( j=0,1,\ldots , l-1 ) , \end{array}\ ] ] where @xmath41 , @xmath42 , and @xmath43 ; * control points @xmath44 @xmath45 are located inside the specified @xmath46-dimensional cube including the edges , i.e. , the following box constraints are fulfilled : @xmath47 where @xmath48 .
notice that in the case of degree reduction of bzier curves , analogical problem was formulated ( cf .
* problem 3.1 ) ) .
[ r : trad ] note that papers @xcite deal with the minimization of , with the conditions , but without the box constraints .
in addition , a reasonable assumption that @xmath49 is made . further on in this paper , such an approach is called the _
traditional merging _
* remark 3.2 ) ) .
as it turns out , the observations on the degree reduction problem ( see @xcite ) also apply to the merging problem . in order to see the issue clearly ,
let us consider the following example .
[ ex:1 ] we give the planar composite bzier curve `` ampersand '' , with three fifth degree bzier segments ( see figure [ figure1a ] ) , defined by the control points @xmath50
@xmath51 @xmath52 @xmath53 @xmath54 @xmath55 , @xmath56 @xmath57 @xmath58 @xmath59 @xmath60
@xmath61 , and @xmath62 @xmath63 @xmath64 @xmath65 @xmath66 @xmath67 @xmath68 , respectively . assuming that the partition of the interval @xmath23 $ ] is given by @xmath69 ; we look for a single bzier curve being the result of the traditional merging for @xmath70 , @xmath71 , @xmath72 ( see remark [ r : trad ] ) .
figure [ figure1b ] shows the original composite curve and the merged curve .
clearly , the result of the approximation is very accurate .
the errors are @xmath73 and @xmath74 , where @xmath75 with @xmath76 for @xmath77 .
observe also that the original control points are quite close to the plot of the curve ( see figure [ figure1a ] ) .
in contrast , the resulting control points are located far away from the plot of the curve ( see figure [ figure1c ] ) . note that we are unable to see the curve and its control points in one figure . because of the non - intuitive location of the control points , further modeling of the merged curve is hard to imagine . a designer that modifies the control points uses a convex hull property , which gives an intuition on shape and location of the curve . as it was stated in @xcite ,
the size of the convex hull is a measure of _ predictability _ of the curve .
furthermore , let us recall that a small convex hull can be helpful while checking that two curves do not intersect , a curve and a surface do not intersect , a point does not lie on a curve .
observe that the convex hull of the resulting curve is huge , therefore , completely useless .
comparing this result with the ones from @xcite , we see that the defect seems to be even more significant ( cf .
* figures 1b , 4b and 5b ) ) .
now , let us impose some box constraints .
we want the searched control points to be inside the specified rectangular area ( including edges of the rectangle ) .
figure [ figure2 ] presents the solution of problem [ p : problem ] for @xmath70 , @xmath71 , @xmath72 , @xmath78 , @xmath79 , @xmath80 , @xmath81 .
notice that the approximation is quite accurate ( errors : @xmath82 and @xmath83 ) .
moreover , in this case , the computed control points are located much closer to the merged curve . as a result ,
the curve can be easily and intuitively modified by moving these points .
what is more , we have obtained much smaller convex hull , which can be used to solve efficiently some important problems .
more examples can be found in section [ sec : ex ] .
see the restricted area ( black dotted - dashed frame).,width=275 ]
now , we give the method of solving problem [ p : problem ] .
first , we notice that some observations concerning the box - constrained degree reduction are also true in the case of the box - constrained merging .
clearly , a bzier curve being the solution of problem [ p : problem ] can be obtained in a componentwise way ( cf .
* remark 3.3 ) ) . therefore , it is sufficient to describe our method in the case of @xmath84 .
further on in this section , we assume that @xmath85 @xmath86 , @xmath87 , and @xmath88 are the lower and upper bounds for the box constraints . next , we recall that the conditions yield the following well - known formulas ( see , e.g. , ( * ? ? ? * theorem 3.1 ) ) : @xmath89 & \displaystyle r_{m - j } = ( -1)^j\binom{n_s}{j}\binom{m}{j}^{\!-1}\,\delta^jp^{s}_{n_s - j}- \sum_{h=1}^{j}(-1)^h\binom{j } { h}r_{m - j+h}\qquad ( j=0,1,\ldots , l-1 )
. \end{aligned}\ ] ] what remains is to minimize @xmath90 subject to the conditions for @xmath91 .
one can see clearly that @xmath90 is a quadratic function .
therefore , in the following subsection , we are dealing with the so - called _ box - constrained quadratic programming problem_. in this subsection , we use the quadratic programming approach to solve problem [ p : problem ] .
quadratic programming is an _ optimization problem _ of minimizing or maximizing a quadratic _ objective function _ of several variables subject to linear constraints on these variables .
taking into account the particular form of the restrictions , let us consider the following quadratic programming problem with box constraints : @xmath92 where @xmath93 \in \mathbb{r}^{m - k - l+1}$ ] and @xmath94 . in our case , we set @xmath95 , where we define @xmath96 and use the notation of .
now , we will adjust @xmath90 to the form .
first , taking into account that @xmath97 is a piecewise polynomial , we have to subdivide the searched polynomial @xmath98 as well .
this can be done by applying the de casteljau algorithm . in @xcite ,
lu gave the following formula : @xmath99 where @xmath100 with @xmath101 [ r : d ] according to ( * ? ? ?
* lemmas 2.5,2.4 ) , @xmath102 \in \mathbb{r}^{(m+1)\times(m+1)}$ ] , where the entries @xmath103 @xmath104 satisfy the following recurrence relation : @xmath105\\ \qquad\quad = ( m - h)d_{j , h+1}^{(i ) } + ( 2h - m)d_{jh}^{(i ) } - hd_{j , h-1}^{(i)}\\ ( 1 \leq j \leq m-1;\ 0 \leq h \leq m).\end{gathered}\ ] ] therefore , one can avoid matrix multiplications and compute the matrices @xmath106 efficiently , with the complexity @xmath107 , using ( * ? ? ?
* algorithm 4.1 ) . observe that the direct use of results in the complexity @xmath108 .
next , assuming that @xmath109 , @xmath110 and using the notation of , we write ( cf .
* ( 11 ) ) ) @xmath111 where @xmath112 and @xmath113 is a certain constant term .
obviously , @xmath114 is meaningless in the minimization process , therefore , the _ significant terms _ of @xmath90 are given by @xmath115 , which is written in the form .
[ r : sol ] matrix @xmath116 is positive definite ( see @xcite ) , therefore , the objective function @xmath117 is strictly convex .
furthermore , the feasible set is nonempty , closed and convex .
we conclude that the quadratic programming problem has a unique solution ( see , e.g. , ( * ? ? ?
* proposition 2.5 ) ) and so does problem [ p : problem ] .
in contrast , a solution of the analogical degree reduction problem may not be unique ( cf .
* theorem 4.1 ) ) .
the difference is that , in the present paper , we consider the _ continuous inner product _
( see ) instead of the _ discrete inner product _
( see ( * ? ? ? * ( 3.1 ) ) ) .
there are many papers dealing with the box - constrained quadratic programming problem . to solve it , one can use a variety of strategies , including _ active set methods _ ( see , e.g. , @xcite ) and _ interior point algorithms _ ( see , e.g. , @xcite ) .
some of the approaches combine the active set strategy with _ gradient projection method _
( see , e.g. , @xcite ) . for extensive lists of references ,
see the mentioned papers .
in this section , we apply our method to the composite bzier curves in @xmath118 . as in @xcite
, we generalize the approach of @xcite and obtain a partition of the interval @xmath119=[0,\,1]$ ] according to the lengths of segments @xmath120 : @xmath121 where @xmath122 integrals are evaluated using @xmath123 ` int ` procedure with the option ` numeric ` .
a solution of the traditional merging problem ( see remark [ r : trad ] ) is computed using ( * ? ? ?
* algorithm 4.2 ) .
the complexity of this algorithm is @xmath107 which , to our knowledge , is significantly less than cost of other methods of merging with the constraints ( cf .
@xcite ) . to solve the quadratic programming problem with box constraints
, we use the matrix version of maple ` qpsolve ` command .
it is worth noting that this procedure implements an iterative active set method and it is suited for the box constraints , i.e. , the vectors of lower and upper bounds can be passed using the optional parameter ` bd ` . according to the documentation provided by @xmath124 , in the case of the convex optimization , a global minimum
is returned ( cf .
remark [ r : sol ] ) . for the initial point
, we choose the lower bounds , i.e. , @xmath125 and @xmath126 .
the results have been obtained on a computer with ` intel core i5 - 3337u 1.8ghz ` processor and ` 8 gb ` of ` ram ` , using @xmath127-digit arithmetic .
@xmath128 worksheet containing programs and tests is available at http://www.ii.uni.wroc.pl/~pgo/papers.html .
[ ex:2 ] we introduce the composite bzier curve `` d '' ( see figure [ figure3a ] ) , formed by three cubic segments which are defined by the control points @xmath129
@xmath130 @xmath131 @xmath132 , @xmath133 @xmath134 @xmath135 @xmath136 and @xmath137 @xmath138 @xmath139 @xmath140
@xmath141 , respectively . formula implies @xmath142 .
figure [ figure3b ] shows the result of the traditional merging for @xmath143 , @xmath144 , @xmath145 .
the merged curve looks like a perfect approximation ( errors : @xmath146 and @xmath147 ) , unfortunately , it suffers from the defect described in section [ sec : motiv ] ( see figure [ figure3c ] ) . to avoid this ,
we solve problem [ p : problem ] for @xmath143 , @xmath144 , @xmath145 , with the following box constraints : @xmath148 ( cf . ) , and obtain the curve shown in figure [ figure3d ] ( errors : @xmath149 and @xmath150 ) . compare figure [ figure3d ] with figure [ figure3c ] to see a big difference in the location of the resulting control points .
obviously , the curve in figure [ figure3d ] is much more satisfying in this regard .
c + + [ ex:3 ] now , we consider the composite bzier curve with four fifth degree bzier segments ( see figure [ figure4a ] ) .
for the original control points , see ( * ? ? ?
* example 3 ) . to place the curve inside the unit box , we have divided each coordinate of the control points by @xmath151 .
according to , we get @xmath152 . as a result of the traditional merging ( @xmath153 , @xmath154 ) , we obtain the bzier curve which is illustrated in figure [ figure4b ] . once again , we get a good approximation ( errors : @xmath155 and @xmath156 ) , however , the resulting control points are located far away from the plot of the curve ( see figure [ figure4c ] )
. taking into account the axis scale in figure [ figure4c ] , we conclude that this example seems to be extremely difficult .
nonetheless , the solution of problem [ p : problem ] for @xmath153 , @xmath154 , with the box constraints @xmath157 is quite decent ( errors : @xmath158 and @xmath159 ) .
see figure [ figure4d ] .
c + [ r : res ] as stated in ( * ? ? ?
* remark 6.3 ) , selection of the restricted area is a difficult issue .
the choice always depends on the considered example and on the precision level that we accept as satisfactory . however , there is a strategy that seems to work quite well for the given examples . to explain this procedure ,
let us revisit example [ ex:2 ] . at the beginning
, we set @xmath160 consequently , the resulting control points will be bounded by the outermost control points of the original curves .
unfortunately , the obtained curve is unsatisfactory ( see figure [ figure5a ] ) .
next , to improve this result , we must expand the restricted area .
intuition tells us that we should try to move the borders with the highest numbers of the control points .
we consider @xmath161 where @xmath162 is the diagonal length of @xmath163-th restricted area .
notice that the error is now lower ( see figure [ figure5b ] and table [ tab : table1 ] ) .
therefore , we should try to make another step in the same direction .
this time , the expansion is greater , i.e. , we set @xmath164 the result can be seen in figure [ figure5c ] .
see also table [ tab : table1 ] . observe that , in example [ ex:2 ] , the restricted area is even larger .
see figure [ figure3d ] .
taking into account that ` qpsolve ` is an iterative method which we apply separately for each coordinate , pairs of numbers of iterations are also given in table [ tab : table1 ] . according to our experiments ,
if the control points of the optimal solution of the traditional merging are located very far away from the plot of the curve ( see figures [ figure1c ] , [ figure3c ] and [ figure4c ] ) , then it is difficult to find a satisfying solution of problem [ p : problem ] .
for that reason , the examples given in this paper are much more demanding than the ones presented in @xcite .
moreover , note that in the case of the box - constrained merging , majority of the resulting control points are located on borders ( see figures [ figure2 ] , [ figure3d ] and [ figure4d ] ) . regardless of choice of the restricted area , one should realize that because of the additional constraints , approximation error must be inevitably larger than for the traditional approach .
c [ r : time ] to solve the box - constrained quadratic programming problem , one can choose a method provided by a software library of a selected programming language or implement one of the algorithms given in @xcite .
for that reason , the running times strongly depend on the implementation of the selected method .
however , regardless of the choice , the box constraints make problem [ p : problem ] more difficult to solve . therefore , the running times of methods dealing with the new problem must be longer than in the case of the traditional merging .
see the comparison given in table [ tab : table2 ] .
the new approach to the problem of merging of bzier curves is introduced .
we propose constraints of the new type and explain the purpose of those restrictions .
a curve being the solution of problem [ p : problem ] is suitable for further modification and applications .
moreover , the resulting convex hull is much smaller than the one obtained using the traditional approach .
consequently , it can be helpful while solving some important problems .
these positive attributes make the new problem worth of consideration , despite the inevitably longer running times and the larger approximation errors , which are also unavoidable .
what is more , the comparison of the results with the ones from @xcite , leads to a conclusion that in the case of the traditional merging , the defect described in section [ sec : motiv ] is even more significant . in the near future ,
the authors intend to study a more general version of problem [ p : problem ] with the geometric continuity constraints instead of the conditions .
furthermore , a different strategy of setting the restricted area could also improve the results .
the authors are grateful to the referees for their remarks which helped to improve the paper .
han , p.m. pardalos , y. ye , computational aspects of an interior point algorithm for quadratic programming problems with box constraints , in : t.f .
coleman , y. li ( eds . ) , proceedings of the workshop on large - scale numerical optimization , siam , philadelphia , 1990 , 92112 .
|
in this paper , we present a novel approach to the problem of merging of bzier curves with respect to the @xmath0-norm . we give illustrative examples to show that the solution of the conventional merging problem may not be suitable for further modification and applications . as in the case of the degree reduction problem
, we apply the so - called restricted area approach proposed recently in ( p. gospodarczyk , computer - aided design 62 ( 2015 ) , 143151 ) to avoid certain defects and make the resulting curve more useful . a method of solving the new problem
is based on box - constrained quadratic programming approach .
bzier curve , merging , multiple segments , parametric continuity , quadratic programming , box constraints .
|
we consider the systematic numerical approximation of a class of allen - cahn type equations describing , for instance , the motion of anti - phase boundaries in crystalline solids @xcite , the evolution of interfaces in more general phase field models @xcite , or the geometric evolution by mean curvature @xcite . for motivation of our considerations ,
let us briefly consider two particular examples which have been studied in detail in @xcite .
as a first model problem , we consider the generalized allen - cahn equation @xmath0 here @xmath1 is an order parameter which takes values close to zero in one phase of the medium and close to one in the other , @xmath2 is the double well potential , @xmath3 is a source term due to external loading , and @xmath4 , @xmath5 are model parameters .
the coefficient @xmath6 plays the role of a regularization parameter that allows to widen the transition zone between the phases and thus to alleviate the numerical simulation . for the choice @xmath7 , @xmath8 , @xmath9 , and @xmath10 , one obtains the standard allen - cahn equation @xmath11 which has been studied intensively in the literature
; see e.g. @xcite and the references given there .
it is well - known @xcite that , and therefore also , can be understood as a gradient flow for an associated free energy functional , which for the problem has the form @xmath12 the multiplicative term @xmath13 on the right hand side of yields a scaling of time that leads to a non - trivial behavior for @xmath14 , which is called the _ sharp interface limit_. it is shown in @xcite that the velocity of the interface @xmath15 can be expanded as @xmath16 \cdot n_\gamma + c \lambda^{1/2 } \kappa_{\gamma } + o(\mu^{1/2}),\end{aligned}\ ] ] where @xmath17 can be expressed expicitly in terms of the problem data .
$ ] denotes the jump of the eshelby tensor @xcite which is related to the external forces @xmath3 and @xmath19 is the normal vector while @xmath20 is the mean curvature of the interface @xmath21 . in the limit @xmath14 , the velocity of the interface motion
can thus be described by the well - known relation @xcite @xmath22 \cdot n_\gamma + c \lambda^{1/2 } \kappa_{\gamma}.\end{aligned}\ ] ] choosing the regularization parameter @xmath23 leads to an incorrect speed of the interface motion , and according to , one should choose @xmath24 to keep the consistency error small .
for parameter @xmath23 the interface is widened to a transition zone of width @xmath25 . in order to resolve the smoothed interface in simulations , the mesh size should therefore be chosen at least as small as @xmath26 .
more restrictive assumptions on the mesh size are required to really prove a good approximation of the interface evolution ; see @xcite for details . if @xmath5 is small , then a very small mesh size is thus required to obtain a good approximation . in order to allow for a larger mesh size and to reduce the computational burden ,
the following model for the interface motion has been proposed and analyzed in @xcite @xmath27 this equation , which has been termed _
hybrid model _ in @xcite , describes the motion of the levelsets of the order parameter @xmath28 with normal speed proportional to the term in parenthesis . following the arguments of @xcite
, the problem can again be interpreted as a gradient flow for an associated energy functional , which here reads @xmath29 the gradient here has to be defined with respect to a metric induced by the solution @xmath30 itselve ; we refer to @xcite and to section [ sec : variational ] for details . using asymptotic analysis ,
the propagation speed of the interface has been shown in @xcite to behave like @xmath31 \cdot n + \nu^{1/2 } \omega_1 \kappa_\gamma + o(\nu^{1/2}),\end{aligned}\ ] ] where the parameter @xmath32 can again be computed from the model data .
a comparison of the two formulas and reveals that for a proper choice of the model and scaling parameters , the hybrid model leads to the same propagation speed for the interface @xmath21 as in the allen - cahn model up to higher order terms . to obtain a good agreement of the results
, one should choose @xmath33 and in this case width of the smoothed interface in the hybrid model can be shown to behave like @xmath34 ; see @xcite for details .
when @xmath5 is small , the smoothed interface of the hybrid model is thus substantially larger then that of the corresponding allen - cahn model and a mesh size of @xmath35 should be sufficient to resolve the interface in computations , instead of @xmath36 required for the corresponding allen - cahn model .
the hybrid model should therefore allow to compute the interface evolution on a much coarser mesh , which was the main motivation for the proposal of this method .
due to the close connection to mean curvature flow , there has been much interest in the numerical approximation for the allen - cahn equation .
various discretization schemes have been proposed and analyzed , see e.g. @xcite and the references given there .
particular aspects of the efficient numerical approximation have been addressed in @xcite and a - posteriori error estimates have been derived in @xcite .
the convergence of numerical approximations of the allen - cahn equation in the sharp interface limit @xmath37 towards the generalized motion by mean curvature has been established in @xcite ; see also @xcite for further numerical tests .
the direct numerical approximation of mean curvature flow has also been investigated intensively ; see for instance @xcite and refer to @xcite for a review about available results and further references .
the focus of the current manuscript is somewhat different to these previous works : here we want to study the systematic numerical approximation of a large class of allen - cahn type models which includes , , and as special cases .
the common feature of all these models is the underlying gradient flow structure with respect to an associated energy functional , and we aim to strictly preserve this structure in the approximation process on the semi - discrete and the fully discrete level . with this in mind , we automatically obtain discretization schemes that are uniformly energy stable and thus consistent with the second law of thermodynamics . in this paper
we establish well - posedness and qualitative properties of a discretization strategy appropriate for a large class of allen - cahn type problems .
the models and introduced above will be considered as particular examples . as a by - product of our investigations
, we also obtain further numerical evidence for the asymptotic expansions and of the interface velocities .
our main arguments can be extended to elastic allen - cahn type models which describe the evolution of phase interfaces in binary alloys @xcite .
the required coupling to the additional elasticity system will be investigated in a forth - comming publication .
the remainder of the manuscript is organized as follows : in section [ sec : variational ] , we first introduce a unified formulation of the problems discussed in the introduction and then present a variational characterization of smooth solutions . in addition , we verify that the respective energies decrease along solution trajectories which is due to the underlying gradient flow structure . in section [ sec : galerkin ] , we then discuss the galerkin approximation of the underlying variational principles .
we establish the well - posedness of the resulting semi - discrete schemes and verify the energy decay on the discrete level . in section [ sec : time ] , we then discuss the subsequent discretization in time by a problem adapted implicit time stepping scheme .
we establish the well - posedness of the fully discrete problem and again prove the decay in energy . a particular space discretization by finite elements is discussed in seciton [ sec : implementation ] , and some further aspects of the implementation are presented . for illustration of our theoretical results ,
we report in section [ sec : numerics ] about some numerical tests , in which we also verify the asymptotic expansions for the interface velocities given in and .
we close with a short discussion of our results and open problems for future research .
for the rest of the manuscript , we consider the following general model problem : @xmath38 here @xmath39 , @xmath40 is a bounded lipschitz domain , @xmath41 are model parameters , and the functions @xmath42 , @xmath43 are given a - priori . for our analysis
later on , we will assume that * @xmath44 and @xmath45 ; * @xmath46 with @xmath47 , @xmath48 , and @xmath49 for all @xmath50 .
some of these technical assumptions are made only for convenience and could further be relaxed without problems .
this will become clear from our analysis below and we will make some remarks in this direction . to complete the model description , we additionally assume that @xmath51 other boundary conditions can also be treated without much difficulty .
the above problem is of quasilinear parabolic type and existence of solutions for appropriate initial values @xmath52 can be proven with standard arguments ; see e.g. @xcite .
a function @xmath53 \times \overline \omega)$ ] is called smooth solution of , if it satisfies the respective equations in a pointwise sense . here
@xmath54 \times \overline \omega)$ ] denotes the space of continuous functions that are continuously differentiable with respect to @xmath55 and twice continuously differentiable with respect to @xmath56 .
note that acording to our definition , smooth solutions and their spatial derivatives are uniformly bounded .
the allen - cahn model can be phrased in the form by setting @xmath57 , @xmath58 , @xmath59 , and defining the potential as @xmath60 .
note that the second derivative @xmath61 of the potential is unbounded here . using maximum principles ,
the solution @xmath30 can however be shown to remain bounded and therefore @xmath62 could be modified for large @xmath30 in order to satisfy ( a2 ) without changing the solution ; see @xcite for other appropriate conditions on the potential .
the hybrid model can also be cast into the form with @xmath63 , @xmath64 , @xmath65 , and @xmath66 . to avoid problems for vanishing gradient
, one can set @xmath67 for some @xmath68 instead , and thus obtain a regularized version of .
the limit @xmath69 may be studied in the framework of @xmath21-convergence @xcite . as a next step ,
let us define an associated energy functional for the problem by @xmath70 equation can then be interpreted as a gradient flow for this energy with respect to the metric @xmath71 .
problem can thus be formally expressed as @xmath72 which , following @xcite , has to be understood in the sense that @xmath73 for all @xmath74 and @xmath75 .
the last term is the negative directional derivative of the energy functional in direction @xmath76 .
these formal arguments motivate the following variational principle .
[ lem : variational ] let @xmath30 be a smooth solution of .
then @xmath77 for all @xmath78 and all @xmath79 . here
@xmath80 is the scalar product of @xmath81 . as a direct consequence of the gradient flow structure ,
we obtain [ lem : decay ] let @xmath30 be a smooth solution of .
then @xmath82 since @xmath30 is smooth , we can apply the chain rule to get @xmath83 by lemma [ lem : variational ] , any smooth solution of also satisfies the variational principle , and testing this variational equation with @xmath84 immediately yields @xmath85 for the last step we only had to use that @xmath42 is non - negative .
motivated by the results of the previous section , we now consider the galerkin approximation of the variational principle to construct semi - discretizations in space .
let us choose some finite dimensional subspace @xmath86 and consider the following discrete variational problem .
[ prob : semi ] @xmath87 + let @xmath88 and @xmath89 be given .
find @xmath90;v_h)$ ] such that @xmath91 and @xmath92 for all test functions @xmath93 and for all @xmath94 . since we employ a conforming galerkin approximation , the gradient flow structure and thus also the energy decay estimates are inherited automatically by the semi - discrete problem .
[ lem : decayh ] let @xmath95 denote a solution of problem [ prob : semi ] .
then @xmath96 the assertion follows with literally the same arguments as in lemma [ lem : decay ] . as a next step ,
let us establish the well - posedness of the semi - discrete problem
. for any @xmath88 and @xmath89 , problem [ prob : semi ] has a unique solution . after choosing a basis for @xmath97
, we can rewrite the discrete problem as initial value problem @xmath98 where @xmath99 here denotes the coordinate vector when expanding @xmath100 in the chosen basis . due to assumptions ( a1)(a2 ) on the coefficients and since @xmath97 is finite dimensional , the matrix functions @xmath101 and @xmath102 can be seen to be lipschitz continuous . from the lower bounds on @xmath42 we further obtain that @xmath103 is symmetric positive definite .
existence of a unique local solution then follows from the picard - lindelf theorem . from and
, we can see that @xmath104 here we used that @xmath105 due to assumption ( a2 ) .
this yields @xmath106 and by the growth condition and the positivity assumption for @xmath107 , we further obtain @xmath108 by the poincar inequality and positivity of @xmath109 , we thus get @xmath110 with constant @xmath111 only depending on the domain , on the bounds in the assumptions , and on the energy of the initial value .
this implies that @xmath99 remains uniformly bounded and hence the solution can be extended uniquely to arbitrarily large @xmath75 . under assumptions ( a1)(a2 )
, we thus obtain global existence of a unique semi - discrete solution @xmath95 .
the constant @xmath111 in the a - priori bound @xmath110 only depends on the initial energy @xmath112 , on the values of @xmath113 and @xmath114 , and on the lower bound @xmath115 for the growth rate of the potential @xmath43 .
let us note that the energy decay estimate also provides a uniform bound for the time derivative of the solution which might be useful for a full convergence analysis for the galerkin approximation ; we refer to @xcite for results in this direction .
as a second step in the approximation process , let us now discuss the discretization in time . given some time step @xmath116
, we define @xmath117 for @xmath118 , and we denote by @xmath119 the backward difference quotient at time @xmath120 .
as will become clear from our analysis , non - uniform time steps could be considered as well . for the time discretization of the galerkin approximation defined in problem [ prob : semi ]
, we then consider the following implicit time - stepping scheme .
[ prob : full ] @xmath87 + set @xmath121 with initial value as in problem [ prob : semi ] . for any @xmath122 ,
find @xmath123 such that @xmath124 for all @xmath93 .
we set @xmath125 wherever @xmath126 .
let us note that a similar treatment of the nonlinear term has been proposed in the context of mean - curvature flow in @xcite .
the particular of the discrete variational problem allows us to establish a decay estimate for the energy of the fully discrete approximation with very similar arguments as used for the analysis on the continuous level .
[ lem : decayhh ] let @xmath127 denote a solution of problem [ prob : full ] .
then @xmath128 using definition , we can decompose the discrete energy as @xmath129 we now utilize the fully discrete variational principle with @xmath130 to replace the first and last term together by @xmath131 which directly yields the result .
the second term on the right hand side of the above energy estimate stems from numerical dissipation of the implicit time - stepping scheme .
the particular time discretization thus enhances the stability of the numerical solution . as further theoretical backup
, we next establish the well - posedness of the fully discrete scheme .
let @xmath132 be given .
then has at least one solution @xmath123 , and the solution is unique for all time steps @xmath133 with some @xmath134 sufficiently small .
existence of a solution @xmath135 follows from the energy estimate given in the previous section and the brouwer fixed point theorem , and we know that @xmath136 .
now let @xmath135 and @xmath137 be two solutions of .
then testing with @xmath138 and using assumption ( a1 ) leads to @xmath139 for ease of notation , we utilized the symbol @xmath140 here to abbreviate the difference quotient . by the fundamental theorem of calculus and using the bounds of assumption ( a2 ) for the coefficient @xmath43
, we further get @xmath141 a combination of the two estimates then directly leads to @xmath142 for @xmath116 small enough , the right hand side can be absorbed by the terms on the left side and we thus also obtain the uniqueness of the solution . before we proceed , let us briefly indicate at this point how the assumption ( a1)(a2 ) could be possibly relaxed , which can be deduced by inspection of the previous proof .
[ rem : ass ] the bounds for @xmath143 are not required for all @xmath144 here , but only for values @xmath30 that are actually attained during the evolution . if a bound @xmath145 is available , which may be proved by a discrete comparison principles @xcite , then one could replace the constant @xmath146 in the above proof by @xmath147 .
if the space @xmath97 admits an inverse inequality @xmath148 then the a - priori estimate @xmath149 implies @xmath150 with @xmath151 .
the constant @xmath152 can be estimated explicitly for many discretization schemes , e.g. , for finite element methods which discussed in the following section . in one space dimension ,
the above inverse inequality holds true for any choice of @xmath97 due to the continuous embedding of @xmath153 into @xmath154 .
more general conditions on the potential @xmath43 have also been considered in @xcite .
the considerations of this section show that we can expect unique solutions and global well - posedness for the numerical approximations of the model problems and , if the time step @xmath155 is chosen sufficiently small .
this is also observed in our computations .
before we turn to numerical tests , let us briefly discuss some details of the implementation that will be used in our computations later on .
this particularly involves the choice of the approximation spaces @xmath97 and the solution of the nonlinear systems in every time step .
let @xmath156 denote a non - overlapping partition of the domain @xmath39 , @xmath157 into triangles or tetrahedra .
we assume that the partition is conforming and uniformly shape - regular in the usual sense @xcite .
we then choose @xmath158 as the space of continuous piecewise polynomials of degree @xmath159 .
in our numerical tests we will actually only utilize the lowest order case @xmath160 , but the methods can in principle be extended to higher order without difficulty .
similar approximations in space have been considered in @xcite . without loss of generality
, one may assume that the potential @xmath43 is a polynomial function .
otherwise , @xmath43 can be replaced by a polynomial and due to the weierstrass theorem , the error of this approximation can be made arbitrarily small on bounded intervals . in our numerical tests
, we will utilize a function of the form @xmath161 the first part is the usual double - well potential and the linear term amounts to external forces . note that the potential @xmath43 does not strictly satisfy the assumptions ( a2 ) , in particular , the upper bounds on the second derivative are violated . as noted in remark [ rem : ass ]
, this assumption was made for convenience and is not required as long as @xmath30 stays uniformly bounded . for the above choice of the potential @xmath62 , the difference quotient arising in can be expressed as @xmath162 the nonlinear problem that has to be solved in every time step of the fully discrete scheme thus has a polynomial nonlinearity and is non - degenerate .
standard iterative methods can therefore be utilized for an efficient solution .
the following fixed - point iteration will be used in our numerical tests for the solution of the nonlinear problem that arises in every time step of the fully discrete scheme .
[ prob : iter ] let @xmath132 and @xmath163 , @xmath116 be given and define @xmath164 .
further set @xmath165 and for @xmath159 , find @xmath166 such that @xmath167 for all @xmath93 .
if convergence is reached after iteration @xmath168 , set @xmath169 .
choosing @xmath170 allows to improve the convergence behavior also in case @xmath171 ; see @xcite for a similar argument . as an immediate consequence of the particular construction ,
we obtain for any @xmath172 , the above fixed - point iteration is well - defined .
if @xmath116 is sufficiently small , then the iteration converges to the unique solution of problem [ eq : variationalhh ] .
first note that the problem that has to be solved in every step of the fixed - point iteration of problem [ prob : iter ] is linear .
moreover , the left hand side in the above iteration defines a positive definite quadratic form .
this implies the existence of a unique solution @xmath173 in every step of the iteration . for @xmath116
sufficiently small , one can show that the map @xmath174 is contractive , and convergence follows by the banach fixed - point theorem . for time step @xmath116
sufficiently small , the contraction constant in the above iteration becomes small and convergence will typically be observed within a few iterations . to speed up convergence , one may
alternatively also utilize newton - type iterations .
for our computations , we consider the allen - cahn equation and the hybrid model .
the model parameters are set to @xmath175 in all tests , and the double - well potential is defined as @xmath176 . as potential for the external forces , we choose @xmath177 with constant @xmath178 to be specified below .
the constant @xmath4 in the allen - cahn equation is chosen as @xmath179 . for this setting ,
the interface velocities of allen - cahn and the hybrid model given in and agree up to higher order terms ; see @xcite for details .
for the space discretization , we utilize the finite element method described in section [ sec : implementation ] with polynomial degree @xmath160 .
the time integration is performed by the implicit time - stepping scheme with constant time step @xmath116 , and the nonlinear systems arising in every time step are solved with the fixed - point iteration outlined above . as a first test case
, we consider a quasi one - dimensional geometric setting .
we choose @xmath180 and set @xmath181 where @xmath182 is the characteristic function of the set @xmath183 .
the interface @xmath21 between the two phases at time @xmath184 is a straight line here and thue the curvature @xmath185 is zero .
this can be shown to remain true for all @xmath75 by a symmetry argument .
the evolution of the interface is therefore only driven by external forces in this example .
for our computations , we choose @xmath177 with @xmath186 which yields a constant driving force @xmath187 acting on the whole domain .
as mentioned above , the solution @xmath188 can be shown to be independent of @xmath189 for all @xmath190 .
we thus obtain @xmath191 with @xmath192 defined by @xmath193 here @xmath194 is the one - dimensional cross - section at arbitrary @xmath189 and @xmath195 . the initial condition for the one - dimensional problem
is @xmath196 . in figure
[ fig:1d ] , we display a few snapshots of the solution @xmath197 obtained with the allen - cahn equation in two dimensions and the corresponding solutions @xmath198 and @xmath199 for the allen - cahn problem and the hybrid model in one space dimension . for the solution @xmath28 of the two - dimensional hybrid model at time @xmath200 ( top ) and solutions @xmath198 and @xmath199 for the corresponding one - dimensional problems ( bottom ) .
the grid lines do not coincide with the mesh but are plotted only to allow for a better comparison.[fig:1d],title="fig:",scaledwidth=20.0% ] for the solution @xmath28 of the two - dimensional hybrid model at time @xmath200 ( top ) and solutions @xmath198 and @xmath199 for the corresponding one - dimensional problems ( bottom ) .
the grid lines do not coincide with the mesh but are plotted only to allow for a better comparison.[fig:1d],title="fig:",scaledwidth=20.0% ] for the solution @xmath28 of the two - dimensional hybrid model at time @xmath200 ( top ) and solutions @xmath198 and @xmath199 for the corresponding one - dimensional problems ( bottom ) .
the grid lines do not coincide with the mesh but are plotted only to allow for a better comparison.[fig:1d],title="fig:",scaledwidth=20.0% ] for the solution @xmath28 of the two - dimensional hybrid model at time @xmath200 ( top ) and solutions @xmath198 and @xmath199 for the corresponding one - dimensional problems ( bottom ) .
the grid lines do not coincide with the mesh but are plotted only to allow for a better comparison.[fig:1d],title="fig:",scaledwidth=20.0% ] + for the solution @xmath28 of the two - dimensional hybrid model at time @xmath200 ( top ) and solutions @xmath198 and @xmath199 for the corresponding one - dimensional problems ( bottom ) .
the grid lines do not coincide with the mesh but are plotted only to allow for a better comparison.[fig:1d],title="fig:",scaledwidth=20.0% ] for the solution @xmath28 of the two - dimensional hybrid model at time @xmath200 ( top ) and solutions @xmath198 and @xmath199 for the corresponding one - dimensional problems ( bottom ) .
the grid lines do not coincide with the mesh but are plotted only to allow for a better comparison.[fig:1d],title="fig:",scaledwidth=20.0% ] for the solution @xmath28 of the two - dimensional hybrid model at time @xmath200 ( top ) and solutions @xmath198 and @xmath199 for the corresponding one - dimensional problems ( bottom ) .
the grid lines do not coincide with the mesh but are plotted only to allow for a better comparison.[fig:1d],title="fig:",scaledwidth=20.0% ] for the solution @xmath28 of the two - dimensional hybrid model at time @xmath200 ( top ) and solutions @xmath198 and @xmath199 for the corresponding one - dimensional problems ( bottom ) .
the grid lines do not coincide with the mesh but are plotted only to allow for a better comparison.[fig:1d],title="fig:",scaledwidth=20.0% ] the results clearly illustrate the relation @xmath201 between the solutions of the one- and the two - dimensional problem .
moreover , the propagation of interface @xmath202 for the one - dimensional allen - cahn and the corresponding hybrid model take place at the same speed , which is in perfect agreement with formulas and . in figure
[ fig : area1d ] , we display the area @xmath203 surrounded by the interface @xmath204 and we depict the decay of the energy functional for the two models illustrating the underlying gradient flow structure . for the one - dimensional allen - cahn model ( blue ) and the one- and two - dimensional hybrid model ( red on top of green ) .
right : energy @xmath205 for the one - dimensional allen - cahn ( blue ) and hybrid ( red ) model .
[ fig : area1d],title="fig:",scaledwidth=70.0% ] + as a second test case , we consider the evolution of a circular interface @xmath21 .
we again choose @xmath180 and define the initial values as @xmath206 here we set @xmath207 which yields @xmath18=0 $ ] and thus the motion is only driven by the curvature of the interface .
the solution of the sharp interface limit can be shown to be radially symmetric and according to , the propagation of the interface is governed by @xmath208 here @xmath209 denotes the radial position of the interface @xmath210 at time @xmath55 .
this allows to compute the area @xmath211 surrounded by the interface @xmath210 analytically , i.e. , @xmath212 in figure [ fig : circle2d ] , we display a few snapshots of the interface @xmath210 for the numerical solutions of the allen - cahn model and the hybrid model . for solutions @xmath1 and @xmath28 of the allen - cahn equation ( top ) and the corresponding hybrid model ( bottom ) for @xmath213.[fig : circle2d].,title="fig:",scaledwidth=20.0% ] for solutions @xmath1 and @xmath28 of the allen - cahn equation ( top ) and the corresponding hybrid model ( bottom ) for @xmath213.[fig : circle2d].,title="fig:",scaledwidth=20.0% ] for solutions @xmath1 and @xmath28 of the allen - cahn equation ( top ) and the corresponding hybrid model ( bottom ) for @xmath213.[fig : circle2d].,title="fig:",scaledwidth=20.0% ] for solutions @xmath1 and @xmath28 of the allen - cahn equation ( top ) and the corresponding hybrid model ( bottom ) for @xmath213.[fig : circle2d].,title="fig:",scaledwidth=20.0% ] + for solutions @xmath1 and @xmath28 of the allen - cahn equation ( top ) and the corresponding hybrid model ( bottom ) for @xmath213.[fig : circle2d].,title="fig:",scaledwidth=20.0% ] for solutions @xmath1 and @xmath28 of the allen - cahn equation ( top ) and the corresponding hybrid model ( bottom ) for @xmath213.[fig : circle2d].,title="fig:",scaledwidth=20.0% ] for solutions @xmath1 and @xmath28 of the allen - cahn equation ( top ) and the corresponding hybrid model ( bottom ) for @xmath213.[fig : circle2d].,title="fig:",scaledwidth=20.0% ] for solutions @xmath1 and @xmath28 of the allen - cahn equation ( top ) and the corresponding hybrid model ( bottom ) for @xmath213.[fig : circle2d].,title="fig:",scaledwidth=20.0% ]
+ a brief visual inspection shows that the motion of the interface is rather similar for both models .
one can further deduce from the plots that the width of the transition zone in the hybrid model is somewhat larger than for the allen - cahn model ; compare with our discussion in the introduction .
a coarser mesh might therefore be used for the simulation of the interface evolution by the hybrid model . for a better comparison of the two models
, we again display in figure [ fig : area_circle2d ] the evolution of the area @xmath211 and of the energy @xmath214 representing one phase of the system . enclosed by the interface @xmath210 for the allen - cahn model ( blue ) , the hybrid model ( red ) , and the exact solution of the sharp inferface limit ( black ) .
right : evolution of the energy @xmath214.[fig : area_circle2d],title="fig:",scaledwidth=70.0% ] + in accordance with formulas and for our choice of parameters , the evolution of the interface here occurs at approximately the same speed .
moreover , the energy @xmath214 is monotonically decreasing for both models which is in perfect agreement with the assertions of lemma [ lem : decayhh ] . as a last test case
, we now consider the evolution of an interface which initially is non - smooth and encloses a non - convex region at time @xmath184 .
some snapshots of the evolution are depicted in figure [ fig : general2d ] .
( top to bottom)[fig : general2d],title="fig:",scaledwidth=20.0% ] ( top to bottom)[fig : general2d],title="fig:",scaledwidth=20.0% ] ( top to bottom)[fig : general2d],title="fig:",scaledwidth=20.0% ] ( top to bottom)[fig : general2d],title="fig:",scaledwidth=20.0% ] + ( top to bottom)[fig : general2d],title="fig:",scaledwidth=20.0% ] ( top to bottom)[fig : general2d],title="fig:",scaledwidth=20.0% ] ( top to bottom)[fig : general2d],title="fig:",scaledwidth=20.0% ] ( top to bottom)[fig : general2d],title="fig:",scaledwidth=20.0% ] + like in the previous test , we set the force potental to @xmath207 which implies @xmath18=0 $ ] , and hence the motion is again driven only by the curvature .
note that in the vicinity of convex corners , the interface moves towards the interior of the enclosed area while at the non - convex corners , the interface propagates in the other direction , which is caused by the change of sign in the curvature . in figure
[ fig : area_general2d ] , we again depict the area surrounded by the interface and the uniform decay of the energy for the numerical solutions obtained with the two models . enclosed by the interface @xmath210 for allen - cahn ( blue ) and the hybrid model ( red ) .
right : evolution of the energies @xmath214.[fig : area_general2d],title="fig:",scaledwidth=70.0% ] + like in the previous examples , the velocity of the interface motion in the allen - cahn and the hybrid model are very similar ; compare with the formulas and .
due to the gradient flow structure of the fully discrete evolution , a strict decrease in the energy @xmath214 can again be observed .
the results are very similar , and the slight discrepancies can be explained by approximation errors .
in this paper , we investigated the systematic numerical approximation of a general class of allen - cahn type equations .
the common feature of these models was a gradient flow structure with respect to an associated energy .
we proposed and analyzed semi - discrete and fully discrete numerical schemes which strictly preserve the underlying gradient flow structure and which therefore automatically yield uniformly energy stable discrete approximations .
well - posedness of the numerical schemes and energy decay was established theoretically and illustrated by numerical tests .
our computational results also provide further numerical evidence for asymptotic expansions of interface motion obtained @xcite . in our analysis , we used some rather strong structural assumptions on the coefficients arising in the equations , which should allow to conduct a full a - priori convergence analysis of the semi- and fully discrete scheme by following the arguments of @xcite .
as indicated in remarks , a detailed inspection of the proofs allows to further relax these assumptions .
our arguments and methods of proof are rather general and seem to be applicable also to phase - field models for the dynamics of solid - solid phase transition which are governed by coupled allen - cahn and elasticity equations .
these topics and further extensions are left for future research .
this work was supported by the german research foundation ( dfg ) via grants irtg 1529 , trr 154 , and eg-331/1 - 1 and by the german excellence initiative via grant gsc 233 .
j. f. blowey and c. m. elliott . curvature dependent phase boundary motion and parabolic double obstacle problems . in _
degenerate diffusions ( minneapolis , mn , 1991 ) _ , volume 47 of _ i m a vol .
_ , pages 1960 .
springer , new york , 1993 .
u. clarenz , f. hauer , m. rumpf , a. voigt , and u. weikard . on level set formulations for anisotropic mean curvature flow and surface diffusion . in _ multiscale modeling in epitaxial growth _ ,
volume 149 of _ internat .
_ , pages 227237 .
birkhuser , basel , 2005 .
o. a. ladyenskaja , v. a. solonnikov , and n. n. uralceva . .
translated from the russian by s. smith .
translations of mathematical monographs , vol .
american mathematical society , providence , r.i . , 1968 .
|
we study the systematic numerical approximation of a class of allen - cahn type problems modeling the motion of phase interfaces .
the common feature of these models is an underlying gradient flow structure which gives rise to a decay of an associated energy functional along solution trajectories .
we first study the discretization in space by a conforming galerkin approximation of a variational principle which characterizes smooth solutions of the problem .
well - posedness of the resulting semi - discretization is established and the energy decay along discrete solution trajectories is proven .
a problem adapted implicit time - stepping scheme is then proposed and we establish its well - posed and decay of the free energy for the fully discrete scheme .
some details about the numerical realization by finite elements are discussed , in particular the iterative solution of the nonlinear problems arising in every time - step .
the theoretical results are illustrated by numerical tests which also provide further evidence for asymptotic expansions of the interface velocities derived by alber et al .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ allen - cahn equation , phase - field models , gradient systems , mean curvature flow , finite elements , implicit time stepping _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 65m60,74n20,35j93,53c44 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
|
production of low momentum particles in the region away from the beam remnant is a key laboratory to study the non - perturbative effect of hadronisation , the process of conversion of the partonic final state into hadrons .
the lack of understanding of this phenomenon leads to a huge variety of phenomenological models , each with a large number of free parameters .
an example of such a model is the lund string model @xcite and its implementation in the jetset monte carlo ( mc ) model@xcite .
comparison of the predictions of this model with accurate data from the lep experiments @xcite allows some of the parameters to be precisely fixed .
it is possible to test the universality of the fragmentation prescription in such models , by comparing the predictions of the mc , tuned with @xmath0 data , to the hera data in photoproduction . in this paper
we present a measurement@xcite of centrally produced low momentum protons and compare the results with the predictions of the pythia mc @xcite . a common behaviour as a function of @xmath2 in the production cross - section for different hadrons of mass @xmath3 and transverse momentum @xmath4 , produced in photoproduction , is also demonstrated .
experimental data for the analysis were collected with the h1 detector @xcite during the 1996 running period , in which hera collided 27.5 gev positrons with 820 gev protons .
the data sample corresponds to an integrated luminosity of 6.0 pb@xmath5 .
the photoproduction events are identified by a first level trigger which requires energy in the small angle electron tagger and the presence in the central tracking chamber of several well measured tracks which originate from the primary vertex .
the former requirement restricts the kinematic region to @xmath6 gev@xmath7 and @xmath8 , where @xmath9 is the virtuality of the exchanged photon and @xmath10 is the inelasticity of the scattered positron .
the average centre of mass energy of the @xmath11 system is 200 gev .
background events coming from the interaction of beam particles with the residual gas in the beam pipe ( beam - gas events ) are characterized by small or zero energy deposition in the direction opposite to the direction of the incoming beam particle .
to suppress such background , only events with energy deposits greater than 2 gev in both the forward region of the liquid argon calorimeter and in the backward calorimeter spacal are accepted .
( 6.68cm,5.5cm)[r ] ( 6.68,5.5)(0.0,0.1 ) ( -5.0,4.7)a ) ( -5.0,2.0)b ) charged particles are measured in the h1 central tracker which covers the polar angle range of @xmath12 , where @xmath13 is defined with respect to the incoming proton direction .
@xmath14 , the measured specific ionization energy loss of the particle , is used to identify the protons in the region of high acceptance and resolution , @xmath15 gev and @xmath16 , where @xmath4 is the transverse momentum relative to the beam - line and @xmath17 is the rapidity of the particle .
the main source of the background is the production of secondary protons by particles from ep collision , interacting in the beam pipe and other material in front of the track chambers . in order to estimate the yield of such protons the distribution of closest approach ( dca ) of the track to the primary vertex is analyzed . for background protons
this distribution is flat , in contrast to the shape for protons and anti - protons from ep interaction , see fig.[fig : dca ] .
this background is subtracted on a statistical basis .
the measurement is made in the laboratory frame of reference and is presented as an average cross - section @xmath18 for photoproduction of protons and anti - protons , as half the sum of the corresponding proton and anti - proton cross - sections . here e and p are the proton energy and momentum , respectively .
this cross section is shown as a function of the proton transverse momentum @xmath4 and rapidity @xmath17 in fig.[fig : data_pythia ] .
the errors are the quadratic sum of the statistical and systematic uncertainties .
the systematic component is dominant . using the mc simulation ,
the resulting cross - sections are corrected for the acceptance cuts discussed in the previous section ( cuts on the energy in forward and backward part of the h1 calorimeters and on the number of tracks coming from the primary vertex ) .
the measured cross - section is compared with the pythia mc model@xcite prediction , including the grv - lo proton and photon parton densities @xcite . in this model leading order ( lo ) qcd
matrix elements are used to describe the hard scattering processes , while parton shower algorithms produce initial and final state parton radiation .
multiple soft and hard parton interactions are also simulated . for the hadronisation step the lund string model @xcite , as implemented in jetset 7.4@xcite , was utilized .
the yield of baryons produced during the hadronisation in the lund string model is proportional to the amount of diquark states created in the color field .
thus the most important parameter of this model related to the production of protons is the diquark suppression factor which determines the probability of producing a ( @xmath19 ) pair relative to the ( @xmath20 ) pair in the color field .
results from @xmath0 collisions at lep favor a value of approximately 0.1@xcite for this parameter . the pythia mc model with the @xmath0-derived value of 0.1 overshoots the data ( fig.[fig : data_pythia ] ) ,
although it provides a fair description of the shape of the spectra . with the parameter value 0.05 both the shape and yield
are well described .
this could indicate the absence of universality of baryon production within this model approach .
more detailed studies of the production of different hadrons over a wider range a phase space than covered in this work are required in order to further understand this phenomenon .
the present measurement of the inclusive proton cross - section adds a new point to the series of measurements on inclusive hadron production in photoproduction at hera .
previous data on the production of long - lived low mass hadrons @xmath21@xcite , @xmath22 , @xmath23 @xcite and the charmed mesons @xmath24 and @xmath25@xcite are available . in the following
we compare these cross - sections with the present new measurement of protons .
for this comparison we convert the published data points to differential cross - sections @xmath18 , correcting for the flux of the photons@xcite where ep cross - sections are given .
the pion data are recalculated from the measured charged particle spectra taking into account a 17@xmath26 admixture of charged kaons and protons to these spectra . in the comparison for different hadrons
the cross - sections are given for one isospin projection and are corrected for the particle spin .
in fig.[fig : pt_mt_dependence]a these cross - sections are plotted as a function of the transverse momentum @xmath4 .
the individual particle cross - section clearly depends strongly on the specific hadron mass .
it was shown in lepton - hadron@xcite and hadron - hadron@xcite collisions that the @xmath4-dependence of the hadron cross - sections is well described by a power - law behavior @xmath27 . in @xmath0 collisions ,
on the other hand , the production rates of hadrons are determined only by the particle spins , isospins and masses@xcite .
we replot in fig.[fig : pt_mt_dependence]b our cross - section data as a function of the modified transverse mass @xmath1 ( @xmath28 , where @xmath3 is the particle mass ) .
the data show that the cross - sections for different hadrons produced in the photoproduction regime at hera have similar behaviour as a function of this variable . in summary
, we have presented a measurement of the production of protons in photoproduction at hera , performed in the central rapidity region and at low proton momentum .
the pythia mc model only describes the normalisation of the cross section if the model parameter ( diquark suppression factor ) is a factor of 2 smaller than the value obtained from @xmath0 collisions .
the cross - sections for different hadrons produced in photoproduction at 200 gev are shown to have similar behaviour as a function of @xmath29 .
i would like to thank grazyna nowak and the whole local organizing committee for the kind hospitality and for the high level of organization of this very fruitful workshop .
i wish to acknowledge jan olsson and paul richard newman for the careful reading of the manuscript and for useful discussions .
b. andersson , phys .
* 97 * ( 1983 ) 31 .
t. sjstrand , comput .
commun .
* 82 * ( 1994 ) 74 .
t. sjstrand , comput .
commun .
* 135 * ( 2001 ) 238 [ arxiv : hep - ph/0010017 ] .
p. abreu , z. phys .
c * 73 * ( 1996 ) 11 .
h1 collaboration , paper submitted to eps , budapest , july 2001 .
( abstract 815 ) .
m. glck , e. reya and a. vogt , z. phys .
c * 53 * ( 1992 ) 127 .
i. abt _ et al . _
[ h1 collaboration ] , phys .
b * 328 * ( 1994 ) 176 . c. adloff
_ et al . _
[ h1 collaboration ] , eur .
j. c * 10 * ( 1999 ) 363 [ arxiv : hep - ex/9810020 ] .
m. derrick _ et al .
_ [ zeus collaboration . ] , z. phys .
c * 67 * ( 1995 ) 227 [ arxiv : hep - ex/9503014 ] . c. adloff _ et al . _
[ h1 collaboration ] , z. phys .
c * 76 * ( 1997 ) 213 [ arxiv : hep - ex/9705018 ] .
j. breitweg _ et al .
_ [ zeus collaboration ] , eur .
j. c * 6 * ( 1999 ) 67 [ arxiv : hep - ex/9807008 ] .
j. breitweg _ et al . _
[ zeus collaboration ] , phys .
b * 481 * ( 2000 ) 213 [ arxiv : hep - ex/0003018 ] . c. adloff
_ et al . _
[ h1 collaboration ] , nucl .
b * 545 * ( 1999 ) 21 [ arxiv : hep - ex/9812023 ] . c. f. von weizsacker , z. phys .
* 88 * ( 1934 ) 612 .
e. j. williams , phys .
rev . * 45 * ( 1934 ) 729 .
s. frixione , m. l. mangano , p. nason and g. ridolfi , phys . lett .
b * 319 * ( 1993 ) 339 [ arxiv : hep - ph/9310350 ] .
g. bocquet , phys .
b * 366 * ( 1996 ) 434 .
p. v. chliapnikov and v. a. uvarov , phys .
b * 345 * ( 1995 ) 313 .
|
a measurement of the inclusive cross - section for the photoproduction of protons in the central fragmentation region at hera is presented .
the measured cross - section is compared with the prediction of the pythia mc model .
the obtained value of the diquark suppression factor in the lund string model is different from that obtained in @xmath0 data .
the dependence of the cross - section as a function of the transverse mass @xmath1 is shown to be similar for different hadrons produced in photoproduction at hera .
|
rotation powered pulsars are known to be sources of magnetized relativistic winds whose interaction with the ambient medium produces spectacular pulsar wind nebulae ( pwne ) , observable from the radio to tev @xmath8-rays .
when a pulsar is moving through the interstellar medium ( ism ) with a speed exceeding the ism sound speed , a cometary bow shock is expected to form . in the hydrodynamical approximation ( see , e.g. , ) the pulsar wind is confined to the interior of the contact discontinuity surface , which separates the shocked ism from the shocked pulsar wind , while the shocked ism is confined between the forward shock and contact discontinuity surfaces . the highly relativistic pulsar wind bulk flow experiences a termination shock , where the flow speed drops below the sound speed in the outflow . in the idealized case of initially isotropic wind
the termination shock has a bullet shape ( see , e.g. , figure 9 in @xcite , and figures 1 and 2 in ) . in reality
, pulsar winds can be highly anisotropic , with a polar component along the pulsar spin axis and an equatorial component , as demonstrated by x - ray observations of pwne created by young pulsars ( see @xcite for a review ) .
therefore , the shape of the shocks and the nebula appearance may depend on the angle between the velocity vector and the spin axis of the pulsar @xcite .
in addition , inhomogeneities in the ambient medium @xcite and ism entrainment @xcite can affect the nebula shape and properties .
the ism matter is compressed and heated while passing through the forward shock , which can lead to excitation of ism atoms followed by radiative de - excitation in the shocked ism .
unless the local ism ahead of the moving pulsar is strongly ionized , one can expect strong emission in lyman and balmer lines caused by collisional excitation of most abundant hydrogen atoms by fast ions and electrons in the shocked gas and by electron transfer from neutral hydrogen atoms to protons . since emission in lyman lines
is strongly absorbed in the ism and the earth atmosphere , bow shocks can be most easily detected in h@xmath0 , the strongest of the balmer lines .
to date the bow - shaped h@xmath0 nebulae have been detected around 9 pulsars ( see table 1 in @xcite ; br14 hereafter ) .
in addition , about two dozen pulsars exhibit elongated or tail - like pwn morphologies in x - rays and radio @xcite , commonly attributed to synchrotron radiation from shocked pulsar winds . however , bow shocks , expected for such pulsars , were detected in h@xmath0 from only a few of them ( psrs b1957 + 20 , j2124@xmath13358 and , possibly , b1951 + 32 ) .
a possible explanation for the scarcity of such `` hybrid '' bow shocks is a high degree of ionization of the ambient medium , which might be caused , in some cases , by pre - ionization of the ism by radiation from the pulsar and/or the x - ray pwn .
one of the first h@xmath0 pulsar bow shocks was detected by @xcite wavelength , @xmath9 .
] ahead of the millisecond ( recycled ) pulsar j04374715 ( j0437 hereafter ) .
j0437 is a @xmath10ms pulsar with spindown energy loss rate @xmath11ergs@xmath4 ( corrected for shklovskii effect ; @xmath12 is the neutron star moment of inertia in units of @xmath13 g @xmath14 ) .
the pulsar is in a wide binary system ( @xmath15 d , companion separation @xmath16 cm ) with a nearly circular orbit inclined by an angle of @xmath17 to the sky plane .
the binary companion is a cool ( @xmath18 k ) white dwarf ( wd ) ( see @xcite and references therein ) . the pulsar and wd masses are @xmath19 and @xmath20 @xcite .
j0437 is the closest known pulsar , @xmath21pc , with an accurately measured proper motion , @xmath22masyr@xmath4 , @xmath23masyr@xmath4 , corresponding to the transverse velocity @xmath24kms@xmath4 @xcite .
@xcite reported the results of _ hubble space telescope _ ( _ hst _ ) observations of the pulsar in far - uv ( fuv ) and near - uv ( nuv ) , and analyzed the pulsar plus wd spectrum from mid - infrared to x - rays and @xmath8-rays .
a sharp image of the j0437 h@xmath0 bow shock was obtained by andrew fruchter in 1995 with the ctio 0.9 m telescope ( a. fruchter , priv . comm . ) .
the image clearly shows a symmetric bow shock structure with the standoff distance of about @xmath2 ( @xmath25 cm ) between the j0437 binary and the leading edge ( apex ) of the bow shock .
@xcite observed the h@xmath0 bow shock with the 2.3 m anu telescope .
they measured the h@xmath0 photon flux , @xmath26@xmath27s@xmath4 , and estimated the pre - shock number density , @xmath28@xmath29 .
br14 observed the j0437 h@xmath0 bow shock with the optical imager on the ctio 4.2 m soar telescope .
they measured a factor of 34 higher flux , @xmath30@xmath27s@xmath4 from an apex zone of the bow shock .
the flux difference can be partly ( but likely not entirely ) due to the different bow shock areas used in the analysis .
@xcite searched for an x - ray pwn around j0437 in the _ chandra x - ray observatory _
( _ chandra _ ) images but did not find it .
in this paper we report on our discovery of a fuv bow shock ahead of j0437 .
we present an analysis of fuv broadband photometry and prism spectroscopy of the bow shock and place constraints on the shape of its spectrum based on our dedicated _ hst _ program and archival _ hst _ data .
we also use archival _ chandra _ data to perform a new search for extended x - ray emission around j0437 .
fuv observations of the j0437 field were carried out in two _ hst _ programs . first program ( go 10568 )
was devoted to studying the spectrum of the pulsar and wd emission .
it involved observations with the high resolution camera ( hrc ) , wide field camera ( wfc ) and solar blind channel ( sbc ) of the advanced camera for surveys ( acs ) with the f140lp ( fuv ) , f300w ( @xmath31 u ) , and f555w ( @xmath31 v ) filters , and pr130l ( fuv ) and pr200l ( nuv ) prisms . the second program ( go 12917 ) was dedicated to bow shock imaging and photometry in fuv ( f115lp , f125lp , f140lp ) and h@xmath0 ( f658n ) with the acs sbc and wfc , respectively .
the images from both campaigns were drizzled with the ` astrodrizzle ` package from pyraf . additionally , we used two archival images , f555w ( @xmath31 v ) and f814w ( @xmath31 i ) , obtained with the wide filed planetary camera 2 ( wfpc2 ; go 6642 , pi foster ) .
we downloaded the wfpc2 data from the _ _ hubble legacy archive _ _ ( _ hla _ ) . for each filter
, the _ hla _ combines the individual flat - fielded exposures using the pyraf ` multidrizzle ` task , which produces co - aligned , geometrically corrected images .
more details about individual observations can be found in table [ tbl - hst ] .
llllr 1996 - 05 - 19 & wfpc2 & foster & f555w & 560 + 1996 - 05 - 19 & wfpc2 & foster & f814w & 560 + 2005 - 12 - 10 & acs / sbc & kargaltsev & f140lp & 2550 + 2005 - 12 - 10 & acs / sbc & kargaltsev & pr130l & 4800 + 2006 - 04 - 30 & acs / hrc & kargaltsev & pr200l & 2637 + 2006 - 04 - 30 & acs / hrc & kargaltsev & pr200l & 2396 + 2006 - 04 - 30 & acs / hrc & kargaltsev & pr200l & 2580 + 2006 - 04 - 30 & acs / hrc & kargaltsev & f300w & 360 + 2006 - 04 - 30 & acs / hrc & kargaltsev & f555w & 200 + 2006 - 05 - 01 & acs / sbc & kargaltsev & f140lp & 1700 + 2006 - 05 - 01 & acs / sbc & kargaltsev & f140lp & 2550 + 2006 - 05 - 01 & acs / sbc & kargaltsev & pr130l & 4800 + 2006 - 05 - 01 & acs / sbc & kargaltsev & pr130l & 3200 + 2006 - 05 - 02 & acs / sbc & kargaltsev & pr130l & 1600 + 2006 - 05 - 02 & acs / sbc & kargaltsev & f140lp & 2550 + 2006 - 05 - 02 & acs / sbc & kargaltsev & pr130l & 4800 + 2006 - 05 - 02 & acs / sbc & kargaltsev & f140lp & 850 + 2013 - 01 - 20 & acs / wfc & durant & f658n & 1997 + 2013 - 08 - 01 & acs / sbc & durant & f115lp & 4800 + 2013 - 08 - 01 & acs / sbc & durant & f140lp & 2198 + 2013 - 08 - 02 & acs / sbc & durant & f125lp & 4800 + 2013 - 08 - 02 & acs / sbc & durant & f140lp & 2198 + the sbc detector , a multi - anode microchannel array ( mama ) type device sensitive in the 1100 to 2000 range , was the primary instrument in both 2006 and 2013 fuv imaging observations .
the details of the spectroscopic 2006 observations and the spectral analysis of the pulsar itself are given by @xcite . to calibrate the wavelength of dispersed light , direct images ( without the dispersing element ) with the f140lp filter
have been taken in addition to slitless spectroscopic exposures with the pr130l prism .
the f140lp ( @xmath322000 ) filter cuts off all prominent background geocoronal lines .
the observations with the pr130l ( @xmath332000 ) , which provides a modest spectral resolution , were taken in the earth s shadow to reduce the geocoronal background .
the 2006 observations also include hrc nuv imaging and spectroscopy with f300w and pr200l , respectively ( see @xcite for details ) . in the 2006 campaign , for the majority of the exposures , the telescope was oriented to keep the pulsar away from the region of enhanced sbc thermal glow .
as a result , the bow shock ( which , at that time , was not expected to be seen in fuv ) was outside the sbc field of view , except for a small part imaged in the detector corner . in one of the visits during the 2006 campaign the observatory experienced loss of guiding stars and simultaneous increase of the detector background .
although most of the data from that visit were lost , the very first f140lp image of that visit ( see figure [ panel ] , middle left panel ) happened to be good , with the telescope being oriented in such a way that most of the bow shock was imaged in the region of low detector background . in the 2013 campaign the bow shock was intentionally placed outside the region of brightest sbc glow ( see figure [ panel ] , middle right and bottom left panels ) .
the observations with the f115lp ( 11502000 ) and f125lp ( 12502000 ) filters were taken in the earth shadow to reduce the geocoronal background , while the f140lp filter was used in the non - shadow parts of the respective orbits .
a shorter wfc observation used the h@xmath34 filter ( f658n ) . unfortunately , because of a very bright geocoronal ly-@xmath34 background ( despite being in the earth shadow ) , we had to discard the f115lp image from further analysis . for the j0437 pwn search , we used archival observations with the acis detector , which is more sensitive than the hrc detector and provides spectral information .
there are 9 acis observations of j0437 in the _ chandra _ archive . in the earliest observation ( obsid 741 , pi pavlov ,
the results are reported by @xcite ) the target was offset by 39 from the optical axis to mitigate the pile - up effect , which considerably distorted the point spread function ( psf ) and strongly complicated the search for a compact pwn .
the other 8 observations were carried out by the _
acis instrument team for calibration purposes .
four of those observations were taken at roll angles too close to @xmath35 or @xmath36 , at which the so - called acis trailed image ( also known as `` readout streak '' ) of the bright pulsar coincides with the pulsar s sky trajectory and can mimic the expected pwn tail behind the pulsar and/or contribute to the pwn head image . among the remaining four , one observation had a very large ,
nonstandard frame time of 6 s ( hence the pulsar image was strongly distorted by pileup ) and another one had a short exposure of 7.8 ks . therefore , we selected two of the calibration observations , obsids 6154 and 6155 ( 22.8 and 24.8 ks exposures , @xmath37 and @xmath38 roll angles , respectively ) for the pwn search .
the data were taken on 2005 february 18 ( obsid 6154 ) and 2005 march 03 ( obsid 6155 ) in very faint telemetry format . in both observations the target
was placed @xmath39 from the optical axis on acis - s3 chip , other acis chips were not activated . in obsid 6154 a 101 pixels ( 497 ) subarray was used ( frame time 0.4 s ) , while the whole chip was read out in obsid 6155 ( frame time 3 s ) .
we processed the data using the _ chandra _ interactive analysis of observations ( ciao ) software ( ver .
4.6 ) and _ cxo _ calibration data base ( caldb ver .
the fuv f140lp and f125lp images in figure [ panel ] clearly reveal a bent extended structure ahead of the pulsar .
its apex distance , @xmath40 ( @xmath25 cm ) , and shape are very similar to those of the h@xmath0 bow shock .
to compare the fuv and h@xmath0 structures , we plotted h@xmath0 surface brightness contours , within which the brighter part of the h@xmath0 bow shock is confined in the middle - right panel of figure [ panel ] ( the selected contour represents the level for which the h@xmath0 bow shock is not broken down into individual parts ) . since the h@xmath0 bow shock is only marginally detected in our 2013 acs / wfc f658n image ( figure [ halpha ] ) , we used an archival soar image obtained on 2012 march 21 ( see top left panel of figure [ panel ] ) to produce the h@xmath0 contours . because the fuv structure very closely follows the h@xmath0 structure , we conclude that it is an fuv counterpart of the h@xmath0 bow shock .
the apparent thickness of the fuv bow shock at its apex is about @xmath41 , somewhat smaller than that of the h@xmath0 bow shock ( but it depends on image depth ) .
the fuv shock is also seen in the ( dispersed ) sbc pr130l image of 2006 ( see bottom right panel of figure [ panel ] ) , but no shock is detected in the hrc pr200l image , nor in the short hrc observations taken with the f300w and f555w filters in the same 2006 campaign . ) applied .
the fuv `` blob '' ( see figure [ panel ] , mid - left panel ) is within the @xmath42 white circle .
the pulsar position is shown by the cross . ]
in addition to the fuv bow shock , the f140lp image from the 2006 may 1 observation reveals a `` blob '' of @xmath31@xmath43 size , overlapping with the bow shock limb ( mid - left panel of figure [ panel ] ) .
the blob , centered at r.a.@xmath44 , decl.@xmath45 , is clearly resolved but appears to be amorphous , with no clear structure .
it is out of the sbc fields of view in the fuv images of 2013 , but it is marginally detected in the 2013 wfc h@xmath0 image ( figure [ halpha ] ) .
the ratios of the surface brightness of the blob to that of the bow shock are @xmath312.5 and @xmath311.5 in the f140lp and h@xmath0 images , respectively .
interestingly , while the bow shock was not detected in the 2006 hrc / pr200l data , the brighter uv blob was .
this proves that the blob is not an artifact of the sbc detector .
furthermore , the blob is not seen in the previous high - resolution h@xmath0 images ( e.g. , in the image by br14 ) , which suggest its variability .
some of the archival optical / uv images ( wfpc2 f555w and f814w , acs / hrc f300w and f555w ; see table 1 for observation dates ) show a faint point source near the eastern boundary of the f140lp blob but no trace of extended emission .
it remains unclear whether the point source is physically connected with the blob or it is just a projection effect .
the inspection of the combined _ chandra _ acis - s image ( shown in figure [ panel ] , top - right panel ) reveals a faint diffuse x - ray emission extending up to about @xmath6 south - east of j0437 ( i.e. , in the direction of pulsar s proper motion ) . for a conservative estimate of the detection significance
, we extracted the source and background counts from the smaller and larger sectors of the @xmath46 annulus around the pulsar position , shown by green lines in the top right panel of figure [ panel ] .
we found @xmath47 background - subtracted counts ( 110 and 230 in the source and background areas , respectively ) in the area of 13.1 arcsec@xmath48 of the smaller sector , i.e. , the extended emission ( presumably an x - ray pwn ) is detected with a @xmath49 significance .
@xcite could not detect such a small , faint nebula in their 25 ks acis - s observation of 2000 may 29 ( obsid 741 ) because of strong psf degradation at the @xmath50 off - axis target position ( see section 2.2 ) , while the 20 ks hrc - i observation ( 2000 february 13 ; obsid 742 ) , also analyzed by these authors , was not deep enough for pwn detection .
we fit the spectrum of the detected extended emission with the absorbed power - law ( pl ) model with fixed @xmath51 @xmath27 ( this value was determined by @xcite from fits of the x - ray pulsar spectrum ) .
the fit yields the photon index @xmath52 and absorbed flux @xmath53ergs@xmath4@xmath27 , which corresponds to luminosity @xmath54ergs@xmath4 ( at the distance of j0437 ) .
we use two independent measurements to constrain the fuv spectrum for the j0437 bow shock : ( 1 ) the f140lp and f125lp imaging photometry obtained in the 2013 program ; and ( 2 ) the pr130l slitless spectroscopy from the 2006 program . for each of the two measurements we test two spectral models .
first model is a simple pl continuum
. it would be directly applicable to synchrotron emission from relativistic electrons with a pl spectral energy distribution , and it could crudely characterize the slope of a more complicated spectrum .
second model is an fuv spectrum ( spectral lines plus continuum ) emitted from ism matter compressed and heated in the forward shock region . limited by the data quality
, we have to assume that the spectrum does not vary throughout the observed bow shock region .
we selected the same source and background extraction regions for the f125lp and f140lp images ( shown in figure [ panel ] , middle - right and bottom - left panels , respectively ) .
one of the two background regions was selected outside the bow shock to avoid the bow shock emission inside the bow . however , the background from this region ( hereafter , 1-sided background ) may be lower than the background at the bow shock location because of the nonuniform thermal glow contribution .
therefore , we also selected another background region inside the bow shock and used the `` inside + outside '' region for an alternative background estimate ( hereafter , 2-sided background ) , which may somewhat exceed the true background because it includes a contribution from the shock emission .
cccccc f125lp & 32.2 & 29.4 & @xmath55 & @xmath56 + f125lp & 32.2 & 61.8 & @xmath57 & @xmath58 + f140lp & 32.2 & 29.4 & @xmath59 & @xmath60 + f140lp & 32.2 & 61.8 & @xmath61 & @xmath62 the net source count rate for each background choice is calculated as @xmath63 , where @xmath64 is the total count rate in the source region of area @xmath65 , and @xmath66 is the background count rate in the region of area @xmath67 . correspondingly , the source count rate uncertainty is @xmath68^{1/2}t_{\rm exp}^{-1/2}$ ] , where @xmath69 is the exposure time .
the f125lp and f140lp count rates and their uncertainties for two background choices are given in table [ tbl - phot ] .
the source count rates estimated with the 1-sided background exceed those estimated with the 2-sided background by a factor of about 1.45 , but the ratio of the source count rates , @xmath70 and @xmath71 , is virtually the same for the two background choices . since the throughputs of the two filters coincide at @xmath72 ( see figure [ throughputs ] ) , the fact that @xmath73 means that both the 12501400 and 14002000 wavelength bands are contributing to the fuv bow shock radiation ,
i.e. , the fuv emission is not concentrated in one spectral line . below we will use the measured count rates to constrain the bow - shock fuv spectrum , with the aid of the pyraf package pysynphot .
( @xmath74 ) for three different values of reddening , @xmath75 , 0.05 and 0.1 .
the horizontal lines mark the measured value of the ratio of the net count rates and its uncertainties ( for the 2-sided background ; see section 3.3.1 for details ) .
the vertical lines show the range of allowed slopes for @xmath76 . ] for the absorbed _ pl model _ , we simulate the count rates @xmath77 and @xmath78 as functions of slope @xmath0 of the pl spectrum , @xmath79 , for three reddening values , @xmath80 , 0.05 , and 0.1 , in a plausible reddening range .
then we compare the simulated count rate ratio ( which does not depend on pl normalization ) with the measured one , which allows us to determine the best - fit slope and its uncertainties for a given reddening ( see figure [ slope ] ) .
we found that the best - fit @xmath81 slope is nearly flat , @xmath82 , corresponding to flux densities , @xmath83 , of @xmath84 and @xmath85ergs@xmath4@xmath27@xmath4 ( in the f125lp and f140lp filters , respectively ) , but the uncertainties of the spectral slope are large .
for instance , @xmath86 and @xmath87 for the 2-sided and 1-sided background , respectively , at the most plausible @xmath88 @xcite . based on this model ,
we calculate the absorbed model fluxes in the f125lp and f140lp wavelength ranges ( which only weakly depend on the assumed reddening ) : @xmath89@xmath90ergs@xmath4@xmath27 and @xmath91@xmath92ergs@xmath4@xmath27 , for the 2-sided background .
they correspond to luminosities @xmath93 and @xmath94 erg s@xmath4 , respectively .
we should note that the quoted flux uncertainties are statistical ones ; systematic uncertainties are substantially larger .
in particular , all the fluxes and luminosities would be a factor of 1.45 larger if we use the 1-sided background .
we also stress that these fluxes and luminosities are for the image area of 32.2 arcsec@xmath48 , which is likely a small fraction of the entire fuv bow shock area , so the full fluxes / luminosities can be higher . the estimated fuv fluxes and luminosities are larger than the h@xmath0 ones from the same region of the bow shock . a crude estimate from our shallow _
hst _ h@xmath0 image ( figure [ halpha ] ) gives the photon flux @xmath95 @xmath27s@xmath4 in the fuv extraction region ( the value strongly depends on background region choice ) .
alternatively , the h@xmath0 flux can be estimated by scaling the br14 flux , @xmath96@xmath27s@xmath4 , measured in the larger `` apex zone '' , using the ratio ( @xmath97 of source counts in the apex zone to that in the fuv extraction region projected onto the soar image .
this gives @xmath98 @xmath27s@xmath4 , slightly higher than ( but within the uncertainties of ) the flux estimated from the _ hst _ h@xmath0 image .
the latter photon flux value , which we consider more reliable , corresponds to the energy flux @xmath99erg@xmath27s@xmath4 , a factor of 8 ( 11 ) lower than the absorbed ( unabsorbed ) flux in the 12502000 range .
we can also use the pl model to estimate the flux of the blob detected in the f140lp image . assuming a flat @xmath81 spectrum
, we obtain @xmath100 erg @xmath27 s@xmath4 ( observed flux ) , which corresponds to the luminosity @xmath101 erg s@xmath4 .
it is larger than the blob s flux / luminosity in the f658n h@xmath0 filter by a factor of about 15 ( this factor depends on choice of the background region ) .
the other model we use to compare with the observed fuv bow shock properties is the _ emission spectrum from shocked ism matter _ , compressed and heated while passing through the shock
. numerical models of collisionless shocks and their emission have been developed by many authors ( see , e.g. , the review by @xcite for references ) .
however , as we are not aware of _ bow shock _ fuv emission models in the regime appropriate for j0437 , we have to use models for a _ plane shock_. for model calculations , we use the shells code described in section 5 of the @xcite review .
this code allows one to calculate radiation spectra ( continuum and lines ) of one - dimensional shocked flows in the far - infrared@xmath1fuv range , simultaneously with the upstream and downstream flow structure for a collisionless shock moving with a @xmath102400 kms@xmath4 velocity through the ism of an @xmath10325@xmath29 upstream number density ( @xmath104 , where @xmath105 is the mass density , and @xmath106 is the mean atomic mass of ions ) .
the model accounts for radiative cooling and heating , non - equilibrium photoionization and recombination , various emission mechanism ( bound - bound , free - bound , free - free , and two - photon transitions ) , radiation transfer , and magnetic fields . of the plane shock with @xmath107kms@xmath4 , @xmath108 @xmath29 .
some of the lines are labeled . ] for the comparison with our observations , we computed a set of shells models for different combinations of the most important parameters @xmath109 and @xmath110 , assuming a solar chemical composition ( @xmath111 ) and an upstream ( ism ) magnetic field of @xmath112 perpendicular to the shock front .
an example of the fuv part of the spectrum , for @xmath107kms@xmath4 , @xmath108@xmath29 , is shown in figure [ shock - model ] .
for each of these combinations , we calculated the expected count rate ratios @xmath113 with the aid of pysynphot package , for plausible reddening @xmath114 , and compared them to the observed ratio , similarly to the absorbed pl model .
the upper panel of figure [ shells_fluxes ] shows that at @xmath88 the model count rate ratio becomes smaller than the observed range , @xmath1151.72 , for @xmath116118kms@xmath4 , depending on preshock density in a plausible range @xmath1170.3 @xmath29 ( see section 4 ) . for @xmath88 ( top ) , @xmath89-@xmath118 ratio ( middle ; the fluxes are corrected for extinction ) and the model flux emitted from unit area of plane shock , @xmath119-@xmath120 in units of @xmath121erg@xmath27s@xmath4 , as functions of shock speed @xmath109 for preshock densities 0.1 , 0.2 , and 0.3 @xmath29 , computed with the shells code .
the dotted horizontal lines in the top and middle panels show the observational bounds on the corresponding quantities . ] ) are shown together with spectral models .
the black dots show the f125lp and f658n spectral fluxes ( the narrower f140lp band is contained within the f125lp one and hence it is not shown for clarity ) .
the arrows show the @xmath122 upper limits for the hrc f330w and f555w observations calculated assuming a flat @xmath123 spectrum .. the horizontal black lines represent the filter widths at the half - maximum .
absorbed flux densities for the pl ( @xmath124 ) and shocked plasma model ( for @xmath107kms@xmath4 and @xmath108@xmath29 ) are plotted as blue and red lines , respectively .
the dot - dashed blue lines represent the @xmath125 uncertainty in the pl slope . ]
another model parameter to compare with observations is the ratio of the fuv and h@xmath0 fluxes .
the middle panel of figure [ shells_fluxes ] shows that the ratio of the model energy fluxes @xmath89@xmath118 increases with decreasing @xmath109 and becomes larger than the observed flux ratio , @xmath12615 ( corrected for extinction @xmath88 ) , at @xmath127106kms@xmath4 , depending on @xmath110 .
thus , comparing the flux ratios in 3 filters , we constrain the shock speed : @xmath128kms@xmath4 , at @xmath129@xmath29 ( @xmath130kms@xmath4 at @xmath108 @xmath29 ) .
we note that these constraints are derived for the effective _ plane shock _ velocity , which can be lower than the bow - shock ( pulsar ) velocity because the velocity component perpendicular to the bow shock is lower than the pulsar velocity except for the bow - shock apex .
although the bow shock was not detected in the f300w and f555w images , it is interesting to compare upper limits at these wavelengths with the spectral models . due to the apparent non - uniformities in the background , clearly exceeding the statistical fluctuations
, we chose to measure the upper limits by sampling the background counts from five different regions in the vicinity of the bow shock .
the standard deviation of these measurements was used as a @xmath125 upper limit .
the corresponding @xmath122 upper limits of the flux densities are @xmath131 and @xmath132ergs@xmath4@xmath27@xmath4 in the f330w and f555w images , respectively .
figure [ limits ] shows the best - fit pl and shocked plasma models together with the @xmath122 limits and the fuv and h@xmath0 flux density measurements .
we see that the optical upper limits are too high to distinguish between the pl and shocked plasma models , but the f555w limit favors smaller values of the pl slope , @xmath133 in the broad range of @xmath0 obtained from the fuv data for the pl model ( section 3.3.1 ) .
a small part of the fuv bow shock was also detected with the sbc / pr130l prism ( see figure 1 , bottom - right panel ) .
even though the sbc prism is designed for spectroscopy of point sources , we attempted to analyze the prism data to check if any additional information about the bow shock spectrum can be obtained .
since the bow shock is an extended feature , we had to stack the data in the cross - dispersion ( vertical ) direction before we could fit the spectrum .
we shifted the image data ( using 10-pixel - wide stripes ; see figure [ prism ] ) following the curvature of the bow shock , and summed the counts along the vertical direction , to make a one - dimensional dispersed spectrum .
we halted the extraction where the dispersed image nears the detector edge . because the bow shock is resolved in the direct image along the dispersion ( horizontal ) direction , the stacked one - dimensional spectra are relatively broad ( significantly broader than the spectrum of a point source would be ) .
the observed dispersed spectrum of the bow shock is thus a convolution of the spatial profile of the bow along the dispersion direction with the known throughput and dispersion of the prism . to fit the observed spectrum with a model , we follow a forward - fitting method , i.e. , calculate the dispersed pr130l spectrum ( in detector coordinates ) for a point source , convolve it with the bow shock profile measured from the direct image , and compare with the observed prism profile .
while stacking multiple spectra ( figure [ prism ] ) increases the signal - to - noise ratio ( @xmath134 ) , it also complicates the background subtraction . unlike a point source extraction , where the background can be estimated in the immediate vicinity of the source ,
the background for each spectral extraction region ( stripes in figure [ prism ] ) is slightly different .
moreover , the thermal glow also changes across different parts of the detector .
in addition , the profile from the direct image ( which is used for the model spectrum convolution ) comes from different detector coordinates than the profile from the dispersed prism image , where the background and thermal glow are different .
finally , any changes in the bow shock profile or spectrum along the bow shock are not accounted for in our approach .
this complicates the uncertainty estimation of our final results , and these systematic errors significantly exceed formal statistical errors obtained during the model fitting .
we explored various background subtraction procedures , including a constant background and a polynomial ( with background levels rising across the dispersion direction ) .
since in the interior of the bow shock we can not distinguish between the background due to the sbc thermal glow and the real emission from the bow shock itself , we decided to adopt a flat background determined from a region outside of the bow shock .
although the emission is present throughout the entire sensitivity range of pr130l ( 1100@xmath11800 ) , the dispersed spectrum overlaps with the thermal glow . therefore , we restrict our fitting to the 100@xmath1300 pixel range ( figure [ model ] ) , corresponding to the stripes shown on figure [ prism ] .
we find that under these assumptions the data can be satisfactory described by pl models with @xmath135 ( figure [ model ] , left panel ) , which is consistent with the @xmath0 values obtained from the observations with broad filters .
again , as an alternative to the simple pl model , we also used a model of emission from a collisionless plane shock @xcite . through multiple trials , we established that the models with strong individual lines produce too narrow profiles that can not fit the observed one .
we concluded that models with weak lines , such as the model with @xmath107kms@xmath4 , @xmath108@xmath29 ( figure [ shock - model ] ) , are preferred by the fits ( figure [ model ] , right panel ) .
the results are not sensitive to the extinction in the @[email protected] range .
thus , the poor statistics and smearing due to the extended nature of the bow shock do not allow us to discriminate between the pl and collisionless shock emission models in the prism spectra , but at least models in which emission is dominated by a few individual lines can be excluded .
sensitive slit spectroscopy or integral field spectroscopy would be needed to reach further progress .
our fuv observations of j0437 have revealed a structure whose shape matches closely ( within 05 for the leading edge ) the brighter part of the h@xmath0 bow shock ( see figure [ panel ] ) .
this strongly suggests that the fuv emission is due to the interaction of the pulsar wind with the ambient ism and most likely comes from the shocked ism matter .
the shocked ism matter is confined between the forward shock ( fs ) and the contact discontinuity ( cd ) , which separates the shocked ism from the shocked pulsar wind . for an isotropic pulsar wind ,
the distance @xmath137 from the pulsar to the fs apex is a factor of @xmath1381.4 larger than the distance @xmath139 to the cd apex , which can be estimated as @xmath140^{1/2},\ ] ] where @xmath141 is the ambient mass density , @xmath142 the barion number density , @xmath143 the pulsar velocity , depends on the orbital phase .
this should lead to periodic variations of the bow shock position with the period @xmath15d .
the amplitude of these variations depends on the angle between the velocity @xmath144 of the binary system and the orbital plane . since this angle is unknown , and the pulsar s orbital velocity , @xmath145
km s@xmath4 , is small compared to the velocity of the binary , @xmath146 , we neglect this effect in our estimates . ] , and @xmath147 is the angle between the pulsar velocity and the line of sight ( e.g. , ) . for the j0437 bow shock , @xmath148 cm ( @xmath2 in the image ) corresponds to @xmath149
cm ( @xmath150@xmath151 in the image ) , in agreement with the observed bow shock thickness near apex . using the approximation by @xcite , @xmath152
, we estimate the spin - down power as @xmath153ergs@xmath4 , and obtain an estimate for the ambient baryon density from equation ( 1 ) : @xmath154@xmath29 , in agreement with the @xmath155@xmath29 estimated from the h@xmath0 bow shock flux ( br14 ) . from the analysis of the h@xmath0
bow shock emission , br14 derived a high neutral h fraction , @xmath156 , in the preshock ambient medium , which implies a low degree of photoionization by uv and soft x - ray photons emitted from the pulsar , its pwn , and the shocked ism .
if the pulsar were not moving , it would ionize the ambient medium up to distances much larger than @xmath137 ( within its strmgren sphere ) .
however , as shown by @xcite ( see also and @xcite ) , the characteristic size @xmath157 of the photoionized region ahead of a _ moving _ source of ionizing radiation can be estimated as @xmath158 where @xmath159 is the source ( pulsar ) velocity , @xmath160 ev is the hydrogen ionization potential , @xmath161 is the photon spectrum of ionizing radiation , @xmath162 is the photoionization cross section , @xmath163 is the fine - structure constant , @xmath164 is the bohr radius , and @xmath165 is the gaunt factor ( @xmath166 at @xmath167 ) . for j0437 ,
the main source of ionizing radiation is thermal emission from the neutron star surface .
we derive @xmath168 @xmath14 s@xmath4 for the j0437 spectrum , which is well fitted with superposition of three blackbody components @xcite .
this gives @xmath169 cm , two orders of magnitude smaller than @xmath137 .
thus , the preshock degree of ionization is essentially the same as in the unperturbed ism , thanks to the high pulsar speed . from our observations we estimated the fuv
bow shock luminosity @xmath3 erg s@xmath4 , a factor of 10 higher than the h@xmath0 luminosity from the same ( small ) region of the bow shock .
the shape of the fuv spectrum remains poorly constrained because the imaging observations were done with only two very broad filters with overlapping passbands , while the prism observation was not suited for spectroscopy of extended sources .
our analysis has shown , however , that the limited data available are consistent with model spectra of shocked ism .
the ism matter is compressed , heated , and partly ionized at the fs front . at a highly supersonic pulsar velocity
the ions are heated up to @xmath170 k , where @xmath171 is the mean ion mass . behind the fs front
, the ions energy is partly transferred to electrons and the flow speed decreases .
therefore , at some postshock distance the plasma temperature and density become high enough to emit a complex uv - optical spectrum comprised of both continuum and line emission . using the ` shells ` code for plane - parallel shocks
, we found that the observed count rate ratio in two fuv filters and the ratio of the fuv and h@xmath0 fluxes are consistent with the model predictions at reasonable values of the preshock density and shock velocity ( section 3.3.1 ) , and the model spectrum is consistent with the observed prism spectrum at the same parameters ( section 3.3.2 ) .
in addition to comparing the count rate and flux ratios ( see figure [ shells_fluxes ] ) , a comparison of the observed shock fluxes with the model fluxes allows us to estimate the emitting area of the plane shock needed to provide the observed flux .
since the model flux is calculated per unit area of the plane shock , the emitting area can be estimated as @xmath172@xmath14 , where @xmath173 and @xmath174 are the observed and model fluxes in units of @xmath175 and @xmath121erg@xmath27s@xmath4 , respectively .
the model flux in the 12502000 range is shown in the lower panel of figure [ shells_fluxes ] as a function of @xmath109 for three @xmath110 values .
the flux weakly depends on @xmath109 , and it is approximately proportional to the upstream density , @xmath176 .
the observed fuv flux , estimated with pysynphot for the calculated model spectra , weakly depends on the assumed model .
its characteristic value is @xmath177 , and its uncertainty is mainly due to systematic errors . using this estimate
, we obtain @xmath178 @xmath14 .
this area corresponds to the radius @xmath179 cm ( angular radius of @xmath180 for @xmath108@xmath29 ) comparable with the fuv extraction region size perpendicular to the pulsar s proper motion direction .
we should note , however , that the size of the downstream fuv emitting region in the plane shock model , @xmath181 a few @xmath182 cm for the plausible @xmath109 and @xmath110 values , considerably exceeds the thickness of the observed bow shock , @xmath183 cm , and even the stand - off distance of the bow shock apex , @xmath184 cm .
this is not surprising because , contrary to the bow shock , the downstream region of the plane shock is not confined by the pulsar wind ram pressure , but it shows that applicability of plane shock models to bow shocks is limited .
thus , to prove that the observed fuv structure is indeed emission from a shocked ism matter , the observational data should be compared to a realistic radiative bow shock model . in some of our observations we detected an extended `` blob '' of @xmath43 ( 460au at @xmath185pc ) size at the bow shock limb ( section 3.1 ) .
the lack of the blob in the wide - band optical / nuv images implies a spectrum that would be very unusual for a galaxy of any kind .
the positional coincidence with the bow shock limb and the fact that the blob is also marginally detected in the acs / wfc image from 2013 suggest that the blob is likely associated with the bow shock ( e.g. , caused by an ism inhomogeneity or an instability in the fs region ) . according to pwn models for supersonically moving pulsars ( e.g. , )
, the emission from the shocked ism behind the fs should be accompanied by synchrotron emission from the shocked pulsar wind confined between the bullet - like termination shock ( ts ) surface and the cd surface .
the synchrotron pwn is expected to have a cometary shape , with a `` head '' in front of the moving pulsar and a `` tail '' behind the pulsar .
such head - tail pwn morphologies have been observed in x - rays from a number of ordinary and recycled pulsars ( see @xcite for examples ) .
our analysis of archival x - ray observations of j0437 has shown an extended emission up to a distance of about @xmath6 ahead of the pulsar ( a factor of @xmath1811.5 smaller than the @xmath139 estimate ) but no tail behind .
we presume that the detected emission represents a brighter part of the pwn head , possibly enhanced due to compression of the magnetic field , while the fainter emission from the rest of the pwn could be detected in deeper high - resolution x - ray observations .
we should note that such a small head would not be resolved if j0437 were not so nearby .
among other nearby pulsars , geminga ( @xmath186 pc ) and psr b105552 ( @xmath187 pc ) show hints of x - ray emission up to a few arcseconds ahead of the pulsar @xcite .
however , geminga also shows three tails whose origin is unclear , while b105552 is likely moving away from us close to the line of sight , which leads to a different shape of the x - ray pwn image .
the x - ray luminosity of the detected extended emission , @xmath188 ergs@xmath4 , corresponds to the x - ray efficiency @xmath189 .
such an efficiency is consistent with those of pwne of supersonically moving pulsars @xcite , which supports our interpretation of the extended x - ray emission .
although the interpretation of the observed fuv emission as due to a collisionless shock in a partially ionized ism is consistent with the data available , we can not provide a direct proof of this interpretation until the emission spectrum ( including spectral lines ) is measured and compared with bow shock model predictions .
therefore , there still remains a possibility that the fuv emission might be due to synchrotron radiation of relativistic electrons leaked from the x - ray pwn region and trapped at the forward bow shock . to escape from the x - ray pwn and reach the bow shock ,
the electron should have a sufficiently large gyroradius , @xmath190 cm for the x - ray emitting electrons , at ( and beyond ) the pwn s outer boundary ( cf . ) .
a fraction of the escaped electrons can be `` trapped '' in the bow shock vicinity via multiple scattering on quasi - resonant magnetic fluctuations created by the current - driven instability ( see , e.g. , ) .
protons with energies of up to @xmath1911 tev , required for developing of this instability , could be accelerated at the moving bow shock .
this scenario may result in a specific spectrum of the high energy electrons trapped in the bow shock vicinity and producing the fuv synchrotron emission . not only their energies should be above some minimal energy to escape from the x - ray pwn and reach the bow shock , but they should also be limited at the high energy end by the confinement condition at the bow shock .
if this maximal energy is close to that of fuv - emitting electrons , it could explain the lack of x - ray emission at the bow shock .
a sharp drop in the fuv surface brightness toward the x - ray pwn , such as the one seen in our sbc images , could be due to a relatively low magnetic field in that region .
of course , this qualitative scenario remains hypothetical until it is supported by quantitative models . to definitively determine the nature of the discovered fuv bow shock of psrj0437@xmath14715 , its spatially resolved spectrum
should be measured with a higher spectral resolution , in fuv as well as in nuv / optical wavelength ranges .
we expect to see fuv shocks from other nearby pulsars , including those for which no h@xmath0 bow shocks were detected .
observations of such fuv shocks , currently possible only with the _ hst _ , would be very useful for studying the ism properties , the interaction of pulsar winds with the ambient medium , and the mechanisms of particle acceleration .
support for programs go12917 and go10568 was provided by nasa through grants from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
a.m. bykov acknowledges support from rsf grant 16 - 12 - 10225 .
the authors are very grateful to giovanni morlino for the insightful discussion .
a.m.b . and a.m.k
acknowledge joint supercomputer center of russian academy of science where some simulations were performed .
we are thankful to the referee for the useful suggestions that helped to improve this paper .
|
pulsars traveling at supersonic speeds are often accompanied by cometary bow shocks seen in h@xmath0 .
we report on the first detection of a pulsar bow shock in the far - ultraviolet ( fuv ) .
we detected it in fuv images of the nearest millisecond pulsar j0437@xmath14715 obtained with the _ hubble space telescope_. the images reveal a bow - like structure positionally coincident with part of the previously detected h@xmath0 bow shock , with an apex at @xmath2 ahead of the moving pulsar .
its fuv luminosity , @xmath3 erg s@xmath4 , exceeds the h@xmath0 luminosity from the same area by a factor of 10 .
the fuv emission could be produced by the shocked ism matter or , less likely , by relativistic pulsar wind electrons confined by strong magnetic field fluctuations in the bow shock .
in addition , in the fuv images we found a puzzling extended ( @xmath5 in size ) structure overlapping with the limb of the bow shock . if related to the bow shock , it could be produced by an inhomogeneity in the ambient medium or an instability in the bow shock .
we also report on a previously undetected x - ray emission extending for about @xmath6 ahead of the pulsar , possibly a pulsar wind nebula created by shocked pulsar wind , with a luminosity @xmath7 erg s@xmath4 .
based on observations made with the nasa / esa hubble space telescope , obtained at the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
these observations are associated with programs go 12917 and go 10568 . ]
|
` cmbology ' ( or cmb - cosmology ) , the study of temperature fluctuations in the cosmic microwave background ( cmb @xcite ) , and helioseismology , the study of acoustic oscillations on the surface of the sun @xcite , are two fields of much experimental and theoretical interest today . in the last decade , our knowledge of both areas has increased dramatically through an active ground and space - based observational programme .
this has led to substantial improvement in the quantification of models for both the early universe and the interior of the sun , and hence given us a deeper understanding of the underlying physics .
these two areas are vastly different in terms of scale .
firstly , consider the difference in physical size .
cosmology literally `
the study of the whole universe ' encompasses scales so vast as to make even our own galaxy seem minuscule .
the study of solar oscillations , on the other hand , is constrained to an object barely 100 earths in diameter virtually non - existent on a cosmological map .
we can also consider the vast temporal disparity between the two fields .
the photons we observe from the sun describe it as it was about 8 minutes ago , while cmb photons give an imprint of the universe as it was more than 13 billion years ago .
although the sun and cmb are very different in terms of scale , the underlying physics enabling us to understand them is essentially the same the physics of sound waves resonating in a cavity .
acoustic waves are a common everyday physical phenomenon , and every undergraduate learns about the use of standing sound waves to understand the interior of a cavity . probing the same kinds of waves in the early universe and solar interior is of great importance , since they can be used to study the insides of objects which are otherwise unreachable . in this article
we will compare the physics of these 2 astrophysical arenas : cmb anisotropies and and helioseismology .
both use similar language , talking about acoustic modes , the photosphere and spherical harmonics , and hence it should come as no surprise that there are very close physical analogues which can be drawn .
however , as we will see , this is only possible _ if one thinks about the universe as an inside - out version of a star_.
all current cosmological data point towards the hot big bang picture , in which the early universe was full of relativistic particles ( baryons and electrons ) and radiation , along with components of dark matter and dark energy .
this extremely dense and hot universe evolved by expanding and cooling as does an expanding gas .
most of the initial energy was in radiation , which drove the expansion according to the equations of general relativity . because of the expansion , radiation from earlier times reaches us with stretched wavelengths , and it is natural to use the observed redshift , @xmath0 , as a label for the epoch we are observing @xmath0 increases monotonically as we look farther away and back to earlier times .
the immensely high temperatures meant that photons whose energy distributions well approximated a black - body spectrum were originally energetic enough to keep the atoms ionized .
the primordial gas , then , must have been optically thick , due to the high cross - section for thomson scattering of the free electrons .
photons and baryons were said to be ` coupled ' during this era ; the continuous scattering off one another linked their temperature and fluid properties .
as the universe cooled , the rate of expansion fell due to the overall gravitational attraction of matter .
a number of important epochs occurred as particle interaction rates fell below the expansion rate .
one example is the formation of the light elements at a temperature of around @xmath1mev ( redshift of several billion ) , as the conversion between neutrons and protons froze out .
this gave a helium mass fraction of around 25% , in fairly close agreement with the value in the solar interior today .
as the universe continued to cool , the ratio of the energy density in massive particles relative to radiation increased , until the epoch of matter - radiation equality was reached . at a redshift several times smaller than this ( @xmath2 ) , very few photons were energetic enough to ionize hydrogen , and so electrons were captured by protons , leading to an optically thin universe a process referred to as cosmological recombination .
photons , having been freed from their electron captors travelled unhindered through a mostly empty space until they are seen today as the cmb . the probability for a photon last scattering with an electron peaks at @xmath3 ,
and so we call this the ` last scattering surface '
. photons which last scattered at this epoch are seen today with a remarkably pure black - body distribution at a cool temperature of about @xmath4k . at a redshift of 1100
this means that recombination occurred at a temperature of about @xmath5k .
this cosmic photosphere is seen at an age of around 400,000 years , while the age of the universe today is about 14 billion years .
the background radiation , as seen by any observer , is remarkably isotropic , but contains the signatures of primordial structure in the form of temperature anisotropies on the order of one part in @xmath6 .
it is these anisotropies which are of most interest to cosmologists , as their measurement promises to constrain the many otherwise free parameters in theoretical models of the evolution of the universe .
moreover , they are caused by the small fluctuations in density which grow in contrast to become the rich structure ( galaxies , stars , people ) that we see today . not unlike the cmb ,
the solar surface is mostly uniform .
when we observe the sun , we see photons escaping from the solar photosphere , the energy distribution of which is crudely that of a black - body with an effective temperature @xmath7k . because of its brightness and angular extent on our sky , solar temperature anisotropies were witnessed long before scientific explanation could be provided for them .
sunspots were first noted by galileo and could be seen with a tool as simple as a pin - hole camera .
however , these large solar ` blemishes ' are very localized on the solar surface , as well as being transient , and as such do not contribute much to the overall variance of angular anisotropies ; in cmb language , sunspots are localized highly non - gaussian cold spots with amplitude @xmath8 . the configuration of the solar interior can be inferred by applying the standard equations of stellar structure , which derive from the principles of thermal and hydrostatic equilibrium .
these are complicated by details of energy generation through nuclear processes and energy transport by radiation and convection .
the convective zone is confined to the outer 30@xmath9 of the solar radius , where no radiative transport occurs .
energy generation is driven by the conversion of hydrogen to helium at similar temperatures to what was achieved in the universe when the primordial helium was formed .
it is the acoustic oscillations anisotropies discovered in 1960 by observing doppler shifts in absorption lines due to the physical movements of atoms in the photosphere which are of greatest interest to helioseismologists .
these oscillations , composed of various pressure modes or ` p - modes ' , are waves sustained by a radial pressure gradient ; they are sound waves trapped in the solar interior .
the principle underlying helioseismology is that the various acoustic modes provide different information about the solar interior .
in particular , modes characterized by different numbers of radial nodes penetrate to different depths within the sun , providing a series of probes allowing one to determine the radially dependent physics of the sun .
for example , by measuring the dispersion relation of a mode , one can estimate the average sound speed it experiences .
using several modes , and knowing their penetration depths , a helioseismologist can determine the functional form of the solar sound speed with respect to radius .
such tests can be used to both confirm and to constrain parameters within the standard solar model , including determining the interior composition and rotation rate .
some similarities between the sun and the universe should already be apparent .
consider , for instance , that an observation of either the cmb or the sun collects photons originating from a spherical surface , and describing a nearly uniform black - body spectrum . in the sun
the photosphere is the surface where gas density has increased sufficiently for photons to be strongly scattered , and its radius is usually defined as @xmath10 , which is around @xmath11 km . in the cmb
the last scattering surface has a distance from the big bang which is given by the recombination time times the speed of light .
_ however _ , the cmb sky surrounds us , and when we look out towards the cosmic photosphere it is like looking _ into _ the surface of a star .
in contrast , the sun is localized on our sky .
that is to say , one can point a finger at the centre of the sun with the knowledge that it is entirely contained within some finite radius of that point ( at a given time ) .
this is not possible for the cmb last scattering surface , where every observer ( potentially in quite different parts of the universe ) has their own last scattering surface .
this is because in the uniformly expanding universe everything is moving away from everything else
thus there is no true centre of the universal expansion or , rather , every point can be considered to be the centre .
so every observer sees the early universe photons arriving from all directions in a spherical shell around them .
we are at the centre of a space in which the cosmic photosphere surrounds us , as if the surface of the star had been wrapped all round us .
the centre of this ` star ' is then located in the very early universe , well beyond the distance of the last scattering surface , and this centre ( the position of the ` big bang ' if you like ) is in every direction as we look out , _ into _ the ` cosmic star ' .
the solar and cosmic photospheres are therefore quite analogous to each other .
it is the fluctuations over each of these spheres which are of most interest , since they probe the nature of the acoustic cavities .
helioseismologists and cosmologists use the same mathematical tools to describe these acoustic modes , namely the spherical harmonics .
any angular function @xmath12 can be expanded in terms of spherical harmonics by f ( , ) = _
= 0^ _ -^ a_m y_m ( , ) .
several of these modes are illustrated in fig [ fig : harmmodes ] .
roughly speaking , the @xmath13 index describes the angular extent of features , @xmath14 , while @xmath15 characterizes the azimuthal dependence .
cosmologists are mainly interested in the power spectrum of @xmath16s , while helioseismologists study their time dependence .
the fact that the universe is like an inside - out star means there is also another important difference to keep clear .
the spherical harmonics describing anisotropies in the sun and the cmb must be projected from a different centre in each cases helioseismologists use the centre of the sun as the origin , while cosmologists use the observer s position as the centre .
the scales and ` inside - out ' geometry are illustrated in fig .
[ fig : inside_out ] .
the other major difference is the colossal distance of the cosmic photosphere on which we see the cmb anisotropies . at a redshift of around 1100 ,
the last scattering surface is about @xmath17gpc away from us ( a bit more than the light travel time in 14 billion years , because it has been expanding during that interval ) .
this is about @xmath18 times the scale of a sun - like star .
so the universe is exactly like a star , except that it is completely turned inside - out , and 100 quadrillion times bigger ! the reason why the inside - out star is an attractive analogy for the cmb is that we observe photons emerging from the cosmic photosphere .
this last - scattering surface is a shell defined by the distance from us in which there is a significant probability for the photons to have suffered their last scattering event .
we see to where the optical depth @xmath19 is around unity , with @xmath20 entirely due to thomson scattering off free electrons .
the last scattering surface is therefore defined by the epoch at which the universe went from being a plasma to being a neutral gas ; the time since this epoch , coupled with the finite speed of light , defines this spherical surface .
for the sun ( or any other star ) the photosphere is also defined by the region where @xmath21 , but the source of opacity is much more complicated , including lines from heavy elements and @xmath22 ion scattering .
but there is a more significant distinction , coming from the vastly different photon - to - atom ratio
approximately 1 billion for the universe , but less than one _ billionth _ for the solar surface !
hence the temperature at which the universe went from plasma to neutral is determined by when the photons allowed the atoms to recombine , and this determines the last - scattering surface .
but the photosphere of the sun is mainly determined by where the _ density _ has fallen off .
so it is much more like an actual edge to the solar material than in the case of the universe , where the density is slowly varying , but the ionization changes .
one might then ask why the cosmic recombination temperature turns out to be within a factor of 2 of the temperature of the solar surface .
the answer is that partly this was just a coincidence .
but the fact that the order of magnitude is similar is _ not _ surprising , since this comes basically from the temperature at which atoms get ionized .
so , despite the chemistry being different , this is @xmath23ev in both cases .
another important feature for the cmb is acoustic damping , which is related to the thickness of the last scattering surface .
this turns out to be approximately @xmath24 , which leads to a smearing of the anisotropies at an angular scale which is about 10 times smaller than the causal scale at the last scattering epoch , corresponding to @xmath25 , as can be seen in fig .
[ fig : modes ] . similarly , in the sun the optical depth does not drop instantaneously , although it much more abrupt than for the cmb .
the thickness of the solar photosphere is a few hundred km or around @xmath26 .
this means that the damping for solar modes is at an angular scale about 100 times higher in @xmath13 than for the cmb .
a crucial property of both cmb and solar fluctuations is that their amplitude is small . as a result
, the equations which describe them can be solved by linear perturbation theory , such that a set of initial conditions can be evolved forward in time exactly to predict the final observed oscillation spectrum ( particularly in the cosmic case ) , greatly simplifying the comparison of theory with observations .
both the fluids in the solar interior and early universe are characterized by several variables , each with an average value and a small perturbation which varies with both time and position the energy density @xmath27 , pressure @xmath28 ( related by an equation of state @xmath29 ) and local velocity @xmath30 .
a continuity equation enforces conservation of mass , euler s equation determines the motion of the fluid , and poisson s equation describes the response of the fluid to gravity .
the full cosmological equations are relativistic generalizations of these , but the physics is the same . in order to understand acoustic waves it is sufficient to initially ignore gravity and consider a density perturbation in each fluid .
the cosmological fluid is adiabatic , a natural outcome of inflationary initial conditions , and this is also an excellent approximation in most of the solar interior .
this means that one can ignore heat exchange between fluid elements only compression increases the temperature , and expansion cools it . in both cases ,
the resulting continuity and euler equations then describe a simple oscillator [ eqn : sound ] + + c_s^2 k^2 = 0 , where @xmath31 , @xmath32 and an overdot denotes the time derivative .
the scale factor @xmath33 describes the cosmological expansion , and is related to redshift by @xmath34 this term is zero in the solar case .
we have written this equation in fourier space , where @xmath35 is the wavenumber of the mode . neglecting the expansion term , the solution to this equation are plane acoustic waves oscillating at the sound speed , defined through @xmath36 . in a relativistic cosmological fluid @xmath37 , where @xmath38 is the speed a light . in the solar interior
the sound speed varies as a function of radius , and for an ideal gas we would just have @xmath39 , where @xmath40 is boltzmann s constant and @xmath41 is the average particle mass . in the standard model of the sun
the values range from about @xmath42 near the solar surface to about @xmath43 near the solar core .
equation ( [ eqn : sound ] ) has several important consequences @xcite .
the dispersion relation of the waves , given by @xmath44 , means that spatial and temporal modes are related by a constant in the cosmological fluid
i.e. a wave with twice the wavelength has twice the oscillation timescale , which turns out to be crucial for understanding the observed cmb acoustic spectrum . in the solar interior
, the oscillation equations must be combined with boundary conditions at the surface . coupled with the fact that the sound speed varies as a function of depth
, this leads to _ refraction _ of waves in the solar interior .
high frequency modes are trapped near the surface , while low frequency modes penetrate closer to the core .
this means that observation of oscillations as a function of frequency can be used to probe the interior .
since the oscillations in density in equation ( 2 ) also involve oscillations in velocity , then there may be more than one physically observable effect .
the velocities will be @xmath45 out of phase , since the velocity maxima and minima occur at the zeros of the density oscillations . for the sun
it is usually the time - varying doppler shifts of the velocity oscillations that are observed directly ( although there are also luminosity variations observable from the lowest multipole modes ) . for the cmb
the amplitudes of the standing waves are frozen at the last scattering epoch
the main effect is from the density variations , although there is a sub - dominant contribution from the velocities .
another important property of the acoustic modes is that they are _ irrotational_.
this means that the fluid velocity is in the direction of the wavevector * k*. the cosmological fluid is dominated by irrotational modes , which are the primary source of anisotropy in the cmb .
however , inflationary initial conditions also predict a small fraction of rotational modes , which are seeded by gravitational waves .
the most characteristic effect of these is through their effect on the pattern of cmb polarization in which ( through analogy with curl in electro - magnetism ) they are usually referred to as ` b - modes ' .
these have not yet been detected , but are one of the main motivations for future cmb missions . in the sun
there is some evidence for ` r - modes ' in the photosphere , which are similar to the rossby waves seen in the earth s atmosphere and oceans .
these are driven by the sun s rotation .
so , although there is a very loose analogy with the cmb ` b - modes ' , there is also a very fundamental difference
there are extremely strict limits on the rotation of the universe , coming from the non - observations of spiral - like patterns in the cmb anisotropies .
inflationary cosmology predicts that quantum fluctuations created in the very early universe seeded gravitational fluctuations within the primordial plasma .
the introduction of gravity means that pressure and gravity are now two competing forces .
the gravitational instability imposed on these perturbations , along with the counter - acting radiation pressure from the energetic and numerically dense photons ( still highly coupled to the baryons ) leads to acoustic oscillations within this early plasma .
after recombination , the photons restoring radiation pressure no longer drives these oscillations and the perturbations in structure are frozen - out as the cmb is released . on the largest scales we simply see a reflection of the initial conditions in the gravitational potential .
this is the main physical effect for angles larger than that subtended by the causal length at last - scattering , which corresponds to about @xmath46 on the sky ( about the width of your thumb held at arm s length , or coincidentally about 4 times the diameter of the sun ) . on smaller scales ,
the temperature fluctuations are due to the density and velocity variations in the oscillating plasma at the time of last scattering .
one observes these as a ` fundamental ' mode , together with a series of harmonic overtones in angular scale on the sky .
this occurs because of the approximately constant sound speed within the plasma up to decoupling .
there therefore exists a scale , which at last scattering had suffered maximal compression so cmb photons have a maximum temperature variance at this angular scale .
this fundamental mode for the cmb is the angle subtended by the ` sound horizon ' at last scattering , i.e. the maximum distance sound could propagate since the initial fluctuations were laid down .
this corresponds to an angular scale of around half a degree on the sky , or @xmath47 .
we observe a set of harmonic overtones at @xmath48 , corresponding to the modes which have undergone further oscillations to reach maximal compression , and we observe peaks at @xmath49 , corresponding to maximal rarefaction modes ( with @xmath50 in each case ) .
these features are most easily seen by plotting the variance of cmb temperature against angular scale on the sky , or more precisely by plotting power versus multipole for spherical harmonics . in the left panel of fig .
[ fig : modes ] , we show the anisotropy power spectrum for current observational data . the conventional quantity which is plotted as power is a scaled version of @xmath51 , where the angled brackets mean an average over all possible realizations of each mode , and each @xmath15 is equivalent ( since there are no preferred directions in the universe ) .
we see a clear detection of the first , second and third acoustic peaks .
this coherence of the cmb power spectrum only occurs because the initial conditions are ` synchronized ' .
this happens naturally in inflationary models for the initial perturbations when modes start oscillating at very early times and over almost arbitrarily large scales ( even those which are apparently acausal ) .
this synchronization means that each cosmological fourier mode has the same temporal phase . in the acoustic cavity analogy
this means that the modes have a node at @xmath52 ( with the fundamental and harmonics having nodes or anti - nodes at the recombination time ) .
this is just like the radial modes in the sun , which all have a node at the centre and at the surface .
_ however _ , for the sun the distance and epoch are not tied together as they are for cosmological distance and look - back time .
the sound waves in the sun may all have a node at the centre , but they are excited at different ( and random ) times .
inflationary initial conditions with amplitude about 1 part in @xmath6 appear to provide an excellent fit to today s cosmological perturbations
they are approximately scale invariant in gravitational potential , are adiabatic in nature and maximally random ( i.e. have gaussian statistics , with no phase correlations between modes ) .
this means that the fluctuations obey purely gaussian statistics , so that the power - spectrum contains all useful cosmological information .
of course this is manifestly not true for the sun , or indeed any object where one can meaningfully point at specific features . in the cmb ,
the oscillation modes are mainly acoustic in nature . in the sun
the modes are acoustic ( ` p - modes ' ) , gravity ( ` g - modes ' ) and surface waves ( ` f - modes ' ) . acoustic oscillations are driven by pressure variations , whereas g - modes are driven by buoyancy of fluid parcels ( gravity provides the restoring force ) .
surface waves arise from discontinuities in density along the surface and propagate along these discontinuities .
neither gravity nor surface waves are possible in the cmb at linear order , due to the isotropy of temperature at zeroth order . in solar models spherical symmetry and boundary conditions at the surface
are enforced .
this leads to a set of discrete oscillation frequencies @xmath53 , where @xmath13 and @xmath15 are the degree and order of spherical harmonics , and @xmath54 is the radial node number .
these frequencies are independent of @xmath15 for a spherically symmetric non - rotating star . however , if the star is rotating ( as is usually the case ) , this splits the frequencies , in a similar fashion to the zeeman or stark effects for spectral lines .
the @xmath55 modes travel at a different speed around the star than the @xmath56 modes .
thus observations of this splitting over many different multiplets enables reconstruction of the solar interior rotation . in the sun
the period of acoustic oscillation is dependent on the radial node number @xmath54 . when a helioseismologist talks about the ` fundamental ' , what is meant is the radial mode which has a node at the centre and an anti - node at the surface .
the harmonics are then the radial modes with extra numbers of nodes .
each of these radial modes can have a whole set of angular harmonics with different @xmath13 ( and @xmath15 ) .
the frequencies can be estimated from the time taken for a sound wave to travel one horizontal wavelength , which gives @xmath57 ( for a wave propagating at depth @xmath58 ) .
the fundamental node ( @xmath59 ) has a period of @xmath60 hour , coming approximately from the free - fall time @xmath61 .
typical observed modes have @xmath6230 , with a period of @xmath63 minutes .
however , there is not really a fundamental _ angular _ mode for the sun in the way that there is for the cmb ( although see section [ sec : sunpower ] ) .
since the solar oscillation timescale is much ( much ! ) shorter than the cosmological timescale , helioseismologists have a more powerful signature of acoustics in the sun than the acoustic peaks frozen onto the cmb they can observe the _ actual _ oscillations and measure their frequencies directly , instead of simply inferring the oscillations from a harmonic imprint .
these observations can be captured in an @xmath13@xmath64 plot , where the angular frequencies are plotted against the temporal frequencies , as shown in fig .
[ fig : modes ] .
the ridges in this plot correspond to node number the ridge with lowest temporal frequency corresponds to the fundamental mode . since modes with small @xmath54 probe deeper into the solar interior , they can be used to constrain radially dependent properties .
two key effects in the sun prevent us from seeing coherent acoustic modes like we observe in the cmb .
firstly the modes are probably generated by turbulent eddies in the convection zone , so that there is stochastic forcing of the oscillations , with random temporal phases .
and secondly the very long lifetime of a star compared with its typical mode timescale ( @xmath65 ) ensures that even if initial conditions were synchronized like for the universe , stochastic excitation would almost immediately lead to loss of coherence . to summarize , in table [ tab : compare ] we contrast some of the features of the the universe with those of the sun .
.summary of similarities and differences between cmbology and helioseismology . [ cols="<,<",options="header " , ] [ tab : compare ] at present , the instrument which has produced the most sensitive full - sky maps of the cmb is the wilkinson microwave anisotropy probe ( _ wmap _ ) @xcite . for the sun ,
the analogous instrument is the michelson doppler imager ( mdi ) on board the solar and heliospheric observatory ( _ soho _ ) @xcite .
_ wmap _ measures temperature differences on the sky with an instrument located at the l2 lagrange point of the earth - sun system , which lies about @xmath66 million km on the opposite side of the earth from the sun .
_ wmap _ has an angular resolution of @xmath67 , which means it can measure anisotropies up to @xmath68 .
mdi measures the doppler shifts of the gas in the solar photosphere on a spacecraft located at the l1 lagrange point , which lies at the same distance from the earth as l2 , but on the sun side .
mdi has an angular resolution of about @xmath69 when observing the full solar disk , such that the pixel resolution on the solar surface is around @xmath70 .
however , since the sun is _ not _ an inside - out star , the origin for the spherical harmonic coordinate system is the solar centre , and this means that mdi can measure anisotropies out to @xmath71 .
this is coincidentally similar to the _ wmap _ resolution . the solar oscillation spectrum in fig . [ fig : modes ]
was produced using 11 hours of nearly continuous high - resolution mdi solar dopplergrams , consisting of one - minute integrations .
helioseismologists studying such data have access to several years of uninterrupted dopplergrams , but the basic information is evident using this limited data - set .
the information extracted from the modes has a different form for each of the 2 fields we are comparing . in cosmology one makes a cmb map , extracts the power spectrum of anisotropies and typically uses a least - squares routine to fit a cosmological model to the data .
it is remarkable that only 6 parameters ( within a simple isotropic , homogeneous framework ) are needed to provide an excellent fit to _ wmap _ data . in helioseismology , fitting to the acoustic frequencies are often computed using direct inversion techniques on the data .
these are then used to tune the solar model , where the ` parameters ' include some unknown radial functions .
cmbologists learn about the universe from the cmb power spectrum , because it contains almost all the useful information , together with the fact that the linear perturbation theory is remarkably simple .
in contrast , the full theory for the amplitudes of helioseismic modes would have to be _ non_-linear , and hence is far from simple .
because of this , the amplitude information of solar modes is often set aside in order to focus on the frequencies .
nevertheless , we can ask what the mode power spectrum would look like for the sun , in analogy with the cmb @xmath72s . in the solar @xmath13@xmath64 plot ( fig . [
fig : modes ] ) the ridges are p - modes with different radial mode number , @xmath54 , while the signal at lower temporal frequencies is dominated by convective and other noise effects .
this is mainly from convective granulation and supergranulation motions , and although these are interesting in their own right , they will obscure the acoustic oscillations .
we therefore remove this low ( temporal ) frequency signal from the mdi data before attempting to make a power spectrum .
we do this by cutting out everything in @xmath13@xmath64 which would lie below the surface sound speed ( following @xcite ) , which is the lowest speed at which acoustic modes can propagate in the interior .
we can make an order of magnitude conversion to temperature using a blackbody model for the luminosity : @xmath73 , where @xmath74 is the stefan - boltzmann constant . if for simplicity we take @xmath75 as constant over any oscillation , then @xmath76 , where @xmath77 is an average of the velocity integrated over time , and @xmath78 is the measured oscillation frequency .
performing this scaling to ` temperature units ' , integrating mdi data over all frequencies , and correcting for some instrumental efficiency effects , gives the curve shown in fig .
[ fig : powcomp ] .
remarkably , the cmb power spectrum needs only to be scaled up by about an order of magnitude in order to be comparable .
this mean that in terms of @xmath79 amplitude the two power spectra are within about a factor of three , although of course the shapes of the 2 curves are quite different .
we have shown interesting analogies between the fields of cmbology and helioseismology .
one could contrast more features for example looking at the polarization information or comparing details of how the sun and the universe have changed over time .
however , we have probably carried this analogy far enough to be useful in understanding more of the physics of both the cmb and the sun . as a final remark
, we note that as well as learning about the sun through its acoustic structure , astrophysicists are also beginning to learn about the interiors of other stars the science of asteroseismology .
for example , since 2003 the microsatellite _ most _ @xcite has been studying low angular degree modes in many nearby stars . if there is a further analogy to be drawn here , it may be that each of these stars is like a separate inside - out universe , and asteroseismology is like studying the multiverse !
this research was supported by the natural sciences and engineering research council of canada .
we thank chris cameron , mark halpern , jaymie matthews and jim zibin for very useful conversations .
see e.g. scott d. , smoot g.f .
, 2006 , ` cosmic microwave background mini - review ' , in yao w .- m . , et al . , `
the review of particle physics ' , j. phys .
g33 , 1 [ astro - ph/0601307 ] and 2007 web update at http://pdg.lbl.gov/ further discussion of the physics of cmb acoustic oscillations can be found in : scott d. , white m. , 1995 , ` echoes of gravity ' , gen .
grav . , 27 , 10231030 [ astro - ph/9505102 ] ; hu w. , sugiyama n. , silk j. , 1997 , ` the physics of microwave background anisotropies ' , nature , 386 , 3743 [ astro - ph/9604166 ] ; hu w. , white , 2004 , ` the cosmic symphony ' , sci .
american , 290 , 4453 .
the web - page of mark whittle also has some interesting visuals and sound files on ` big bang acoustics ' : http://www.astro.virginia.edu/@xmath80dmw8f/ d. georgobiani , j. zhao and a. .g .
kosovichev , 2007 , ` local helioseismology and correlation tracking analysis of surface structures in realistic simulations of solar convection ' , astrophys .
j. 657 , 11571161
|
acoustic modes can be used to study the physics of the interior of a cavity , and this is especially useful when the inside region is inaccessible .
many astrophysicists use such sound waves as an essential tool in their research . here
we focus on two separate sub - fields in which oscillations on the surface of a sphere are studied helioseismology and cmbology the surface being either the solar or cosmic photosphere .
both research areas use the language of spherical harmonics , as well as sharing many close similarities in the underlying physics . however , there are also many fundamental differences , which we explain in this pedagogical article .
|
the comprehensive formulation for loop quantum cosmology ( lqc ) in the spatially flat - isotropic model has been constructed @xcite . with a massless scalar field serving as the _ emergent time _ , the result shows that the quantum evolution is _ deterministic across the deep planck regime _ and in the backward evolution of the states which are semiclassical at late times , _ the big bang is replaced by a big bounce_. based on the same principles , the construction was further improved by a more direct implementation of the underlying physical ideas of loop quantum gravity ( lqg ) @xcite . in the improved dynamics ,
_ the big bounce occurs precisely when the matter density enters the planck regime _ , regardless of the value of the momentum @xmath0 of the scalar field .
both the precursor strategy ( `` @xmath1-scheme '' ) and the improved strategy ( `` @xmath2-scheme '' ) were applied and reconstructed for the bianchi i model to include anisotropy @xcite .
the analytical investigation shows that the state in the kinematical hilbert space associated with the classical singularity is _ completely decoupled _ in the difference evolution equation , indicating that the classical singularity is resolved in the quantum evolution and the big bounce may take place when any of the area scales undergoes the vanishing behavior .
while a thorough numerical investigation remains to be done to draw the definite conclusion for the details of the quantum evolution in the bianchi i model , this paper studies its effective dynamics with lqc discreteness corrections .
not only does the result affirm the anticipations in @xcite but more intuitive pictures are also obtained in this semiclassical approach , giving an insight into how and why the big bounces take place . in accordance with the formulation in @xcite , this paper focus specifically on the model with a massless scalar field . in the context of effective dynamics with lqc discreteness corrections , the similar analysis for more generic bianchi i models with the inclusion of arbitrary matter with equation of state @xmath3
is also investigated in @xcite , which gives the similar results for the occurrence of big bounces with only difference in detail . with arbitrary matter sources , however ,
the equations of motion are very complicated and a proper approximation has to be used . by contrast , in the special case of a massless scalar field , the equations of motion can be solved analytically and therefore the underlying physics is more transparent . this paper is organized as follows . in section [ sec : classical dynamics ] , the classical dynamics of the bianchi i cosmology with a massless scalar source is solved in terms of ashtekar variables in hamiltonian formulation .
the effective dynamics with lqc corrections in @xmath2-scheme is constructed and solved in section [ sec : mubar dynamics ] .
its phenomenological ramifications are discussed in section [ sec : discussion ] . as a comparison to the @xmath2-scheme , the effective dynamics in @xmath1-scheme
is also included in appendix [ sec : muzero dynamics ] .
the spacetime metric of bianchi type i is given as : ds^2=-dt^2+a_1 ^ 2(t)dx^2+a_2 ^ 2(t)dy^2+a_3 ^ 2(t)dz^2 . in terms of ashtekar variables , the phase space of bianchi i models is given by the diagonal triad variables @xmath4 and diagonal connection variables @xmath5 for @xmath6 , which satisfy the canonical relation : \{c_i , p_j}=8g_ij .
the triad variables @xmath4 are related with the length scale factors @xmath7 via : [ eqn : p and a ] p_1=a_2a_3,p_2=a_1a_3,p_3=a_1a_2 . in the presence of a massless scalar field @xmath8 , ( which is independent of the spatial coordinates with homogeneity assumed ) ,
the classical dynamics is govern by the hamiltonian constraint : [ eqn : cl hamiltonian ] & & c = c_grav+c _ + & = & - + , where @xmath0 is the conjugate momentum of @xmath9 and has the canonical relation with @xmath9 : \{,p_}=1 .
we can simplify the hamiltonian by choosing the lapse function @xmath10 and thus introducing the new time variable @xmath11 .
the rescaled hamiltonian constraint is given by [ eqn : cl rescaled hamiltonian ] h=- + .
+ the equations of motion are governed by the hamilton s equations : [ eqn : cl eom 1 ] & = & \{p_,h}=0p _ + [ eqn : cl eom 2 ] & = & \{,h}=p _ , + [ eqn : cl eom 3 ] & = & \{c_1,h}=8 g + & = & -^-1c_1(c_2p_2+c_3p_3 ) , + [ eqn : cl eom 4 ] & = & \{p_1,h}=-8 g + & = & ^-1p_1(c_2p_2+c_3p_3 ) , and so on for @xmath12 , @xmath13 , @xmath14 , @xmath15 in the cyclic manner .
in addition to the hamilton s equations , the constraint that the hamiltonian must vanish yields [ eqn : cl eom 5 ] & & h(c_i , p_i)=0 + p_^2&= & ( c_2p_2c_3p_3+c_1p_1c_3p_3+c_1p_1c_2p_2 ) . combining ( [ eqn : cl eom 3 ] ) and ( [ eqn : cl eom 4 ] ) gives [ eqn : const ki ] ( p_ic_i)=0,p_ic_i=8gk_i , where @xmath16 are dimensionless constants , which will be used to parameterize the solutions of evolution . taking ( [ eqn : const ki ] ) into ( [ eqn : cl eom 5 ] ) , we have [ eqn : p_ph and k ] p_^2=16g^2 \{k_2k_3+k_1k_3+k_1k_2 } or equivalently [ eqn : k ] k_^2=2 ( k_2k_3+k_1k_3+k_1k_2 ) , if we define [ eqn : def of p_ph ] p_:=k_. putting ( [ eqn : const ki ] ) into ( [ eqn : cl eom 4 ] ) gives = 8 g ( k_2+k_3 ) , by referring to ( [ eqn : cl eom 2 ] ) , this leads to [ eqn : cl diff eq 2 ] = 8 g = ( ) , where we scale the parameters @xmath17 , @xmath18 such that [ eqn : para constraint ] _ 1+_2+_3=1 , _
1 ^ 2+_2 ^ 2+_3 ^ 2+_^2=1 . regarding @xmath9 as the _ emergent time _
, the solutions of evolution are given by [ eqn : cl sol 1 ] p_i()=p_i(_0)e^()(-_0 ) , or equivalently [ eqn : cl sol 2 ] a_i()=a_i(_0)e^(-_0 ) .
the classical bianchi i model with a massless scalar field admits both `` kasner - like '' ( two of @xmath19 positive and the other negative ) and `` kasner - unlike '' ( all @xmath19 positive ) solutions .
the kasner - like solution , which has two expanding and one contracting directions ( say @xmath20 ) , eventually encounters the `` kasner - like singularity '' ( a given regular cubical cell stretches as an infinitely long line ) in the far past and the `` planar collapse '' ( a regular cubical cell stretches as an infinitely large plane ) in the far future . on the other hand , the kasner - unlike solution , with all directions expanding , encounters the `` kasner - unlike singularity '' ( a regular cubical cell vanishes to a point ) in the far past and no planar collapse .
we will see that with lqc discreteness corrections , both kasner - like and kasner - unlike singularities are resolved and replaced by the _
big bounces _ , whereas the planar collapse remains its destiny even one of the three diagonal directions approaches infinitely small length scale .
in lqc , the connection variables @xmath5 do not exist and should be replace by holonomies . in the effective theory , to capture the quantum corrections , following the procedures used in the isotropic case @xcite , we take the prescription to replace @xmath5 by @xmath21 , introducing discreteness variables @xmath22 . in the improved strategy ( @xmath2-scheme )
used in bianchi i lqc @xcite , @xmath22 are not fixed constants but given by |_1=,|_2=,|_3= , where @xmath23 is the _ area gap _ in the full theory of lqg .
imposing this prescription plus the loop quantum correction to the inverse triad on ( [ eqn : cl hamiltonian ] ) , we have the effective hamiltonian constraint to the leading order : [ eqn : qm hamiltonian original ] c_eff&=&f(p_1)f(p_2)f(p_3 ) - + & & \ { p_2p_3 + } , where @xmath24 is the eigenvalue of the inverse triad operator @xmath25 .
the loop quantization gives the quantum corrections : f(p_i)~\ { cr ( 1+o(_pl^2/p_i ) ) & p_i_pl^2 + ^n/_pl^2n+1 & p_i_pl^2 . with the planck length @xmath26 and a positive @xmath27 .
the corrections to @xmath24 is significant only in the planckian region in the vicinity of @xmath28 . from now on
, we will ignore the quantum corrections to @xmath24 by simply taking its classical function @xmath29 .
[ we will see that in the backward evolution the big bounce takes place much earlier before the discreteness correction on the inverse triad operator becomes considerable , and it is the `` non - locality '' effect ( i.e. , using the holonomies ) that accounts for the occurrence of the big bounce . ] with @xmath29 , by choosing @xmath11 , the hamiltonian constraint ( [ eqn : qm hamiltonian original ] ) can be rescaled as [ eqn : qm hamiltonian ] & & h_|= + & & - \ { p_2p_3 + } .again , the equations of motion are given by the hamilton s equations and the constraint that the hamiltonian must vanish : [ eqn : qm eom 1 ] & = & \{p_,h_|}=0p _ + [ eqn : qm eom 2 ] & = & \{,h_|}=p _ , + [ eqn : qm eom 3 ] & = & \{c_1,h_|}=8 g + & = & -^-1 ( - ) + & & ( p_2 + p_3 ) , + [ eqn : qm eom 4 ] & = & \{p_1,h_|}=-8 g + & = & ^-1p_1(|_1c_1 ) + & & ( p_2 + p_3 ) , and [ eqn : qm eom 5 ] & & h_|(c_i , p_i)=0p_^2= + & & \ { p_2p_3 + } . by ( [ eqn : qm eom 3 ] ) and ( [ eqn : qm eom 4 ] )
, we have [ eqn : qm dpc / dt ] & & ( - ) + p_i(|_ic_i ) + & & = = 0 , which gives [ eqn : qm pc ] p_i = 8gk_i . + taking ( [ eqn : qm pc ] ) into ( [ eqn : qm eom 5 ] ) again gives the same constraints on the constant parameters as in ( [ eqn : p_ph and k ] ) or ( [ eqn : k ] ) . substituting ( [ eqn : qm pc ] ) into ( [ eqn : qm eom 4 ] ) yields [ eqn : qm diff eq 1 ] = 8 g ( |_1c_1)(k_2+k_3 ) . by regarding @xmath9 as the emergent time via ( [ eqn : qm eom 2 ] ) and expressing @xmath30 , ( [ eqn : qm diff eq 1 ] ) then gives [ eqn : qm diff eq 2 ] = ( ) ^1/2 , where we define the _ directional density _ : _
i:= for the @xmath31-direction and its critical value is given by the _ planckian matter density _
@xmath32 times a numerical factor : _
i , crit:=()^2_pl , _
pl:=(8 g ^2)^-1 .
( 400,150)(0,0 ) ( -60,0 ) ( 460,150)(0,0 ) ( 10,133)(a ) ( 175,133)(b ) ( 345,133)(c ) ( 400,150)(0,0 ) ( -60,0 ) ( 460,150)(0,0 ) ( 10,133)(a ) ( 175,133)(b ) ( 345,133)(c ) ( 400,150)(0,0 ) ( -60,0 ) ( 460,150)(0,0 ) ( 10,133)(a ) ( 175,133)(b ) ( 345,133)(c ) ( 400,150)(0,0 ) ( -60,0 ) ( 460,150)(0,0 ) ( 10,133)(a ) ( 175,133)(b ) ( 345,133)(c )
as opposed to the classical equation ( [ eqn : cl diff eq 2 ] ) , in which @xmath4 continues to decrease toward the classical singularity in the backward evolution , the effective equation in ( [ eqn : qm diff eq 2 ] ) flips sign exactly at the moment when @xmath33 approaches its critical value @xmath34 .
note that by ( [ eqn : p and a ] ) @xmath4 can be regarded as the _ area _ scale factors . therefore , with the lqc discreteness corrections , ( [ eqn : qm diff eq 2 ] ) shows that the singularities ( both kasner - like and kasner - unlike ) are resolved and replaced by the big bounces in the backward evolution when any of the area scales undergoes the vanishing behavior . across the bounces ,
the equation of motion again comes closer and closer to the classical counterpart .
hence , the semiclassicality is retained on both asymptotic sides of the evolution .
furthermore , the detailed evolutions of @xmath4 are decoupled in different diagonal directions and evolve independently of one another once the initial conditions ( @xmath35 , @xmath0 and @xmath19 ) are specified .
thus , the bounces occur up to three times , once in each direction , whenever each of the directional densities @xmath33 approaches its critical value .
as expected , in @xmath2-scheme , the critical values @xmath34 are in the planck regime of @xmath36 and _ independent of the value of @xmath0 _
( @xmath34 depend on @xmath0 only through the ratio @xmath37 ) . , the old precursor strategy ( @xmath1-scheme ) is presented and it shows that the critical value of @xmath33 can be made arbitrarily small by increasing @xmath0 . ] note that @xmath33 have the same dimension as the matter density @xmath38 and @xmath33 play the same role as @xmath39 does in the isotropic case , signaling the occurrence of big bounces . on the other hand ,
the planar collapse is _ not _ resolved but one of the length scale factors @xmath7 continues the vanishing behavior in the kasner - like case .
this is expected since the classical solutions ( [ eqn : const ki ] ) and ( [ eqn : cl sol 1 ] ) yield @xmath40 ( and @xmath41 in @xmath1-scheme ) toward the planar collapse and therefore the quantum corrections become more and more negligible ( in both schemes ) . for given initial conditions ,
the differential equation ( [ eqn : qm diff eq 2 ] ) can be solved numerically .
the behaviors of @xmath42 , @xmath43 and @xmath44 are depicted in parts ( a ) , ( b ) and ( c ) respectively in fig .
[ fig : fig1 ] and fig .
[ fig : fig2 ] for kasner - unlike solutions and in fig .
[ fig : fig3 ] and fig .
[ fig : fig4 ] for kasner - like solutions .
the fact that smallness of @xmath4 ( not of @xmath7 ) is an indication of the occurrence of big bounces seems to support the suggestion that `` area is more fundamental than length in lqg '' , although whether this is simply a technical artifact or reflects some deep physics is still not clear .
( see section vii.b of @xcite for some comments on this aspect and @xcite for more details . ) meanwhile , as the length operator has been shown to have a discrete spectrum @xcite , the fact that the vanishing of the length scale factor in the planar collapse is not stopped seems to contradict the discreteness of the length spectrum .
whether we miss some important ingredients when imposing the fundamental discreteness of lqg in the lqc construction or indeed area is more essential than length remains an open question for further investigation .
it is also noteworthy that ( [ eqn : qm diff eq 2 ] ) remains invariant if we rescale @xmath45 and @xmath46 at the same time .
this is reminiscent of the idea as suggested in @xcite that area is measurable only if the surface is _ coupled with the material reference_. the scaling invariance , however , breaks down in the full lqc theory since the quantum evolution is governed by a difference equation @xcite , in which the step size of difference introduces an additional scale in the deep planck regime . and @xmath4 .
accordingly , we put big values of @xmath0 and @xmath35 in the figures to make sense of the semiclassical approach for the effective dynamics .
the figures are trivially rescaled under the scaling . ]
meanwhile , related to the above observation , the physical meaning of the directional densities @xmath33 can be interpreted as the ( inverse of ) area scales , again , _ measured by the reference of the matter content_. the big bounces take place whenever one of the area scales becomes very small by the reference of the matter momentum .
it is then attempting to regard not only @xmath9 as the `` internal clock '' ( emergent time ) but also @xmath0 as the `` internal rod '' namely , the measurement of both temporal and spatial geometries makes sense only in the presence of matter content .
this observation may support the ideas of the relational interpretation of quantum mechanics with real rods and clocks such as studied in @xcite ( see also @xcite ) , although the link is far from clear .
if this concept is taken seriously , in return , we might be able to further improve the @xmath2-scheme to better reflect the underlying physics of lqg such that the difference equation of evolution in the full lqc theory also respects the scaling invariance mentioned above .
the author would like to thank abhay ashtekar , golam hossain , tomasz pawlowski , parampreet singh for useful discussions and especially kevin vandersloot for sharing his private notes and important ideas .
this work was supported in part by the nsf grant phy-0456913 .
one of the virtues of the improved strategy ( @xmath2-scheme ) in the isotropic model is to fix the serious drawback in the old precursor strategy ( @xmath1-scheme ) that the critical value of the matter density at which the bounce occurs can be made arbitrarily small by increasing the momentum @xmath0 . as the directional densities @xmath33 play the same role in the bianchi i model as the matter density @xmath39 does in the isotropic case
, we expect that the critical value of @xmath33 at which the bounce occurs can be made arbitrarily small by increasing the momentum @xmath0 in @xmath47-scheme but is independent of @xmath0 in @xmath2-scheme .
the latter is what is shown in the main text of this paper . for comparison ,
the effective dynamics in @xmath1-scheme is presented here . in the effective theory of @xmath1-scheme
, we take the prescription to replace @xmath5 by @xmath48 with the _ fixed _ numbers @xmath49 for discreteness . analogous to ( [ eqn : qm hamiltonian ] )
, we have the effective ( rescaled ) hamiltonian constraint : & & h__o= + & & - \ { p_2p_3 + } .
again , the equations of motion are given by the hamilton s equations and the constraint that the hamiltonian must vanish : [ eqn : qm0 eom 1 ] & = & \{p_,h__o}=0p _
+ [ eqn : qm0 eom 2 ] & = & \{,h__o}=p _ , + [ eqn : qm0 eom 3 ] & = & \{c_1,h__o}=8 g + & = & -^-1 ( ) + & & ( p_2 + p_3 ) , + [ eqn : qm0 eom 4 ] & = & \{p_1,h__o}=-8 g + & = & ^-1p_1(^o_1c_1 ) + & & ( p_2 + p_3 ) , and [ eqn : qm0 eom 5 ] & & h__o(c_i , p_i)=0p_^2= + & & \ { p_2p_3 + } . from ( [ eqn : qm0 eom 3 ] ) and ( [ eqn : qm0 eom 4 ] ) , we have [ eqn : qm0 dpc / dt ] = 0 , which gives [ eqn : qm0 pc ] p_i = 8gk_i . taking ( [ eqn : qm0 pc ] ) into ( [ eqn : qm0 eom 5 ] ) gives the same constraints on the constant parameters as in ( [ eqn : p_ph and k ] ) or ( [ eqn : k ] ) . substituting ( [ eqn : qm0 pc ] ) into ( [ eqn : qm0 eom 4 ] ) yields [ eqn : qm0 diff eq 1 ] = 8 g ( ^o_1c_1)(k_2+k_3 ) . by ( [ eqn : qm0 eom 2 ] ) and @xmath50 , ( [ eqn : qm0 diff
eq 1 ] ) leads to [ eqn : qm0 diff eq 2 ] = ( ) ^1/2 , which gives the bouncing solutions with the behaviors similar to those given by ( [ eqn : qm diff eq 2 ] ) except that the critical value of @xmath33 at which the big bounce takes place is given by ^_o_i , crit:= ^3/2 , which can be made arbitrarily small by increasing the value of @xmath0 .
this tells that @xmath1-scheme gives wrong semiclassical behavior and should be improved by @xmath2-scheme .
|
the detailed formulation for loop quantum cosmology ( lqc ) in the bianchi i model with a massless scalar field was recently constructed . in this paper , its effective dynamics with the lqc discreteness corrections
is studied and the equations of motion are analytically solved , showing that the big bang singularity is replaced by the big bounces , which take place up to three times , once in each diagonal direction , whenever each of the area scale factors approaches its critical value in the planck regime measured by the reference of the scalar field momentum .
|
* the main result . * the classical function fields version of chebotarev s theorem states that the local frobenius substitutions associated with a galois covering of algebraic varieties over a finite field are equidistributed with respect to a suitably defined dirichlet density .
it can be proved using the lang - weil estimate for the number of points of varieties over finite fields , together with an ` untwisting trick ' .
we encourage the interested reader to compare the original number - theoretic theorem and the function field version by consulting @xcite , and to find a beautiful unification in @xcite .
we prove an analogue of the function fields version of chebotarev s theorem in _ difference algebraic geometry_. suppose @xmath0 is an _ tale galois covering _ of finite - dimensional difference schemes over a finite field with a power of frobenius . intuitively , @xmath1 is a set of endomorphisms of @xmath2 closed under a binary operation reminiscent of conjugation and a finite group with operators @xmath3 acts on @xmath4 in a particular fashion so that @xmath5 identifies @xmath6 with the quotient @xmath7 and @xmath8 identifies with @xmath9 . let @xmath10 be a conjugacy domain in @xmath1 . for a point @xmath11 with values in the algebraic closure of a finite field equipped with a power of frobenius @xmath12 , the _ frobenius substitution _ at @xmath13 is the element @xmath14 which matches the action of the frobenius power @xmath12 on @xmath13 , i.e. , @xmath15 for a point @xmath16 , the _ frobenius substitution _ at @xmath17 is the conjugacy class @xmath18 of any @xmath19 with @xmath20 .
the following is an informal restatement of theorem [ chebotarev - density ] .
[ main - ch - th ] the dirichlet density of the set of @xmath16 with varying @xmath12 , with the property that @xmath21 is equal to @xmath22
. * motivation and historical overview .
* there is a significant body of work related in one way or another to counting solutions of difference polynomial equations over algebraic closures of finite fields equipped with powers of the frobenius automorphism .
firstly , since counting the number of solutions of polynomial equations over finite fields is a special case , it subsumes the amazing achievements of grothendieck s circle around the weil conjectures and deligne s proof of the riemann hypothesis over finite fields .
given a well - known translation mechanism between the languages of difference equations and algebraic correspondences ( as expounded in ( * ? ? ?
* [ s : diff - pro ] ) ) , the work of pink @xcite , fujiwara @xcite and varshavsky @xcite on deligne s conjecture regarding the number of fixed points of correspondences twisted by powers of frobenius is highly relevant .
however , due to the strong properness assumptions these authors require to prove a very precise trace formula , these results can not be applied to a general difference polynomial system . without the restrictive assumptions , hrushovski produced an ingenious yet very difficult proof of a difference analogue of the lang - weil estimate for the number of points on a difference scheme over fields with frobenii @xcite .
inspired by these considerations , we embarked on a progamme to develop _ difference algebraic geometry _ to the level where it reveals the fine number - theoretic information regarding numbers of points of difference schemes over fields with frobenii . in the first instance
, we aim to generalise the techniques developed by fried , jarden et al .
@xcite , @xcite , @xcite over finite fields , collectively known under the name of _ galois stratification_. the techniques around the theorem of chebotarev developed in this paper are crucial for this work , and the development of _ twisted _ galois stratification is described in the follow - up paper @xcite .
one of the main obstacles was that , apart from the pioneering work in @xcite and @xcite , which we quickly review in section [ strict - dga ] , there are no other attempts of a systematic study of difference algebraic geometry , so all prerequisites would have to be developed from first principles .
monograph @xcite and a recent book by levin @xcite are sources of some of the difference algebra needed .
we must emphasise that a typical difference scheme that arises when studying difference polynomial equations is of _ finite transformal type _ over a difference field ( [ finsigmatype ] ) , but its ambient scheme is of infinite type and thus it falls just beyond the reach of tools and methodology of the classical algebraic geometry . * a need for generalised difference algebraic geometry . *
the first key observation we made was that the context of _ ordinary _ ( or _ strict _ ) difference schemes ( with a single endomorphism ) is too rigid and does not allow meaningful galois actions , coverings or quotients .
thus we are led into a study of _ generalised difference schemes , _ endowed with a set of ( not necessarily commuting ) endomorphisms , closed under a binary operation of ` conjugation ' .
various authors made attempts to generalise the framework of ordinary difference algebra and geometry .
the treatment of partial difference equations with respect to several commuting endomorphisms in @xcite is too restrictive for our requirements , because we are exactly interested in phenomena arising in the case of _ non - commuting _ endomorphisms .
the authors of @xcite went in the direction of considering not just fixed points of a single endomorphism @xmath23 but also those of its powers @xmath24 .
a similar approach is taken in @xcite , where the author proves a chevalley - type theorem on ( near-)constructibility of images of morphisms of difference schemes ( our version is ( * ? ? ?
* [ chevalley ] ) ) . intuitively , the object that results from a consideration of higher powers @xmath24 is closer to the ambient scheme and is therefore better - behaved .
our approach generalises both of these , and has a very interesting interaction with the latter , our framework being slightly more precise when dealing with galois actions .
section [ gen - dga ] contains the extensive development of difference algebraic geometry of generalised difference schemes needed to formulate the key notion of a _ galois covering _ of difference schemes in section [ sect : galois ] .
section [ ch : ttch ] contains the main results of the paper .
the _ dirichlet density _ is introduced as an analytic density defined by means of _ zeta _ and _ @xmath25-functions _ associated with _
constructible functions _ on galois coverings of difference schemes over fields with frobenii .
this density statement [ main - ch - th ] in its precise formulation [ chebotarev - density ] follows from the trace formula [ trace - formula ] , which is an approximative difference avatar of the classical lefschetz trace formula .
it is proved using hrushovski s twisted lang - weil estimate .
* applications . *
we consider our approach to generalised difference algebra a major advance in its own right , and it should be of intrinsic interest in the difference algebra community , with possible applications to galois theory of difference equations along the lines of @xcite . moreover , there are other naturally - occurring contexts where it may be advantageous to study objects with several endomorphisms .
in particular , as tom scanlon pointed out , a key step in most approaches to manin - mumford conjecture ( @xcite , for example ) is to find a difference polynomial equation that captures all the relevant torsion points , which is very hard to achieve with a single endomorphism . on the other hand ,
it is much easier to capture all of the torsion if one uses _ two _ endomorphisms .
apart from number theory and algebraic geometry , our results ( especially [ trace - formula ] ) should be of interest in model - theory and logic since , in conjunction with the description of definable sets in terms of galois stratifications from @xcite , they reduce the problem of counting points on definable sets over fields with powers of frobenius to a calculation of ( twisted ) character sums . thus , the present paper can be viewed as a conceptualisation of the ideas of @xcite and @xcite for fields with frobenii . as already mentioned , [ trace - formula ] and its consequence [ twisted - cebotarev ] are used in our subsequent work on galois stratifications in @xcite .
before embarking on a development of _ generalised difference schemes _ in the next section , we include a summary of known results ( @xcite , @xcite ) for difference schemes in the _ strict sense _ for the benefit of the reader .
a _ difference ring _ is a ring @xmath26 together with a distinguished monomorphism @xmath23 . given an element @xmath27 we may also write @xmath28 for @xmath29 .
a difference ring homomorphism @xmath30 is a ring map making the following diagram commutative given a difference ring extension @xmath31 , the difference subring of @xmath32 generated by a set @xmath33 over @xmath26 is denoted by @xmath34 or @xmath35_\sigma$ ] .
similarly , given a difference field extension @xmath36 , the difference subfield of @xmath25 generated by a set @xmath37 over @xmath38 is denoted by @xmath39 or @xmath40 .
let @xmath41 be an ideal in a difference ring @xmath42 .
we say that : 1 .
@xmath41 is a @xmath23-ideal if @xmath43
@xmath41 is _ well - mixed _
if @xmath44 implies @xmath45 ; 3 .
@xmath26 itself is well - mixed if the zero ideal is ; 4 .
@xmath41 is _ perfect _ if @xmath46 implies @xmath47 and @xmath28 are both in @xmath41 .
given a difference ring @xmath42 , let @xmath48 as a locally ringed space , together with the topology induced by the zariski topology of @xmath49 and the induced structure sheaf @xmath50 .
the following notation is useful when discussing the induced topology . for @xmath51 and an ideal @xmath41 in @xmath26 ,
we let @xmath52 and @xmath53 .
since the endomorphism of @xmath49 induced by @xmath23 gives a morphism @xmath54 and @xmath23 is the identity on @xmath55 , we obtain a sheaf morphism @xmath56 .
it defines a morphism of locally ringed spaces since for @xmath57 , the corresponding morphism of stalks is just the morphism @xmath58 induced by @xmath23 , which is local .
this also makes the residue field @xmath59 into a _ difference field . _
it is clear that a prime @xmath23-ideal @xmath60 is in @xmath61 if and only if it is perfect .
every @xmath23-ideal @xmath41 has a _ perfect closure _ @xmath62 .
if @xmath41 is well - mixed , its perfect closure is clearly given by @xmath63 , a^\nu\in i\}.\ ] ] given a @xmath23-ideal @xmath41 , the set @xmath64 only depends on the perfect closure @xmath62 and henceforth we adopt the notation @xmath65 for it .
the closed sets in the topology on @xmath66 are the sets @xmath65 and we have the following .
let @xmath42 be a difference ring . 1 .
@xmath67 if and only if @xmath68 .
if @xmath26 is well - mixed , an element @xmath69 defines a zero section in @xmath70 if and only if @xmath71 is @xmath23-nilpotent .
if @xmath41 is perfect , @xmath65 is irreducible if and only if @xmath72 .
@xmath66 is a quasi - compact topological space .
[ wmaffi ] suppose @xmath42 is well - mixed . 1
. the canonical morphism @xmath73 is injective . in particular , if @xmath26 is nontrivial , @xmath66 is non - empty .
2 . for @xmath51 , the map @xmath74_\sigma\to h^0(d^\sigma(r),{\mathcal{o}}_{{{\rm spec}}^\sigma(r)})$ ]
is injective .
the induced morphism @xmath75 is an isomorphism . in every difference ring @xmath42
there exists a smallest well - mixed ideal @xmath76 and thus we have the largest well - mixed quotient @xmath77 , with the universal property that every morphism from @xmath42 to a well - mixed @xmath78 factors through @xmath79 .
the closed immersion @xmath80 induces a homeomorphism @xmath81 . in view of this discussion and the merits of [ wmaffi ]
, we shall not hesitate to assume well - mixedness when necessary . 1 . an _ affine difference scheme _ @xmath82 consists of a locally ringed space @xmath83 with a morphism @xmath84 , which is isomorphic to @xmath85 for some well - mixed difference ring @xmath42 .
2 . a _ difference scheme _
@xmath82 is a locally ringed space which is locally isomorphic to an affine difference scheme .
3 . a _ morphism of difference schemes _
@xmath86 is a morphism of locally ringed spaces which respects the difference structure , @xmath87 . for a point @xmath17 on a difference scheme @xmath6 ,
we denote by @xmath88 the local ( difference ) ring at @xmath17 , and by @xmath89 the residue ( difference ) field at @xmath17 .
the following is an important consequence of [ wmaff ] .
[ strembedcat ] the ` global sections ' functor @xmath90 is left adjoint to the contravariant functor @xmath91 from the category of well - mixed difference rings to the category of difference schemes . for any difference scheme @xmath92 and any well - mixed difference ring @xmath42 , @xmath93 for an overly enthusiastic reader , it is worth remarking that , unlike in the algebraic case , @xmath91 and @xmath90 do not determine an equivalence of categories of difference rings and affine difference schemes , see [ rem - embedcat ] .
moreover , the global sections functor @xmath90 on the category of quasi - coherent sheaves of @xmath94-modules on an affine difference scheme @xmath6 is not necessarily exact .
in this section we wish to broaden the class of difference schemes in order to allow certain galois actions .
we shall henceforth refer to difference schemes with a single endomorphism as discussed above as _ strict difference schemes _ , or difference schemes in the strict sense , and we shall expand the term ` difference scheme ' to include objects with multiple endomorphisms . let us consider the category @xmath95 as follows .
an object of @xmath95 is a ( finite ) set @xmath1 , equipped with a map @xmath96 , @xmath97 such that @xmath98 .
a morphism @xmath99 is a function such that for all @xmath100 , @xmath101 [ diffcat ] let @xmath102 be a category .
the _ difference category _ over @xmath102 , denoted @xmath103 , is defined as follows .
its objects are of form @xmath104 , where @xmath6 is an object of @xmath102 , and @xmath1 is a set of @xmath102-endomorphisms of @xmath6 such that there exists a function @xmath96 , @xmath97 such that : 1 . for every @xmath100 , @xmath105 2 .
[ dvaa ] @xmath106 is an object of @xmath95 ; 3 .
[ trii ] for every @xmath107 , @xmath108 is a @xmath95-morphism .
a morphism @xmath109 consists of a @xmath95-morphism @xmath110 and a @xmath102-morphism @xmath111 such that for every @xmath107 , @xmath112 for the _ strong difference category _ over @xmath102 ,
we require that , additionally , for every object @xmath104 , all endomorphisms in @xmath1 must be @xmath102-epimorphisms of @xmath6 .
[ opdifcat ] let @xmath113 be a category .
the ( strong ) _ dual difference category over _ @xmath113 , is the opposite category of the ( strong ) difference category of the opposite category of @xmath113 , @xmath114 . to avoid misunderstandings ,
let us specify the details in the strong case .
the objects are of form @xmath115 , where @xmath116 is an object of @xmath113 , and @xmath1 is a set of @xmath113-monomorphisms @xmath117 such that for every @xmath118 , there exists a ( necessarily unique ) @xmath119 such that @xmath120 it also follows that @xmath98 for every @xmath107 .
a morphism @xmath121 consists of an @xmath113-morphism @xmath122 and a map @xmath110 such that @xmath123 moreover , we require that @xmath124 1 . in the definition of a
_ strong _ difference category , the conditions [ dvaa ] and [ trii ] from [ diffcat ] are superfluous by arguments analogous to the discussion for the dual case below .
if @xmath115 is an object of the strong dual difference category over @xmath113 , then every @xmath107 automatically defines an endomorphism of @xmath115 .
indeed , the equality @xmath125 for arbitrary @xmath126 follows from the fact that all the relevant arrows are monomorphisms so the following diagram can be completed in only one way .
@xmath127 \matrix(m)[matrix of math nodes , row sep=.05 cm , column sep=.6 cm , text height=1.2ex , text depth=0.25ex ] { & |(a)|{a } & & |(b)|{a}\\[.3 cm ] & |(c)|{a } & & |(d)|{a } \\ } ; \path[->,font=\scriptsize,>=to , thin , inner sep=1pt ] ( a)edge node[pos=0.5,above]{$\rho^\sigma$}(b ) ( b)edge node[pos=0.3 , right]{$\tau^\sigma$}(d ) ( a ) edge[dashed ] node[pos=0.5,left]{$(\tau^\sigma)^{(\rho^\sigma)}=(\tau^\rho)^\sigma$}(c ) ( c)edge node[pos=0.25,above]{$\rho^\sigma$}(d ) ( a)edge[cross line ] node[pos=0.25,above]{$\rho$}(b ) ( b)edge[cross line ] node[pos=0.5,right]{$\tau$}(d ) ( a)edge[cross line ] node[pos=0.62,right]{$\tau^\rho$}(c ) ( c)edge[cross line ] node[pos=0.5,below]{$\rho$}(d ) ( a)edge node[pos=0.5,above right]{$\sigma$}(a ) ( c)edge node[pos=0.5,above right]{$\sigma$}(c ) ( b)edge node[pos=0.5,above right]{$\sigma$}(b ) ( d)edge node[pos=0.5,above right]{$\sigma$}(d ) ; \end{tikzpicture}\ ] ] 3 .
the last requirement from [ opdifcat ] shows that the composite of an arbitrary morphism @xmath12 and the structure morphism @xmath107 in @xmath114 is well defined , @xmath128 . in case @xmath12
is a monomorphism itself , the condition is superfluous by a diagram similar to the one above .
[ fibrediff ] it is quite illuminating to view the construction of @xmath103 in the language of fibred categories .
let us consider the functor @xmath129 , such that @xmath130 , the category of @xmath102-objects with distinguished endomorphisms @xmath1 , with @xmath1-equivariant @xmath102-morphisms . to a @xmath95-morphism @xmath131
we assign @xmath132 , which maps a @xmath133-object @xmath134 to the @xmath1-object @xmath135 which is just @xmath134 considered with the @xmath1-action where @xmath23 acts as @xmath136 .
the split fibration @xmath137 associated with @xmath138 is exactly the functor @xmath139 assigning to each @xmath104 its ` structure ' @xmath1 . the canonical arrows @xmath140 are cartesian and a @xmath103-morphism @xmath141 is a pair consisting of a @xmath95-morphism @xmath131 and a @xmath142-morphism @xmath143 . the dual fibration of @xmath137 is the fibration @xmath144 which in our language corresponds to @xmath145 . 1 . a difference object @xmath104 is called _ inversive _ , if every @xmath107 is an automorphism of @xmath6 .
2 . a difference object
@xmath104 is _ almost - strict _
, if there exists a finite subgroup @xmath146 of @xmath147 such that for all @xmath148 , there exists a @xmath149 such that @xmath150 , i.e. , @xmath151 for some @xmath107 .
[ def - regular ] 1 .
a @xmath95-structure @xmath1 is _ regular _ , if for every @xmath152 , the map @xmath108 is bijective .
2 . a regular @xmath95-structure @xmath1 is _ full _ , if it is equipped with _
generalised conjugation _ in the sense that for any @xmath153 , there is an bijective assignment @xmath154 on @xmath1 which has the property that @xmath155 . in a given @xmath1-object @xmath6
, this element gets interpreted as a morphism satisfying @xmath156 intuitively , in an inversive structure , @xmath157 should be thought of as @xmath158 . as shahn majid pointed out , a regular @xmath95-object
is usually called a _
quandle _ in the literature . 1 .
if @xmath104 is inversive , then @xmath1 is regular .
2 . if @xmath104 is almost - strict , then @xmath1 is full .
[ fc ] if @xmath104 is inversive and almost - strict , then the group of automorphisms @xmath159 of @xmath6 generated by @xmath1 is finite - by - cyclic , i. e. , for any @xmath107 , we have an exact sequence @xmath160 { |(1)|{1 } & |(2)|{g } & |(3)|{\langle\sigma\rangle } & |(4)|{\langle\sigma\rangle } & |(5)|{1}\\ } ; \path[->,font=\scriptsize,>=to , thin ] ( 1 ) edge ( 2 ) ( 2 ) edge ( 3 ) ( 3 ) edge ( 4 ) ( 4 ) edge ( 5 ) ; \end{tikzpicture}\ ] ] clearly , @xmath146 can be thought of as a group with an operator @xmath161 and a morphism @xmath162 of inversive almost - strict objects gives rise to a homomorphism of groups with operators , or , equivalently , a commutative diagram @xmath163 { |(1)|{1 } & |(2)|{g } & |(3)|{\langle\sigma\rangle } & |(4)|{\langle\sigma\rangle } & |(5)|{1}\\ |(l1)|{1 } & |(l2)|{h } & |(l3)|{\langle t\rangle } & |(l4)|{\langle\sigma^\varphi\rangle } & |(l5)|{1}\\ } ; \path[->,font=\scriptsize,>=to , thin ] ( 1 ) edge ( 2 ) ( 2 ) edge ( 3 ) ( 3 ) edge ( 4 ) ( 4 ) edge ( 5 ) ( l1 ) edge ( l2 ) ( l2 ) edge ( l3 ) ( l3 ) edge ( l4 ) ( l4 ) edge ( l5 ) ( 2 ) edge ( l2 ) ( 3 ) edge ( l3 ) ( 4 ) edge ( l4 ) ; \end{tikzpicture}\ ] ] [ definvring ] every ( strong ) difference ring @xmath164 with @xmath1 finite and regular has an _ inversive closure _ with the following universal property .
there is an embedding @xmath165 such that every morphism @xmath166 to an inversive difference ring factors uniquely through @xmath167 .
the proof of @xcite can be lifted to our framework in spite of having to deal with non - commuting endomorphisms .
let @xmath168 , and let @xmath169 denote the bijection @xmath161 , for @xmath107 .
we know how to take inversive closures of strict difference rings , so let @xmath170 .
we must show that @xmath171 can be endowed with a @xmath1-structure .
indeed , if @xmath172 , and @xmath173 , then there is a positive @xmath71 such that @xmath174 .
we define @xmath175 and it can be checked that this is independent of the choice of @xmath71 .
we continue inductively , by letting @xmath176 and endowing it with @xmath1-structure for @xmath177 .
it is clear that @xmath178 is the inversive closure of @xmath164 .
clearly , for every @xmath179 , there exists a @xmath180 in the free semigroup generated by @xmath1 such that @xmath181 ( see [ diff - ops ] ) . as a consequence ,
every difference scheme @xmath104 can be dominated by an inversive difference scheme @xmath182 with the dual universal property , where the morphism is bijective at the level of points .
[ defspecsigma ] let @xmath164 be an object of a difference category over the category of commutative rings with identity .
we shall consider each of the following subsets of @xmath49 as locally ringed spaces with the zariski topology and the structure sheaves induced from @xmath49 : 1 .
@xmath183 , for any @xmath107 ; 2 .
@xmath184 ; 3 . @xmath185 . in discussions of induced topology
, we shall use the notation @xmath64 , @xmath186 , @xmath187 , @xmath188 , @xmath189 , @xmath190 for the traces of @xmath191 and @xmath192 on @xmath66 , @xmath193 , @xmath194 , respectively . to be more precise ,
let us denote @xmath195 , @xmath196 , @xmath197 and we define the sheaf @xmath198 . using the adjunction @xmath199
, we get a morphism @xmath200 , i.e. , for every @xmath201 open in @xmath6 , we get a map @xmath202 moreover , for each @xmath203 , the stalks @xmath204 and @xmath205 are both isomorphic to @xmath206 . for each @xmath107
we have the morphism of locally ringed spaces @xmath207 induced by @xmath23 , where @xmath208 .
for an arbitrary @xmath209 , @xmath210 , so we have @xmath211 and we get a diagram @xmath212 { x^\sigma & x\\ x^\sigma & x\\ } ; \path[->,font=\scriptsize,>=to , thin ] ( m-2 - 1 ) edge node[auto ] { $ i$ } ( m-2 - 2 ) edge node[left ] { $ { { { \vphantom{\sigma}}^{\lower0.0ex\hbox{\smash{${\phantom{a}\llap{$\scriptstylea$ } \kern.01ex\lower.75ex\hbox{$\smash{\mathring{}}$}}$}}}\kern-.1ex \sigma}}$ } ( m-1 - 1 ) ( m-2 - 2 ) edge node[right ] { $ { { { \vphantom{\sigma}}^{a}{\kern-.1ex\sigma}}}$ } ( m-1 - 2 ) ( m-1 - 1 ) edge node[auto ] { $ i$ } ( m-1 - 2 ) ; \end{tikzpicture}\ ] ] the ` restriction ' of the sheaf morphism @xmath213 to @xmath214 is the composite @xmath215 where the base change morphism ( bc ) is derived from the adjunction maps @xmath199 and @xmath216 as the composite @xmath217 in order to avoid cumbersome notation , we may write @xmath218 for @xmath219 , when considered on @xmath220 .
thus , the locally ringed space @xmath220 is equipped with a set of endomorphisms of locally ringed spaces @xmath221 which , by functoriality ( as in the case of general morphisms below ) , happens to be closed under ` conjugation ' , i.e. , has a @xmath95-structure . in other words
, @xmath222 is an object of the difference category over the opposite category of locally ringed spaces .
suppose we are given a morphism @xmath223 in the difference category over the category of commutative rings with identity .
let us write @xmath195 , @xmath196 , @xmath197 , @xmath224 , @xmath225 , @xmath226 and we let @xmath227 be the morphism of locally ringed spaces @xmath228 induced by @xmath12 , where @xmath229 .
one easily verifies that for any @xmath107 , @xmath230 , so @xmath231 , and we get a diagram @xmath212 { x & y\\ x^\sigma & y^t\\ } ; \path[->,font=\scriptsize,>=to , thin ] ( m-2 - 1 ) edge node[auto ] { $ i$ } ( m-1 - 1 ) edge node[below ] { $ { { { \vphantom{\varphi}}^{\lower0.0ex\hbox{\smash{${\phantom{a}\llap{$\scriptstylea$ } \kern.01ex\lower.75ex\hbox{$\smash{\mathring{}}$}}$}}}\kern-.1ex \varphi}}$ } ( m-2 - 2 ) ( m-2 - 2 ) edge node[right ] { $ j$ } ( m-1 - 2 ) ( m-1 - 1 ) edge node[auto ] { $ { { \vphantom{\varphi}}^{a}{\kern-.1ex\varphi}}$ } ( m-1 - 2 ) ; \end{tikzpicture}\ ] ] the ` restriction ' of the sheaf morphism @xmath232 to @xmath233 is the composite @xmath234 where the base change morphism ( bc ) is derived from the adjunction maps @xmath199 and @xmath235 as the composite @xmath236 in order to show that @xmath237 defines a morphism @xmath238 in the difference category over the opposite category of the locally ringed spaces , for each @xmath107 one must chase through the following diagram , where we write @xmath239 for ease of notation . @xmath127
\matrix(m)[matrix of math nodes , row sep=.6 cm , column sep=.6 cm , text height=1.5ex , text depth=0.25ex ] { & |(u4)| { x } & & |(u3)| { y } \\
|(u1)| { x^\sigma } & & |(u2)|{y^t } & \\ & |(d4)| { x } & & |(d3)| { y } \\ |(d1)| { x^\sigma } & & |(d2)|{y^t } & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( d1 ) edge node[left]{${{\vphantom{\sigma}}^{\lower0.0ex\hbox{\smash{${\phantom{a}\llap{$\scriptstylea$ } \kern.01ex\lower.75ex\hbox{$\smash{\mathring{}}$}}$}}}\kern-.1ex \sigma}$ } ( u1 ) edge node[below]{${{\vphantom{\varphi}}^{\lower0.0ex\hbox{\smash{${\phantom{a}\llap{$\scriptstylea$ } \kern.01ex\lower.75ex\hbox{$\smash{\mathring{}}$}}$}}}\kern-.1ex \varphi}$ } ( d2 ) edge node[auto]{$i$ } ( d4 ) ( d4 ) edge node[pos=0.25,left]{${{\vphantom{\sigma}}^{a}{\kern-.1ex\sigma}}$ } ( u4 ) edge node[pos=0.25 , below]{${{\vphantom{\varphi}}^{a}{\kern-.1ex\varphi}}$ } ( d3 ) ( u1 ) edge [ cross line ] node[pos=.8 , above]{${{\vphantom{\varphi}}^{\lower0.0ex\hbox{\smash{${\phantom{a}\llap{$\scriptstylea$ } \kern.01ex\lower.75ex\hbox{$\smash{\mathring{}}$}}$}}}\kern-.1ex \varphi}$ } ( u2 ) edge node[auto]{$i$ } ( u4 ) ( d2 ) edge [ cross line ] node[pos=0.75 , right]{${{\vphantom{\tau}}^{\lower0.0ex\hbox{\smash{${\phantom{a}\llap{$\scriptstylea$ } \kern.01ex\lower.75ex\hbox{$\smash{\mathring{}}$}}$}}}\kern-.1ex \tau}$ } ( u2 ) edge node[below]{$j$ } ( d3 ) ( u4 ) edge node[above]{${{\vphantom{\varphi}}^{a}{\kern-.1ex\varphi}}$ } ( u3 ) ( u2 ) edge node[pos=0.4,above]{$j$ } ( u3 ) ( d3 ) edge node[right]{${{\vphantom{\tau}}^{a}{\kern-.1ex\tau}}$ } ( u3 ) ; \end{tikzpicture}\ ] ] the crucial property that needs to be verified is that @xmath240 this is unpleasant to verify directly due to a number of base change morphisms , but it is straightforward to check the equivalent condition for the relevant morphisms obtained by adjunction .
indeed , let us write @xmath241 , @xmath242 , @xmath243 for the morphisms corresponding to @xmath244 , @xmath245 , @xmath246 by adjointness .
up to some canonical identifications , @xmath247 , @xmath248 and @xmath249 .
the condition @xmath250 is now readily verified , using the fact that @xmath251 to avoid the above discombobulation with the notation , we may choose to write @xmath227 in place of @xmath237 when it is clear from the context that we are referring to a morphism @xmath252 induced by @xmath253 .
[ specsigma ] we would like to state now that the functor @xmath254 mapping an object @xmath164 to the object @xmath255 , and a morphism @xmath256 to the morphism @xmath257 is a ` contravariant ' functor from the strong dual difference category of commutative rings with identity to the difference category of locally ringed spaces which respects the difference structure .
this statement can be made precise in the language of fibered categories . using the terminology from [ fibrediff ] ,
let @xmath258 denote the strong dual difference category over commutative rings with identity , and let @xmath259 denote the difference category of locally ringed spaces .
we have fibrations @xmath260 and @xmath261
. then @xmath254 defines a cartesian functor @xmath262 from the opposite fibration of @xmath263 to @xmath264 . when all @xmath193 are quasi - compact , ( e.g. when @xmath1 is finite ) , the target difference category is also strong .
[ diff - ops ] given a difference structure @xmath1 , we shall write @xmath159 for the semigroup generated by @xmath1 , which is the free semigroup generated by @xmath1 modulo the relations @xmath265 for @xmath100 .
we let the _ difference operators _ rig @xmath266 $ ] be the set of @xmath267-linear combinations of the elements of @xmath159 .
if @xmath164 is a difference ring , every @xmath268 $ ] can be thought of as a difference operator on @xmath26 via its natural action @xmath269 .
we have : 1 .
@xmath270 for @xmath107 ; 2 .
@xmath271 ; 3 .
@xmath272 ; 4 .
if @xmath273 , @xmath274 . for an element @xmath268 $ ] given by @xmath275 with @xmath276 , it will be convenient to write @xmath277 .
given a difference ring extension @xmath278 , the difference subring of @xmath32 generated by a set @xmath33 over @xmath26 is denoted by @xmath35_\sigma$ ] , and it equals @xmath279t]$ ] .
similarly , given a difference field extension @xmath280 , the difference subfield of @xmath25 generated by a set @xmath37 over @xmath38 is denoted by @xmath281 and it equals @xmath282t)$ ] .
let @xmath41 be an ideal in a difference ring @xmath164 .
we say that : 1 .
@xmath41 is a _ @xmath1-ideal _ if @xmath43 for every @xmath107 ; 2 .
@xmath41 is _
@xmath1-reflexive _ if @xmath283 for every @xmath107 ; 3 .
@xmath41 is _ @xmath1-well - mixed _ if @xmath44 implies @xmath45 for any @xmath107 ; 4 .
@xmath26 itself is well - mixed if the zero ideal is ; 5 .
@xmath41 is _
if for every @xmath107 , @xmath46 implies @xmath47 and @xmath28 are both in @xmath41 . for a set @xmath133 ,
we denote by @xmath284 the least @xmath1-perfect ideal containing @xmath133 , for which we have the following construction . for a subset @xmath32 of @xmath164 , let @xmath285\}$ ] .
starting with @xmath286 , let @xmath287'_\sigma$ ] .
it is clear that @xmath288 .
[ perfwm ] 1 . clearly , @xmath41 is @xmath1-perfect if and only if @xmath289 whenever @xmath290 for @xmath268\setminus\{0\}$ ] , and the notion of a perfect ideal is a natural generalisation of the notion of a radical ideal in the difference context .
if @xmath41 is @xmath1-perfect , then @xmath41 is @xmath1-well - mixed . indeed , if @xmath44 and @xmath107 , then @xmath291 so @xmath292 and thus @xmath45 .
if @xmath41 is @xmath1-well - mixed , clearly @xmath293\}$ ] . [ perf1 ] for @xmath294 , @xmath295 .
consequently , @xmath296 .
let @xmath297 and @xmath298 be @xmath1-invariant sets ( such as @xmath299 , @xmath300 ) . for @xmath301 , @xmath302 , there exist @xmath303 $ ] such that @xmath304_\sigma$ ] and @xmath305_\sigma$ ] .
by invariance of @xmath297 and @xmath298 , we conclude that @xmath306_\sigma$ ] .
if we take some @xmath268 $ ] such that @xmath307 divides @xmath308 ( @xmath309 , for example ) , we will have that @xmath310_\sigma$ ] and thus @xmath311'$ ] .
let @xmath312s$ ] and @xmath313t$ ] be the @xmath1-invariant closures of @xmath32 and @xmath133 .
we can write arbitrary @xmath314 , @xmath315 as @xmath316 , @xmath317 for some @xmath318 , @xmath319 , @xmath303 $ ] . in order to show that there exist @xmath320 $ ] such that @xmath321 divides @xmath322 , we may reduce to the case where @xmath273 .
then @xmath323 and @xmath324 suffice : @xmath325 therefore , @xmath326'_\sigma=(st)_1 $ ] and from the property established above for invariant sets , @xmath327 [ perf2 ] let @xmath32 and @xmath133 be subsets of a difference ring @xmath164 . 1 . [ jen ] @xmath328 for @xmath329 ; 2 . [ dva ] @xmath330 . as a consequence , @xmath331 .
the inclusion ( [ jen ] ) immediately follows from the fact that @xmath332_\sigma$ ] is contained in both @xmath333_\sigma$ ] and @xmath334_\sigma$ ] . for ( [ dva ] ) , if @xmath335 , then , using [ perf1 ] @xmath336 so @xmath337 . [ perfintrs ]
let @xmath164 be a difference ring and let @xmath41 be a @xmath1-perfect ideal .
then @xmath338 suppose @xmath41 is a @xmath1-perfect difference ideal and @xmath339 .
it suffices to find a @xmath340 such that @xmath341 .
by zorn s lemma , there exists a maximal @xmath1-perfect ideal @xmath342 , @xmath343 .
suppose that @xmath344 is not prime , i. e. there exist @xmath345 , @xmath346 , @xmath347 , @xmath348 . from maximality of @xmath344 , @xmath349 and @xmath350 .
but then @xmath351 which is a contradiction .
thus @xmath344 is prime and for every @xmath107 , @xmath352 since @xmath344 is @xmath1-perfect .
[ wknulsatz ] @xmath353 if and only if @xmath354 . with some extra care , a similar string of results to the above can be established even in the case where @xmath1 is just a set of monomorphisms on @xmath26 , without @xmath164 being a difference ring , but we decided to abandon this generality since it did not contribute to the geometrical picture we are describing .
the results are as follows .
if we denote by @xmath355 and @xmath356 the @xmath1-invariant closures of @xmath32 and @xmath133 , from the proof of [ perf1 ] we obtain @xmath357 .
the analogue of [ perf2 ] is @xmath358 , and the statement of [ perfintrs ] is the same .
the proof of [ perfintrs ] is modified as follows .
if @xmath346 and @xmath344 is perfect , then [ perfwm ] implies that @xmath344 is well - mixed so @xmath359 for any @xmath360 , and we have : @xmath361 and we obtain the same conclusion .
[ qcompact ] let @xmath164 be a difference ring , and let @xmath51 .
then @xmath362 is quasi - compact .
it is clear from [ wknulsatz ] that @xmath363 , @xmath364 , cover @xmath362 if and only if @xmath365 , which holds if and only if @xmath366 for some finite @xmath367 . clearly , if @xmath115 is a difference ring with @xmath1 _ finite _ , and @xmath368 , [ qcompact ] allows us to conclude that @xmath369 is quasi - compact .
let @xmath115 be a difference ring , and let @xmath368 .
the _ multiplicative system of generalised powers _ of @xmath370 is the set @xmath371\}$ ] .
the _ saturated multiplicative system _ associated with @xmath370 is the set @xmath372 in view of [ wknulsatz ] , @xmath373 if and only if @xmath374 .
lemma [ perf2 ] shows that @xmath375 is a multiplicative set and we define the difference ring localisations @xmath376 and @xmath377 .
[ wmaff ] let @xmath115 be a well - mixed difference ring ( or even ring with a set of monomorphisms ) , @xmath368 . 1 . [ jennn ]
both canonical morphisms @xmath378 are injective .
if moreover @xmath369 is quasi - compact , we have the following . 1 . [ dvaaa ] for each @xmath379 , there exist @xmath380 such that @xmath381 and there is a section @xmath382 such that @xmath383 for @xmath384 .
[ dvaipol ] let @xmath379 .
the ideal @xmath385 is well - mixed .
[ triii ] suppose @xmath379 and @xmath386 such that @xmath387 .
then there is a @xmath388 such that @xmath389 ( on @xmath369 ) .
[ cetriii ] let @xmath379 and @xmath386 .
there exist @xmath388 and @xmath27 such that @xmath390 .
[ pettt ] let @xmath391 such that @xmath392 .
then there exists a @xmath393 $ ] such that @xmath394 ( on @xmath214 ) .
[ sesttt ] there exist canonical injections @xmath395 inducing an isomorphism of difference schemes @xmath396 .
( [ jennn ] ) recall that , for an element @xmath397 , and @xmath398 , @xmath399 .
if @xmath400 , then for every @xmath401 , @xmath402 in @xmath403 , i. e. , there exists an @xmath404 , @xmath405 in @xmath116 .
let @xmath406 .
thus , for every @xmath401 , @xmath407 witnesses @xmath408 . in other words ,
@xmath409 so @xmath410 .
note that the assumption that @xmath116 is @xmath1-well - mixed implies that @xmath411 is a @xmath1-well - mixed ideal . by [ wknulsatz ]
, we conclude that @xmath412 and by [ perfwm ] , there exists a @xmath268 $ ] such that @xmath413 .
thus , @xmath414 so @xmath402 in @xmath415 . had we started with @xmath416 and @xmath417 , we would have finished with an equality in @xmath418 , so this not only shows that @xmath419 is injective but also that the composite map is injective .
( [ dvaaa ] ) by definition , @xmath420 , where @xmath421 is the inclusion . unravelling the definition of the inverse image sheaf in the case of an inclusion ( @xcite , 0 , 3.7.1 )
we get that if @xmath379 , then for every @xmath384 there exists a @xmath422 with @xmath423 and a section @xmath424 such that @xmath425 for every @xmath426 . using the assumption that @xmath369 is quasi - compact
, we find a finite subcovering @xmath381 of @xmath427 and @xmath428 is represented by a section @xmath429 which glue together to form a section @xmath382 , as required .
( [ dvaipol ] ) suppose @xmath430 . with the notation of ( [ dvaaa ] ) ,
let @xmath431 for some @xmath432 and we have that for every @xmath433 , @xmath434 on @xmath363 . by ( [ jennn ] ) , there exists a @xmath435 such that @xmath436 . by well - mixedness
, we get that @xmath437 and since @xmath438 is invertible on @xmath439 , we conclude that @xmath440 on @xmath439 ( and therefore on @xmath441 ) for every @xmath433 , so @xmath442 .
( [ triii ] ) assume that @xmath443 is represented by an @xmath444 as in ( [ dvaaa ] ) and that @xmath60 falls in some @xmath441
. then @xmath445 is represented by @xmath446 for some @xmath432 , and @xmath447 in @xmath403 , so there exists a @xmath448 such that @xmath449 in @xmath116 . alternatively , since @xmath450 is invertible on @xmath439 , we get that that @xmath451 on @xmath439 . by replacing @xmath443 by @xmath452
, we may assume @xmath453 .
for any @xmath454 , on @xmath455 , @xmath456 ( represented by @xmath457 ) coincides with @xmath458 ( represented by @xmath459 ) so there exists an @xmath460 such that @xmath461 in @xmath116 . since @xmath462 is invertible on @xmath463
, we get that @xmath464 on @xmath463 .
letting @xmath465 , @xmath466 on @xmath467 so @xmath468 on @xmath369 .
( [ cetriii ] ) if @xmath469 , apply ( [ triii ] ) to @xmath470 .
( [ pettt ] ) claim ( [ triii ] ) provides , for each @xmath401 a witness to @xmath471 .
thus , @xmath472 so we conclude that @xmath473 , and , since ( [ dvaipol ] ) guarantees that @xmath474 is well - mixed , according to [ perfwm ] there exists a @xmath393 $ ] with @xmath475 .
( [ sesttt ] ) to see that @xmath476 is injective , note that the composite of the given canonical maps is injective by ( [ jennn ] ) while the injectivity of the right map is guaranteed by ( [ pettt ] ) .
consider @xmath477 , given by @xmath478 .
this map is surjective since for every @xmath107 and @xmath479 , the ideal @xmath480 maps to @xmath60 . to see that it is injective , fix a @xmath481 and take some @xmath482 such that @xmath483 . given @xmath484 , by ( [ cetriii ] ) , there exist @xmath388 and @xmath27 such that @xmath390 on @xmath485 .
thus @xmath486 if and only if @xmath487 so @xmath488 is uniquely determined by @xmath60 . to see that @xmath433 moreover defines an isomorphism of locally ringed spaces
, it suffices to note that if @xmath489 , @xmath490 is an isomorphism . using [ qcompact ]
, the conditional statements ( [ dvaaa])([sesttt ] ) become unconditional if we replace every occurrence of @xmath214 and @xmath369 by @xmath491 and @xmath362 , respectively .
[ univgs ] a commutative diagram of well - mixed difference rings @xmath212 { \bar{a } & \bar{b}\\ a & b\\ } ; \path[->,font=\scriptsize,>=to , thin ] ( m-2 - 1 ) edge node[auto ] { $ i$ } ( m-1 - 1 ) edge node[below ] { $ \alpha$ } ( m-2 - 2 ) ( m-2 - 2 ) edge node[right ] { $ j$ } ( m-1 - 2 ) ( m-1 - 1 ) edge node[above ] { $ \beta$ } ( m-2 - 2 ) ; \end{tikzpicture}\ ] ] is uniquely completed to a square by @xmath492 , i.e. , @xmath493 .
indeed , this follows from @xmath212 { \bar{\bar{a } } & \\
\bar{a } & \bar{b}\\ a & b\\ } ; \path[->,font=\scriptsize,>=to , thin ] ( m-1 - 1 ) edge node[above]{$\bar{\beta}$ } ( m-2 - 2 ) ( m-3 - 1 ) edge node[auto ] { $ i$ } ( m-2 - 1 ) edge node[below ] { $ \alpha$ } ( m-3 - 2 ) ( m-3 - 2 ) edge node[right ] { $ j$ } ( m-2 - 2 ) ( m-2 - 1 ) edge node[above ] { $ \beta$ } ( m-3 - 2 ) edge node[above]{$\bar{\alpha}$ } ( m-2 - 2 ) edge node[left]{$\bar{\imath}$ } ( m-1 - 1 ) ; \end{tikzpicture}\ ] ] and the fact that @xmath494 and @xmath495 .
of course , @xmath496 is an isomorphism .
[ intrmed ] the proof of [ wmaff]([sesttt ] ) in fact shows that for any difference ring @xmath497 such that @xmath498 , we obtain an isomorphism of difference schemes @xmath499 .
this observation will prove invaluable for proving certain finiteness properties later on .
[ embedcataff ] let @xmath164 and @xmath500 be well - mixed difference rings whose spectra are quasi - compact and let @xmath501 be the ring of global sections .
there is a natural isomorphism @xmath502 where the first @xmath503 is in the difference category over the category of locally ringed spaces and the second is in the category of difference rings . in other words , the functor @xmath90 is left adjoint to @xmath254 .
let @xmath504 , @xmath505 , @xmath506 , @xmath507 , and write @xmath508 and @xmath509 for the inclusions from [ wmaff]([sesttt ] ) .
we know by [ wmaff]([sesttt ] ) that @xmath510 is an isomorphism @xmath511 and that @xmath512 is an isomorphism @xmath513 .
we have already shown in [ specsigma ] that for every @xmath514 , the pair @xmath227 is a morphism @xmath515 , and we can produce a morphism @xmath516 as the composite @xmath517 .
suppose that we are given @xmath518 a morphism of locally ringed spaces . by definition , @xmath519 , so taking global sections we get @xmath520 and precomposing with the inclusion @xmath521 from [ wmaff]([jennn ] ) we obtain a morphism @xmath522 , @xmath523 .
we claim that @xmath524 . by assumption , for @xmath525
, @xmath526 is a local homomorphism of local rings .
writing @xmath527 , @xmath528 , the diagram @xmath529 { s & \bar{s } & & \bar{r } \\ s_{y } & & r_{x } & \\ & \bar{s}_{\bar{y } } & & \bar{r}_{\bar{x}}\\ } ; \path[->,font=\scriptsize,>=to , thin ] ( m-1 - 1 ) edge node[below ] { $ i$ } ( m-1 - 2 ) edge ( m-2 - 1 ) edge [ bend left=20 ] node[pos=0.6,above]{$\varphi$}(m-1 - 4 ) ( m-1 - 2 ) edge node[below ] { $ \bar{\varphi}$ } ( m-1 - 4 ) edge ( m-3 - 2 ) ( m-2 - 1 ) edge node[pos=0.7,above ] { $ \theta^\sharp_x$ } ( m-2 - 3 ) edge node[left ] { $ \tilde{\jmath}^\sharp_{\bar{y}}$ } ( m-3 - 2 ) ( m-1 - 2 ) edge ( m-2 - 1 ) ( m-1 - 4 ) edge ( m-2 - 3 ) edge ( m-3 - 4 ) ( m-2 - 3 ) edge node[pos=0.35,right ] { $ \tilde{\imath}^\sharp_{\bar{x}}$ } ( m-3 - 4 ) ( m-3 - 2 ) edge node[auto ] { $ \bar{\varphi}_{\bar{x}}$ } ( m-3 - 4 ) ( m-2 - 1 ) edge [ dashed ] node[above ] { $ \varphi_{\bar{x}}$ } ( m-3 - 4 ) ; \end{tikzpicture}\ ] ] is commutative , where the top parallelogram follows from the definition of stalks and the discussion after [ defspecsigma ] which allows the identifications @xmath530 and @xmath531 , and the vertical arrows are the localisation maps .
it follows that @xmath532 , i.e. , @xmath533 and that , writing @xmath534 for the morphism obtained by localisation of @xmath12 at @xmath535 , @xmath536 . since a morphism is characterised by its action on stalks , we conclude that @xmath524 , as required .
[ rem - embedcat ] it is worth remarking that , unlike in the algebraic case , @xmath254 and @xmath90 do not determine an equivalence of categories , only a weaker notion which we might dub temporarily ` an embedding of categories ' for the lack of a reference : the unit of the adjunction @xmath537 is a natural isomorphism , while the counit @xmath538 is only a natural injection by [ wmaff]([sesttt ] ) . 1 .
an _ affine difference scheme _ is an object @xmath539 of the difference category over the category of locally ringed spaces , which is isomorphic to some @xmath193 for some well - mixed @xmath164 .
2 . a _ difference scheme _
is an object @xmath539 of the difference category over the category of locally ringed spaces , which is locally an affine difference scheme .
3 . a _ morphism of difference schemes _
@xmath540 is just a morphism in the difference category over the category of locally ringed spaces .
given a difference scheme @xmath539 and @xmath107 , we define a locally ringed space @xmath541 , together with the topology and structure sheaf induced from @xmath83 .
since @xmath542 , clearly @xmath543 and @xmath544 is a strict difference scheme .
we have the following properties : 1 .
2 . for every @xmath546
, there is a unique element @xmath547 such that @xmath548 is a morphism of difference schemes in the strict sense .
if @xmath549 is a morphism , then for every @xmath152 there exists a @xmath550 such that @xmath551 is a morphism of difference schemes in the strict sense .
[ fulfinorb ] if @xmath104 is a full difference scheme with @xmath1 finite , then the @xmath159-orbit of any @xmath525 is finite of bounded length . as a direct consequence ,
the morphisms @xmath552 discussed above are bijective at the level of points .
indeed , let @xmath525 , and consider the set @xmath553 .
suppose @xmath554 for some @xmath555 . for an arbitrary @xmath556 ,
take a @xmath557 with @xmath558 .
then @xmath559 so we conclude that @xmath560 .
this is consistent with our intuition that a full ( or almost - strict ) difference scheme is a finitary perturbation of a strict difference scheme . [ embedcat ] the ` global sections ' functor @xmath90 is left adjoint to the contravariant functor @xmath254 from the category of well - mixed difference rings to the category of difference schemes . for any difference scheme @xmath104 and any well - mixed difference ring @xmath500 ( with @xmath133 finite ) , @xmath561 a _ quasi - coherent sheaf _
@xmath562 on a difference scheme @xmath539 is a quasi - coherent sheaf @xmath563 on the locally ringed space @xmath83 so that @xmath562 is an @xmath564-module in a natural sense .
difference subscheme _ of @xmath539 is the locally ringed space equipped with @xmath1-structure associated with a quasi - coherent sheaf @xmath565 of @xmath564-ideals .
let us give definitions of various versions of reduced schemes associated with a difference scheme @xmath539 . in each case ,
the underlying topological space is that of @xmath6 , and we specify the defining sheaf of ideals by giving its stalks : 1 .
@xmath566 , the _ reduced subscheme _ of @xmath6 , defined by @xmath567 , where @xmath568 is the nilradical of @xmath88 ; 2 .
@xmath569 , the _ well - mixed subscheme _ of @xmath6 , defined by @xmath570 , where @xmath571 is the least well - mixed ideal of @xmath88 ; 3 .
@xmath572 , the _ reflexively reduced subscheme _ of @xmath6 , defined by @xmath573 , where @xmath574 is the reflexive closure of @xmath575 in @xmath88 ; 4 .
@xmath576 , the _ perfectly reduced subscheme _ of @xmath6 , defined by @xmath577 , where @xmath578 is the perfect closure of @xmath575 in @xmath88 .
we shall say that @xmath6 is _ perfectly reduced _ if @xmath579 , and similarly for other notions above .
let @xmath104 be a difference scheme . 1 .
we say that @xmath6 is irreducible ( resp .
connected ) if its underlying topological space is .
we say that @xmath6 is integral ( resp . transformally integral ) if it is irreducible and reduced ( resp .
perfectly reduced ) . 1 .
let @xmath580 be a difference scheme .
the category of _ @xmath580-difference schemes _ has difference scheme morphisms @xmath581 as objects ( considered as structure maps ) , while a morphism between @xmath581 and @xmath582 is a commutative diagram of difference scheme maps @xmath583 { |(11)|{(x,\sigma ) } & & |(12)|{(y , t)}\\[.7 cm ] & |(22)|{(s,\sigma_0 ) } & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( 11 ) edge ( 12 ) edge(22 ) ( 12 ) edge ( 22 ) ; \end{tikzpicture}\ ] ] 2 .
let @xmath584 be a difference ring .
the category of _ @xmath584-difference schemes _ consists of difference schemes which are locally of the form @xmath485 , for a difference @xmath584-algebra @xmath115 .
morphisms are required to locally preserve the @xmath584-algebra structure . 1 .
let @xmath104 be a difference scheme and @xmath585 a difference field .
a _ @xmath585-rational point _ of @xmath104 is a morphism @xmath586 .
when @xmath587 , this means we have a point @xmath588 and a local map @xmath589 , where @xmath590 is the image of @xmath12 in @xmath1 by the difference structure map @xmath591 .
alternatively , we have an inclusion @xmath592 .
2 . let @xmath104 be a difference scheme over a difference field @xmath593 and let @xmath594 .
the set of _
@xmath585-rational points _ of @xmath104 , henceforth denoted by @xmath595 , is the set of all @xmath593-morphisms @xmath596 .
modulo a small technical condition , fibre products exist in the category of difference schemes .
[ fibreprod ] let @xmath597 , @xmath598 , be morphisms of difference schemes .
if the difference structure map @xmath599 is surjective , then the fibre product @xmath600 exists , i.e. , the functor @xmath601 from the opposite category of @xmath602-schemes to sets is representable . indeed ,
suppose we have morphisms @xmath603 and @xmath604 , with @xmath599 surjective . our task is to show the existence of a cartesian square :
@xmath605 { & |(p)|{x\times_s y } & \\
|(1)|{x } & & |(2)| { y}\\ & |(h)|{s } & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( p ) edge ( 1 ) edge ( 2 ) ( 1 ) edge ( h ) ( 2 ) edge ( h ) ; \end{tikzpicture}\ ] ] standard reductions allow us to assume that @xmath6 , @xmath134 , @xmath32 are affine , difference spectra of @xmath606 , @xmath607 , @xmath608 , respectively .
assume we have a diagram @xmath609 { & |(p)|{(z , t ) } & \\
|(1)|{(x_1,\sigma_1 ) } & & |(2)| { ( x_2,\sigma_2)}\\ & |(h)|{(s,\sigma ) } & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( p ) edge ( 1 ) edge ( 2 ) ( 1 ) edge ( h ) ( 2 ) edge ( h ) ; \end{tikzpicture}\ ] ] using [ embedcat ] , we obtain @xmath610 { & |(p)|{(h^0(z),t ) } & \\
|(1)|{(\bar{a}_1,\sigma_1 ) } & & |(2)|{(\bar{a}_2,\sigma_2)}\\ & |(h)|{(c,\sigma ) } & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( h ) edge ( 1 ) edge ( 2 ) ( 1 ) edge ( p ) ( 2 ) edge ( p ) ; \end{tikzpicture}\ ] ] writing @xmath611 , by the universal property of tensor products as well as the induced difference structure map @xmath612 , we get a unique morphism @xmath613 . since @xmath614 is well - mixed
, we obtain a unique morphism @xmath615 .
thus @xmath616 can play the role of @xmath600 .
extra care will be needed when dealing with products in our subsequent work in order to avoid the following undesirable situations . 1 . even in the category of ordinary difference schemes ,
the ( fibre ) product can be @xmath617 .
that is the case when @xmath618 and @xmath619 are _ incompatible _ , so we can not find a @xmath620 which admits a morphism to @xmath618 and @xmath619 at the same time ( see @xcite for examples ) .
if we omit the surjectivity assumption from [ fibreprod ] , the product need not exist .
for example , if @xmath104 is a difference scheme with @xmath621 , pick @xmath622 and consider the morphisms @xmath623 for @xmath598 .
the product @xmath624 will not exist in general .
[ glsprod ] suppose we have morphisms of well - mixed difference rings @xmath625 and @xmath626 , let @xmath627 , @xmath628 and let @xmath629 be as above .
then we have an isomorphism of difference schemes @xmath630 in the diagram @xmath631 { \hspace{6em } & & |(a)|{\overline{(\bar{a}\otimes_c\bar{b})_w } } & & \hspace{5em } \\[2em ] & & |(b)|{\overline{(a\otimes_cb)_w } } & & \\[2em ] & |(d)|{\bar{a}\otimes_c\bar{b } } & & |(c)|{(\bar{a}\otimes_c\bar{b})_w } & \\
|(f)|{\bar{a } } & & & & |(g)|{\bar{b}}\\ & |(h)|{a\otimes_cb } & & |(e)|{(a\otimes_cb)_w } & \\
|(i)|{a } & & & & |(j)|{b}\\ & & |(k)|{c } & & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( k ) edge ( i ) edge ( f ) edge ( g ) edge ( j ) ( i ) edge node[left]{$\iota_1 $ } ( f ) edge node[above]{$\epsilon_1 $ } ( h ) edge[bend left=0 ] node[above]{$\alpha_1 $ } ( e ) ( j ) edge node[pos=0.6,below]{$\epsilon_2 $ } ( h ) edge node[pos=0.5,above]{$\alpha_2 $ } ( e ) edge node[right]{$\iota_2 $ } ( g ) ( f ) edge[bend left=10 ] node[above]{$\bar{\alpha}_1 $ } ( b ) edge node[below]{$\epsilon_1'$}(d ) edge[bend left=10 ] ( a ) ( h ) edge node[pos=0.5,right]{$\varphi$ } ( d ) edge node[above]{$\pi$ } ( e ) ( g ) edge node[pos=0.6,below]{$\epsilon_2'$ } ( d ) edge[bend right=10 ] node[right]{$\bar{\alpha}_2 $ } ( b ) edge[bend right=10 ] ( a ) edge ( c ) ( e ) edge[dashed ] node[pos=0.4,right]{$\varphi_w$ } ( c ) edge[dashed ] node[pos=0.65,left]{$\gamma$ } ( b ) ( d ) edge[bend left=0 ] node[pos=0.5,right]{$\psi$ } ( b ) edge node[above]{$\pi'$ } ( c ) ( c ) edge[dashed ] node[right]{$\delta$ } ( a ) edge[dashed ] node[left]{$\psi_w$ } ( b ) ( b ) edge[dashed ] node[pos=0.4,left]{$\bar{\varphi}_w$ } ( a ) ; \end{tikzpicture}\ ] ] the morphism @xmath12 is obtained from the initial setup by using [ wmaff]([sesttt ] ) and the universal property of tensor products , and @xmath632 is obtained from it by the universal property of passing to the well - mixed quotient .
the morphisms @xmath633 and @xmath634 are obtained by precomposing with the quotient @xmath635 , and @xmath636 and @xmath637 result from them by functoriality of passing to global sections .
we define @xmath638 via the universal property of tensoring , and we use the universal property of well - mixed quotients to get @xmath639 from it . finally , by passing to global sections we derive @xmath640 from @xmath632 .
we wish to show that the diagram commutes , with particular emphasis on the dashed part .
the only commutativity relations which are not a priori clear are @xmath641 and @xmath642 . in order to show the first relation , since @xmath643 is onto , it suffices to verify @xmath644 , i.e. , that @xmath645 .
since @xmath646 is the unique morphism @xmath647 satisfying @xmath648 , @xmath598 , the above follows from the construction since @xmath649 . for the second relation ,
since @xmath650 is onto , it suffices to show that @xmath651 .
since @xmath652 is the unique morphism @xmath647 satisfying @xmath653 , @xmath598 , it is enough to verify that @xmath654 , but say the first one these follows by applying [ univgs ] to @xmath116 , @xmath655 , @xmath656 and @xmath657 .
from the dashed part of the above diagram , we deduce : @xmath658 { { { \rm spec}}^{\sigma_{\times}}({\overline{(\bar{a}\otimes_c\bar{b})_w } } ) & { { \rm spec}}^{\sigma_{\times}}({\overline{(a\otimes_cb)_w}})\\ { { { \rm spec}}^{\sigma_{\times}}((\bar{a}\otimes_c\bar{b})_w ) } & { { { \rm spec}}^{\sigma_{\times}}((a\otimes_cb)_w } ) \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( m-1 - 1 ) edge ( m-1 - 2 ) edge ( m-2 - 1 ) ( m-2 - 1 ) edge ( m-2 - 2 ) ( m-1 - 2 ) edge ( m-2 - 2 ) edge ( m-2 - 1 ) ; \end{tikzpicture}\ ] ] the vertical arrows are isomorphisms by [ wmaff]([sesttt ] ) and thus the diagonal is also an isomorphism .
when working in a relative setting over a base @xmath580 , it is natural to think of a morphism @xmath581 as a _ family _ of difference schemes parametrised by parameters from @xmath32 .
we make the notion of a _
fibre _ of the morphism precise .
[ d : fibre ] let @xmath581 be a morphism of difference schemes , let @xmath585 be a difference field and let @xmath659 .
the _ fibre _
@xmath660 is the @xmath585-difference scheme obtained by base change via the morphism @xmath661 , @xmath662 let @xmath297 be a property of difference schemes .
if @xmath581 is a difference scheme over a given base , we shall say that @xmath6 is _ geometrically _
@xmath297 , if every base change of @xmath6 has the property @xmath297 .
if @xmath584 is a difference ring and @xmath663 is a @xmath95-morphism , the _ difference polynomial ring _
@xmath664_\sigma$ ] in @xmath665 variables over @xmath584 is defined as the polynomial ring @xmath666,\ ] ] where @xmath159 is the semigroup generated by @xmath1 defined in [ diff - ops ] , together with the unique @xmath1-structure whereby a @xmath107 acts as @xmath667 on @xmath26 and maps @xmath668 to @xmath669 .
[ finsigmatype ] let @xmath584 be a difference ring . 1 .
an @xmath584-algebra @xmath602 is of _ finite @xmath1-type _ if it is an equivariant quotient of some difference polynomial ring @xmath664_\sigma$ ] .
equivalently , there exist elements @xmath670 such that @xmath671_\sigma$ ] .
an @xmath584-difference scheme @xmath104 is of _ finite @xmath1-type _ , or of _ finite transformal type _ if it is a finite union of affine difference schemes of the form @xmath672 , where @xmath602 is of finite @xmath1-type over @xmath584 .
[ finstypmor ] a morphism @xmath673 is _ of finite @xmath1-type _
if @xmath134 is a finite union of open affine subsets @xmath674 such that for each @xmath433 , @xmath675 is of finite @xmath1-type over @xmath676 .
4 . a morphism @xmath677 is _ integral _ ( resp . _ finite _ ) if @xmath134 is a finite union of open affine subsets @xmath674 such that for each @xmath433 , @xmath675 is @xmath678 , where @xmath679 is integral ( resp .
finite ) over @xmath680 .
a morphism is _ transformally finite _ if it is integral and of transformally finite type .
a morphism @xmath677 is _ quasi - finite _ if if is of finite transformal type and for every @xmath681 , the fibre @xmath682 is finite over @xmath683 .
[ fullfintyp ] let @xmath602 be an @xmath584 algebra , with @xmath1 full , and let @xmath663 be the corresponding difference structure morphism .
the following statements are equivalent : 1 .
@xmath602 is of finite @xmath1-type over @xmath584 ; 2 .
@xmath78 is of finite @xmath23-type over @xmath684 for all @xmath107 ; 3 .
@xmath78 is of finite @xmath23-type over @xmath684 for some @xmath107 .
writing @xmath685 for the bijection @xmath686 , we have that @xmath687 suppose @xmath688_\sigma$ ] for some tuple @xmath47 .
let @xmath689 . by the above generalised conjugation relation ,
we conclude that for any @xmath148 , @xmath690 and that indeed @xmath691_\sigma$ ] for any @xmath107 .
[ fintyp ] suppose @xmath504 and @xmath505 are two affine difference schemes of finite transformal type over some difference field @xmath593 and let @xmath692 be a morphism .
we can not conclude that @xmath370 is the spectrum of some morphism @xmath693 . on the other hand , by [ embedcat ]
, we have a map @xmath694 , and even though @xmath70 may not be of finite @xmath1-type over @xmath695 , by [ intrmed ] , we can get @xmath370 as the spectrum of the induced map @xmath696 $ ] , which is a map of algebras of finite transformal type over @xmath695 . therefore , any given commutative diagram of affine difference schemes of finite transformal type can be assumed to arise from a ( dual ) commutative diagram of difference rings of finite transformal type ( over a given base ) .
moreover , this shows that a morphism between affine difference schemes of finite transformal type is of finite transformal type , justifying [ finsigmatype]([finstypmor ] ) .
any base change of a morphism of finite transformal type is again of finite transformal type .
suppose that we have a diagram @xmath605 { & |(p)|{x\times_s y } & \\
|(1)|{x } & & |(2)| { y}\\ & |(h)|{s } & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( p ) edge ( 1 ) edge ( 2 ) ( 1 ) edge ( h ) ( 2 ) edge ( h ) ; \end{tikzpicture}\ ] ] where @xmath697 is of finite @xmath1-type and reduce to the affine case with notation from the proof of [ fibreprod ] . in view of [ fintyp ] and [ glsprod ]
, we can find a difference ring @xmath698 of finite @xmath1-type over @xmath10 such that @xmath699 can be realised as @xmath700 , where @xmath701 and this is clearly of finite @xmath23-type over @xmath702 .
an easy corollary of the proposition is that if both @xmath697 and @xmath703 are of finite transformal type , then @xmath704 is again of finite transformal type .
thus the _ category of difference scheme of finite transformal type over a given base has fibre products .
_ the following result shows that the various finiteness properties of morphisms are very close to being independent of the choice of an open affine covering in [ finsigmatype ] . on the other hand , the notions of having _
almost _ ( with terminology from ( * ? ? ?
* section [ local - properties ] ) ) the corresponding finiteness property we can derive from it are intrinsic .
let @xmath705 be a homomorphism of well - mixed difference rings , let @xmath706 be an @xmath707-module and let @xmath708 be such that @xmath709 . 1 .
[ jj ] if each @xmath710 is ( algebraically ) finite over @xmath711 , then @xmath706 is almost finite over @xmath707 .
[ dd ] if each @xmath712 is integral over @xmath713 , then @xmath714 is almost integral over @xmath116 .
[ tt ] if each @xmath715 is of finite @xmath23-type over @xmath711 , then @xmath716 is almost of finite @xmath23-type over @xmath707 .
[ cc ] if each @xmath715 is @xmath23-finite over @xmath711 , then @xmath716 is almost @xmath23-finite over @xmath707 .
( [ jj ] ) by assumption , for every @xmath433 there exists a finite free @xmath116-module @xmath717 and a morphism @xmath718 whose localisation @xmath719 is surjective .
we can lift the operator @xmath23 to @xmath717 by freeness , and thus achieve that @xmath720 be a morphism of @xmath707-modules .
let @xmath721 and let @xmath722 . by construction , for every @xmath433 , @xmath723 is surjective so , in particular , for every @xmath479 , @xmath724 is surjective .
thus , by ( * ? ? ?
* [ locwksurj ] ) , @xmath725 is almost surjective and @xmath726 is almost finite over @xmath116
. we can give a more direct proof of the above as follows . by assumption
, there is an @xmath727 such that for every @xmath728 , @xmath729 generate @xmath730 , for some @xmath731 and @xmath732 $ ] .
let @xmath733 be generated by @xmath734 .
then , for every @xmath735 there exists a @xmath736 $ ] such that for every @xmath433 , @xmath737
. let @xmath738_w\}$ ] , which is clearly a well - mixed difference ideal . then , since @xmath739 , we have that @xmath740 , so @xmath741 and we conclude that @xmath742_w$ ] .
( [ tt ] ) we use the template from the first proof of ( 1 ) . by assumption , for every @xmath433 there exists a @xmath23-polynomial @xmath707 algebra @xmath743 and a morphism @xmath744 such that @xmath745 is surjective .
let @xmath746 .
then the associated morphism ( which is in particular a morphism of @xmath707-modules ) @xmath747 has the property that for every @xmath479 , the localisation @xmath748 is surjective and ( * ? ? ?
* [ locwksurj ] ) implies that @xmath714 is almost of finite @xmath23-type over @xmath116 .
a difference ring @xmath164 is _ ritt _ if it has the ascending chain condition on @xmath1-perfect ideals , or equivalently , if @xmath194 is topologically noetherian .
the following results are known ( @xcite ) in the strict difference case .
[ strictritt ] 1 .
every difference ring of finite @xmath23-type over a ritt difference ring is ritt .
2 . difference schemes of finite @xmath23-type over a ritt difference ring are topologically noetherian .
3 . every open subset of a difference scheme of finite @xmath23-type over a ritt difference ring if of finite @xmath23-type . [ strictcomp ]
let @xmath42 be a ritt difference ring .
if @xmath41 is @xmath23-perfect , we have a unique irredundant decomposition @xmath749 with @xmath750 .
in other words , @xmath751 . by generalising techniques of @xcite , it can be shown that if @xmath602 is of finite @xmath1-type over a ritt difference ring @xmath584 , then @xmath602 is again ritt , i.e. , @xmath752 is topologically noetherian .
we are interested , however , in the study of @xmath672 , which might be unfathomable using these techniques , and that is why we restrict our attention to rings with full difference structures ( which notably includes the almost - strict case ) .
let @xmath602 be a full algebra of @xmath1-finite type over a ritt difference ring @xmath753 , and @xmath1 finite .
then @xmath672 is topologically noetherian and @xmath32 has an ascending chain condition on ideals which are perfect with respect to any @xmath107 . by [ fullfintyp ]
, @xmath78 is of finite @xmath23-type over @xmath26 for every @xmath107 .
thus @xmath754 is clearly noetherian by [ strictritt ] .
working with generalised difference schemes , we obtain an analogue of [ strictcomp ] , as well as identify a new phenomenon where difference schemes can decompose topologically but not structurally .
we shall return to the study of this behaviour in ( * ? ? ?
* section [ sect : bi - fib ] ) .
suppose @xmath164 is a full difference ring such that each @xmath42 is ritt .
let @xmath755 and suppose @xmath41 is a @xmath1-reflexive ideal which is @xmath756-well mixed .
then we have an irredundant decomposition @xmath757 where @xmath758 , and @xmath759 the @xmath1-difference schemes @xmath760 should be thought of as ` structural components ' of @xmath187 , even though they decompose further into topological components which are shuffled by @xmath1 .
since @xmath41 is @xmath756-well - mixed , using the ascending chain condition on @xmath756-perfect ideals and techniques of [ perfintrs ] , we can find a decomposition @xmath761 , with @xmath758 .
applying @xmath209 to the equality , using [ fulfinorb ] , yields @xmath762 , and it suffices to take the union over all @xmath209 . when necessary
, we shall assume the following * finiteness condition*. a _ difference scheme _ will mean an almost strict difference scheme @xmath104 which can be covered by a finite number of affine difference schemes which are themselves almost strict of finite @xmath1-type over a difference field .
for a point @xmath17 on a difference scheme @xmath104 , we denote by @xmath763 the local ( difference ) ring at @xmath17 , and by @xmath764 the residue ( difference ) field at @xmath17 .
[ def - limdeg ] let @xmath36 be an extension of difference fields . 1 .
an element @xmath765 is _
@xmath23-algebraic _ over @xmath38 if the set @xmath766 is algebraically dependent over @xmath38 .
2 . the @xmath23-algebraic closure over @xmath38 defines a _ pregeometry _ on @xmath25 and the dimension with respect to this pregeometry is called the @xmath23-transcendence degree .
alternatively , @xmath767 is the supremum of numbers @xmath665 such that the difference polynomial ring @xmath768 in @xmath665 variables embeds in @xmath25 .
3 . @xmath25 is _
@xmath23-separable _ over @xmath38 if @xmath25 is linearly disjoint from @xmath769 over @xmath38 , where the _ inversive closure _ @xmath770 , as in [ definvring ] , is the unique ( up to @xmath38-isomorphism ) difference field extension of @xmath771 where @xmath23 is an automorphism of @xmath769 and @xmath772 4 .
suppose @xmath25 is @xmath23-algebraic of finite @xmath23-type over @xmath38 , @xmath23-generated by a finite set @xmath116 .
let @xmath773 and let @xmath774 $ ] .
it is shown in @xcite that for every @xmath695 , @xmath775 and we may define the _ limit degree _ as @xmath776 this definition is independent of the choice of the generators .
when @xmath777 is @xmath23-algebraic but not necessarily finitely @xmath23-generated , one defines @xmath778 as the maximum of @xmath779 where @xmath780 runs over the extensions of finite @xmath23-type contained in @xmath25 .
[ fulllimdeg ] let @xmath781 be an extension of difference fields , with @xmath1 full and finite and let @xmath663 be the associated @xmath95-morphism .
then , for any @xmath148 , @xmath782 we may suppose @xmath783 for a finite tuple @xmath784 .
let @xmath689 .
as in the proof of [ fullfintyp ] we conclude that @xmath690 for any @xmath148 .
let @xmath785 since the limit degree does not depend on the choice of generators , @xmath786={\mathop{\rm dl}\nolimits}((l,\sigma')/(k,\pi(\sigma')).\ ] ] the following definition makes sense by [ fulllimdeg ] .
let @xmath781 be an extension of difference fields with @xmath1 full and finite .
let @xmath787 , for any @xmath107 .
before introducing the various dimensions and degree invariants of difference schemes , it is useful to define an auxiliary structure where some of those invariants will take values . the _ rig _
( ring without negatives ) @xmath788 $ ] admits a natural lexicographic polynomial ordering @xmath789 , and an equivalence relation @xmath790 , where @xmath791 if @xmath792 $ ] have the same degree in @xmath793 and and their leading coefficients are equal .
we will consider the rig @xmath788/{\approx}$ ] .
[ dimtot - ring ] let @xmath794 be a difference field , @xmath795 a full difference field extension and let @xmath164 be a full @xmath593-algebra . 1 .
let the _ transformal degree _ of @xmath795 be @xmath796 in @xmath788/{\approx}$ ] .
2 . let the _ effective transformal degree _ of @xmath6 be @xmath797 .
3 . let @xmath798 and analogously for @xmath799 .
4 . the _ limit degree _ @xmath800 and _ total dimension _
@xmath801 are defined through @xmath802 and analogously for the _ effective total dimension_. [ trdim - sch ] let @xmath593 be a difference field , and consider a morphism @xmath803 of @xmath593-difference schemes . 1
. the _ @xmath23-dimension _ of @xmath6 is @xmath804 2 .
the _ relative @xmath23-dimension _
@xmath805 where @xmath806 is the fibre over @xmath807 .
[ dimtot - sch ] let @xmath593 be a difference field , and consider a morphism @xmath808 of full @xmath593-difference schemes . 1 .
let the _ transformal degree _ of @xmath6 be @xmath809 and analogously for @xmath810 .
2 . the _ limit degree _ @xmath811 and _ total dimension _
@xmath812 are defined through @xmath813 and analogously for the _ effective total dimension_. 3 .
the _ relative transformal degree _ @xmath814 and analogously for @xmath815 . from these
we derive the notions of _ relative limit degree _ and _ relative total dimension_. 1 . clearly ( cf .
@xcite , @xcite ) , @xmath816 if and only @xmath812 and @xmath817 are finite , and analogously for the relative dimensions . in this case , if in addition @xmath12 is of finite @xmath23-type , @xmath818 $ ] , i.e. , the limit degree is finite .
when @xmath819 , @xmath820 , so the above definition is consistent .
indeed , as remarked in @xcite , the inequality @xmath821 is obvious . in the other direction ,
let @xmath822 such that @xmath823 .
then @xmath60 induces a @xmath23-ideal in @xmath824 , where @xmath825 is the perfect closure of @xmath60 , and the opposite inequality follows .
when @xmath826 is a @xmath23-separable @xmath23-algebraic extension of @xmath771 , @xmath827 is an algebraic extension of @xmath828 and @xmath829 thus , when @xmath111 is _
@xmath23-separable _ in the sense that for every @xmath525 , the extension @xmath830 is @xmath23-separable , we get that @xmath831 4 .
[ tower ] thanks to the corresponding property of the limit degree and the additivity of total dimension , the transformal degree is multiplicative in towers .
[ comp ] let @xmath819 . by the ritt ascending chain condition for perfect ideals in @xmath26 ( @xcite ) ,
@xmath6 is a noetherian topological space and therefore we get a decomposition of @xmath6 into irreducible components , @xmath832 where @xmath833 for some @xmath834 .
equivalently , the zero ideal in @xmath26 can be represented as @xmath835 since @xmath6 is of transformal dimension @xmath575 ( equivalently , of finite total dimension ) , for @xmath836 , @xmath837 and the results of @xcite entail @xmath838 an analogous statement holds for @xmath839 .
let @xmath116 be an integral domain with fraction field @xmath38 and let @xmath840 be a finite dimensional vector space over @xmath38 .
an @xmath116-submodule @xmath726 of @xmath840 is called an _ @xmath116-lattice _ in @xmath840
if @xmath726 contains a basis for @xmath840 as a @xmath38-module ( i.e. @xmath841 ) and @xmath726 is a submodule of a finitely generated @xmath116-submodule of @xmath840 .
[ norm - lattice ] let @xmath116 be a normal domain with fraction field @xmath38 and let @xmath25 be a finite separable extension of @xmath38 .
let @xmath714 be the integral closure of @xmath116 in @xmath25 .
choose an integral basis @xmath842 for @xmath38 over @xmath25 and let @xmath843 be its discriminant .
then @xmath844 in particular , @xmath714 is an @xmath116-lattice in @xmath25 .
suppose @xmath845 is a normal domain with fraction field @xmath846 and let @xmath847 be an extension of @xmath846 with @xmath848 $ ] finite .
let @xmath714 be the integral closure of @xmath116 in @xmath25 .
then @xmath849 for every @xmath107 so @xmath850 is a difference ring extension of @xmath845 .
[ cor : norm - lattice ] with above notation , assume @xmath25 is separable over @xmath38 .
then there is a @xmath756-localisation @xmath851_{\sigma_0}$ ] of @xmath116 , for an element @xmath852 , such that the corresponding normalisation @xmath853_{\sigma}$ ] is a finite @xmath698-module .
perform a transformal localisation at the discriminant .
if @xmath707 is of finite @xmath23-type over @xmath716 with both @xmath116 , @xmath714 domains such that the fraction field of @xmath116 is finite separable over the fraction field of @xmath714 , there is a @xmath23-localisation @xmath854 of @xmath714 such that @xmath855 is a finite @xmath854-module .
suppose a finite group @xmath146 acts on a difference scheme @xmath104 by automorphisms from the right , through a homomorphism @xmath856 .
we will not require a notational device for distinguishing an element @xmath857 of @xmath146 from the corresponding automorphism , since it only makes sense to compose @xmath857 with other morphisms of difference varieties when @xmath857 is considered as an automorphism . for any difference scheme @xmath858 , @xmath146 acts on the set @xmath859 on the left
and we can consider the set @xmath860 of @xmath146-invariant morphisms .
a natural question to ask is whether the functor @xmath861 is representable , i.e. , isomorphic to a functor @xmath862 . in other words , is there a difference scheme @xmath863 and a @xmath146-invariant morphism @xmath864 such that for every @xmath858 , the function @xmath865 defined by @xmath866 is bijective .
if this is the case , we say that @xmath863 is a _ quotient _ of @xmath104 by @xmath146 , and it is determined up to a unique isomorphism .
it would be difficult to consider the existence of a completely general categorical quotient ( in the above sense ) in the difference context , but if we assume some additional reasonable ( universal categorical or geometric ) properties from the quotient , such as that the fibres of @xmath5 are in fact @xmath146-orbits , we uncover the existence of a richer structure by the following heuristic .
suppose , in the best possible case , that for any ( structural ) morphisms @xmath867 , @xmath868 implies the existence of a ( unique ) @xmath869 such that @xmath870 .
then , since @xmath871 , there must be an @xmath869 such that @xmath872 . by the assumption that @xmath23 is an epimorphism
, it follows that @xmath873 is unique and we denote it @xmath874 .
thus we obtain a homomorphism @xmath875 for every @xmath107 , with the property that for every @xmath149 , @xmath876 if @xmath23 is invertible , it follows that @xmath161 is in fact a group automorphism . in this context
, we will say that @xmath3 acts on @xmath104 , where @xmath877 .
let us prove the existence of quotients for affine difference schemes , where @xmath3 acts ( on the left ) by automorphisms on the difference ring @xmath115 , i.e. , for every @xmath107 and @xmath149 , @xmath878 [ quotaff ] suppose a finite group @xmath3 acts on a difference ring @xmath115 so that @xmath879 and let @xmath880 be the subring of invariants of @xmath116 , @xmath881 , @xmath882 and @xmath883 the canonical ( @xmath146-invariant ) morphism
. then the following holds . 1 .
[ ajn ] @xmath116 is integral over @xmath714 .
[ cvaj ] the morphism @xmath5 is surjective , its fibres are @xmath146-orbits and the topology of @xmath134 it the quotient of the topology of @xmath6 .
[ draj ] let @xmath525 , @xmath884 , let @xmath885 be the stabiliser of @xmath17 and let @xmath886 .
let @xmath887 and @xmath888 , where @xmath889 is induced by @xmath890 for every @xmath891 . then @xmath89 is a quasi - galois algebraic extension of @xmath683 and the canonical map @xmath892 is surjective .
[ funf ] the natural homomorphism @xmath893 is an isomorphism .
[ fir ] @xmath894 is a quotient difference scheme of @xmath104 by @xmath146 .
let @xmath895 and let @xmath896 be the restriction map .
we need to check that the elements of @xmath897 are endomorphisms of @xmath714 . for @xmath107 and @xmath898 , @xmath899 for every @xmath149 , which shows that @xmath849 , as required .
thus @xmath5 is the morphism associated to the inclusion @xmath900 .
( [ ajn ] ) it is well - known that @xmath116 is integral over @xmath714 since every @xmath27 is a root of the monic polynomial @xmath901 $ ] .
( [ cvaj ] ) let us denote @xmath902 the morphism of ambient affine schemes @xmath903 and @xmath904 induced by @xmath905 , so that @xmath906 .
the statement of ( [ cvaj ] ) is known for @xmath907 ( @xcite , v.1.1 ) , so it suffices to prove that @xmath908 .
pick an @xmath681 , say @xmath909 , for some @xmath910 . then , for any @xmath107 restricting to @xmath911 ( @xmath912 ) , we have @xmath913 on @xmath914 , so given any @xmath915 such that @xmath916 , @xmath917 so there exists a @xmath149 with @xmath918 . by assumption ,
@xmath919 , so @xmath920 .
( [ draj ] ) the fact that the natural homomorphism @xmath921 is surjective is known ( loc .
cit ) , and the difference superstructure is a bookkeeping exercise .
( [ funf ] ) it is known ( loc .
that @xmath922 is an isomorphism .
let us write @xmath923 and @xmath924 .
by applying the exact functor @xmath925 , we obtain an isomorphism @xmath926 on the other hand , for @xmath840 open in @xmath134 , @xmath927 where the limit is taken over open subsets @xmath928 of @xmath914 . from the proof of ( [ cvaj ] ) , we know that @xmath929 is a @xmath146-invariant set and we can replace the above limit by the limit over the cofinal system of invariant neighbourhoods @xmath930 of @xmath931 with @xmath932 an open neighbourhood of @xmath840 in @xmath933 , allowing the identifications @xmath934 thus providing the desired isomorphism .
( [ fir ] ) this statement is immediate from ( [ cvaj ] ) and ( [ funf ] ) .
[ propadmissibl ] let @xmath104 be a difference scheme with a finite group of automorphisms @xmath3 , let @xmath935 an affine invariant morphism such that @xmath936
. then the conclusions ( [ ajn ] ) , ( [ cvaj ] ) , ( [ draj ] ) , ( [ fir ] ) of [ quotaff ] still hold . for ( [ ajn ] ) , ( [ cvaj ] ) , ( [ draj ] ) we may assume that @xmath134 and thus @xmath6 are affine , and if @xmath937 and @xmath115 are their rings , the hypothesis implies through [ embedcat ] that @xmath938 , so we can apply [ quotaff ] again .
the statement ( [ fir ] ) follows from ( [ cvaj ] ) and @xmath939 . with assumptions of [ propadmissibl ] , for every open @xmath201 in @xmath134
, @xmath201 is the quotient of @xmath940 by @xmath146 .
indeed , the morphism @xmath941 induced by @xmath5 satisfies the same assumptions as @xmath5 .
in addition to the conditions of [ propadmissibl ] , let @xmath6 be a difference scheme over @xmath2 and suppose the action of @xmath146 is by @xmath2-automorphisms .
then @xmath134 is again a difference scheme over @xmath2 .
moreover , @xmath6 is affine over @xmath2 if and only if @xmath134 is .
if @xmath6 is of finite @xmath1-type over @xmath2 , then @xmath6 is @xmath1-finite over @xmath134 . in order to prove the finiteness statements , we reduce to the case @xmath881 , @xmath942 , @xmath943 , where both @xmath115 and @xmath944 are @xmath945-algebras , with @xmath116 of finite @xmath1-type over @xmath10 .
let @xmath946 be the @xmath1-generators of @xmath116 over @xmath10 , i.e. , @xmath947_\sigma$ ] . as already noted in the proof of [ quotaff]([ajn ] ) , each @xmath948 is a root of the monic polynomial @xmath949 $ ] , and let @xmath950 denote the tuple of its coefficients .
firstly , @xmath116 is clearly @xmath1-finite over @xmath714 , being integral and of finite @xmath23-type over @xmath714 ( as it was of finite @xmath1-type already over @xmath10 ) . moreover , if we let @xmath951_{\sigma_0}$ ] , @xmath116 is again @xmath23-finite over @xmath952 , being both integral and of finite @xmath23-type over it , but now @xmath952 is of finite @xmath1-type ( equivalently , @xmath756-type ) over @xmath10 .
note that @xmath714 is a well - mixed @xmath716-submodule of @xmath116 . indeed ,
if @xmath953 for some @xmath954 and @xmath27 , then @xmath955 for every @xmath149 , so @xmath956 . by well - mixedness of @xmath116 , we get that @xmath957 . since @xmath958 again , @xmath959 for every @xmath149 so @xmath960 .
let @xmath104 be a difference scheme with a finite group of automorphisms @xmath3 .
we say that @xmath146 acts in an _ admissible _ way , if there exists a morphism @xmath961 with properties listed in [ propadmissibl ] .
this implies that @xmath962 is isomorphic to the quotient difference scheme of @xmath963 .
if @xmath3 acts admissibly on @xmath104 , so does every subgroup - with - operators @xmath964 .
suppose @xmath3 acts admissibly on @xmath104 and suppose that @xmath965 is a @xmath858-difference scheme .
taking a base change morphism @xmath966 , denote @xmath967 , @xmath968 , so that @xmath969 acts on @xmath970 by transport of structure and @xmath971 is invariant .
if @xmath972 is flat over @xmath2 , then @xmath973 satisfies the hypotheses of [ propadmissibl ] .
thus , @xmath969 acts on @xmath974 in an admissible fashion and we have that @xmath975
. we can clearly reduce to the case where @xmath976 and @xmath134 , @xmath977 affine .
it remains to prove that if @xmath937 is the subring of invariants of @xmath3 acting on @xmath115 , and if @xmath978 is a flat @xmath937-algebra , then @xmath978 is the subalgebra of invariants of @xmath979 , but this follows from the fact that the exact sequence @xmath980 where the last term is a power of @xmath116 and @xmath981 is the tuple with entries @xmath982 for @xmath983 , remains exact upon tensoring by @xmath854 , while the compatibility of the difference structure is an easy exercise .
in addition to the assumptions of [ quotaff ] , suppose @xmath116 is integral .
then @xmath984 .
suppose @xmath622 with @xmath985 .
consider the maps @xmath986 , @xmath598 , where @xmath987 is the natural inclusion . by
* v 2.2 , corollaire to th .
2 ) , there exists a @xmath149 such that @xmath988 , so @xmath989 .
suppose @xmath3 acts on @xmath104 . for @xmath525 , the _ decomposition group
_ at @xmath17 is the stabiliser @xmath990 of @xmath17 . with the notation of [ quotaff ] , @xmath991 acts on @xmath992 and
the _ inertia group _ @xmath993 at @xmath17 is the set of elements of @xmath994 which act trivially on @xmath89 .
assume that @xmath3 acts admissibly on @xmath104 and that @xmath965 is a @xmath858-difference scheme .
fix a @xmath995 and an embedding of @xmath996 into a model @xmath997 of acfa , whose saturation is sufficient to accommodate all extensions @xmath998 where @xmath525 is above @xmath13 .
we can consider @xmath999 as a @xmath858-difference scheme and @xmath997-valued points of @xmath6 correspond to @xmath996-algebra homomorphisms @xmath1000 , where @xmath17 is a point of @xmath6 above @xmath13 ( since @xmath1001 is large enough , every point @xmath17 above @xmath13 is a locus of some point in @xmath1002 ) .
thus we deduce a natural map @xmath1003 which is invariant under the action of @xmath146 on @xmath1002 . by the conclusions ( [ cvaj ] ) , ( [ draj ] ) of [ quotaff ] , this map is surjective and @xmath1004 . moreover ,
if @xmath17 is the locus of @xmath1005 , the stabiliser of @xmath47 in @xmath146 is exactly the inertia group @xmath993 .
bearing this in mind , in case @xmath1006 it makes sense to define the _ local @xmath1007-substitution @xmath1008 at @xmath47 _ as the unique element @xmath1009 satisfying @xmath1010 in other words , @xmath1008 is the element of @xmath1011 corresponding ( via the conclusion of [ quotaff]([draj ] ) ) to the image @xmath1012 of @xmath1007 by the morphism of difference structure @xmath1013 .
note , if @xmath1014 , there exists a @xmath149 with @xmath1015 and @xmath1016 so we conclude that @xmath1008 and @xmath1017 are @xmath146-conjugate .
therefore , we can define the _
local @xmath1007-substitution at @xmath1018 _ as the @xmath146-conjugacy class ( equivalently , the @xmath1-conjugacy class ) @xmath1019 of any @xmath1008 with @xmath1020 .
with hindsight , all of this holds even without the assumption of saturation or largeness of @xmath1001 , and when @xmath1021 is already given , we can even discuss @xmath1022 when @xmath1001 is just algebraically closed since this alone already guarantees the existence of some @xmath1005 above @xmath1018 .
equivalently , if we fix a section @xmath1023 , writing @xmath1024 for the image of @xmath1025 , we could consider as the relevant part of the datum for the local @xmath1007-substitution the unique element @xmath1026 with the property @xmath1027 .
as explained above , for @xmath1028 , there is a @xmath149 such that @xmath1029 so we can define @xmath1030 as the @xmath1031-conjugacy class in @xmath146 of any @xmath1032 with @xmath1020 . in the special case when @xmath1033 is the algebraic closure of a finite field @xmath695 and @xmath1034 is the frobenius automorphism generating @xmath1035 , and @xmath1036 is a @xmath1037-rational point mapping onto @xmath1018 in @xmath134 ,
we obtain the notion of the _ local frobenius substitution _ at @xmath1018 , denoted by @xmath1038 when considered as a conjugacy class in @xmath1 , or @xmath1039 when considered as a twisted conjugacy class in @xmath146 . from now on ,
a morphism satisfying the equivalent conditions of the corollary will be called a _ galois covering _ of @xmath1040 with group @xmath3 . for the purposes of this paper , a galois covering
will be called _ tale _ if all the inertia groups are trivial .
it is shown in ( * ? ? ?
* [ trivin - et],[et - trivin ] ) that the adjective ` tale ' is justified .
note that for tale galois coverings , the notion of local substitutions is well defined , as discussed above .
[ fig1 ] a recurring situation in this chapter is a morphism of two ( families of ) galois coverings , and it is convenient to fix the notation for use throughout the chapter . in the diagram @xmath1041 \matrix(m)[matrix of math nodes , minimum size=1.7em , nodes={circle } , inner sep=0pt , row sep=1em , column sep=.75em , text height=1.5ex , text depth=0.25ex ] { & & & [ -1em ] & |(uz)|{z_s } & & |(uw)|{w_s } \\[2em ] & & & & |(ux)|{x_s } & & |(uy)| { y_s } \\[.3em ] & & & & & |(us)|{\tilde{s } } & \\[-7em ] |(z)|{z } & & |(w)|{w } & & & & & \\[2em ] |(x)|{x } & & |(y)| { y } & & & & & \\[.3em ] & |(s)|{s } & & & & & & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( uz ) edge ( z ) edge ( ux ) edge ( uw ) ( ux ) edge node[above]{$f_s$ } ( uy ) edge node[below=-1pt , pos=0.05]{$p_s$ } ( us ) ( uw ) edge ( uy ) ( uy ) edge node [ right]{$q_s$ } ( us ) ( ux ) edge node[above]{$r_s$}(x ) ( z ) edge ( x ) edge ( w ) ( x ) edge node[above=-1pt , pos=0.7]{$f$ } ( y ) edge node[left]{$p$ } ( s ) ( w ) edge[cross line ] ( y ) ( y ) edge node[left , pos=0.3]{$q$ } ( s ) ( uw ) edge [ cross line](w ) ( uy ) edge [ cross line ] node [ below , pos=0.75]{$t_s$}(y ) ( us ) edge node[below]{$s$ } ( s ) ; \end{tikzpicture } \ ] ] we assume that @xmath1042 and @xmath1043 are galois coverings of @xmath78-difference schemes . for a given point @xmath318 , we consider @xmath1044 and , by slightly abusing the notation , we denote by @xmath444 the natural morphism @xmath1045 , and we let @xmath1046 , @xmath1047 , @xmath1048 , @xmath1049 be the relevant fibres above @xmath444 . [ udilw ]
let @xmath78 be a normal connected difference scheme of finite @xmath23-type over @xmath1050 , and let @xmath1051 be a morphism of finite transformal type with geometrically transformally integral fibres of finite total dimension @xmath1052 , limit degree @xmath1053 and purely inseparable dual degree @xmath1054 , and let @xmath1055 .
there is a constant @xmath1056 and a localisation @xmath1057 of @xmath78 such that for every closed @xmath1058 , and every finite field @xmath695 with @xmath1037 extending @xmath1059 , @xmath1060 [ geomgaleq ] let @xmath78 be a normal connected difference scheme of finite @xmath23-type over @xmath1050 , and let @xmath1061 be an tale galois covering with galois group @xmath3 of normal @xmath78-difference schemes of finite transformal type such that the fibres of @xmath1062 are geometrically transformally integral of total dimension @xmath1052 .
there is a constant @xmath1056 and a localisation @xmath1057 of @xmath78 such that for every closed @xmath1058 , and every finite field @xmath695 with @xmath1037 extending @xmath1059 , and every @xmath209 , @xmath1060 let us assume for simplicity that the purely inseparable dual degree @xmath1063 .
since we only need an estimate , we can immediately reduce to the affine case .
let @xmath1064 , @xmath1065 , @xmath1066 for some fixed @xmath107 , and @xmath1067 .
since @xmath1068 , @xmath1069 up to @xmath1070 . in view of [ udilw ] , it suffices to show that the relative limit degree of @xmath1071 over @xmath78 does not depend on @xmath149 .
since we are allowed to localise the base @xmath32 , we can reduce to the consideration of the limit degree of the associated function fields .
the situation is reminiscent of [ fulllimdeg ] , but we find it informative to proceed with the proof in the present context .
write @xmath1072 , @xmath771 , @xmath847 for the function fields of @xmath78 , @xmath92 and @xmath4 , respectively .
we are given that @xmath1073 and @xmath1066 and for any @xmath149 , @xmath1074 .
suppose @xmath1075 , and let @xmath1076 .
let @xmath1077 be a tuple of @xmath23-generators of @xmath38 over @xmath263 , @xmath1078 .
then @xmath1079 , and since the definition of limit degree does not depend on the choice of generators by [ def - limdeg ] , @xmath1080,\ ] ] where @xmath1081 on the other hand , @xmath1082 so @xmath1083 for any @xmath149 and @xmath1084 for any @xmath149 . 1 .
a function @xmath1085 is called _ central _ if @xmath1086 for every @xmath100 .
a subset of @xmath10 of @xmath1 is called a _ conjugacy domain _ when @xmath1087 if and only if @xmath1088 for all @xmath100 .
we denote by @xmath142 the algebra of all central functions on @xmath1 .
it is a trivial but useful observation that @xmath142 is spanned by the characteristic functions of conjugacy domains in @xmath1 .
when the underlying set of @xmath1 is finite , we can equip @xmath142 with an inner product as follows . 1 .
the _ pullback _ of @xmath1097 along @xmath638 is the central function @xmath1098 2 . if @xmath1099 is injective , we let the _ pushforward _ of @xmath1100 along @xmath638 be the central function @xmath1101 for @xmath1102 .
3 . if @xmath1103 is surjective , we let the _ pushforward _ of @xmath1100 along @xmath638 be the central function @xmath1104 for @xmath1102 .
an arbitrary @xmath1095 decomposes as @xmath1105 , with @xmath1106 a surjection and @xmath1107 an injection , so we can define @xmath1108 [ pushpullfunct ] the operation @xmath1109 defines a contravariant functor from the category of finite difference structures to the category of ( inner product ) algebras , @xmath1110 while @xmath1111 defines a covariant functor from the category of finite difference structures to the category of ( inner product ) vector spaces , @xmath1112 to explicate the functoriality : the only statement which requires checking due to our unconventional framework is the projection formula .
in view of the above functoriality , it suffices to check it separately for cases where @xmath638 is injective or surjective .
[ repbasech ] suppose we have a cartesian diagram @xmath609 { & |(p)|{\sigma_1\times_t\sigma_2 } & \\
|(1)|{\sigma_1 } & & |(2)| { \sigma_2}\\ & |(h)|{t } & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( p ) edge node[left , pos=0.35]{$\pi_1 $ } ( 1 ) edge node[right , pos=0.35]{$\pi_2 $ } ( 2 ) ( 1 ) edge node[left , pos=0.65]{$\psi_1 $ } ( h ) ( 2 ) edge node[right , pos=0.65]{$\psi_2 $ } ( h ) ; \end{tikzpicture}\ ] ] of surjective @xmath95-morphisms , and let @xmath1100 be a central function on @xmath1 .
then @xmath1123 suppose that there exists a finite group @xmath146 acting faithfully on @xmath1 such that @xmath638 can be identified with the canonical projection @xmath1134 .
then the assumption ( [ twosurj ] ) from [ frobrecip ] is satisfied .
[ def : basic - constr ] with notation of [ fig1 ] , let @xmath1135 be an tale galois covering of @xmath78-difference schemes with group @xmath3 ( such that @xmath1136 ) and let @xmath1085 be a central function .
we shall say that a pair @xmath1137 defines a _ basic @xmath78-constructible function _ on @xmath6 , in the sense that for any @xmath318 and every @xmath1138 extending @xmath1059 with @xmath263 algebraically closed , we obtain an actual function @xmath1139 we reserve the possibility of writing the last term in an oversimplified manner as @xmath1140 .
let @xmath92 be an @xmath78-difference scheme .
a _ constructible function _
@xmath1141 on @xmath92 is defined by a partition of @xmath92 into a finite set of integral normal locally closed difference @xmath78-subschemes @xmath1142 of @xmath92 , each equipped with an tale galois covering @xmath1143 with group @xmath1144 , and a central function @xmath1145 . in other words , @xmath1100 is determined by a normal stratification of @xmath6 and a basic constructible function on each stratum .
[ geomequid ] let @xmath78 be a normal connected difference scheme of finite @xmath23-type over @xmath1050 , and let @xmath1147 be an tale galois covering with galois group @xmath3 of normal connected @xmath78-difference schemes of finite transformal type such that the fibres of @xmath1062 are geometrically transformally integral of total dimension @xmath1052 .
let @xmath1100 be a @xmath78-constructible function on @xmath6 associated with the covering @xmath1148 .
there is a constant @xmath1056 and a localisation @xmath1149 of @xmath32 such that for any closed @xmath1058 , and any finite field @xmath695 with @xmath1037 extending @xmath1059 , @xmath1150 [ def : inflat ] let @xmath1100 be an @xmath78-constructible function on @xmath92 given by @xmath1151 and suppose that for each @xmath433 we have an tale galois @xmath78-covering @xmath1152 which dominates @xmath1153 .
thus , we have a diagram ( over @xmath78 ) @xmath1154 { |(11)|{z'_i } & & |(12)|{z_i}\\ & |(22)|{x_i } & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( 11 ) edge ( 12 ) edge(22 ) ( 12 ) edge ( 22 ) ; \end{tikzpicture}\ ] ] and let us denote by @xmath1155 the @xmath95-morphism part of @xmath1156 . with the above notation
, it is clear that @xmath1158 and @xmath1100 can be thought of as the same @xmath78-constructible function on @xmath6 , since for every @xmath1159 and every algebraically closed @xmath1138 extending @xmath1059 , @xmath1160 as functions @xmath1161 .
[ def : refin ] let @xmath1162 be an @xmath78-constructible function on @xmath92 and assume we have a further stratification of each @xmath1163 into finitely many integral , normal , locally closed @xmath78-subschemes @xmath1164 . for each @xmath1165 , let @xmath1166 be a connected component of @xmath1167 , and let @xmath1168 be its decomposition subgroup in @xmath1153
. moreover , let @xmath1169 .
then @xmath1170 is a galois covering over @xmath78 with group @xmath1171 and denote @xmath1172 .
we define the _ refinement of @xmath1100 to the stratification @xmath1173 of @xmath6 _ to be the @xmath78-constructible function @xmath1174 [ rem : refin ] again , the refinement procedure does not fundamentally change the constructible _ function _ , since for each @xmath318 , and algebraically closed @xmath1138 extending @xmath1059 , the realisations of @xmath1100 and @xmath1158 are the same : @xmath1160 as functions @xmath1175 .
[ def : pullbk ] let @xmath1176 be a morphism , and let @xmath1177 be a constructible function on @xmath134 , where each @xmath1178 is an tale galois covering with group @xmath1179 such that @xmath1180 is a central function on @xmath1181 .
let @xmath1182 be a component of @xmath1183 , and let @xmath1184 be its decomposition subgroup in @xmath1185 .
moreover , let @xmath1186 .
then @xmath1187 is a galois covering with group @xmath1188 and denote @xmath1189 . morally speaking , we would like to define the _ pullback of @xmath1097 along @xmath370 _ to be @xmath1190 but the strata @xmath1191
need not be normal . thus , to be precise , we must choose ( non - canonically , using ( * ? ? ?
* [ locnorm ] ) ) a normal stratification @xmath1164 which refines the stratification of @xmath6 into @xmath1191 and we define @xmath1192 to be the refinement of the above data . in spite of the non - canonical choice of a normal refinement , the notation @xmath1192 can be justified via [ rem : refin ] .
[ constr - algebra ] let us denote by @xmath1195 the set of @xmath78-constructible functions on @xmath92 .
we wish to endow @xmath1195 with a _
@xmath1090-algebra _ structure .
suppose @xmath1196 , and let @xmath1197 , @xmath1198 . upon a refinement of the underlying stratifications
, we may assume that @xmath1199 , and upon an inflation , we may even assume that @xmath1200 and @xmath1201 are the same galois coverings , and @xmath1202 .
then we can define : for the sake of simplicity , we will only define the pushforward along morphisms of a very special kind .
note that we will never use it for more general morphisms .
for simplicity of notation , the definition is given in an absolute case , but it is clear that all operations can be performed in a relative setting , over a given base @xmath78 .
[ defpushfwd ] let @xmath1176 be a morphism of finite transformal type of normal connected difference schemes with geometrically integral fibres , and let @xmath1100 be a basic constructible function on @xmath6 ( we leave the sorites of the definition for a general constructible function to the reader ) .
let @xmath1042 be an tale galois covering with group @xmath3 such that @xmath1100 is associated with a central function on @xmath1 . 1
. [ geomconn ] assume that @xmath2 is connected ( i.e. , integral ) and let @xmath1205 be the normalisation of @xmath1206 in the algebraic closure of @xmath1207 in @xmath1208 as depicted in the diagram @xmath1209 \matrix(m)[matrix of math nodes , minimum size=1.7em , nodes={circle } , inner sep=0pt , row sep=.5em , column sep=2.5em , text height=1.5ex , text depth=0.25ex ] { |(3)|{z } & & \\ & |(p)|{w_x } & |(2)| { w } \\[2em ] & |(1)|{x } & |(h)|{y } \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( p ) edge ( 1 ) edge ( 2 ) ( 1 ) edge ( h ) ( 2 ) edge ( h ) ( 3 ) edge ( 1 ) edge ( 2 ) edge ( p ) ; \end{tikzpicture}\ ] ] where @xmath1210 .
then @xmath1205 is a galois cover of @xmath1206 with some group @xmath1211 .
however , @xmath1212 is not necessarily tale , but in view of [ cor : norm - lattice ] we can find a finite @xmath23-localisation @xmath977 to achieve that the corresponding covering @xmath1213 is finite tale galois .
thus , we may continue to define @xmath1214 on the stratum @xmath977 and we postpone the definition of the pushforward by the morphism @xmath1215 for the next stage of devissage , remarking that this morphism shares the required properties of @xmath370 .
we henceforth relabel @xmath1216 and @xmath977 back to @xmath1217 and @xmath134 .
we obtain an exact sequence @xmath1218 moreover , @xmath1219 acts faithfully on @xmath1 so that @xmath1220 .
if @xmath1221 denotes the difference structure part of the morphism @xmath1222 , we define @xmath1223 as a central function on @xmath133 associated with the covering @xmath1212 .
if @xmath840 is a representation of @xmath1 , we define @xmath1224 , as a representation of @xmath133 .
[ notgeomconn ] if @xmath4 is not connected , @xmath1225 as a topological space , but @xmath1 is transitive on the set of components @xmath41 , we let @xmath1226 be the decomposition subgroup of @xmath1227 , and let @xmath1228
. then each @xmath1229 is a galois covering of @xmath92 with group @xmath1230 and maps onto @xmath1205 .
let us denote by @xmath1231 the restriction of @xmath1100 on @xmath1230 , associated with the covering @xmath1153 .
we define , resorting to the previous case , @xmath1232 as a central function on @xmath133 associated with the covering @xmath1233 .
[ disconn ] if @xmath1234 , let @xmath1231 be a restriction of @xmath1100 to @xmath1227 , and we define @xmath1235 where the sum of constructible functions is defined in terms of [ constr - algebra ] .
the case [ notgeomconn ] is a peculiarity of the difference framework , since in the algebraic case one need consider only geometrically connected / disconnected dichotomy .
one might argue that our definition in [ notgeomconn ] ignores the full structure of @xmath4 by breaking it up into @xmath1229 , but the key underlying principle is that we are dealing with fixed - point sets , and the structure morphisms which shuffle components too wildly can not have fixed points .
let us illustrate by showing a case where the definitions [ geomconn ] and [ notgeomconn ] agree
. with notation of [ notgeomconn ] , suppose we have a representation @xmath1236 of @xmath1 such that @xmath1237 , where each @xmath1238 is associated with a representation @xmath1239 of @xmath1240 . in other words , if @xmath1241 , @xmath1242 . a justification for the above choice of a particular form for @xmath1244 is beyond the scope of this paper , and it can be clarified in the framework of difference tale sheaves from the forthcoming work of the present author @xcite .
intuitively , suppose we have a locally constant tale sheaf @xmath563 on @xmath92 .
the scheme @xmath4 can be thought as induced from @xmath1247 via @xmath1248 .
a consequence of the appropriate sheaf condition is that @xmath1249 , where @xmath1250 above should be thought as the value @xmath1251 and @xmath840 as the value @xmath1252 .
[ bc ] ket @xmath1253 be a morphism of finite transformal type of normal connected difference schemes with geometrically integral fibres .
let @xmath807 be a point of @xmath134 , and let @xmath682 be a fibre over @xmath807 .
writing @xmath1254 and @xmath1255 , for any constructible function @xmath1100 on @xmath92 , @xmath1256 we may assume that @xmath1100 is basic , associated with an tale galois covering @xmath1042 and let @xmath1205 be as in [ defpushfwd ] , and we may also assume that @xmath807 does not fall in the ramification locus of @xmath1212 , so without any loss , we shall continue with the assumption that @xmath1212 is finite tale galois . by the compatibility of the statement with the case [ notgeomconn ] of [ defpushfwd ] , we may further reduce to the case where @xmath2 is integral over @xmath32 .
let @xmath1257 .
choose an @xmath1258 , let @xmath1259 and let @xmath1260 be the fibre of @xmath6 over @xmath807 .
moreover , let @xmath1261 and let @xmath1262 and @xmath1263 denote the fibres of @xmath1264 and @xmath2 above @xmath1265 .
@xmath1209 \matrix(m)[matrix of math nodes , minimum size=1.7em , nodes={circle } , inner sep=0pt , row sep=1em , column sep=1em , text height=1.5ex , text depth=0.25ex ] { & & [ 2em]&[-3.2em ] |(u3)|{z_{y ' } } & & [ 2em ] \\ & & & & |(up)|{x_{y ' } } & |(u2)| { \tilde{y } ' } \\[2em ] & & & & |(u1)|{x_y } & |(uh)|{\tilde{y } } \\ [ -7em ] |(3)|{z } & & & & & \\ & |(p)|{x_w } & |(2)| { w } & & & & \\[2em ] & |(1)|{x } & |(h)|{y } & & & & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( up ) edge ( u1 ) edge ( u2 ) ( u3 ) edge ( u1 ) edge ( u2 ) edge ( up ) ( u2 ) edge ( uh ) ( p ) edge ( 1 ) edge ( 2 ) ( 1 ) edge ( h ) ( u3 ) edge ( 3 ) ( up ) edge[cross line ] ( p ) ( u1 ) edge ( 1 ) ( u2 ) edge[cross line ] node[below=-2pt , pos=0.3]{$y'$}(2 ) ( uh ) edge node[below=-1pt]{$y$}(h ) ( u1 ) edge[cross line ] ( uh ) ( 2 ) edge[cross line ] ( h ) ( 3 ) edge ( 1 ) edge[cross line ] ( 2 ) edge ( p ) ; \end{tikzpicture}\ ] ] where the left vertical arrow is an isomorphism . a bit of diagram chasing yields that @xmath1267 .
consequently we obtain a cartesian diagram of difference structures @xmath1268 { & |(p)|{\sigma_{\smash{\mathrlap{y ' } { } } } } & \\
|(1)|{\sigma } & & |(2)| { \sigma^{\smash{\mathrlap{y'}{}}}}\\ & |(h)|{t } & \\ } ; \path[->,font=\scriptsize,>=to , thin ] ( p ) edge ( 1 ) edge ( 2 ) ( 1 ) edge ( h ) ( 2 ) edge ( h ) ; \end{tikzpicture}\ ] ] so the result follows from [ repbasech ] .
[ unfibc ] with notation from [ fig1 ] , let @xmath78 be a normal connected difference scheme and let @xmath1253 be a morphism of finite transformal type of normal connected difference @xmath78-schemes with geometrically integral fibres .
for any @xmath78-constructible function @xmath1100 on @xmath92 , @xmath1269 let @xmath1274 be a difference scheme of finite total dimension over a finite difference field @xmath1275 and let @xmath1100 , @xmath1097 be two constructible functions on @xmath6 ( not necessarily associated with the same galois covering ) .
let @xmath695 be a finite field such that @xmath1037 extends @xmath1275 , when @xmath1276 , we define a pairing @xmath1277 when @xmath1278 , we stipulate the expression to be @xmath575 .
[ adjpushpull ] let @xmath78 be a normal connected difference scheme of finite @xmath23-type over @xmath1050 , and let @xmath1253 be a morphism of finite transformal type of normal connected finite - dimensional difference @xmath78-schemes with geometrically transformally integral fibres .
let @xmath1100 be a @xmath78-constructible function on @xmath92 and let @xmath1097 be a @xmath78-constructible function on @xmath1206 .
there is a constant @xmath1056 and a localisation @xmath1057 of @xmath78 such that for every closed @xmath1058 and every finite field @xmath695 with @xmath1037 extending @xmath1059 , @xmath1279 by the usual tricks with refinement , inflation and pullback , we may assume that we have a situation as in [ defpushfwd ] , and that @xmath1100 is associated with the covering @xmath1148 and @xmath1097 is associated with @xmath1212 and we can continue using the notation of [ fig1 ] .
fix @xmath1056 and a localisation @xmath1149 of @xmath32 which depend only on the above data , which make all subsequent hrushovski s estimates work .
choose an @xmath1058 .
we proceed in several steps and reductions .
_ assuming momentarily that @xmath1280 is geometrically integral , for any central @xmath1158 associated with @xmath1281 and any central @xmath1282 associated with @xmath1283 , @xmath1284 indeed , using [ geomequid ] , we have @xmath1285 similarly , @xmath1286 the result follows from frobenius reciprocity [ frobrecip ] and hrushovski s estimate [ udilw ] . denoting topological components of @xmath1280 by @xmath1227 , @xmath364 and assuming @xmath695 is large enough to yield some rational points , we obtain @xmath1287 where @xmath790 denotes equality up to @xmath1288 .
[ trace - formula ] with notation of [ fig1 ] , let @xmath1290 be a morphism of finite transformal type of normal connected difference schemes with geometrically transformally integral fibres of finite relative total dimension @xmath1052 , and suppose @xmath78 is of finite @xmath23-type over @xmath1050 .
let @xmath1100 be an @xmath32-constructible function on @xmath134 .
there is a constant @xmath1056 and a localisation @xmath1057 of @xmath78 such that for every closed @xmath1058 and every finite field @xmath695 with @xmath1037 extending @xmath1059 , @xmath1291 in the next section we will need the following corollary , which conceptually states that every conjugacy class in @xmath1 which is not explicitly banned , is achieved as a local frobenius substitution of some rational point over some field with a high enough power of frobenius .
[ twisted - cebotarev ] let @xmath1061 be a galois covering over @xmath78 with group @xmath3 such that @xmath1294 has geometrically transformally integral fibres , and let @xmath10 be a conjugacy class in @xmath1 .
there exists a @xmath23-localisation @xmath1149 of @xmath32 such that for every @xmath1058 , for every large enough finite field @xmath695 with @xmath1037 extending @xmath1059 , if @xmath10 restricts to @xmath1295 , there exists a point @xmath1296 with @xmath1297 .
if @xmath6 is of finite relative total dimension over @xmath32 , the statement is immediate from [ trace - formula ] .
otherwise , one can consider a sitation over a suitable sub-@xmath78-scheme of finite relative total dimension .
let @xmath92 be a geometrically integral difference scheme of finite transformal type over @xmath1298 , where @xmath1299 is a finite field and @xmath1300 is a power of frobenius on @xmath695 , @xmath1301 , and @xmath1302 divides @xmath1303 .
we shall write @xmath1304 for the appropriate power of frobenius on @xmath1305 .
let @xmath1100 be a constructible function on @xmath92 . 1 .
the _ zeta function _ of @xmath92 over @xmath1298 is defined as the formal power series @xmath1306 2 .
the _ @xmath25-function _ associated with @xmath1100 is defined as the formal power series @xmath1307 using hrushovski s estimate [ udilw ] , there exists a constant @xmath1056 and numbers @xmath1311 with @xmath1312 so that @xmath1313 } \\=-\mu q_0^d\log(1-q^dt)+q_0^{d-1/2}\sum_n\epsilon_n\frac{(tq^{d-1/2})^n}{n},\end{gathered}\ ] ] and it is a simple exercise to calculate the radius of convergence for the latter series .
let @xmath1100 be a constructible function on a difference scheme @xmath92 of total dimension @xmath1052 over a finite @xmath1298 .
we define the integral of @xmath1100 with respect to the _ dirichlet density _
@xmath1053 to be the expression @xmath1314 [ chebotarev - density ] let @xmath92 be a geometrically transformally integral normal difference scheme of finite transformal type over a finite @xmath1320 and let @xmath1100 be a constructible function on @xmath92 associated with an tale galois covering @xmath1042 and a central function @xmath1085 .
then @xmath1321 where the latter denotes the integral with respect to the normalised counting measure on @xmath1 .
in other words , the local frobenius elements @xmath1322 are _ equidistributed _ on @xmath6 with respect to the dirichlet density . by additivity of the statement
, we can reduce to the case of connected @xmath2 .
writing @xmath1323 and @xmath1324 for the structure morphism , let us compute @xmath1325 .
denoting by @xmath1326 the algebraic closure of @xmath1320 inside @xmath1208 , @xmath1325 is just @xmath1327 , where @xmath1221 is the difference structure part of the morphism @xmath1328 . note that @xmath133 is just the coset @xmath1329 , and
thus the sequence @xmath1330 is periodic with period of length @xmath1129 .
thus , @xmath1331 } { \sum_n\frac{t^n}{n}\left[\mu(q^nq_0)^d+\epsilon'_n(q^nq_0)^{d-1/2}\right]}\\ & \stackrel{\ref{lem - dirichl}}{= } \lim_{t\nearrow q^{-d}}\frac{\sum_n\frac{(tq^d)^n}{n}p_*\alpha(\varphi_{n , s } ) } { \sum_n\frac{(tq^d)^n}{n } } \stackrel{\ref{lem - dirichl}}{= } \frac{1}{|t|}\sum_{\tau\in t}\pi_*\alpha(\tau)\\ & \stackrel{\ref{defpushpull}}{= } \frac{1}{|t|}\sum_{\tau\in t}\frac{1}{|\pi^{-1}(\tau)|}\sum_{\sigma\in\pi^{-1}(\tau)}\alpha(\sigma)=\frac{1}{|\sigma|}\sum_{s\in\sigma}\alpha(\sigma ) , \\ \end{split}\ ] ] as required .
alexandre grothendieck . .
documents mathmatiques ( paris ) , 3 .
socit mathmatique de france , paris , 2003 .
sminaire de gomtrie algbrique du bois marie 196061
. directed by a. grothendieck , with two papers by m. raynaud , updated and annotated reprint of the 1971 original [ lecture notes in math .
, 224 , springer , berlin ; ( 50 # 7129 ) ] .
|
we prove a function - field version of chebotarev s density theorem in the framework of difference algebraic geometry by developing the notion of galois coverings of generalised difference schemes , and using hrushovski s twisted lang - weil estimate .
|
observational data coming from the type ia supernovae ( sneia ) surveys , large scale structure ( lss ) , and cosmic microwave background ( cmb ) anisotropy spectrum indicate that the expansion of our present universe is accelerating rather than slowing down @xcite .
this cosmic acceleration can not be explained by the four known fundamental interactions in the standard models , which is the greatest challenge today in the modern physics .
the proposals that have been put forward to explain this observed phenomenon can basically be classified into two categories .
one is to assume that in the framework of einstein s general relativity ( gr ) , an exotic component with negative pressure called dark energy ( de ) is necessary to explain this observed phenomena . for a good review on the dynamics of different de models
see @xcite and references therein .
another alternative to account for the current accelerating cosmic expansion is to modify gr theory .
the well - known modified gravity theories are , for examples , @xmath4 theory , scalar - tensor theory ( including brans - dicke theory ) , braneworld scenarios ( such as dgp , rsi and rsii ) , @xmath5 theory ( @xmath6 is the gauss - bonett term ) , hoava - lifshitz theory , modified newtonian dynamics ( mond ) , and so forth . for some relevant reviews
see @xcite .
recently , a new modified gravity theory namely the so - called @xmath0 theory @xcite-@xcite was proposed to describe the present accelerating expansion of the universe without resorting to de .
it is a generalization of the teleparallel gravity ( tg ) @xcite by replacing the so - called torsion scalar @xmath3 with @xmath0 .
tg was originally developed by einstein in an attempt of unifying gravity and electromagnetism .
the basic variables in tg are tetrad fields @xmath7 , where the weitzenbock connection rather than the levi - civita connection was used to define the covariant derivative . as a result
, the spacetime has no curvature but contains torsion .
a vector @xmath8 in tg is parallel transported along a curve if its projection @xmath9 remains constant , this is the so - called teleparallelism .
the main advantage of @xmath0 theory is that its field equations are the second order which are remarkably simpler than the fourth order equations of @xmath4 theory @xcite .
the other interesting issue in modern cosmology is the thermodynamical description of the accelerating universe driven by de or modified gravity .
it was shown that by applying the first law of thermodynamics ( clausius relation ) @xmath10 to the apparent horizon @xmath11 , the friedmann equation in the einstein gravity can be derived if we take the hawking temperature @xmath12 and the entropy @xmath13 on the apparent horizon , where @xmath14 is the area of the horizon @xcite . here , @xmath15 is the amount of energy flow through the fixed apparent horizon .
the equivalence between the first law of thermodynamics and the friedmann equation was also found for gravity with gauss - bonnet term , the lovelock gravity theory and the braneworld scenarios @xcite . note that in the thermodynamics of the apparent horizon in the einstein gravity , the geometric entropy is assumed to be proportional to its horizon area , @xmath13 @xcite .
however , this definition is changed for other modified gravity theories . for instance , the geometric entropy in @xmath4 gravity is given by @xmath16 @xcite , where the subscript @xmath17 denotes a derivative with respect to the curvature scalar r. in @xmath0 gravity , it was shown that when @xmath18 is small , the first law of black hole thermodynamics is satisfied approximatively and the entropy of horizon is @xmath19 @xcite , where the subscript @xmath3 denotes a derivative with respect to the torsion scalar @xmath3 . besides examining the validity of the thermodynamical interpretation of gravity by expressing the gravitational field equations into the first law of thermodynamics in different spacetimes , it is also of great interest to investigate the validity of the generalized second law ( gsl ) of thermodynamics in the accelerating universe @xcite-@xcite .
the gsl of thermodynamics like the first law is an accepted principle in physics . here
, our aim is to investigate the gsl of thermodynamics in the framework of @xmath0 gravity for a spatially flat friedmann - robertson - walker ( frw ) universe filled with the pressureless matter . to do this , in section 2 , we briefly review the @xmath0 gravity . in section 3 ,
we investigate the gsl of thermodynamics on the dynamical apparent horizon with the hawking temperature . in section 4 , for two viable @xmath0 models we first calculate the effective equation of state and deceleration parameters . then , the null and strong energy conditions are investigated .
also the viability of both models is checked by cosmography .
finally , the validity of the gsl is examined .
section 5 is devoted to conclusions .
in the framework of @xmath0 theory , the action of modified tg is given by @xcite @xmath20,\label{action}\ ] ] where @xmath21 and @xmath22 is the lagrangian density of the matter inside the universe .
also @xmath23 is the vierbein field which is used as a dynamical object in tg .
the modified friedmann equations in the framework of @xmath0 gravity in the spatially flat frw universe are given by @xcite @xmath24 @xmath25 where @xmath26 @xmath27,\label{pt}\ ] ] @xmath28 and @xmath29 is the hubble parameter . here
@xmath30 and @xmath31 are the energy density and pressure of the matter , respectively .
also @xmath32 and @xmath33 are the torsion contributions to the energy density and pressure .
the energy conservation laws are still given by @xmath34 @xmath35 note that if @xmath36 , from eqs .
( [ rot ] ) and ( [ pt ] ) we have @xmath37 and @xmath38 then eqs .
( [ ft11 ] ) and ( [ ft22 ] ) transform to the usual friedmann equations in gr .
the effective equation of state ( eos ) parameter due to the torsion contribution is defined as @xcite @xmath39 note that for the de sitter universe , i.e. @xmath40 , we have @xmath41 behaving like the cosmological constant .
using eqs .
( [ ft11 ] ) , ( [ rot ] ) and ( [ t ] ) we have @xmath42 here , we consider a spatially flat frw universe filled with the pressureless matter , i.e. @xmath43 .
we ignore the contribution of radiation which is compatible with the observations @xcite . from eqs .
( [ ft11 ] ) to ( [ pt ] ) for @xmath43 , one can obtain @xmath44 substituting eq .
( [ rhom ] ) into ( [ hdot ] ) and using @xmath45 , one can get @xmath46 with the help of above relation , the effective eos parameter ( [ omegat ] ) can be rewritten as @xmath47 for the deceleration parameter @xmath48 using eqs .
( [ t ] ) and ( [ tdot ] ) one can obtain @xmath49 for the case @xmath36 , we get @xmath50 corresponding to the matter dominated phase .
here , we investigate the validity of the gsl for a spatially flat frw universe filled with the pressureless barionc matter ( bm ) and dark matter ( dm ) . according to the gsl ,
the entropy of matter inside the horizon plus the entropy of the horizon must not decrease in time @xcite .
we assume that the boundary of the universe to be enclosed by the dynamical apparent horizon @xmath11 . also the hawking temperature on the apparent horizon @xmath11
is given by @xcite @xmath51 where @xmath52 ensures that the temperature is positive .
the entropy of the universe including bm and dm inside the dynamical apparent horizon is given by gibb s equation @xcite @xmath53 where @xmath54 and @xmath55 is the volume containing the pressureless matter with the radius of the dynamical apparent horizon @xmath11 .
also @xmath56 taking time derivative of both sides of eq .
( [ em ] ) and using ( [ ceqm ] ) , gibb s equation ( [ tdsm1 ] ) yields @xmath57 where the dot denotes a time derivative . substituting eq .
( [ rhom ] ) into ( [ tdsm2 ] ) gives @xmath58 now we need to calculate the contribution of the apparent horizon entropy .
following @xcite , when @xmath18 is small , the horizon entropy in @xmath0 gravity is given by @xmath59 where @xmath60 .
for the special case @xmath36 , eq .
( [ sa ] ) recovers the bekenstein - hawking entropy @xmath13 in the einstein gravity . using eq .
( [ ta1 ] ) , the evolution of horizon entropy ( [ sa ] ) is obtained as @xmath61 inserting eq .
( [ tdot ] ) into ( [ tadsa1 ] ) gives @xmath62.\label{tadsa2}\ ] ] for the flat frw metric , the dynamical apparent horizon is same as the hubble horizon given by @xcite @xmath63 taking time derivative of eq .
( [ ra ] ) , using ( [ t ] ) , ( [ rhom ] ) and ( [ hdot ] ) one can get @xmath64 replacing eqs .
( [ ra])-([rdota ] ) into ( [ tdsm3 ] ) and ( [ tadsa2 ] ) and using ( [ t ] ) , one can obtain @xmath65 @xmath66 .\label{tadsa3}\ ] ] adding eqs .
( [ tdsm4 ] ) and ( [ tadsa3 ] ) yields the gsl in @xmath0 gravity as @xmath67 , \label{tasdot}\ ] ] where @xmath68 is the total entropy due to different contributions of the matter and the horizon .
note that in the einstein tg , i.e. @xmath36 , the gsl ( [ tasdot ] ) yields @xmath69 which always holds . in what follows
, we investigate the validity of the gsl , i.e. @xmath70 , for two viable @xmath0 models introduced in the literature .
here , we consider two viable @xmath0 models proposed by @xcite to explain the present cosmic accelerating expansion of the universe .
the first model is a power law @xmath71 where @xmath72 and @xmath73 are constants @xcite .
the second model has an exponential dependence on the torsion scalar as @xmath74 where @xmath75 and @xmath76 are two model parameters , and @xmath77 @xcite .
the parameters @xmath72 and @xmath75 can be obtained by inserting eqs .
( [ model1 ] ) and ( [ model2 ] ) into the modified friedmann eq .
( [ ft11 ] ) . solving the resulting equations for the present time gives @xmath78 @xmath79 where @xmath80 is the dimensionless matter energy density and the index 0 denotes the value of a quantity at the present . note that both models can unify a number of interesting extensions of gravity beyond the standard gr .
for instance , model 1 for the cases @xmath81 and @xmath82 reduces to the @xmath83cdm and dgp @xcite models , respectively .
model 2 acts like the @xmath83cdm for @xmath84 .
also both models can satisfy the condition @xmath85 at high redshift which is compatible with the primordial nucleosynthesis and cmb constraints @xcite . for model 1 to be a viable model compared to current data one needs @xmath86 @xcite .
the joint analysis of the astronomical data from sneia+bao+cmb gives the best fit values ( @xmath87 ) for model 1 and ( @xmath88 ) for model 2 at the 95% confidence level ( cl ) @xcite .
using the statefinder geometrical analysis and @xmath89 diagnostic method , it was shown that both model 1 and model 2 evolve from the standard cold dm ( scdm ) to a de sitter phase @xcite . also the effective de for model 2 with @xmath90
is similar to the @xmath83cdm both in the high redshift regimes and in the future , while for model 1 with @xmath91 this similarity occurs only in the future @xcite . the evolution of the effective eos parameter , eq .
( [ omegat2 ] ) , for model 1 and model 2 is plotted in figs .
[ wt - model1 ] and [ wt - model2 ] , respectively .
figures show that from the early times to the present epoch , both models behave like quintessence de , i.e. @xmath92 , and their effective eos parameters can not cross the phantom divide line @xcite .
this result has been already obtained by @xcite . at early times ( @xmath93 ) , for model 1 and model 2 we have @xmath94 and @xmath95 , respectively .
model 2 behaves like the @xmath83cdm at high redshift . at the present time ( @xmath96 ) , for model 1 and model 2 we obtain @xmath97 and @xmath98 , respectively . an interesting result which is absent in @xcite is that in the future ( @xmath99 ) for model 1 and model 2 at @xmath100 and @xmath101 , respectively , we have a transition from the quintessence state , @xmath92 , to the phantom regime , @xmath102 .
the evolution of the deceleration parameter , eq .
( [ q2 ] ) , for model 1 and model 2 is plotted in figs .
[ q - model1 ] and [ q - model2 ] , respectively .
figures illustrate that for model 1 and model 2 in the past @xmath103 and @xmath104 , respectively , we have a cosmic deceleration @xmath105 to acceleration @xmath106 transition which is compatible with the observations @xcite . at early times ( @xmath93 ) , for both models we obtain @xmath107 which indicates that the universe was experiencing a phase of deceleration at the early stage of its evolution due to the domination of the matter component . at the present time ( @xmath96 ) , for model 1 and model 2 we get @xmath108 and @xmath109 , respectively , which are in good agreement with the recent observational result @xmath110 @xcite .
the evolution of @xmath111 versus @xmath112 for model 1 and model 2 is plotted in figs .
[ nec - model1 ] and [ nec - model2 ] , respectively .
figures show that the null energy condition ( nec ) , i.e. @xmath113 , is violated for model 1 when @xmath114 and for model 2 when @xmath115 and @xmath116 . when the eos parameter of de is less than @xmath95 , the universe reaches a big rip singularity within a finite time . in this case
, the nec is violated .
barrow @xcite showed that a different type of future singularity , the so - called sudden singularity , can appear at a finite time even when the strong energy condition ( sec ) , i.e. @xmath117 and @xmath113 , is satisfied .
this type of future singularity corresponds to the one in which the pressure density @xmath33 diverges at @xmath118 , but the energy density @xmath32 is finite . the properties of different future singularities in the de universe have been investigated in ample detail by @xcite .
our numerical results show that the sudden future singularity appears in model 1 and model 2 when @xmath119 and @xmath120 , respectively .
also the sec for model 1 and model 2 is satisfied only when @xmath121 and @xmath122 , respectively . here ,
following capozziello et al .
@xcite we use a cosmographic analysis to check the viability of model 1 and model 2 without the need of explicitly solving the field equations and fitting the data . in this approach
, the parameters of a given @xmath0 model must be chosen in such a way that the model - independent constraints on the cosmographic parameters ( @xmath123 ) obtained by fitting to sneia hubble diagram and bao data are satisfied . here
, @xmath124 is the hubble constant and @xmath125 are the deceleration , jerk , snap , and lerk parameters , respectively .
let us first start with model 2 . using eqs .
( 4.21 ) and ( 4.22 ) in @xcite one can find @xmath126 now the @xmath127 value for @xmath128 where @xmath129 can be expressed as function of @xmath76 only when we fix @xmath130 from the wmap7 data . following @xcite for each @xmath131 value of the sample obtained above from the cosmographic parameters analysis , we solve @xmath132 .
this yields @xmath133 which takes place in the 95% cl from the model - dependent constraints @xcite .
then , we estimate the theoretically expected values for the other derivatives @xmath134 . the median and 68% and 95% confidence ranges
are obtained as @xmath135 now we compare the above results with the model - independent constraints on the @xmath136 values given in table ii in @xcite . following @xcite
we use only the 68% cl which we compare the above constraints to .
our comparison shows that the values of @xmath134 take place in the 68% cl in table ii in @xcite .
therefore , we conclude that model 2 is favored by the observational data . for model 1
, its cosmographic analysis has been already done by capozziello et al .
they showed that for the best fit value of the cosmographic parameters , solving @xmath137 yields @xmath138 .
note that the value of @xmath73 takes place in the 95% cl from the model - dependent constraints @xcite .
capozziello et al .
@xcite found quite small values for @xmath134 as expected for the @xmath83cdm model . here , the disagreement with the constraints in table ii in @xcite may be due to this fact that depending on the value of @xmath73 , assuming @xmath139 for @xmath140 in taylor expanding @xmath0 can fail for model 1 so that the constraints on @xmath136 should not be considered reliable @xcite . here
, we examine the validity of the gsl , eq .
( [ tasdot ] ) , for both models .
first , we need to check the validity of the horizon entropy relation ( [ sa ] ) .
because as we already mentioned , eq . ( [ sa ] ) is valid only when @xmath18 is small @xcite . to check this , we plot @xmath18 versus @xmath141 for model 1 and model 2 in figs .
[ fttf - model1 ] and [ fttf - model2 ] , respectively .
figures show that the @xmath18 is very small throughout history of the universe .
this confirms the validity of eq .
( [ sa ] ) for both models .
now we can calculate the gsl , eq .
( [ tasdot ] ) , for both models . for model 1 ,
the resulting gsl is @xmath142\nonumber\\&&\times\big[1 + \big(1-n(2+n)\big)\mu_1{(-t)}^{n-1}+n(1 - 2n)\big(4-n(9 - 4n)\big)\mu_1 ^ 2{(-t)}^{2n-2}\big ] , \label{tsdotmodel1}\end{aligned}\ ] ] and for model 2 we obtain @xmath143}^2},\label{tsdotmodel2}\end{aligned}\ ] ] where @xmath144e^{\frac{\beta t_0}{t}}.\end{aligned}\ ] ] the evolutions of the gsl ( [ tsdotmodel1 ] ) and ( [ tsdotmodel2 ] ) are plotted in figs .
[ gsl - model1 ] and [ gsl - model2 ] , respectively .
figures clear that for both model 1 and model 2 , the gsl is satisfied from the early times to the present epoch . at early times ( @xmath93 ) , for both models we have @xmath145 1.125 . at the present time ( @xmath96 ) , for model 1 and model 2 we get @xmath146 0.116 and 0.010 , respectively .
note that in the future ( @xmath99 ) , fig .
[ gsl - model1 ] clears that the gsl for model 1 is violated , i.e. @xmath147 , for @xmath148 and @xmath149 .
also in the future , according to fig .
[ gsl - model2 ] the gsl for model 2 , like model 1 , is violated for @xmath150 , @xmath151 and @xmath152 .
figures [ gsl - model1 ] and [ gsl - model2 ] do nt show the negative values of the gsl , because the axis of @xmath153 is based on a logarithmic scale .
here , we studied the gsl in the framework of @xmath0 gravity . among other approaches related with a variety of de models , a very promising approach to de
is related with the modified tg known as @xmath0 gravity , in which de emerges from the modification of torsion .
the class of @xmath0 gravity theories is an intriguing generalization of einstein s new gr , taking a curvature - free approach and using a connection with torsion .
it is analogous to the @xmath4 extension of the einstein - hilbert action of standard gr , but has the advantage of the second order field equations @xcite .
we investigated the gsl on the hubble horizon with the hawking temperature for a spatially flat frw universe filled with the pressureless matter .
for two viable @xmath0 models containing @xmath1 and @xmath2 , we first calculated the effective eos and deceleration parameters .
interestingly enough , we found that for both models there is a transition from the quintessence state , @xmath92 , to the phantom regime , @xmath102 , in the future . also both models showed a cosmic deceleration @xmath105 to acceleration @xmath106 transition in the near past which is compatible with the observations @xcite .
furthermore , we investigated the nec and sec and concluded that both models show a sudden future singularity .
also our cosmographic analysis cleared that model 1 is unable to predict the observationally motivated @xmath154 values but model 2 is favored by the observational data .
finally , we examined the validity of the gsl for the selected @xmath0 models .
we concluded that the gsl is satisfied for both models from the early times to the present epoch .
but in the future , the gsl is violated for model 1 when @xmath148 and @xmath149 and for model 2 when @xmath150 , @xmath151 and @xmath152 .
the authors thank the anonymous referee for a number of valuable suggestions .
the work of k. karami has been supported financially by center for excellence in astronomy & astrophysics ( ceaa - riaam ) , maragha , iran .
riess , et al .
j. * 116 * , 1009 ( 1998 ) ; + s. perlmutter , et al .
, astrophys .
j. * 517 * , 565 ( 1999 ) ; + p. de bernardis , et al . ,
nature * 404 * , 955 ( 2000 ) ; + s. perlmutter , et al .
, astrophys .
j. * 598 * , 102 ( 2003 ) .
t. padmanabhan , phys . rep . * 380 * , 235 ( 2003 ) ; + p.j.e .
peebles , b. ratra , rev .
phys . * 75 * , 559 ( 2003 ) ; + e.j .
copeland , m. sami , s. tsujikawa , int . j. mod
d * 15 * , 1753 ( 2006 ) ; + m. li , x.d .
li , s. wang , y. wang , commun .
* 56 * , 525 ( 2011 ) .
s. capozziello , int .
d * 11 * , 483 ( 2002 ) ; + s. capozziello , m. de laurentis , v. faraoni , open astron . j. * 3 * , 49 ( 2010 ) ; + t. p. sotiriou , v. faraoni , rev .
* 82 * , 451 ( 2010 ) ; + s. nojiri , s.d .
odintsov , phys . rep . * 505 * , 59 ( 2011 ) .
yang , eur . phys .
j. c * 71 * , 1797 ( 2011 ) ; + s.c .
ulhoa , j.f .
da rocha neto , j.w .
maluf , int . j. mod
d * 19 * , 1925 ( 2010 ) ; + k. bamba , et al .
, jcap * 01 * , 021 ( 2011 ) ; + r. zheng , q.g .
huang , jcap * 03 * , 002 ( 2011 )
. j.b .
dent , s. dutta , e.n .
saridakis , phys .
d * 83 * , 023508 ( 2011 ) ; + j.b .
dent , s. dutta , e.n .
saridakis , jcap * 01 * , 009 ( 2011 ) ; + c.q .
geng , c.c .
lee , e.n .
saridakis , y.p .
wu , phys . lett .
b * 704 * , 384 ( 2011 ) ; + c.q .
geng , c.c .
lee , e.n .
saridakis , jcap * 01 * , 002 ( 2012 ) .
a. einstein , sitzungsber .
( 1928 ) ; + a. einstein , sitzungsber .
( 1930 ) ; + a. einstein , math . ann . * 102 * , 685 ( 1930 ) ; + k. hayashi , t. shirafuji , phys .
d * 19 * , 3524 ( 1979 ) ; + k. hayashi , t. shirafuji , phys .
d * 24 * , 3312 ( 1981 ) .
a. sheykhi , b. wang , phys .
b * 678 * , 434 ( 2009 ) ; + a. sheykhi , b. wang , mod .
lett . a * 25 * , 1199 ( 2010 ) ; + a. sheykhi , phys .
d * 81 * , 104011 ( 2010 ) ; + a. sheykhi , eur . phys .
j. c * 69 * , 265 ( 2010 ) ; + a. sheykhi , class .
quantum grav .
* 27 * , 025007 ( 2010 ) .
k. karami , jcap * 01 * , 015 ( 2010 ) ; + k. karami , s. ghaffari , phys .
b * 685 * , 115 ( 2010 ) ; + k. karami , s. ghaffari , phys . lett .
b * 688 * , 125 ( 2010 ) ; + k. karami , s. ghaffari , m.m .
soltanzadeh , class .
quantum grav .
* 27 * , 205021 ( 2010 ) ; + k. karami , s. ghaffari , m.m .
soltanzadeh , astrophys .
space sci .
* 331 * , 309 ( 2011 ) ; + k. karami , a. sheykhi , n. sahraei , s. ghaffari , europhys .
lett . * 93 * , 29002 ( 2011 ) ; + k. karami , a. abdolmaleki , n. sahraei , s. ghaffari , jhep * 08 * , 150 ( 2011 ) .
cai , h. li , y.s .
piao , x. zhang , phys . lett .
b * 646 * , 141 ( 2007 ) ; + y.f .
cai , t. qiu , y.s .
piao , m. li , x. zhang , jhep * 10 * , 071 ( 2007 ) ; + y.f .
cai , t. qiu , r. brandenberger , y.s .
piao , x. zhang , jcap * 03 * , 013 ( 2008 ) .
|
we investigate the validity of the generalized second law ( gsl ) of gravitational thermodynamics in the framework of @xmath0 modified teleparallel gravity .
we consider a spatially flat frw universe containing only the pressureless matter .
the boundary of the universe is assumed to be enclosed by the hubble horizon .
for two viable @xmath0 models containing @xmath1 and @xmath2 , we first calculate the effective equation of state and deceleration parameters .
then , we investigate the null and strong energy conditions and conclude that a sudden future singularity appears in both models . furthermore ,
using a cosmographic analysis we check the viability of two models .
finally , we examine the validity of the gsl and find that for both models it is satisfied from the early times to the present epoch .
but in the future , the gsl is violated for the special ranges of the torsion scalar @xmath3 .
= 16truecm = 23truecm = -1.3truecm = -2truecm = 1.20 cm = 1.60 cm * keywords : * cosmology of theories beyond the sm , modified gravity
|
multiscale analysis , based on the wavelet approach , has been successfully used for many different physical applications , including global solar activity ( ochadlick et al .
@xcite ; lawrence et al .
intermittence features have been well analysed through the use of a continuous wavelet transforms on the monthly sunspot numbers time series . more recently , different authors used longer daily sunspot numbers time series to investigate more accurately new subtle periodicities and their evolutions ( frick et al .
@xcite ; ballester et al .
@xcite ; fligge et al .
@xcite ; sello @xcite ) . a recent work on wavelet applications to biomedical signals , shows the utility of a new parameter , derived from wavelet formalism , to quantify the degree of _ disorder _ in the time series .
this quantity , called _ wavelet entropy _ , essentially gives an information on the scales extension involved in the energy distribution , i.e. a dynamical complexity parameter ( quian quiroga et al .
the main aim of the present work is the application of the wavelet entropy approach to the solar activity , through the wavelet analysis of the monthly mean sunspot numbers .
fourier analysis is an adequate tool for detecting and quantifying constant periodic fluctuations in time series . for intermittent and transient multiscale phenomena ,
the wavelet transform is able to detect time evolutions of the frequency distribution .
the continuous wavelet transform represents an optimal localized decomposition of time series , @xmath1 , as a function of both time @xmath2 and frequency ( scale ) @xmath3 , from a convolution integral : @xmath4 where @xmath5 is called an analysing wavelet if it verifies the following admissibility condition : @xmath6 where : @xmath7 is the related fourier transform . in the definition , @xmath3 and @xmath8
denote the dilation ( scale factor ) and translation ( time shift parameter ) , respectively .
we define the local wavelet spectrum : @xmath9 where @xmath10 denotes the peak frequency of the analysing wavelet @xmath5 . from the local wavelet spectrum we can derive a mean or global wavelet spectrum , @xmath11 :
@xmath12 the relationship between the ordinary fourier spectrum @xmath13 and the mean wavelet spectrum @xmath11 is given by : @xmath14 indicating that the mean wavelet spectrum is the average of the fourier spectrum weighted by the square of the fourier transform of the analysing wavelet @xmath5 shifted at frequency k. here we used the family of complex analyzing wavelets consisting of a plane wave modulated by a gaussian , called morlet wavelet , ( torrence et al .
@xcite ) : @xmath15 where @xmath16 is the non dimensional frequency here taken to be equal to 6 in order to satisfy the admissibility condition , eq.(2 ) . following quian quiroga et al .
( @xcite ) , we define a _ wavelet entropy _ as a function of time : @xmath17 where : @xmath18 is the energy probability distribution for each scale level . from the definition
, follows that an ordered activity corresponds to a narrow frequency distribution of energy , with low wavelet entropy , and a random activity corresponds to a broad frequency distribution , with high wavelet entropy .
of course , higher values for wavelet entropy means higher dynamical complexity , higher irregular behaviour , lower predictability .
the application of the wavelet entropy is optimal for non - stationary signals .
we considered here the record of solar activity given by the monthly mean number of sunspots , from sidc archive ( url : http://www.oma.be/ksb-orb/sidc/index.html ) , covering the time interval : 1749 - 2000.04 and consisting of 3013 observations . using a proper wavelet analysis we found the results displayed in fig.1 .
following the notation described in sello ( @xcite ) , the upper part shows the original time series in its natural units .
time is here expressed in years .
the central part shows the amplitudes of the wavelet local power spectrum in terms of an arbitrary greyscale contour map .
white higher values are strong energetic contributions to power spectrum , while black lower values are weak energetic contributions .
horizontal time axis corresponds to the axis of time series and vertical scale ( frequency ) axis is , for convenience , expressed in log values of cycles per year@xmath19 .
thus the range analyzed is between 148 years ( value -5 ) and 134 days ( value 1 ) .
the right part shows the mean global wavelet power spectrum ( solid line , sm ) obtained with a time integration of the local wavelet power spectrum , and the @xmath20 significance level using a red noise autoregressive lag-1 background spectrum with @xmath21 ( dotted line , sc ) more precisely , we assumed here that different realizations of the considered stochastic process , given by sunspot activity , will be randomly distributed about a red noise background spectrum ( i.e. increasing power with decreasing frequency ) , which is used to verify a _
null hypothesis _ for the significance of a given peak in the wavelet power spectrum .
in particular , if a peak in the wavelet power spectrum is significantly above the assumed red noise background spectrum , then it can be considered a real feature of the original signal with a given confidence level .
this statement appears well justified from different applications to many complex physical signals , e.g. geophysical data . in order to obtain a simple model for red noise we considered a univariate lag-1 markov process : @xmath22 , where @xmath23 is the assumed lag-1 autocorrelation , and
@xmath24 is a gaussian process .
note that : @xmath25 corresponds to a white noise .
( for more details see : torrence et al .
@xcite ) . from the figure , it is possible to distinguish the main region corresponding to the well defined schwabe ( @xmath26 years ) cycle , with a dominant energetic contribution above the red noise background spectrum . in particular ,
the complex evolution of the cycle , both for scales and amplitudes , is well evident .
the irregular finite extension in the map represents a continuous interaction of frequencies which generate complex dynamical behaviour . in order to better quantify the evolution of complexity , or disorder for the solar cycle time series ,
we computed the time evolution of the wavelet entropy . an interesting feature of this approach is that the usual assumed hypothesis of stationarity of time series is here not necessary . .
[ lab1 ] fig.2 ( lower part ) shows the time evolution of the wavelet entropy , eq.(8 ) . for comparison , an ordered deterministic ( harmonic ) signal gives the numerical result displayed by the dashed line .
it is interesting to note that the degree of disorder reaches maximum local values , often in phase with solar maxima , where the magnetic activity results higher , but with many important exceptions ( see , for example , cycles 20 and 22 ) .
the recorded maximum entropy value has been reached during the maximum of solar cycle 4 ( 1789 ) . moreover
, the entropy results quite low for long periods , including solar cycles from 5 to 7 ( 1798 - 1828 ) and , in particular , for solar cycles from 9 to 14 ( 1848 - 1912 ) . for the last three cycles , 21 , 22 and 23 , we found a clear average increase of the wavelet entropy , suggesting a more complex dynamics , with a higher level of disorder due to a broad frequency energy distribution , connected to evolution of the global magnetic activity . this property forces a reduction in the accuracy level reachable by prediction methods , which are based on some strong deterministic hypothesis . for nonlinear chaotic dynamics methods
this means , at least , a lower value for lyapunov s predictability time , ( sello @xcite ) . furthermore , we note that the level of wavelet entropy can help to quantify the intermittence degree in solar activity caused by all the stochastic sources and , in particular , by stochastic fluctuations in meridional circulation of plasma , which are able to influence significant unpredictable variation in amplitude and phase of the solar cycle , as recently evidenced by dikpati & charbonneau ( @xcite ) .
finally , we point out that the new information , added by the wavelet entropy , it is not useful for sunspot predictions . in order to investigate more accurately the periodicities which drive the evolution of the wavelet entropy , we performed a wavelet analysis on the above derived time series . for this analysis we used a wavelet software provided by c. torrence and g. compo , and available at url : http://paos.colorado.edu/research/wavelets .
fig.3 depicts the wavelet map with @xmath20 confidence level regions . here
the scale axis shows the periods in log values .
the u - shaped line corresponds to the so - called cone of influence , i.e. the region of the wavelet spectrum in which edge effects , due to the finite - length of signal , become important .
( torrence et al .
the global power spectrum ( right part , pow ) indicates that there is mainly one significant persistent periodicity related to the wavelet entropy , and localized around 40 years , a possible period doubling value of the main schwabe cycle ( lawrence et al .
@xcite ) . other detected lower periodicities are more localized and energetically weaker , below the @xmath20 significance level of the assumed red - noise background spectrum ( right part , sl ) .
the time evolution of the wavelet entropy suggests also the existence of other longer modulations , not resolvable in the available time series .
it is interesting to point out possible relations existing between wavelet entropy behaviour and solar excursion phases , recently investigated by mursula et al .
( @xcite ) .
the excursion phases of the solar dipole are linked to the strong half solar rotation periodicity and arise when the heliospheric current sheet is flat and tilted .
in particular , the authors found a clear connection between sunspot activity and the solar dipole tilt .
more precisely , the overall occurrence of excursions remained very weak during solar cycles from 9 to 14 , where the solar dipole moment was lower with a high stability of the coronal evolution , and this is coherent with the lowest values of wavelet entropy .
instead , during the most recent cycles ( 15 - 22 ) we detected largest excursions of the solar dipole , with a different cycle distribution , larger instabilities and a more disturbed heliosheet .
this behaviour is again coherent with the overall increase of the wavelet entropy for the last cycles .
the suggested correlation between wavelet entropy and solar excursion phases is here limited to a pure qualitative viewpoint .
further investigations , with a more quantitavive evaluation based on geomagnetic indices , are thus needed in order to support a reliable strict link between the degree of disorder in the sunspot activity , as measured by wavelet entropy , and the solar dipole mechanism .
from wavelet analysis approach , we considered a measure of the _ disorder _ content of solar activity , following the temporal evolution of the so - called _
wavelet entropy_. the analysis is performed on the monthly time series of sunspot numbers and shows an interesting multiscale evolution of the wavelet entropy . in particular , we found a clear increase in the disorder content of solar activity for the current @xmath0 solar cycle
this result suggests general low accuracies reachable for current solar cycle prediction methods , which are based on some strong deterministic hypothesis . finally , we suggested a possible strict correlation between the wavelet entropy qualitative behaviour and the solar excursion phases of solar dipole , linked to coronal complexity evolution
|
using wavelet analysis approach , we can derive a measure of the _ disorder _ content of solar activity , following the temporal evolution of the so - called _
wavelet entropy_. the interesting feature of this parameter is its ability to extract a dynamical complexity information , in terms of frequency distribution of the energy content , avoiding restrictions , common in the nonlinear dynamics theory , such as stationarity .
the analysis is performed on the monthly time series of sunspot numbers . from the time behaviour of the wavelet entropy
we found a clear increase in the disorder content of solar activity for the current @xmath0 solar cycle .
this result suggests general low accuracies for current solar cycle prediction methods .
moreover , we pointed out a possible connection between wavelet entropy behaviour and solar excursion phases of solar dipole .
|
the discovery of an unexpected diminution in the observed energy fluxes coming from supernovae type ia @xcite , which are thought of as standard candles , has been interpreted in the context of the standard cosmological model as evidence for an accelerating universe dominated by something called dark energy .
this fact is one of the most puzzling and deepest problems in cosmology and fundamental physics today .
although the cosmological constant seems to be the simplest explanation for the phenomenon , several dynamical scenarios have been tried out ( see , for instance , @xcite and references therein ) .
it is worthwhile to emphasize that the evidence for an accelerating universe mainly relies on energy flux measurements for type ia supernovae at different values of cosmological redshifts .
they provide the most direct and consistent way to determine the recent expansion history of the universe .
nevertheless the relation between the cosmological redshift and the energy flux for a point - like source involves not only the evolution of the universe during the light journey but some assumptions about the nature of the light itself .
customarily , one accepts the linear maxwell theory to describe the light propagation , where light propagates without interacting with other electric or magnetic fields .
however , in the context of non - linear electrodynamics the interaction between the light emitted from such distant sources and cosmological magnetic backgrounds modifies the relation between the redshift and the flux of energy .
if this kind of effect were not correctly interpreted it could lead to an erroneous conclusion about the expansion history of the universe . concretely
, an effect coming from non - linear electrodynamics could explain the curves of luminosity distance vs redshift for type ia supernovae without invoking dark energy .
this remark drives us to study the propagation of non - linear electromagnetic waves in an expanding universe with a magnetic background
. we will benefit from recently obtained results for born - infeld electromagnetic plane waves propagating in a magnetic uniform background in minkowski space - time @xcite .
born - infeld electrodynamics @xcite is a non - linear theory for the electromagnetic field @xmath0 governed by the lagrangian @xmath1 where @xmath2 is a new fundamental constant and @xmath3 and @xmath4 are the scalar and pseudoscalar field invariants @xmath5 ( in minkowski space - time it is @xmath6 and @xmath7 , @xmath8 and @xmath9 being the electric and magnetic fields respectively ) .
born - infeld electrodynamics goes to maxwell electromagnetism when @xmath10 .
in particular , the born - infeld field of a point - like charge reaches the finite value @xmath2 at the charge position and becomes coulombian far from the charge .
born - infeld theory is the only non - linear spin 1 field theory displaying causal propagation and absence of birefringence @xcite .
nowadays , born - infeld theory is reborn in the context of superstrings because born - infeld - like lagrangians emerge in the low energy limit of string theories @xcite .
born - infeld - like lagrangians have been also proposed to describe a matter dynamics able to drive the universe to an accelerated expansion @xcite . in spite of this revival of born - infeld s ideas , there is no experimental evidence for born - infeld effects in electrodynamics : the value of @xmath2 remains unknown ( see an upper bound for @xmath2 in @xcite ) . in the next section
we will summarize the recently obtained results on born - infeld waves propagating in minkowski space - time in the presence of a uniform magnetic background @xcite .
these results properly adapted will be used in section iii to understand the born - infeld energy flux coming from a point - like source in a spatially flat friedman - robertson - walker ( frw ) expanding universe . in section iv we will reformulate the relation between the luminosity distance @xmath11 and the redshift @xmath12 within the framework of born - infeld electrodynamics .
we will show that the presence of magnetic backgrounds modify the curves @xmath11 vs @xmath12 , and we will analyze the consequences for the measurements of the luminosity distance of supernovae type ia .
free born - infeld electromagnetic plane waves do not differ from maxwell plane waves . however , if the wave propagates in the presence of background fields then the propagation velocity becomes lower than @xmath13 as a consequence of the non - linearity of the theory .
this issue has been studied in @xcite by considering the propagation of discontinuities .
recently the exact solution for a born - infeld electromagnetic plane wave propagating in a static uniform magnetic background has been obtained @xcite .
the result retrieves the value for the phase velocity obtained in the above mentioned references : @xmath14 where @xmath15 is the background field and @xmath16 is its component along the propagation direction .
furthermore , the results of @xcite exhibit an up to now unknown feature : the wave develops a longitudinal electric field which depends on @xmath16 and @xmath17 ( the projection of the background field @xmath18 on the polarization direction ) .
in fact , the total electric and magnetic fields result @xmath19 where @xmath20 is the ( usual ) transversal electric field of the wave , and @xmath21 is its phase . the longitudinal electric field of the wave is @xmath22 ( of course @xmath23 vanishes when @xmath24 ) . as a consequence ,
the poynting vector @xmath25 acquires a component transverse to the propagation direction . in born - infeld electrodynamics
the poynting vector results @xmath26 for a wave @xmath27 the time averaged values of the transversal and longitudinal components of the poynting vector associated with the field ( [ field ] ) are @xmath28 @xmath29\ + \
{ \cal o}(b^{-4})\label{poynting2}\ ] ] notice that the transversal part of @xmath30 is parallel to the transversal background field @xmath31 .
the angle @xmath32 between the ray and the propagation direction is @xmath33 although this effect resembles the behavior of the extraordinary ray in anisotropic media , it should be emphasized that no birefringence exists in this case , which is a distinctive feature of born - infeld non - linear electrodynamics @xcite . in born - infeld electrodynamics
the energy density is @xmath34\ ] ] then , the time averaged energy density associated with the wave is @xmath35\\ & & + \frac{{b}^2}{8\ , \pi } \bigg[1 -\frac{{b}^2}{4 b^2 } \bigg]+\ , { \cal o}(b^{-4 } ) \label{rho}\end{aligned}\ ] ] the above enumerated properties provides the rules to propagate born - infeld light rays in the presence of static uniform magnetic backgrounds .
in this section we will study the energy flux of born - infeld waves propagating in a spatially flat frw expanding universe , @xmath36 the conformal time @xmath37 is related with the cosmological time @xmath38 ( the proper time of the comoving fluid ) through the equation @xmath39 .
the energy - momentum conservation laws , @xmath40 , can be written as @xmath41_{,\nu}+\gamma^{\mu}_{\nu\rho}\:\sqrt{-g}\:t^ { \rho\nu}=0\ ] ] for the geometry ( [ metric ] ) it is @xmath42 , thus the energy balance becomes @xmath43_{,0 } + a^2\ , { t_0^\alpha}_{,\alpha}-a_{,0}\ , a\ t = 0\label{conservacion}\ ] ] where @xmath44 and @xmath45 is the trace of the energy - momentum tensor
. we will use ( [ conservacion ] ) to study electromagnetic waves propagating in a magnetic background ; so , @xmath46 in ( [ conservacion ] ) includes both the wave and the background fields . in maxwell electromagnetism
the trace @xmath47 is identically null . as a consequence ,
if @xmath48 solves ( [ conservacion ] ) in minkowski space - time ( @xmath49 ) then @xmath50 will solve ( [ conservacion ] ) in the spatially flat frw universe .
since maxwell energy - momentum tensor is quadratic in the fields , the former assertion means that the maxwell fields scale with the factor @xmath51 .
this general conclusion is applicable to the particular case of a wave propagating in a magnetic background . on the contrary ,
@xmath47 is non - null in born - infeld electrodynamics .
thus , the scaling of @xmath52 with the factor @xmath53 does not guarantee the fulfillment of ( [ conservacion ] ) .
an additional correction is needed , which must vanish in the limit @xmath24 .
we will search this correction to the lowest order in @xmath54 for the born - infeld plane wave propagating in a magnetic background , whose main features were depicted in section ii .
of course , we will assume that the magnetic background do not sensitively affects the homogeneity and isotropy of the space - time dominated by matter and ( presumably ) dark energy , because the energy density of the background field is negligible compared with matter and cosmological constant densities ( a typical value for this field is @xmath55 ) .
since we are going to consider the lowest order correction in @xmath54 for the magnitudes described in section ii such as ( [ elong ] ) , ( [ poynting1 ] ) , ( [ poynting2 ] ) and ( [ rho ] ) , we remark that any magnitude of this order or bigger will only require the maxwellian scaling of the fields with the factor @xmath51 . in the case of the propagation velocity ( [ beta ] ) the correction can also be obtained from the results in @xcite , where it is shown that the equation accomplished by the wave four - vector can be understood as if the rays propagate along null geodesics of an effective metric @xmath56 . in the case of the born - infeld electrodynamics
the effective metric @xmath56 is : @xmath57 where @xmath58 is the space - time metric and @xmath59 is the electromagnetic background where the ray propagates , in this case the components of the background are , @xmath60 for a ray propagating along the @xmath61 direction it is @xmath62 when the effective metric ( [ geomef ] ) is evaluated for the magnetic background ( [ campob ] ) it results @xmath63 thus , it is obtained @xmath64 by integrating the ray path in ( [ geoefe ] ) we recognize that the phase @xmath65 has to be replaced with @xmath66 ( compare with the adiabatic treatment for an oscillator with slowly variable frequency @xcite ) . notice that , since @xmath67 and @xmath68 as in minkowski space - time , then the derivatives of @xmath69 in ( [ conservacion ] ) will preserve their minkowskian structure .
the trace of the born - infeld energy - momentum tensor is ( see for instance @xcite ) @xmath70=\frac{s^2+p^2}{2\pi\ , b^2}+{\cal o}(b^{-4 } ) \label{trace}\ ] ] and scales with @xmath71 .
let us consider a solution @xmath48 for the born - infeld field in minkowski space - time ; i.e. , @xmath72 ( for the case under consideration , @xmath73 is the energy density ( [ rho ] ) , and @xmath74 are the components of the poynting vector ( [ poynting1]-[poynting2 ] ) ) . as stated above
, the scaling of the fields with @xmath51 is not enough to get a solution for ( [ conservacion ] ) , since the term associated with the trace is now non - null .
as the trace is of order @xmath54 , only @xmath52 has to be corrected in ( [ conservacion ] ) .
so , let us try with the scaling @xmath75 . by replacing in ( [ conservacion ] )
we obtain @xmath76 therefore @xmath77 since the trace is of order @xmath54 , then @xmath3 , @xmath4 and @xmath78 can be computed with the maxwellian fields . the time averaged trace for a plane wave traveling in a magnetic background is @xcite @xmath79 thus @xmath80 results @xmath81
according to the results of previous section , we should correct the minkowskian poynting vector ( [ poynting1 ] ) , ( [ poynting2 ] ) with the factor @xmath82 @xcite .
thus the energy flux along the propagation direction is @xmath83\ + \ { \cal o}(b^{-4})\\ = \ \left(1-\frac{\delta_b}{a(\eta)^4\ , b^2}\right)\frac{e_o^2}{8\ , \pi\ , a(\eta)^4}\ + \ { \cal o}(b^{-4 } ) \label{poyntingfrw}\end{aligned}\ ] ] where @xmath84 is positive .
actually we are interested in light rays emitted from point - like sources in the universe . when a light ray travels in the intergalactic space , it undergoes a background magnetic field in different regions of the traveled path . in such case , the values of @xmath85 , @xmath17 , etc .
implied in the propagation should be taken as representative values of the intergalactic background field along the ray trajectory ; so @xmath86 is a magnitude of the same order as the representative squared intergalactic field . for a radial propagation
the minkowskian spherical energy flux is obtained from the plane flux by dividing by the square of the radial distance to the source @xmath87 . therefore the radial energy flux in a frw
spatially flat universe becomes @xmath88 in maxwell electrodynamics the background field does not interfere with the wave , but in non - linear born - infeld electrodynamics the first correction in ( [ tetarpro ] ) displays a coupling between the wave and the background magnetic field .
the luminosity of an object results from integrating the flux at the time of emision @xmath89 , this integration normalizes @xmath90 to fit the value of the luminosity @xmath91 . by using ( [ tetarpro ] ) one obtains @xmath92 combining ( [ tetarpro ] ) and ( [ lum2 ] ) the energy flux can be written as @xmath93\ ] ] the luminosity distance @xmath11 is defined as ( see for instance @xcite ) @xmath94 where @xmath95 is the flux measured at time @xmath96 at the position of the observer @xmath97 ( the source is at @xmath98 ) .
therefore @xmath99\ + \ { \cal o}(b^{-4})\ ] ] in order to relate the luminosity distance with the redshift let us consider the motion of a ray : @xmath100 .
so , if a ray is emitted at time @xmath89 from a source located at @xmath98 , and arrives at time @xmath96 to the position @xmath97 of the observer , then the wavecrest emitted at time @xmath101 will arrive at the observer at time @xmath102 , in such a way that @xmath103 or , equivalently @xmath104 thus @xmath105 , i.e. @xmath106 . therefore the redshift is @xmath107\ + \
{ \cal o}(b^{-4})\end{aligned}\ ] ] by inverting this relation one obtains @xmath108\:\frac{b_{\bot}^2}{2 a_o^4 b^2}\bigg]\ ] ] this quotient is one of the components in the luminosity distance ( [ dl ] ) .
the other one is the proper distance @xmath109 .
this distance depends on how the universe evolves .
in fact , following the motion of the ray , it results @xmath110 where @xmath111 is the redshift of a wave emitted at time @xmath112 .
one can replace @xmath113 in terms of the hubble parameter @xmath114 by performing the derivative of the logarithm of ( [ ao / ae ] ) , @xmath115\:\frac{dz}{dt}\ ] ] thus @xmath116\ : \frac{\beta(z')\ , dz'}{(1+z')h(z')}\\ \nonumber & & \simeq \int_0^z \frac{a_o}{a(z')}\:\bigg[1+\big(1+z'\big)^4\ : \frac{3 b_{\bot}^2}{2 a_o^4 b^2}\bigg]\ : \frac{dz'}{(1+z')h(z')}\\ & & \simeq \int_0^z \bigg[1+\big(4(1+z')^4 - 1\big)\ : \frac{b_{\bot}^2}{2 a_o^4 b^2}\bigg]\ : \frac{dz'}{h(z')}\label{aoro}\end{aligned}\ ] ] einstein equations say that the hubble parameter for a spatially flat universe dominated by matter and cosmological constant is @xcite @xmath117\ ] ] where @xmath118 and @xmath119 are the contributions from matter and cosmological constant to the total density of the universe ( so we are considering here @xmath120 ) . from (
[ ao / ae ] ) one knows that @xmath121\frac{3 b_{\bot}^2}{2 a_o^4 b^2}\bigg]+\omega_\lambda\ ] ] therefore the proper distance ( [ aoro ] ) times the present hubble parameter results @xmath122\\\end{aligned}\ ] ] by replacing this integral in ( [ dl ] ) and combining it with ( [ ao / ae ] ) , the luminosity distance turns out to be @xmath123 f_l(z)\end{aligned}\ ] ] where @xmath124dz\ ] ] and @xmath125 in ( [ hodl ] ) the last two terms containing @xmath126 and @xmath127 characterizes the correction to the luminosity distance coming from the background magnetic field at the considered order @xcite .
the factor multiplying the third term , @xmath128 , is negative . the functions @xmath129 and @xmath130 depends on @xmath118 and @xmath119 .
in order to analyze the extra terms in the luminosity distance we will display the case where the three components of the representative magnetic background are equal .
calling these components @xmath85 and replacing in ( [ hodl ] ) we obtain : @xmath131 where @xmath132 f_l(z)\ ] ] figure [ correccion ] shows the behavior of @xmath133 for three different models .
if @xmath134 then @xmath133 has a maximum at @xmath135 and becomes negative when @xmath136 , if @xmath137 @xmath133 has a maximum at @xmath138 and becomes negative when @xmath139 and if @xmath140 then @xmath133 has a maximum at @xmath141 and becomes negative when @xmath142 .
it can be seen that the lower @xmath118 the higher @xmath143 . for three different sets of @xmath144 . if @xmath145 the correction becomes negative ( @xmath143 is the zero of @xmath133).,width=302 ] figure [ distancia ] displays the luminosity distance times the present hubble parameter ( [ hodl ] ) for the three cases considered in fig.[correccion ] ( we have chosen @xmath146 ) . when @xmath147 the leading terms in ( [ hod5 ] ) are @xmath148 .
so the slope at low redshift gets a contribution coming from the background field .
( @xmath149).,width=302 ] figure [ onlymatter ] compares the low redshift behavior of standard cosmology ( @xmath150 , @xmath151 , @xmath24 ) with the one of born - infeld electrodynamics models without cosmological constant ( @xmath140 ) .
notably the observations @xmath11 vs @xmath12 can be well fitted without using cosmological constant by choosing @xmath152 . in order to better appreciate the role of born - infeld electrodynamics in these curves , fig.[onlymatter ] also includes the curve resulting from an ordinary ( maxwellian ) cosmology with @xmath140 .
although non - linear electrodynamics effects could explain the curves of luminosity distance vs redshift for type ia supernovae @xcite without invoking dark energy , one should be cautious .
accepting a typical value of @xmath55 @xcite for the cosmological background magnetic field @xmath153 @xcite , together with the constraint @xmath154 for the born - infeld parameter @xcite , then the corrections to the standard cosmology would be negligible .
even so , non - linear electrodynamics should be considered as a source of degeneration in the curve @xmath11 vs @xmath12 .
figure [ comparacion ] compares the curve of the standard cosmology ( @xmath137 , @xmath155 , @xmath156 ) with the one resulting from born - infeld electrodynamics for the same values of @xmath118 and @xmath119 .
the curves intersect at @xmath157 . if @xmath158 the curves get their maximum separation at @xmath138 . if @xmath159 the luminosity distance predicted by born - infeld electrodynamics becomes smaller than the one from standard cosmology . in this last case
the curves seem to go dramatically apart ( however this feature should be confirmed by extending the calculus up to a higher order of approximation ) .
the degree of separation of both curves is governed by the value of @xmath160 .
future observations could allow a test of these features to obtain a constraint for this value .
, @xmath155 .
comparison between maxwell ( dashed line ) and born - infeld ( solid line ) electrodynamics ( @xmath161).,width=302 ]
in this paper we have solved the born - infeld equations for electromagnetic plane waves propagating in a background magnetic field . in the absence of a background field ,
the born - infeld plane waves are equal to the maxwell ones . on the contrary , in the presence of a background magnetic field
@xmath18 the non - linear effects modify both the phase and the amplitude of the wave with corrections that depend on the combination @xmath162 , where @xmath163 is the scale factor of the universe .
it is remarkable that born - infeld electrodynamics depends on @xmath163 and @xmath2 only through the combination @xmath164 .
this means that the maxwellian approximation ( @xmath24 ) also corresponds to the limit @xmath165 .
so , although the electromagnetic field is presently well described by maxwell equations for a wide range of phenomena , the non - linear born - infeld electrodynamics could have an influence in the past when the scale factor was smaller .
therefore the expanding universe is a good laboratory to test born - infeld electrodynamics ; many non - linear aspects of its equations could be relevant when highly redshifted objects are observed . in this work
we have begun the search for this kind of effects .
we found that the influence of born - infeld electrodynamics on the luminosity distance ( [ hod5 ] ) exhibits interesting features that could be experimentally established by means of more precise supernova observations and a better knowledge of the cosmological background fields .
firstly , the experimental data for @xmath11 vs @xmath12 could be fitted without invoking dark energy , although there is no observational evidence of the background field that would be required .
secondly , the shape of the curve @xmath11 vs @xmath12 predicted by the standard cosmology ( @xmath137 , @xmath155 , @xmath10 ) for high redshifts differs appreciably from the one predicted by born - infeld electrodynamics , which opens the possibility of detecting non - linear electrodynamics effects in a future .
fradkin and a.a .
tseytlin , phys .
b * 163 * ( 1985 ) , 123 .
a. abouelsaood , c. callan , c. nappi and s. yost , nucl .
b * 280 * ( 1987 ) , 599 .
leigh , mod .
lett . a * 4 * ( 1989 ) , 2767 .
metsaev , m.a .
rahmanov and a.a .
tseytlin , phys .
b * 193 * ( 1987 ) , 207 .
tseytlin , nuc .
b * 501 * ( 1997 ) , 41 .
tseytlin , in _ the many faces of the superworld _ , ed .
m. shifman , world scientific ( singapore ) 2000 .
v. dyadichev , d. galtsov , a. zorin and m. zotov , phys .
d * 65 * , 084007 ( 2002 ) .
r. garca - salcedo and r. breton , int .
a * 15 * , 4341 ( 2000 ) .
m. sami , n. dadhich and t. shiromizu , phys .
b * 568 * , 118 ( 2003 ) .
e. elizalde , j. lidsey , s. nojiri and s. odintsov , phys .
b * 574 * , 1 ( 2003 ) .
a. howard and r. kulsrud , astrophys .
j. * 483 * , 648 ( 1997 ) , m. giovannini , int . j. mod
d * 13 * , 391 ( 2004 ) , p. p.
kronberg , rep .
prog . phys . *
57 * , 325 ( 1994 ) , l. m. widrow , rev .
mod . phys . * 74 * , 775 ( 2003 ) . in curved space - time , @xmath166 ( one covariant index and one contravariant index ) coincide with the tensor components in an orthonormalized basis ; thus , they effectively give the measure of the energy flux .
|
born - infeld electromagnetic waves interacting with a static magnetic background are studied in an expanding universe .
the non - linear character of born - infeld electrodynamics modifies the relation between the energy flux and the distance to the source , which gains a new dependence on the redshift that is governed by the background field .
we compute the luminosity distance as a function of the redshift and compare with maxwellian curves for supernovae type ia .
|
phase competition is a hallmark of strongly correlated electron systems that exhibit ground states with rather distinct order yet of comparable energy .
changing parameters in the hamiltonian by applying external fields , stress , or chemical composition allows one to tune from one state to another .
complex phase diagrams divulge , on the one hand , our limitation to make quantitative predictions for a given compound . on the other hand ,
the nature of the competing ordered states reflects the relevant degrees of freedom in a low energy description .
an important aspect of phase competition that is crucial for a realistic description of correlated materials is the role of disorder and impurities .
iron - based superconductors display a phase diagram characterized by antiferromagnetism , nematic order , and superconductivity along with regions in the parameter space where the non - magnetic normal state displays quantum critical and more conventional behavior@xcite .
numerous arguments support a sign - changing superconducting state , where the @xmath1 state , with opposite sign of the cooper pair wave function on hole and electron pockets , is the most prominent example@xcite . among the strongest evidence in favor of this state
are the emergence of a spin - resonance mode in inelastic neutron scattering experiments@xcite and the field - dependence of the quasi - particle interference pattern in scanning tunneling spectroscopy@xcite . in refs .
it was argued that the nature of the phase competition between antiferromagnetism and superconductivity can be another powerful tool to distinguish between sign - changing and sign - preserving superconducting states .
this conclusion was based on two key ingredients : i ) the same electrons that contribute to the ordered antiferromagnetic moment also contribute to the cooper pair condensate and ii ) there exists at least an approximate particle - hole symmetry between the electron and hole bands shifted by the ordering vector of the magnetic order @xmath2 : @xmath3 this latter condition seems to be reasonably well satisfied in many iron - based superconductors , allowing one to make the connection between phase stability and pairing state , a conclusion that can not be drawn so easily in other systems .
if both conditions are fulfilled , refs . concluded that antiferromagnetism and conventional ( @xmath0 ) superconductivity will be separated by a strong first - order transition as sketched in fig .
afm and sc phase transition lines meet at a bicritical point .
in contrast , in the case of @xmath1 pairing ( and similarly for @xmath4-wave pairing ) , the system is at the verge between a first - order transition and crossing second - order transitions with a regime of homogeneous and simultaneous order of both states as depicted in fig .
depending on details of the electronic structure and interactions , afm and sc phase transition lines meet at a bicritical point or at a tetracritical point , respectively .
the observation of crossing second - order lines in some , but not all , systems was then argued to be strong evidence for unconventional pairing .
thus , one can `` read off '' the superconducting pairing state from the phase diagram of the iron - based superconductors . at the heart of this conclusion
was the fact that the magnetic order parameter has a typical momentum @xmath2 which couples the superconducting condensates in electron and hole sheets of the fermi surface , similar to an internal josephson coupling in momentum space .
the effect is rooted in the same coherence factors that lead to the resonance mode enhancement at the transferred momentum @xmath2 in inelastic neutron scattering experiments . the theory of refs .
was based on the assumption that disorder plays no role in the phase competition . however , the widely studied system ba(fe@xmath5co@xmath6)@xmath7as@xmath7 and a number of related systems are clearly affected by disorder .
this insight was revealed in first - principle calculations@xcite , and very clearly demonstrated in recent nexafs experiments that identified the spin and valence state of co@xcite .
a rigid band calculation of a perfectly clean system as performed in refs .
is therefore not sufficient .
disorder has been shown to strongly affect the phase competition between antiferromagnetism and superconductivity .
it is therefore crucial to investigate the role of disorder on the interplay between superconducting pairing and phase competition .
more generally , the investigation of the role of disorder in unconventional superconductors has proven to yield important clues with regard to the competition between alternative states of order@xcite . in addition
, the more general question of how one can manipulate the degree of competition between different phases clearly deserves more detailed attention . in this paper
, we investigate the role of disorder on the phase competition between magnetism and superconductivity . in our analysis
, we therefore consider the regime where the ordering temperatures of both states remain finite , i.e. , are not suppressed due to disorder . by analyzing several experimentally motivated models for the microscopic nature of disorder we demonstrate that the distinct phase competition between itinerant antiferromagnetism and either @xmath0 or @xmath1 superconductivity is further enhanced if one includes disorder .
thus , the statements of refs . , relating the phase diagram and the nature of the pairing state , apply as well to disordered systems .
before we enter a microscopic analysis of the phase competition , we phenomenologically describe the situation of two second order phase transitions , one antiferromagnetic and one superconductive , that meet in a multicritical point .
the resulting phase diagrams can be divided in two classes : either superconductivity is able to microscopically coexist with antiferromagnetism , or superconductivity and antiferromagnetism mutually exclude each other . in fig .
[ fig : phasediagrams ] , these two types of phase diagrams found in iron pnictide materials are sketched .
the first class shows a bicritical point in the phase diagram , and a first - order transition line between superconductivity and antiferromagnetism , fig .
the second class shows a tetracritical point in the phase diagram , a region where both order parameters are nonzero ( referred to as _
homogeneous coexistence _
@xcite ) , and all phase - transitions are second - order , fig
. 1b . around the multicritical point in the phase diagram
, the free energy can be expanded simultaneously in both order parameters , in the spirit of ginzburg - landau theory of superconductivity @xcite .
the most generic form of the free energy , allowed by the symmetry , in terms of the antiferromagnetic @xmath8 and the superconductive @xmath9 order parameters reads as @xmath10 .
\label{eq : free - energy - expansion}\end{aligned}\ ] ] the last quartic term in eq .
( [ eq : free - energy - expansion ] ) has to be positive , @xmath11 , in order to ensure competition between the two ordered phases .
gradient terms accounting for temporal and spatial fluctuations of the order parameters can be included in this expansion of the free energy as well , and will be discussed later . whether the phase transitions are first or second order , and thereby the shape of the phase diagram near the multicritical point , is determined by the quartic coefficients .
the analysis of the quadratic form associated with the quartic terms suggests to introduce the quantity @xmath12 which determines the shape of the phase diagram .
for @xmath13 we encounter a phase diagram with second - order phase transitions only , and thus a tetracritical point . for @xmath14 , antiferromagnetism and superconductivity
are separated by a first - order phase transition line that terminates in a bicritical point .
below , we will determine microscopic expressions for the coefficients in eq .
for a disordered system with electron electron interactions .
the resulting values of these coefficients determine the location of the multicritical point as a function of microscopic parameters .
changing the disorder strength or other parameters in the hamiltonian will then affect the location of the multicritical point .
is then valid in the vicinity of this multicritical point . in ref .
, using a weak coupling analysis , it was found for the clean system that @xmath15 and @xmath16 . in case of perfect particle - hole symmetry ,
the latter result is exact , as shown in ref . .
in what follows , we derive the coefficients of the expansion from a microscopic model describing the essential features of the iron pnictides in presence of disorder .
the model is introduced in section [ sec : model ] and its implications in the presence of disorder are studied in the remainder of this paper .
point is a circular band whereas the electron band centered around @xmath17 could in principle be of elliptic form .
additionally we could introduce a finite chemical potential @xmath18 to fit our model to more realistic band structures , and introduce @xmath19 . , scaledwidth=35.0% ] we consider the two - band model that is illustrated in fig .
[ fig : two - band - model ] .
it consists of a circular hole band at the @xmath20 point , and an elliptical electron band shifted by vector @xmath17 .
this is the minimal model in which phase competition of antiferromagnetism and @xmath0 or @xmath1 superconductivity in the iron pnictides can be studied .
the hamiltonian @xmath21 of this two - band model contains the usual non - interacting part @xmath22 where we label the two bands by the index @xmath23 . @xmath24 and @xmath25 are the creation and annihilation operators in the respective band . since small changes in the band structure
lead to small changes of @xmath26 , we focus on the particle - hole symmetric case in the following . deviations from particle - hole symmetry were investigated in refs . .
within the assumption of particle - hole symmetry , no further details of the dispersion @xmath27 are needed to calculate @xmath26 . in our analysis of the coefficients of the momentum dependence
we assume a parabolic dispersion , i.e. , @xmath28 , and @xmath29 in case of particle - hole symmetry .
note that band 2 is centered around @xmath17 which in our notation is included in the band index .
the electrons are subject to an effective electron
electron interaction leading to superconductivity and antiferromagnetism . here , we do not attempt to find microscopic expressions for the pairing interaction but are investigating the consequences of alternative pairing states .
so , we assume for simplicity that the superconductive and antiferromagnetic couplings are described by two microscopic hamiltonians of different nature
. superconductivity is described by a bcs - like hamiltonian , @xmath30 where @xmath31 refers to the opposite of @xmath23 .
the electron
electron interaction leading to superconductivity is present for electrons with energies within a shell of width @xmath32 around the fermi energy . for phonon - mediated electron
electron interaction leading to conventional superconductivity , this energy cut - off would be given by the debye frequency . in case of an electronic pairing mechanism
, the cut - off is expected to be of the order of the fermi energy .
the same electrons that form the cooper pairs are subject to an interaction that might lead to magnetic order .
antiferromagnetism shall be described in an itinerant picture , by @xmath33 where @xmath34 is the characteristic energy cutoff for magnetic interactions .
its nature is unimportant for our results , and in what follows we assume that it is of the order of or smaller than the fermi energy , @xmath35 .
weak quenched disorder is represented by the potential @xmath36 of @xmath37 randomly distributed identical impurities , @xmath38 where @xmath39 are the field operators in position space . here
@xmath40 is the impurity potential matrix element that may describe intraband scattering ( @xmath41 ) as well as interband scattering processes ( @xmath42 ) . in the following
we assume the impurity potential @xmath43 to be short - ranged .
we describe impurity scattering in the born approximation , in which the rate of electron collisions with impurities is characterized by @xmath44 where @xmath45 is the fourier component of the potential of a single impurity and @xmath46 the impurity concentration .
we consider the model with a sufficiently smooth disorder , which leads to significantly smaller interband than intraband scattering .
this hierarchy of scattering rates in iron pnictides is also supported by experiment@xcite . without loss of generality
we may assume @xmath47-wave scattering which corresponds to @xmath48-correlated disorder when considering the scattering in one band .
more general models of disorder will not change the results qualitatively , leading only to the replacement of the elastic scattering time by the transport scattering time .
then , the scattering amplitudes are characterized by constants @xmath49 ( @xmath50 ) for scattering within band 1 ( 2 ) , and by a constant @xmath51 for scattering between the bands .
impurity scattering in this two - band model is therefore characterized by intraband scattering rates @xmath52 and @xmath53 , and the interband scattering rate @xmath54 , which can be also assumed momentum - independent . from eq .
follows that the scattering rate @xmath55 is real , whereas @xmath56 is allowed to have a nontrivial phase .
both scattering rates have equal magnitude , @xmath57 .
note , the generalization of this approach to more extended impurities , discussed in refs . and , will be done elsewhere@xcite . in this paper
, we consider several models of disorder to describe the corresponding physically relevant limits . _
a. dominant hole - band scattering .
_ several experiments and first - principles calculations have demonstrated that intraband scattering in the hole band ( labeled by 1 ) is significantly stronger than in the electron band ( band 2 ) .
this is supported for example by transport measurements @xcite , scanning tunneling microscopy@xcite , as well as first - principles density functional theory calculations@xcite .
the physically relevant limit is therefore @xmath58 for the corresponding intraband scattering rates .
however , in this paper , we consider a more general model where the values of the intraband scattering rates are arbitrary , @xmath59 .
furthermore , the transition temperature @xmath60 is suppressed with increasing impurity concentration . in case of @xmath1 pairing , this is caused by interband scattering processes@xcite .
the suppression of the sc transition temperature is an order of magnitude smaller than theoretical prediction based on the scattering rates obtained from transport experiments@xcite .
hence the interband scattering rate @xmath61 that leads to suppression of @xmath60 is smaller than the intraband scattering rate in the band that dominates the transport properties**???**. thus , in iron pnictides @xmath62 seems a reasonable starting point .
since the intraband scattering rate @xmath63 in the hole band is clearly the largest scattering rate in iron pnictide materials , we neglect interband scattering in the first model .
since without interband scattering , the intraband scattering rates in the two bands simply add up in the physical observables calculated in this paper , a finite intraband scattering rate in the electron band yields qualitatively similar results , and we consider our model in the limit , @xmath64 where the physically relevant limit to iron pnictides is @xmath58 .
this model , summarized in eq .
, will be referred to as model a throughout this paper . _
b. investigation of interband scattering .
_ although the interband scattering seems to be significantly weaker than the intraband scattering in the hole band , setting the interband scattering rate to zero is an oversimplification with respect to some aspects .
the weak suppression of the sc transition temperature is one example of the consequences of a finite interband scattering rate .
therefore , we analyze the influence of a finite interband scattering rate as well , and use the model @xmath65 for the investigation of interband scattering on the phase competition in iron pnictides .
this model of disorder in iron pnictides has already been considered by ref . in a slightly different context .
we argued , that the interband scattering rate is smaller than the largest intraband scattering rate @xmath63 , thus the range @xmath66 is the limit interesting for the ratio of interband to intraband scattering rates in iron pnictides .
the model itself however is not limited to this parameter range , and allows for the analysis of arbitrary ratios @xmath67 .
we will refer to this model with a finite interband scattering rate , summarized in eq .
, as model b in the following .
in this section , we derive as an illustration the full ginzburg - landau expansion for model a of disorder in iron pnictides , based on the dominant scattering mechanism .
thus we consider the interband scattering rate @xmath61 to be zero , and finite intraband scattering rates @xmath63 and @xmath68 in bands 1 and 2 , respectively .
experimental evidence suggests that the hole band is more severely affected by impurities than the electron band , so for the sake of clarity we concentrate on the limit @xmath69 in the following derivation of the free energy .
we also calculated the expansion of the free energy for the more general case of arbitrary intraband scattering rates @xmath70 .
the results will be presented at the end of this section .
one can calculate the coefficients in the ginzburg - landau expansion using the eilenberger@xcite approach or straightforwardly using perturbation theory based on the vertices depicted in fig .
[ fig : diagram - elem ] .
the quadratic coefficients correspond to the diagrams presented in fig .
[ fig : quadratic_coefficients ] , and the quartic coefficients correspond to those shown in figs .
[ fig : quartic - diagrams1 ] and [ fig : quartic - diagrams2 ] . in order to arrive at an expansion of the free energy of the model a introduced in section [ sec : model ] , we may write the partition function as @xmath71+\mathcal{h}_\mathrm{afm}\left[\psi^\dagger\left(\tau\right),\psi^{}\left(\tau\right)\right ] } \right>_{\tilde{0}}\ ] ] in matsubara interaction representation where the average @xmath72/\mathcal{z}_{\tilde{0}}$ ] refers to the non - interacting part @xmath73 .
we follow the usual procedure to make the action quadratic in the fermionic operators by introducing hubbard - stratonovich fields and find an effective action in terms of these fields . in momentum and frequency space , we use complex scalar fields @xmath74 and @xmath75 to decouple @xmath76 , and a three - component vector field @xmath77 to decouple @xmath78 , where @xmath79 are bosonic matsubara frequencies . thereby we find for the effective action in terms of these fields @xmath80 \label{eq : eff - hamiltonian - sc}\end{aligned}\ ] ] and @xmath81{~ .}\label{eq : eff - hamiltonian - afm}\end{aligned}\ ] ] from the effective action we may construct the elements of a diagrammatic technique to derive the expansion of the free energy , and the superconducting and antiferromagnetic part of the action contain different types of vertices associated with the fields @xmath82 , @xmath83 and @xmath84 , see fig . [ fig : diagram - elem ] . in order to embed impurity scattering in the diagrammatic technique , we include impurity lines as new diagrammatic elements in our formalism , each associated with a factor @xmath85 .
impurity scattering gives rise to a finite self energy and vertex corrections . within this model of impurity scattering ,
the propagator in the electron band is given by the bare electron propagator , @xmath86 .
the propagator in the hole band is given by @xmath87{./propagator}}}=g_{1,{\boldsymbol{\mathbf{k}}}}(\nu_n ) = \frac{1}{\mathrm{i}\nu_n-\xi_{1,{\boldsymbol{\mathbf{k}}}}+\frac{\mathrm{i}}{2\tau_1}\operatorname{sgn}\nu_n } { ~ , } \ ] ] with a finite self energy due to impurity scattering which we here treat in the lowest non - vanishing order ( born approximation ) .
contributions with crossed impurity lines are neglected since they are suppressed by a small factor@xcite @xmath88 where @xmath89 is the mean free path , and @xmath90 is the fermi velocity .
therefore we construct our diagrams from vertices that already contain the full propagators .
additionally , vertices are renormalized due to impurity scattering which is indicated by a shaded region in the diagrams
. there are no vertex corrections of the vertices associated with @xmath82 and @xmath91 to consider here since we set @xmath92 and @xmath69 , respectively .
vertex corrections of vertices associated with @xmath93 are given by the cooperon ladder in band 1 which leads to a frequency - dependent factor @xmath94 to leading order in @xmath95 and @xmath79 . here the restriction of @xmath96 is implied in every summation that contains this cooperon ladder .
again , in the calculation of the vertex corrections , and also in the construction of the diagrams , all contributions from crossed impurity lines can be neglected due to the small factor @xmath88 .
these prerequisites enable us to derive the full ginzburg - landau expansion of the free energy of our two - band model in presence of weak impurity scattering which reads @xmath97 in frequency and momentum space .
we note , that eq .
( [ a1 ] ) accounts for the gradient terms @xmath98 , @xmath99 , @xmath100 and @xmath101 in the free energy , that characterize spatial and temporal fluctuations of the order parameters which is reflected in the dependence of the quadratic coefficients on finite incoming frequency @xmath79 and momenta @xmath95 . in our analysis
, we implicitly assumed that the sole effect of disorder is to change the values of the coefficients of the order parameters . in the critical regime , it is well established that disorder may change the universality class of the transition , lead to griffiths and quantum griffiths effects , or even cause glassy behavior close to the transition point .
these effects , however , only become important in the very close vicinity of the critical point@xcite as a consequence of the weakness of the interaction sufficient to induce the ordered state . thus , on the one hand , we consider our model sufficiently far away from the multicritical point to ignore these effects , while at the other hand sufficiently close to the multicritical point such that an expansion of the free energy is justified . refs . demonstrated that this intermediate regime covers a wide range if the pairing and magnetic interactions are sufficiently weak . the diagrams for the quadratic terms in the free energy
are shown in fig .
[ fig : quadratic_coefficients ] , and the leading - order behavior in the limiting cases of vanishing and strong disorder is @xmath102 where @xmath103 \nonumber \\ & \approx \tfrac{4}{v_\mathrm{m}}-4{\rho_{\mathrm{f}}}\ln\tfrac{\lambda_\mathrm{m}}{\max\left(t , { \tau_1}^{-1}\right ) } { ~ , } \\ a_{\mathrm{s},11 } & = a_{\mathrm{s},22 } = -2{\rho_{\mathrm{f}}}\left[\operatorname{\psi_0}\left(\tfrac{3}{2}+\tfrac{\lambda_\mathrm{s}}{t}\right)-\operatorname{\psi_0}\left(\tfrac{1}{2}\right)\right ] \nonumber \\ & \approx -2{\rho_{\mathrm{f}}}\ln\tfrac{\lambda_\mathrm{s}}{t } { ~ , } \\ a_{\mathrm{s},12 } & = a_{\mathrm{s},21 } = -\tfrac{2}{v_\mathrm{s}}{~ .}\end{aligned}\ ] ] the magnetic critical points @xmath104 are the points of the phase diagram ( as sketched in fig .
[ fig : phasediagrams ] ) where @xmath105 holds , and analogously , the superconducting critical points are defined by @xmath106 . note that doping can affect the values of the coupling constants @xmath107 and @xmath108 .
the intersection of the two critical lines then defines the multicritical point of the phase diagram .
the renormalization of the superconductive vertex describes diffusion of cooper pairs in band 1 , which , however , does not affect the sc transition temperature because the coefficient @xmath109 does not depend on the scattering rate @xmath63 anymore , and it coincides with the respective result in band 2 which is not affected by impurity scattering . for nonmagnetic impurities in a usual @xmath47-wave superconductor , this constitutes the anderson theorem@xcite . from eq .
follows that the magnetic ordering temperature vanishes for @xmath110 , where @xmath111 is the corresponding transition temperature of the clean system .
the phase competition discussed in this paper is , of course , only sensible for @xmath112 , where both competing states order . .
the characteristic length scales of the order parameter fluctuations , obtained from the ginzburg - landau expansion under the assumption of second - order phase transitions at the respective critical points . [ cols="^,^,^",options="header " , ] from the quadratic coefficients @xmath113 and @xmath114 at finite @xmath95 and @xmath79 we also find the typical length scales of fluctuations of the magnetic and superconducting order parameters to be affected differently by disorder . they are summarized in table [
table : length - scales ] . in the clean case
, we find the same characteristic length scale for both order parameters .
it corresponds to the result for the coherence length of a superconductor@xcite , and is independent of disorder strength . for strong disorder ,
both lengths are reduced with increasing scattering rate @xmath63 .
however , the magnetic length is stronger suppressed by disorder than the superconductive coherence length , cf .
table [ table : length - scales ] .
thus , even if the spatial variation of both order parameters is the same in the clean limit , it is different if one includes disorder .
then , the characteristic length scales for the magnetic degrees of freedom become shorter .
the coefficients of the quartic terms of pure sc and afm , which are not due to phase competition , are depicted in fig .
[ fig : quartic - diagrams1 ] . the resulting coefficients are @xmath115 again , the coefficients associated with superconductivity do not depend on the strength of disorder whereas the quartic coefficient associated with the magnetic order parameter does .
the diagrams contributing to the coefficients of the quartic terms reflecting the phase competition between magnetic order and superconductivity are depicted in fig .
[ fig : quartic - diagrams2 ] , and the results are @xmath116\nonumber \\ & = \left\{\begin{matrix } \frac{7\zeta(3){\rho_{\mathrm{f}}}}{2\pi^2t^2 } { ~ , } & t\tau_1\gg 1 \\ \frac{\pi{\rho_{\mathrm{f}}}\tau_1}{t } { ~ , } & t\tau_1\ll 1 \end{matrix}\right . { ~ , } \\
\gamma_{12 } & = \gamma_{21 } = -16{\rho_{\mathrm{f}}}\tau_1 ^ 2 \left[\operatorname{\psi_0}\left(\tfrac{1}{2}+\tfrac{1}{8\pi t\tau_1}\right ) -\operatorname{\psi_0}\left(\tfrac{1}{2}\right)\right ] \nonumber \\ & \qquad + \tfrac{2{\rho_{\mathrm{f}}}\tau_1}{\pi t}\operatorname{\psi_1}\left(\tfrac{1}{2}\right ) \nonumber \\ & = \left\ { \begin{matrix } \frac{7\zeta(3){\rho_{\mathrm{f}}}}{4\pi^2t^2 } { ~ , } & t\tau_1 \gg 1 \\ \frac{\pi{\rho_{\mathrm{f}}}\tau_1}{t } { ~ , } & t\tau_1\ll 1 \end{matrix } \right .
{ ~ .}\end{aligned}\ ] ] depending on the assumption for the underlying symmetry of the superconducting order parameter , these results can be summarized into @xmath117 where the indices refer to @xmath0 and @xmath1 symmetry of the order parameter @xmath9 , and we omitted the expressions for arbitrary @xmath118 for the sake of brevity . in the case of strong disorder , we find @xmath119 to leading order , i.e. , superconducting and magnetic order parameters completely decouple in the limit @xmath120 , thus competition between magnetism and superconductivity ceases to exist in this limit . as the consideration of finite interband scattering shows , this complete decoupling only occurs in the limit @xmath121 , but even at finite interband scattering rates , the competition between sc and afm order is mitigated by the intraband scattering .
since we are in the regime of weak disorder , the limit of large @xmath122 is understood in the sense that @xmath123 , where @xmath124 denotes the fermi energy . in the case of zero interband scattering rate
, these calculations can be easily generalized to arbitrary finite intraband scattering rates @xmath63 and @xmath68 in band 1 and 2 , respectively .
our calculations show that the intraband scattering rates in the absence of interband scattering simply add up to a total scattering rate @xmath125 .
therefore , the structure of the resulting coefficients remains the same , and the corresponding coefficients can be obtained by substitution of @xmath126 for @xmath63 in the previously discussed expansion of the free energy .
the full ginzburg - landau expansion now allows us to calculate @xmath127 which determines the nature of the phase diagram . here
, the index refers to the respective symmetry of the sc order parameter . in fig .
[ fig : g ] , we plotted @xmath26 as a function of @xmath128 for the two pairing symmetries under consideration , since we expanded the free energy around the multicritical point , where @xmath129 .
in the two limiting cases of vanishing disorder and strong disorder , we find @xmath130 for the two respective cases . as a function of @xmath131 for @xmath0 and @xmath1 pairing symmetries.,title="fig:",scaledwidth=23.0% ] as a function of @xmath131 for @xmath0 and @xmath1 pairing symmetries.,title="fig:",scaledwidth=23.0% ] in the clean limit ( @xmath132 ) we recover the results obtained from a model disregarding disorder@xcite .
@xmath15 means that @xmath0 superconductivity can not coexist with antiferromagnetism , whereas @xmath16 allows for both possible types of phase diagrams since a more detailed band structure may lead to a small positive or negative @xmath133 .
this was taken as evidence against @xmath0 superconductivity to be realized in the iron pnictides .
the consideration of disorder supports this reasoning since @xmath134 increases with disorder and the @xmath0 pairing state is even more inconsistent with the observed variety of phase diagrams .
the @xmath1 pairing state is driven towards the regime of phase coexistence by increasing disorder but the consideration of a more detailed band structure would still allow for both types of phase diagrams .
the result of @xmath135 in the limit of strong disorder results from the complete decoupling of afm and sc orders since in this limit @xmath136 . note that our findings also imply the possibility of a disorder - induced transition from mutual exclusion to coexistence of superconductivity and antiferromagnetism in the iron pnictides .
these findings are summarized in fig .
[ fig : result ] . in case of @xmath1 and @xmath0 pairing.,scaledwidth=40.0% ]
our treatment of the problem with a finite interband scattering rate @xmath61 in eilenberger formalism@xcite is analogous to the approach described in ref . , and we consider a model with finite interband scattering rate @xmath61 and equal intraband scattering rates @xmath137 in the hole and electron bands as well .
this approach allows for the expansion of the sc and afm gap equations which can , up to a factor , be identified with the first derivative of the free energy with respect to the order parameters .
we extract the coefficients of the free energy expansion from the expanded gap equations . in presence of interband scattering , the coefficients contributing to the quantity @xmath138 which determines the nature of the multicritical point , read @xmath139 \nonumber \\ & \qquad -\tfrac{2{\rho_{\mathrm{f}}}\tau_0}{\pi t}\left[\left(1+t\right)\operatorname{\psi_1}\left(\tfrac{1}{2}+\tfrac{1+t}{4\pi t\tau_0}\right)+t\operatorname{\psi_1}\left(\tfrac{1}{2}+\tfrac{t}{4\pi t\tau_{0}}\right)\right ] { ~ , } \\ \gamma_{++}&= -\tfrac{8{\rho_{\mathrm{f}}}\tau_0 ^ 2}{(1+t)^2}\left[\operatorname{\psi_0}\left(\tfrac{1}{2}+\tfrac{1+t}{4\pi t\tau_0}\right)-\operatorname{\psi_0}\left(\tfrac{1}{2}\right)\right ] -\tfrac{2{\rho_{\mathrm{f}}}\tau_0}{\pi t}\tfrac{1}{1+t}\left[\operatorname{\psi_1}\left(\tfrac{1}{2}+\tfrac{1+t}{4\pi t\tau_0}\right)-2\operatorname{\psi_1}\left(\tfrac{1}{2}\right)\right ] { ~ , } \label{eq : interband - gamma++ } \end{aligned}\ ] ] where @xmath140 is the ratio of interband to intraband scattering rate . in the limit of @xmath92 ,
this corresponds to the results discussed in section [ sec : intraband ] , but with the intraband scattering rate doubled , since now the scattering rate in the electron band is finite , and equal to the scattering rate in the hole band .
again , we find that afm and @xmath1 sc decouple in the limit of @xmath141 ( @xmath135 ) , resulting in a phase diagram exhibiting a region where afm and sc coexist microscopically whereas for @xmath0 sc , @xmath142 , and afm and sc exclude each other . for finite interband scattering rates , afm and sc no longer decouple completely in the limit of strong disorder but the coupling @xmath143 is reduced in the case of @xmath1 pairing compared to the clean case .
the stronger interband scattering is , the closer @xmath133 is to zero , but for all ratios @xmath144 we found @xmath145 , implying a phase diagram with a tetracritical point and coexistence of sc and afm .
for @xmath0 pairing , in contrast , we find @xmath146 and @xmath147 , irrespective of the ratio @xmath67 .
furthermore , we found that in case of @xmath0 pairing , the interband scattering simply adds to the intraband scattering rate .
thus , the qualitative behavior does not depend on @xmath67 , and we find @xmath148 , thus @xmath0 sc is not able to coexist microscopically with afm , not even in the presence of inter- and/or intraband scattering .
these results are summarized in fig .
[ fig : result - interband ] . in conclusion , for the @xmath1 pairing state , intraband scattering and interband scattering are antagonistic processes , but the effect of intraband scattering is always stronger , even in the limit of @xmath149 , whereas for the @xmath0 pairing state , the rates of these two scattering processes simply add up .
thus in the presence of interband scattering , we find the @xmath0 pairing state to be inconsistent with phase diagrams revealing a regime of microscopic coexistence of afm and sc .
the analysis of the @xmath1 state , in contrast , suggests that this order parameter symmetry will always result in a phase diagram exhibiting a regime of coexistence of afm and sc .
still , a more detailed analysis including finite ellipticity and/or chemical potential could also lead to @xmath14 , thus allowing for both types of phase diagrams . since ellipticity and chemical potential yield only small corrections
, they could not bring the @xmath0 state to coexists with magnetic order .
therefore , the analysis of finite inter- and intraband scattering supports the reasoning based on phase competition against the @xmath0 as a suitable candidate for the pairing state in iron pnictides .
we studied a model of iron pnictides and related iron - based superconductors and included impurity scattering in the microscopic model .
we developed two complementary simplified models for impurity scattering in the iron pnictides motivated by experimental observations .
model a concentrates on the most important scattering process in the materials under consideration which is intraband scattering in the hole band .
therefore , we neglected intraband scattering in the electron band and all types of interband scattering processes , since @xmath150 . model b focuses on the interband scattering rate , neglected in model a , yet makes the simplifying assumption that the intraband scattering rates of both bands are the same .
we derived the full ginzburg - landau expansion of the free energy from this microscopic model . from the quadratic coefficients we find that the transition temperature of neither @xmath0 nor @xmath1 superconductivity is influenced by impurity scattering if we take only intraband scattering into account .
this is in accordance with the anderson theorem . we further compared the coherence lengths obtained from the gradient terms in the expansion and found the coherence length of the magnetic order parameter more strongly reduced by impurity scattering than the coherence length of the superconducting order parameter . from the quartic coefficients we concluded how impurity scattering affects the phase competition in the iron pnictides .
our analysis supports the argument obtained in the clean case@xcite that @xmath0 superconductivity is inconsistent with phase diagrams that show microscopic coexistence of antiferromagnetism and superconductivity .
this behaviour occurs in models with and without interband scattering .
thus the consideration of disorder provides an even stronger argument against the @xmath0 state to be realized in the iron pnictides and supports @xmath1 superconductivity or other sign - changing superconducting states .
we thank a. v. chubukov , r. m. fernandes , and p. hirschfeld for helpful discussions .
this work was supported by the deutsche forschungsgemeinschaft through dfg - spp 1458 ` hochtemperatursupraleitung in eisenpniktiden ' .
the work of svs was partially supported by the alexander von humboldt foundation through feodor lynen research fellowship and by the nsf grants dmr-1001240 and phy-1125844 .
to illustrate the calculation of the diagrams for the coefficients in the free energy , let us provide here a detailed computation of the coefficient @xmath151 in the framework of our simplified model . in the absence of interband scattering
there is one diagram that contributes to the coefficient @xmath151 . in the following ,
we use the abbreviation @xmath152 and use that the cooperon ladder only depends on the absolute value of @xmath153 . the diagram , fig .
6c , evaluates to @xmath154 this sum may be conveniently evaluated approximately in the limits @xmath155 and @xmath156 , as well as exactly . to calculate the coefficients from the diagrams we have to include the proper symmetry factor which is 2 in the case of @xmath151 .
the resulting coefficient is then given by @xmath157 \nonumber \\ & = \left\{\begin{matrix } \frac{7\operatorname{\zeta}(3){\rho_{\mathrm{f}}}}{4\pi^2 t^2 } & { ~ , } \quad t\tau_1\gg 1{~ , } \\
\frac{\pi{\rho_{\mathrm{f}}}\tau_1}{t } & { ~ , } \quad t\tau_1\ll 1{~ .}\end{matrix } \right.\end{aligned}\ ] ]
the coefficients of the ginzburg - landau expansion for model b , given in eq . to
, have been obtained from the equation of state using the eilenberger formalism @xcite .
the application of the eilenberger formalism to a system showing antiferromagnetism and @xmath1 superconductivity can be done in complete analogy to ref . , using the same parametrization of the eilenberger green s function . in this appendix , we sketch the respective procedure for @xmath0 superconductivity . the mean - field hamiltonian can be written as @xmath158 where we summarized the fermionic operators in the two bands into @xmath159 and introduced the hamiltonian matrix consisting of noninteracting and mean - field parts , @xmath160 where @xmath161 , @xmath162 , and @xmath163 are the pauli matrices in nambu , band , and spin space , respectively .
the matrix green s function is defined by @xmath164 where the self energy of model a ( intraband scattering with rate @xmath137 in both bands , interband scattering with rate @xmath61 ) is given by @xmath165 since in the gap equations as well as in the self energy , the matrix green s function only appears integrated over momenta , it is convenient to introduce the eilenberger ( or quasiclassical ) green s function @xmath166 and rewrite all the equations in terms of @xmath167 . from eqs . to
, we find self - consistently that for our model , the eilenberger green s function must be of the form @xmath168 this parametrization can be used to obtain an expansion of the gap equations for the order parameters @xmath9 and @xmath169 .
such an expansion corresponds , up to an overall prefactor , to the first derivative of the free energy with respect to the order parameter , i.e. , the equation of state .
therefore , we can extract the quadratic and quartic coefficients of the free - energy expansion . by using the parametrization in the self energy ,
we find that the self energy depends on intraband and interband scattering rate only via the total scattering rate @xmath170 for the @xmath0 pairing state .
thus we expect no qualitatively new effects due to interband scattering here that were not already captured in the analysis of intraband scattering .
|
we analyze the effect of weak disorder on the competition between antiferromagnetic order and superconductivity in a model for iron - based superconductors . under the assumption of an approximate particle - hole symmetry
we show that conventional @xmath0 superconductivity can not be realized in the case of coexisting magnetic and superconductive orders , observed experimentally at intermediate doping levels .
this result holds for arbitrary impurity concentrations , and , in particular , in the clean limit .
the inclusion of disorder further amplifies the phase competition between itinerant antiferromagnetism and conventional superconductivity .
in addition , we analyze the effect of disorder on the characteristic length scales of the two order parameters , and find that in a disordered sample the staggered moment fluctuates on shorter scales than the superconductive order parameter , even if both length scales are the same in the clean limit .
|
in phenomenologically interesting models the effects of broken supersymmetry in the visible sector are conveniently parameterized , working in an off - shell formulation , by the expectation values of the auxiliary components of some hidden sector supermultiplets . among the auxiliary fields , the scalar @xmath1 , belonging to the graviton supermultiplet , @xmath2 , stands out as special .
indeed , unlike for auxiliary fields belonging to matter and gauge hidden sector multiplets , the coupling of @xmath1 is completely fixed ( at the leading relevant order ) once the masses and self - couplings of the low energy effective theory , prior to supersymmetry breaking , are specified .
this property just follows from @xmath1 being a partner of @xmath3 whose coupling is equally well specified by the energy momentum tensor of the low energy effective theory .
the scenario of ` anomaly mediated ' ( am ) supersymmetry breaking corresponds to the limiting case in which the contribution of @xmath1 dominates over all of the others @xcite .
the name ` anomaly mediated ' is due to the fact that in the mssm @xmath1 only couples to the visible fields at the quantum level , via a supersymmetric analogue of the dilatation anomaly of non - supersymmetric field theory .
the purpose of this paper will not be to build phenomenological models based on am , but rather to investigate some of its more amusing theoretical aspects .
in fact , far away from the domain of phenomenology , we shall be working in four dimensional supersymmetric anti - de - sitter ( ads ) space .
we nonetheless believe that our study provides interesting additional insight into the properties of am , in particular its being uv insensitive , in spite of being uv generated . to set the stage ,
it is convenient to derive am terms via the superconformal approach to supergravity @xcite . at tree level
, the most general two - derivative lagrangian may be written as @xmath4_d+ \left \{\bigl [ s^3 w(\phi_i)+ f(\phi_i ) w^{\alpha}w_{\alpha})\bigr ] _ f + { { \rm h.c.}}\right \ } , \ ] ] where @xmath5 and @xmath6 are superconformally invariant densities , provided that the chiral superfield , @xmath7 , and the matter fields , @xmath8 , have weyl weights 1 and 0 , respectively .
interesting actions are obtained by consistently taking the lowest component of @xmath7 with non - vanishing expectation value .
this breaks the superconformal group down to poincar supergravity and turns @xmath7 into a purely auxiliary field , formally restoring scale invariance , hence the name ` superconformal compensator ' .
indeed a suitable superconformal gauge can be chosen where @xmath9 .
the couplings of the auxiliary field @xmath1 are thus fixed by dilations and r - symmetry .
in particular a classically scale invariant subsector , like the mssm , couples to @xmath1 only at the quantum level . for a massless gauge theory the coupling of @xmath7
is easily read off by demanding formal scale- ( and r- ) invariance of the 1pi action at 1-loop @xmath10_f + h.c.\ ] ] by expanding in components , one finds a gaugino mass term which is proportional to the @xmath11-function @xmath12 the dependence of @xmath13 on @xmath7 is local , compatibly with its being uv generated . however
, it belongs to a non - local supergravity invariant ` structure ' ( involving @xmath14 ) , and this is why it is convenient to use the 1pi action to determine it .
this is just the supersymmetric generalization of a dilaton coupling to the trace anomaly , hence the name ` anomaly mediation ' . in models with broken supersymmetry and vanishing cosmological constant , @xmath15 , implying a 1-loop contribution of order @xmath16 to gaugino and sfermion masses .
however , one may also have @xmath17 , with unbroken supersymmetry on ads . in that case
, the expectation value is given by the superpotential : @xmath18 , where @xmath19 is the ads radius .
indeed , at tree level , @xmath20 generates the mass splittings , of order of the ads curvature , that are required by supersymmetry in ads .
the rle of a loop effect like anomaly mediation is less clear in this case , though it ought to be easy to understand , given that the theory still enjoys unbroken supersymmetry .
the purpose of this note is to explain the rle played by anomaly mediation in supersymmetric ads .
this issue was briefly considered in @xcite , in the context of a general discussion in which the short distance origin of am was emphasized .
however our explanation for the rle of am in ads space differs from the one proposed in @xcite .
we will argue that the existence of am is a necessary consequence of supersymmetry , given the large - distance properties of ads space , in particular the presence of a ( conformal ) boundary . in this sense ,
our work represents yet another way of deriving am masses , purely via consideration of ir saturated quantities .
the outline is as follows . in section [ adssusy ] ,
we review supersymmetry in ads and supersymmetric qed therein . in section [ mass ] , we compute the 1-loop contributions to the gaugino self - energy in sqed with massless matter , and discuss the implications for the gaugino mass . in section [ out ] , we present conclusions . the case of sqed with massive matter
is relegated to the appendix .
in this section , we briefly review some basic features of supersymmetry in four - dimensional ads space which will be relevant for the following discussion . for more details , see @xcite and refs . therein .
the isometry group of ads@xmath0 is @xmath21 , whose unitary , infinite - dimensional representations are denoted by @xmath22 , where @xmath23 and @xmath24 represent respectively the energy and spin of the lowest energy state in the representation .
the lagrangian mass parameter of the corresponding fields ( in units of @xmath25 ) are functions of @xmath23 and @xmath24 .
for instance , for the simplest cases of @xmath26 , we have @xmath27 just as in flat space , the simplest irreducible representations of the super - group @xmath28 correspond to chiral and vector supermultiplets .
a chiral supermultiplet decomposes into the following representations of @xmath21 : @xmath29 note that the supersymmetry generators raise and lower @xmath23 by a half - integer .
then , according to eqs .
( [ scalar]),([fermion ] ) , the mass terms for fermions and scalars within the same supermultiplet are not , in general , the same .
these splittings are mandated by @xmath28 and originate within the lagrangian from two sources .
one source is the non - vanishing ricci scalar and the other source is @xmath20 .
notice , finally , that in the special case of the conformally - coupled supermultiplet , with @xmath30 , the two scalars have the same mass , even though they belong to different representations : namely @xmath31 and @xmath32 .
turning now to the massless vector supermultiplet , the @xmath21 representation content is @xmath33 this multiplet is both conformally coupled and ` short ' , corresponding to its being related to a gauge invariant lagrangian .
a massive vector multiplet , on the other hand , is characterized by @xmath34 , and decomposes as @xmath35 this is a long multiplet that can be viewed as arising from a higgs mechanism .
indeed , it has the same state multiplicity as the direct sum of the massless vector supermultiplet and the goldstone supermultiplet , whose content is @xmath36 .
since it corresponds to multiplet shortening , the masslessness condition must be stable in perturbation theory .
in particular , the gaugino mass , for an unbroken gauge symmetry , must be zero to all orders .
the presence of the anomaly mediated contribution to the mass ( [ gauginomass ] ) is , navely , at odds with the previous observation that the gaugino should be massless . to clarify the rle of am , we shall focus on the simplest non - trivial example , that is the mass of the gaugino in supersymmetric qed .
our theory consists of @xmath37 supergravity with a vector superfield @xmath38 , and two chiral superfields @xmath39 , with opposite charges @xmath40 .
the khler and superpotential functions are given by ( throughout the paper we use the conventions of wess and bagger @xcite ) @xmath41 since we shall be working in the neighbourhood of @xmath42 , we neglect the higher order terms indicated by @xmath43 .
the constant term in the superpotential gives rise to the ads@xmath0 background and to the expectation value of the compensator , @xmath44 we will find it technically convenient to work in the poincar patch , with metric @xmath45 the co - ordinates @xmath46 ( @xmath47 ) and @xmath48 cover only one of an infinite set of similar poincar patches of the full ads space .
however poincar co - ordinates cover the whole euclidean ads ( eads ) , which can be obtained just by the substitution @xmath49 ( see for instance the discussion in ref .
this last property indicates that , if properly interpreted , computations on the poincar patch yield informations about the properties of qft on full ads . assuming @xmath19 to be positive , in these co -
ordinates the four unbroken supersymmetries are parameterized by the killing spinors @xmath50+z^{-1/2 } x_\mu \sigma^\mu [ \epsilon_0+i \sigma^3 \bar{\epsilon}_0 ] , \label{unbrokensusy}\ ] ] where @xmath51 is a two - component constant spinor .
notice that the killing spinors naturally decompose into two real spinors of @xmath52 .
the first of these corresponds to the standard @xmath37 in 2 + 1 dimensions , while the other corresponds to the conformal supersymmetry .
in fact , for our purposes it will suffice to consider the flat supersymmetries , as the others are implied by the ads isometries . by taking the limit @xmath53 with
@xmath19 fixed , we decouple gravity and focus on quantum effects that are purely due to sqed on ads@xmath0 .
the relevant lagrangian is , therefore , @xmath54 -m ( \psi_+\psi_-+\bar{\psi}_+\bar{\psi}_-)\nonumber \\ & -&(m^2 - \frac 2 { l^2 } ) ( |\phi_+|^2+|\phi_-|^2)+\frac m l ( \phi_+\phi_- + \phi_+^*\phi_-^ * ) \nonumber \\ & + & i g \sqrt{2}\lambda(\psi_+ \phi_+^*-\psi_-\phi_-^*)- i g
\sqrt{2}\bar{\lambda}(\bar{\psi}_+ \phi_+-\bar{\psi}_- \phi_-),\end{aligned}\ ] ] where , without loss of generality , we have taken @xmath55 to be real .
one sees that the scalars acquire non - holomorphic mass terms , originating from the non - vanishing ricci scalar , and holomorphic ( b - type ) masses , arising from the compensator f - term .
( the fermionic mass and interaction terms , by contrast , retain the same form as in flat space . ) the scalar mass eigenstates and their masses are given by @xmath56 eqs .
( [ gauginomass]),([compensator ] ) imply the presence of an am contribution to the gaugino mass , given by @xmath57 as explained above and emphasized in @xcite , a gaugino mass would be incompatible with supersymmetry in ads@xmath0 . indeed , for @xmath58
, there is an additional contribution to @xmath59 , corresponding to a finite threshold effect at the scale @xmath55 , where matter is integrated out .
this is due to the presence of both a fermion mass and an r - breaking b - type mass for the scalars . by the well known property of am in flat space ,
we can directly conclude that , at least for @xmath60 , the threshold effect cancels eq .
( [ gauginobilinear ] ) , at least up to subleading effects of @xmath61 .
however , it would be nice to see the exact cancellation in an explicit computation . moreover , in the limit @xmath62
, corresponding to conformal multiplets , there seems to be a puzzle , in that all sources of r - symmetry breaking disappear from the matter lagrangian ! in other words , for @xmath62 there is , at first sight , no obvious contribution in addition to eq .
( [ gauginobilinear ] ) . in @xcite
, it was concluded that the contribution in eq .
( [ gauginobilinear ] ) does not affect the physical mass ( defined in the sense of the representation of ads ) , since @xmath63 runs to zero in the infrared .
this explanation is , however , puzzling , as it requires an all - orders resummation of diagrams , while we expect the supersymmetry algebra to be satisfied at each finite order in perturbation theory .
furthermore , this argument can not be applied to the non - abelian case . in
actual fact , the resolution of the gaugino mass puzzle has to do with the boundary conditions in ads , which shall be discussed in the next section .
what we shall find there is that boundary effects provide a calculable , ir saturated , contribution to the gaugino bilinear in the 1-loop 1pi effective action .
this contribution corresponds to a mass @xmath64 which exactly cancels the uv one @xmath65 the most relevant feature of ads space , for our discussion , is the presence of a ( conformal ) boundary located at @xmath66 in the poincar patch ( [ ads@xmath0 ] ) .
one immediate consequence of the presence of a 2 + 1-dimensional boundary is that chiral symmetry is always broken in ads@xmath0 @xcite .
this is fully analogous to what happens in a field theory on half of flat space : when a fermion travelling towards the boundary is reflected , the momentum flips sign , while @xmath68 is conserved .
thus , helicity is not conserved .
more formally , chiral symmetry is broken by the boundary conditions that are necessary to define the theory .
this can be seen by considering a two component spinor propagating on half of flat space , with action @xmath69 .
\label{flat}\ ] ] the variation of the action is @xmath70_{z=0}.\ ] ] in order to obtain sensible boundary conditions ( _ i.e. _ not over - constraining ) , a boundary term @xmath71 must be added to the action , where @xmath72 is an arbitrary phase .
the variational principle then demands that @xmath73 implying that chiral symmetry is broken even for vanishing bulk mass , without loss of generality one can choose @xmath74 . ] .
the generalization to ads requires some care , because of the divergent scale factor at @xmath66 .
the boundary conditions in this case can be derived by considering the behavior of the solutions close to @xmath66 . without loss of generality
, we can choose @xmath75 .
normalizability of the solution requires that @xmath76 and again chiral symmetry is necessarily broken .
note that for the ads case , there is no freedom to chose the phase @xmath72 .
this is basically because the bulk mass operator itself plays the rle of a boundary mass term .
this is easily seen by performing a weyl rescaling , @xmath77 : the lagrangian for @xmath78 is just given by eq .
( [ flat ] ) , but with a position dependent mass @xmath79 , which blows up at @xmath66 .
the exponent in the asymptotic behavior is precisely the index @xmath23 of the corresponding representation .
note that for @xmath80 , two inequivalent boundary conditions are possible , corresponding to a double quantization , as happens for scalars in ads @xcite .
the existence of one and two solutions respectively for @xmath81 and @xmath82 , nicely matches eq .
( [ fermion ] ) and the unitarity bound @xmath83 . in the qed case ,
the boundary condition ( [ adsfermion ] ) for a single charged spinor would break electric charge ; in order to conserve electric charge , the boundary conditions must relate @xmath84 to @xmath85 .
the charge preserving boundary condition is forced on us by normalizability . for @xmath86 ,
compatibly with normalizability , there exist two other , inequivalent , charge - breaking boundary conditions .
we will consider these other possibilities elsewhere .
] repeating the exercise above with the two spinors , normalizability of the solutions requires @xmath87 given the boundary conditions for the fermions , supersymmetry then determines the boundary conditions for the scalars . by acting with the unbroken supersymmetries ( [ unbrokensusy ] ) on the fermionic boundary conditions
, one finds @xmath88\nonumber\\
0\leq ml < \frac 1 2 : \qquad & & z\to 0 \quad\implies\quad\phi_+=\pm \phi_-^*\left[1+o(z)\right ] \label{scalarads}\end{aligned}\ ] ] where the sign in the second eq . is correlated with the sign for the fermions .
we can see that this is consistent with the equations of motion for the scalars : in the scalar sector , by solving the wave equation for the two mass eigenstates , @xmath89 and @xmath90 , we find that @xmath91 for @xmath92 , normalizability alone implies that @xmath93 , corresponding to the first solution in eq .
( [ scalarads ] ) . for @xmath94 ,
the mass of the two scalars is in the range where double quantization is allowed , and so we can choose @xmath95 ( consistently with supersymmetry ) , corresponding to the second solution in ( [ scalarads ] ) .
note that , as a combined effect of the boundary conditions for fermions and scalars , @xmath96symmetry is broken in the matter sector even for @xmath62 .
finally , we can also fix the boundary condition for the vector multiplet . by taking neumann boundary conditions for the gauge field and acting with the supersymmetry transformations , we find that the appropriate sign of the gaugino boundary condition is @xmath97 to summarize , the presence of the boundary in ads@xmath0 always breaks chirality and @xmath96symmetry , even when there is no source of explicit breaking in the bulk action .
the physics is essentially that of half of flat space .
what is special to ads@xmath0 is that the chiral symmetry is broken , while the maximal number of isometries is preserved .
this is , of course , crucial to give a meaning to a mass smaller than the curvature of the space .
the boundary conditions derived above provide the necessary ` mass insertions ' to give rise to an ir contribution to the gaugino mass .
focussing on the case of massless sqed , let us now compute the gaugino mass at 1-loop order .
the computation is particularly transparent in the case of massless matter , where the chiral symmetry breaking is entirely due to the boundary effects .
( we present the massive case in the appendix . )
when @xmath62 , the chiral matter supermultiplet is conformally coupled . as a consequence , the full sqed action in this case is invariant under weyl transformations at the classical level .
this allows us to map the theory in ads space to one living on half of flat space and perform all the computations using familiar flat space formulae .
this is achieved via the superconformal rescaling @xmath98 after the rescaling , @xmath99 and @xmath100 . since sqed is weyl invariant ( at tree level ) ,
the compensator decouples , and we are left with the tree level action for massless , sqed in half of flat space , with a boundary at @xmath66 .
the boundary conditions on the fields are most easily implemented by performing an orbifold projection . from the results in the previous section , we have ( dropping the circumflexes on the fields ) , @xmath101 where @xmath102 is the position of the image point . the flat space propagators can be written down directly using the method of image charges .
for the scalars , one has @xmath103 similarly , for the fermions , @xmath104 ^ 2 } , \\
\langle \psi_{+\alpha}(x_1){\psi}_{-}^{\beta}(x_2)\rangle & = & -\frac 1{2 \pi^2 } \frac{(x_1-\tilde{x}_2)_m ( \sigma^m\bar{\sigma}^3)_\alpha^\beta } { [ ( x_1-\tilde{x}_2)^2+i\epsilon ] ^2}. \nonumber \label{fermionprop}\end{aligned}\ ] ] one can see that the @xmath105 prescription in feynman s propagator selects implicitly boundary conditions at @xmath106 : these are the hartle - hawking boundary conditions , appropriate to the poincar patch @xcite . the off - diagonal propagators determine the chiral - breaking contribution to the gaugino self - energy in fig .
[ diag ] @xmath107 where @xmath108 and where our convention on the self - energy is defined by @xmath109 .
performing the wick contractions , we have @xmath110 ^ 3}. \label{selfenergy}\end{aligned}\ ] ] notice that this contribution is non - local , and comes from long - distance physics , as opposed to eq .
( [ gauginobilinear ] ) . in order to extract from @xmath111 the correction to the gaugino mass
, we must evaluate it on a solution of the massless ( tree level ) wave equation .
this is the analogue of computing the self - energy at zero momentum in flat space .
the general solution of the bulk dirac equation for a massless gaugino is @xmath112 physical states must also satisfy the boundary condition in eq .
( [ gaugeboundary ] ) . in order to achieve that , two solutions with opposite velocity , @xmath113 , in the @xmath48-direction should be superimposed
. however , as we shall explain in a moment , the correct procedure we must follow in the poincar patch in order to study the 1-loop corrected wave equation is to work with solutions of the dirac equation that satisfy boundary conditions at the horizon @xmath114 rather than at the boundary @xmath66 .
this is closely related to the ads / cft prescription .
alternatively we could overcome this issue by performing an euclidean computation , as in this case poincar co - ordinates cover the whole space , but we find it more physical to address directly the lorentzian point of view . to obtain the ir contribution to the gaugino mass , we must convolute eq .
( [ selfenergy ] ) with ( [ dirac4 ] ) .
we thus find , @xmath115 ^ 2}\right)\lambda_{0 \beta}(x_2)\nonumber\\ & = & \frac { i g^2}{8 \pi^4}\int d^3 x_2 \frac{1}{[(x_1-x_2)^2+z_1 ^ 2+i\epsilon ] ^2}e^{ip_\mu x_2^\mu } \xi_{\alpha } \label{integral}\end{aligned}\ ] ] where in the last step we integrated by parts and used @xmath116 . in the resulting boundary
integral , we used the explicit expression for @xmath117 in ( [ dirac4 ] ) . notice that @xmath118 are coordinates on the boundary . performing the last integral
explicitly we thus find , @xmath119 where the @xmath105 in the original integral fixes the sign of @xmath120 to be positive .
the right hand side of eq .
( [ finite ] ) is proportional to the original spinor if this satisfies the hartle - hawking boundary conditions : positive frequencies purely outgoing and negative frequencies purely incoming .
this means that when evaluated on this class of solutions of the bulk dirac equation , the ir contribution to the self - energy @xmath121 , acts like a mass term @xmath64 which is precisely equal and opposite to the anomaly mediated contribution ( see eq .
( [ gauginobilinear ] ) after performing the weyl rescaling in eqs .
( [ rescaling0]),([rescaling ] ) ) .
thus an exact cancellation between uv and ir effects arises , as promised in eq .
( [ uvir ] ) .
it is the clever relation among these two contributions that ensures the masslessness of the gaugino , as demanded by supersymmetry .
this is the main result of our paper .
it remains to be explained why our computation works only for the class of solutions of the form ( [ finite ] ) .
these solutions correspond to the creation of incoming particles at the past horizon @xmath122 and to the destruction of outgoing particles at the future horizon @xmath123 that separate the poincar patch from the rest of ads .
intuitively such processes can be described by causality using solely the fields in the poincar patch .
other solutions correspond to processes that are not captured by the poincar patch alone and probe other regions of global ads . in this case
there will be extra - contributions from the rest of the space and a computation in global coordinates would be required . that such contributions exist follows from the fact that the feynman propagator is non - vanishing between a point inside the poincar patch and one outside . had we worked in global coordinates we could have directly checked that the cancellation of the gaugino mass occurs for arbitrary physical states ( i.e. solutions of the wave equation that satisfy the boundary conditions ) .
our result can however be readily interpreted from the viewpoint of the ads / cft correspondence @xcite .
even though lorentzian ads / cft is not nearly as developed as on euclidean space , we do not see obvious obstructions in the case at hand . from this perspective
, the boundary field combination @xmath124 should be viewed as an external source probing the system ( the dual cft ) .
notice that @xmath125 is precisely the combination that is set equal to zero for the ads quantum fields .
performing a path integral over the bulk fields with vacuum boundary conditions at @xmath126 one obtains a functional @xmath127 which generates the correlators of the associated dual operator in the cft .
given @xmath125 , a classical source localized at the boundary , the choice of initial and final vacuum states for our path integral fixes the boundary condition for the corresponding bulk field at @xmath114 .
working with plane waves , this prescription corresponds precisely to the hartle - hawking boundary condition we encountered previously .
this gives a prescription for finding a unique extension of @xmath125 into the bulk , by requiring that its ( effective ) action be stationary . at tree level
, we have the boundary effective action @xmath128 corresponding to the correlator of a dual fermionic current of scaling dimension @xmath129 : @xmath130 the 1-loop computation we have performed is directly translated into a 1-loop computation of the boundary effective action .
the only difference from before is that we need to consider also solutions with euclidean boundary momenta @xmath131 . in this case
the solution in the bulk corresponds to the unique regular solution at @xmath132 as prescribed by euclidean ads / cft .
needless to say the previous computation can be continued to the euclidean region so that the self energy is diagonal on these solutions . working at 1-loop accuracy , the corrected boundary effective action
is simply obtained by substituting the tree level bulk solution into the 1pi bulk effective action .
however our previous result was precisely that the total ( uv + ir ) 1pi vanishes on the very solution of the massless dirac equation that satisfied the ads / cft boundary conditions at @xmath114 ( that is with the same exponent as in eq .
( [ finite ] ) ) .
thus we conclude that at the 1-loop level the boundary action is unaffected and thus the dimension of the cft operator dual to the gaugino field is not renormalized , consistently with supersymmetry . what we have learned is an amusing lesson on the rle of the anomaly mediated gaugino mass .
the basic reason for its existence is that ads@xmath0 behaves as 2 + 1-dimensional field theory as far as chirality is concerned .
the mass of fermions is thus additively renormalized by calculable boundary effects . on the other hand
, supersymmetry mandates the gaugino to be exactly massless .
the simple sqed case , in the end , shows that the only way to achieve this is via the existence of suitable short distance effects , in one - to - one correspondence with the long distance effects .
this is yet another illustration of the uv insensitivity of anomaly mediation . in the previous section
we have shown that the chiral breaking part in the 1-loop self energy does not correct the gaugino mass , nor , similarly , does it correct the boundary effective action . however , strictly speaking there is yet another contribution to the gaugino self - energy that we need to consider .
this is the ` chirality - preserving ' contribution , @xmath133 , the one associated with wave - function renormalization .
the issue at hand arises even in the absence of supersymmetry .
we will show that this contribution vanishes when acting on a massless spinor .
this result may seem obvious at first sight , based on our usual flat space intuition .
indeed , in flat minkowsky space , lorentz invariance constrains this term to be proportional to @xmath134 , which vanishes on - shell as long as @xmath135 is not too singular ( in fact , @xmath135 is a logarithmic function ) .
however , the situation is more subtle in ads , since , at the quantum level , the boundary makes itself felt even inside the bulk , and therefore the @xmath48 direction is not manifestly equivalent to the others .
the purpose of this section is to clarify this issue .
an extra complication comes from the need to regularize the divergent part of @xmath133 .
we shall again focus on massless sqed , for which we can work in the conformally rescaled basis ( [ rescaling ] ) . the general case is briefly considered in the appendix . working in position space ,
we find it convenient to use the method of differential regularization @xcite .
the unregulated @xmath133 is given by , @xmath136 this corresponds to the following correction to the effective action @xmath137 where @xmath138 .
this expression has , however , a non - integrable singularity at @xmath139 , which must be regulated .
navely , using differential regularization amounts to replacing @xmath140 where @xmath141 plays the rle of the renormalization mass scale .
this can not , however , be the full story , since the explicit mass scale @xmath141 breaks dilatation invariance @xmath142 . in the rescaled basis
, @xmath143 arises as the subgroup of @xmath144 which is left unbroken by the compensator background @xmath145 @xcite . consequently the regulated self - energy in eq .
( [ reg ] ) does not respect the ads isometries . as the lack of invariance follows from the regularization ,
the counterterm needed to restore the symmetry must be local , and must of course involve the compensator . by simple reasoning one can quickly derive the unique form of this counterterm . in order to do so ,
let us imagine that we had regulated the loop in a manifestly covariant fashion , by introducing pauli villars fields with mass @xmath141 .
the crucial aspect of pauli - villars fields is that , being massive , their quadratic lagrangian depends directly on the compensator , @xmath146 , via the substitution @xmath147 which formally restores conformal invariance .
however it does not make any sense to simply perform this replacement in eq .
( [ reg ] ) . to find out how eq .
( [ reg ] ) is modified we must be a tad more careful .
we just need to focus on the @xmath141-dependent part of the regulated self - energy .
using the identity @xmath148 the @xmath141-dependent part of the effective action is given by @xmath149 whose unique local covariantization is\bar \lambda\bar\sigma^3\lambda\,.\ ] ] this term must however be discarded as it explicitly breaks cp ( the regulated theory is formally cp - invariant , even though parity is , of course , ` spontaneously ' broken by the expectation value of @xmath146 ) . ]
@xmath150 . \label{amwf}\ ] ] the local @xmath151 term gives the following correction to @xmath152 @xmath153 on the other hand , from eq .
( [ reg ] ) the ` ir ' contribution to the equation of motion is @xmath154 to investigate how this non - local contribution affects the gaugino mass we must compute it on the solution @xmath117 of the massless wave equation specified by ( [ dirac4 ] ) . integrating by parts and using @xmath155 , eq .
( [ nonlocwave ] ) becomes @xmath156 where the final integral is identical to the one computed in the previous section , eq .
( [ integral ] ) .
again the last identity is only valid for solutions satisfying the hartle - hawking boundary conditions .
we thus find that the contributions in eqs .
( [ uv ] ) and ( [ ir ] ) again cancel so that the propagation of the gaugino is not affected .
in particular the gaugino remains massless .
note that , while the cancellation in the previous section relies on supersymmetry , this effect is independent of supersymmetry .
this cancellation between uv and ir contributions , dictated by the ads isometry ( a subgroup of the conformal group ) , can be viewed as an @xmath157 counterpart of the one found previously .
this is perhaps not surprising , as anomaly mediation is itself the supersymmetric counterpart of the trace anomaly . indeed , in a superfield formalism
, these two separate cancellations would be manifestly related .
we studied the rle played by anomaly mediated ( am ) mass terms in @xmath158 theories on ads@xmath0 with unbroken supersymmetry . for simplicity we focussed on the gaugino mass term in sqed with massless matter .
we showed that the am gaugino mass term is required by the super - ads algebra in order to exactly cancel another 1-loop contribution , of infrared origin and associated with the ads boundary .
the latter effect originates because chirality ( @xmath96symmetry in this case ) is necessarily broken by reflection at a 2 + 1-dimensional boundary .
indeed , by computing first this finite ir effect ( which does not require the introduction of a regulator ) and by using the fact that the algebra dictates a massless gaugino , we could have argued the need for a local , uv generated , am contribution .
since the latter is independent of whether the theory lives in flat or curved space , that would have provided yet another derivation of am gaugino masses .
the possibility of relating the am mass to purely ir quantities illustrates the `` uv insensitivity '' of this effect , a property which makes it potentially relevant in phenomenological applications . the fact that am effects represent local parts of non - local structures in the 1pi action is well known . our result provides a new twist on that perspective
: the am gaugino mass is just a reflection of the breakdown of chirality at the 2 + 1-d boundary of ads@xmath0 .
there are several directions in which one might extend and improve our result .
one obvious possibility is to perform the same computation in the non - abelian case , where , unlike in the abelian case , proper gauge - fixing will be needed .
another problem concerns the rle of all other am terms , such as sfermion masses and a - terms " : it should be possible to derive them from consistency conditions as well , but probably in a more subtle way than for the gaugino mass . in this paper we worked on the poincar patch .
this procedure is clean for the euclidean case and from the ads / cft standpoint : our computation corresponds to checking that , as expected by supersymmetry , the scaling dimension of the operator dual to the gaugino field is not renormalized .
the lorentzian computation is more delicate , as we have to deal with boundary conditions at the horizons which separate the chosen patch from the rest of ads
. it would then be interesting to try to perform the same computation in global coordinates , and check that , in that case , the 1-loop self energy does vanish when convoluted with the normalizable solutions .
finally , it would be interesting to understand the rle of anomaly mediation purely from the cft viewpoint .
the ads bulk picture is that the gaugino must be massless even though chirality is broken , corresponding to non - vanishing @xmath121 off - shell . in the cft picture ,
the non - vanishing of @xmath121 , shows up in the 4-point function of operators dual to the ads matter fields .
however it is not immediately obvious how to translate the bulk picture to the boundary , since there is no notion of chirality in 2 + 1-d field theory .
we would like to thank m. bianchi , s. giddings , m. porrati , s. sybiriakov , a. wulzer and a. zaffaroni for useful discussions .
hk was supported by the cquest of sogang university with grant number r11 - 2005 - 021 and cs by the swiss national science foundation . the work of r.r .
is partially supported by the swiss national science foundation under contract no .
200021 - 116372 .
thanks the aspen center for physics where part of this work was carried out .
the cancellation of the uv and ir contributions to the gaugino mass , being a consequence of the algebra , is a general effect which must hold for any mass of the matter fields . in this appendix
we check explicitly the cancellation for arbitrary values of @xmath55 in the superpotential .
this computation can also be interpreted as the derivation of the anomaly mediated uv contributions ( [ gauginobilinear ] ) , ( [ amwf ] ) using pauli - villars fields . for massless matter ,
the only source of chiral symmetry breaking is due to the presence of the boundary , while when @xmath159 , chiral symmetry is broken also in the bulk . in this case , the matter is not conformally - coupled and , therefore , the propagators can not be obtained by simply rescaling the flat space results .
a full ads computation is required .
we will need the propagators for a chiral multiplet with arbitrary mass .
the scalar propagator associated to the representation @xmath160 ( @xmath161 ) is given by where both scalars in the chiral multiplet have standard boundary conditions .
this is the range where a single quantization is possible . ]
@xmath162\gamma[e-1]}{\gamma[2e-2 ] } \left(\frac 2 u\right)^{e}\ , _ 2f_1\left(e ,
e-1;2e-2 , -\frac 2 { u}\right ) , \nonumber\end{aligned}\ ] ] where we have introduced the ads invariant length , @xmath163 the fermion propagator associated to the representation @xmath164 can be found in ref .
@xcite , @xmath165\gamma[e+1]}{(32 \pi^2 l^3)\gamma[2e-1]}\left(\frac { 2}{u+2}\right)^{e+1 } \ , _ 2f_1\left(e+1,e-1 ; 2 e-1,\frac 2 { u+2}\right)\times \gamma_{\alpha}^{\ \beta } , \nonumber \\
\langle \psi_{\pm\alpha}(x_1)\bar{\psi}_{\pm\dot{\beta}}(x_2 ) \rangle & = & \frac { i \gamma[e]\gamma[e+1]}{(32\pi^2 l^3)\gamma[2e-1]}\left(\frac { 2}{u+2}\right)^{e+1 } \ , _ 2f_1\left(e+1,e ; 2 e-1,\frac 2 { u+2}\right)\times \gamma_{\alpha\dot{\beta}}\nonumber \\\end{aligned}\ ] ] where , @xmath166 as in the massless case , the contribution of the matter loop to the gaugino mass arises from the the self - energy ( [ chiralself ] ) , @xmath167 where now @xmath168 and the fermion belongs to the representation @xmath164 . in order to compute the contribution to the gaugino mass , we evaluate the self - energy on the solution of the massless gaugino equation as in section 3.1 .
this highly non - trivial integral of hypergeometric functions can be evaluated numerically by choosing the simplest solution of the massless equation of motion , @xmath169 , @xmath170 following the discussion in section 3.1 we expect the same to hold for any solution satisfying the appropriate boundary conditions .
this contribution as expected does not depend on the mass and cancels the anomaly - mediated uv contribution , proving for general @xmath55 that this term is necessary for the consistency of the supersymmetric theory . as a check of this result , one can consider the limit @xmath171 , as done in @xcite . in this limit
, the curvature is a small effect and the loop can be computed using flat - space propagators , but with the ads mass splitting . for completeness
we also checked the wave functions contribution .
the chiral preserving contribution to self - energy in general reads , @xmath172
\langle \psi_{+\alpha}(x_1)\psi_{+\dot{\beta}}(x_2 ) \rangle\ ] ] repeating the same steps as in section 3.2 , we find numerically , @xmath173 independently of the mass .
this calculation also proves that by regulating the theory with pauli - villars fields there is an @xmath157 anomaly mediated contribution of the form considered before . in this case the contribution of the heavy fields with @xmath174 can not be obtained with the flat space propagators since this effect is entirely due to the fact that the theory lives in curved space .
h. nicolai , `` representations of supersymmetry in anti - de sitter space , '' in supersymmetry and supergravity : proceedings of the trieste spring school on supersymmetry and supergravity , world scientific ( 1984 )
. j. m. maldacena , `` the large n limit of superconformal field theories and supergravity , '' adv .
* 2 * , 231 ( 1998 ) [ int .
j. theor .
phys . * 38 * , 1113 ( 1999 ) ] [ arxiv : hep - th/9711200 ] ; s. s. gubser , i. r. klebanov and a. m. polyakov , `` gauge theory correlators from non - critical string theory , '' phys .
b * 428 * , 105 ( 1998 ) [ arxiv : hep - th/9802109 ] ; e. witten , `` anti - de sitter space and holography , '' adv .
theor . math .
* 2 * , 253 ( 1998 ) [ arxiv : hep - th/9802150 ] .
|
we study supersymmetric qed in ads@xmath0 with massless matter . at 1-loop
the ultra - violet regulator of the theory generates a contribution to the gaugino mass that is navely inconsistent with unbroken supersymmetry .
we show that this effect , known in flat space as anomaly mediated supersymmetry breaking , is required to cancel an infra - red contribution arising from the boundary conditions in ads space , which necessarily break chiral symmetry .
we also discuss an analogous uv / ir cancellation that is independent of supersymmetry .
|
quantum transport in semiconductor quantum dots and molecular systems is a subject of intense study nowadays.@xcite these nanoscaled devices provide a formidable environment to study fundamental aspects of quantum physics , involving many - body correlations and light - matter interaction in regimes out of equilibrium.@xcite these systems have a great potential to form a new generation of optoelectronic devices based on the unique electronic structure that arises from the quantum confinement . for instance , quantum dots can produce a wealth of visible colors depending upon its size , even white light with relatively high efficiency@xcite and potential to integrability with nanoelectronics.@xcite additionally , with the great technological advances in the manufacturing of semiconductor quantum dot system , it became possible to coherent monitor and control electron populations in two - level systems via different pump - probe techniques.@xcite in all these experiments the main signature of quantum coherent nonlinearity is rabi oscillations , which has no classical counterpart .
more recently , rabi oscillations was also reported in organic light - emitting diode.@xcite such coherent optical manipulations constitute a fundamental ingredient to quantum information processing in solid state devices that use electron - spin or excitonic states as qubits.@xcite interestingly , holes in semiconductor quantum dots have been revealed as an alternative to electrons in the manufacturing of spin qubits.@xcite it was originally demonstrated by zrenner _
_ et al.__@xcite that coherent rabi oscillations in a two - level quantum dot photodiode can be monitored by photocurrents .
additionally , it was proposed that a photocurrent in a self - assembled quantum dot photodiode can become spin - polarized due to an effective exchange interaction via biexciton state.@xcite this result points out the potentiality of the present system to future spintronic devices .
it was also observed that the double dot structures present the ability to increase the coherence time of indirect excitons.@xcite recently , thermal effects on the excitonic rabi rotations in a quantum dot system were investigated experimentally.@xcite it was evidenced acoustic phonons as the main source of damping of the rabi oscillations . in the present work
we analyze how the temperature of nearby contacts tunnel coupled to a single quantum dot affects the coherent optical dynamics . applying nonequilibrium green function technique to a microscopic hamiltonian model , we write a set of coupled integrodifferential equations that describes the coherent evolution of the electron - hole populations in the dot , enabling the study of non - markovian effects .
a resonant laser field drives the electron - hole dynamics and generates a photocurrent .
we find two contributions to the current .
the first one comes from electrons in the dot that tunnel to a contact .
this current is positive and is mainly induced by the laser field .
the second current component is related to electrons in the reservoir that acquires enough thermal energy to tunnel into the dot .
this second contribution charges the dot , thus generating additional features not yet reported in the literature .
our main findings include the suppression of the rabi oscillation as the temperature of the nearby contact increases , a negative photocurrent due to a backwards charge flow , and a maximum photocurrent value achieved when the laser intensity is comparable to the mismatch between the excited dot level and the contact chemical potential .
moreover , we analyze how the initial electron population of the conduction band , which is controlled by the temperature , affects the rabi oscillations .
figure ( 1 ) illustrates the system considered .
it is composed of a quantum dot attached to a left and to a right electron reservoirs in the presence of a source - drain bias voltage .
a laser field shines the dot , thus generating electron - hole pairs in it .
the electrons in the conduction band and the holes in the valence band can tunnel out from the quantum dot to the left and to the right reservoirs , respectively .
this results in a photocurrent signal in the system . in the experimental point of view
this system can be implemented in a structure of the kind n - gaas
i schottky contact , as described in ref .
alternatively , a @xmath0-@xmath1-@xmath2 junction can also be applied as describe in ref .
[ ] , with self - assembled quantum dots in the intrinsic region . and temperature @xmath3
. due to the proximity of the conduction band level @xmath4 to @xmath5 , the dot population thermally fluctuates .
this induces nonlinearities in the photocurrent driven by a laser field with resonant energy @xmath6 . ]
the modeling hamiltonian is given ( per spin ) by @xmath7 , with @xmath8 where @xmath9 is the dot level for spin @xmath10 in the valence ( @xmath11 ) or the conduction ( @xmath12 ) band .
the operators @xmath13 ( @xmath14 ) annihilates ( creates ) one electron in level @xmath1 with spin @xmath10 .
the parameter @xmath15 gives the optical transition between valence band and conduction band in the quantum dot .
this parameter can be controlled by the intensity of the incident radiation . in our model electron - electron interaction
is not accounted for in order to allow analytical treatment . to couple the dot to fermionic reservoirs we use the tunneling hamiltonian @xmath16 where @xmath17 ( @xmath18 ) annihilates ( creates ) one electron in the right ( @xmath11 ) or the left ( @xmath12 )
lead.@xcite the parameter @xmath19 gives the dot - leads coupling strength .
finally , the free - electron energies of the electrons in both leads are given by @xmath20 in the present model we assume that the tunneling rates are larger than spontaneous emission rates , so that electron - hole recombination will be neglected . our main task is to explore the effects of the reservoirs temperature on the rabi oscillations and the photocurrent . to this goal we must find the lesser and retarded green functions of the quantum dot , i.e. , @xmath21 and @xmath22 .
note that the occupations of the levels @xmath9 are given by @xmath23 , while the photocurrent is defined as @xmath24 \rangle$ ] , ( @xmath25 ) with @xmath26 being the total number of particles operator .
following ref .
[ ] one can show that @xmath27 where @xmath28 and @xmath29 with @xmath30 . here
@xmath31 is the fermi function to @xmath1-th reservoir and @xmath32 is the tunneling rate with @xmath33 being the density of states of the corresponding reservoir for spin component @xmath10 .
the present formalism allows the inclusion of ferromagnetic leads by considering spin - dependent tunneling rates @xmath34.@xcite according to eq .
( [ i2s ] ) the current at time @xmath35 has two contributions , one that is instantaneous and proportional to the dot occupation @xmath36 ( @xmath37 ) and a second one that involves the whole history of the system ( @xmath38 ) . in this second term , a time integral of the correlation function @xmath39 weighted by a thermal dependent function @xmath40 should be carried on , ranging from @xmath41 to the present time .
all the thermal effects arise via this memory integral .
this contrasts to the density matrix approach used in quantum optics formulation that in general does not account for thermal and memory effects in the standard markov approximation.@xcite calculating the time derivative of @xmath42 we arrive at @xmath43 where the lesser green function is written in a vector - like form @xmath44^t$ ] and @xmath45^t$ ] . here
@xmath46 , with @xmath47 and @xmath48 the matrix in eq . ( [ dglesserdt ] )
is given by @xmath49,\ ] ] with @xmath50 and @xmath51 .
it is yet valid to note that in the absence of the reservoirs , eq .
( [ dglesserdt ] ) reduces to the well known semiconductor bloch equations.@xcite in order to determine @xmath52 we need the retarded and advanced green functions @xmath53 . taking the time derivative with respect to @xmath54 we obtain @xmath55 + \mathbf{p}_\s(t ' ) \mathbf{g}_\s^r(t , t'),\ ] ] where @xmath56^t$ ] , @xmath57 and @xmath58 are the two - component pauli spinors , @xmath59^t$ ] and @xmath60^t$ ] , and the matrix @xmath61 is defined according to @xmath62,\ ] ] with @xmath63 . solving numerically eqs .
( [ dglesserdt ] ) and ( [ dgrdt ] ) we obtain the occupation @xmath36 and the photocurrent . in what follows we present our results .
in order to keep the generality of our results , we express the time in units of @xmath64 , where @xmath65 is proportional to the tunneling rate between dot and reservoirs . for simplicity
we assume the wideband limit , where the tunneling rates are energy independent and we set @xmath66 .
the current unit is given by @xmath67 and all energies are in units of @xmath65.@xcite experimentally , @xmath68 depends on the tunnel barrier and it can be easily controlled by an external gate voltage .
we find for quantum dot systems @xmath69,@xcite which results in @xmath70.@xcite the time @xmath71 ranges in the interval @xmath72 ( @xmath73 ) - @xmath74 ( @xmath75).@xcite additionally , the quantum dot valence and conduction band levels are given by @xmath76 and @xmath77 , respectively.@xcite both levels are measured with respect to the chemical potential @xmath78 , which is taken as our energy reference .
the chemical potential of the right side @xmath79 is given by @xmath80 , where @xmath81 is the bias voltage . in what follows
we adopt @xmath82 .
finally , we assume @xmath83 and @xmath84,@xcite except when those parameters explicitly change in the plots .
figure [ fig2](a)-(b ) shows the evolution of the electron and hole occupations in the quantum dot for differing temperatures @xmath85 . at @xmath86 the valence band level is fully occupied with @xmath87 while the occupation of the conduction band is @xmath88.@xcite this small occupation comes from the proximity of the level @xmath4 to the fermi energy of the left reservoir , which allows thermal excited electrons to tunnel into the dot .
initially ( @xmath86 ) the quantum dot occupation is calculated using the equation @xmath89 where the lesser green function is given by the keldysh equation @xmath90 , where @xmath91 is the retarded ( advanced ) green function of the dot attached to the leads without laser field and @xmath92 .
as @xmath85 increases , electrons in the left electrode acquire enough thermal energy to enhance the population @xmath93 at @xmath86 , while @xmath94 remains the same due to @xmath95 . when the system starts to evolve in the presence of a laser field , the occupations @xmath94 and
@xmath93 develop the characteristic rabi oscillations . in the small temperature regime these oscillations
are more pronounced for small times and become suppressed as the time increases .
this is due to the decoherence imposed by the tunneling between dot and reservoirs . for large enough times
both @xmath94 and @xmath93 reach constant values . as the temperature increases
, the amplitude of the rabi oscillations shrinks for all times .
this is directly related to the enhancement of the initial population @xmath93 with temperature . with the level
@xmath4 becoming more populated , the pauli exclusion principle makes it more difficult to one electron with the same spin in the valence band ( @xmath96 ) to jump to the conduction band ( @xmath4 ) .
so we observe two sources of suppression to the coherent rabi oscillations : _
( i ) _ coupling to reservoirs and _ ( ii ) _ thermal activated pauli blockade . and
@xmath4 and ( c ) photocurrent as a function of time for differing temperatures .
as time evolves all these quantities exhibit coherent rabi oscillations .
the amplitude decreases with time due to the incoherent tunneling between dot and electrodes .
the oscillation amplitude is also suppressed as @xmath85 increases due to a thermal induced pauli blockade .
for @xmath97 the photocurrent is dominated by a backwards current , thus becoming negative . in the inset
we show separately the in and out current components . ]
the photocurrent seen in fig .
[ fig2](c ) , at least to some extent , reflects the @xmath93 behavior .
it oscillates in time with a decreasing amplitude , tending to a stationary nonzero value .
additionally , the amplitude of the rabi oscillations is also reduced as @xmath85 increases , following the behavior of @xmath93 .
interestingly , for small enough temperatures and shorter times , the photocurrent oscillations attain negative values , which corresponds to an unusual flow of electrons from the reservoir into the dot ( see solid line , @xmath98 , around @xmath99 ) . in order to gain further insight of this effect , in the inset of fig .
[ fig2](c ) we show separately the current components @xmath37 and @xmath38 . the outgoing current is positive which means that electrons are flowing from the dot to the reservoir .
the incoming current gives a negative contribution to the current , which corresponds to electrons flowing in the opposite way , i.e. , from the electrode into the dot .
around @xmath100 , the @xmath38 component presents a dip , which pulls down the total photocurrent , making it negative .
when @xmath85 increases , this dip is suppressed and the photocurrent assumes positive values for all times . and @xmath93 and ( b ) current components against @xmath101 . while @xmath93 increases monotonically with @xmath15 ,
@xmath94 is initially suppressed , reaches a minimum and then increases slightly .
oppositely , the photocurrent increases with @xmath15 , reaches a maximum and then becomes slightly suppressed .
this nonmonotonic behavior in both @xmath94 and @xmath102 can be understood looking at the current component @xmath38 .
when @xmath103 , electrons in the reservoir can tunnel into the dot , thus populating @xmath4 .
this enhancement ( in modulus ) of the backwards current suppresses the photocurrent for @xmath104 and increases further @xmath93 . in the inset we plot @xmath105 .
note that the dot charges for @xmath104 . ]
figure [ fig3](a ) shows how @xmath94 and @xmath93 evolves as a function of the parameter @xmath15 , for differing temperatures .
for all @xmath85 values we observe @xmath106 for @xmath107 , while @xmath93 increases with @xmath85 for @xmath107 .
this enhancement of @xmath93 with temperature comes from the thermal excited electrons in the reservoir that acquires enough energy to jump into the dot as @xmath85 increases .
both @xmath94 and @xmath93 are obtained via eq .
( [ ni ] ) for @xmath108 . in the presence of the laser field
, @xmath93 increases monotonically with @xmath101 , while @xmath94 is initially suppressed and then it is enhanced , thus developing a minimum around @xmath109 .
note also that @xmath93 presents a further enhancement near @xmath110 . in the inset of fig .
[ fig3](a ) we show the sum @xmath111 against @xmath101 .
when the resonant condition @xmath112 is attained , the total occupation presents a steeper enhancement for low temperature . for larger
@xmath85 a more broaden increasing is found .
it is valid to note that the sum @xmath105 is not limited to one , as expected in a standard two - level system with one level being initially occupied and the other one being initially empty . here
the level @xmath4 is not populated only by the laser field , but also by the left reservoir .
the occupation profiles will be more clearly understood looking at the current components in the next plot . in fig .
[ fig3](b ) we plot separately the current components @xmath37 , @xmath38 and the total current @xmath113
. while the outgoing current follows @xmath93 , the incoming current is strongly increased ( in modulus ) around @xmath114 . as a result
, the photocurrent is suppressed due to this backwards current , thus developing a peak close to @xmath112 .
increasing even further the temperature , the thermal fluctuations of the reservoirs yield to a more effective injection of electrons into the dot .
this makes @xmath38 starts at higher absolute values for @xmath115 .
this amplification of @xmath38 suppresses the photocurrent when compared to its low temperature profile . in the presence of a laser field in resonance with the difference @xmath116 ,
doublets emerge in the spectrum of the system,@xcite as illustrated in the drawn of fig .
[ fig3](b).@xcite as the laser intensity increases , the lower energy peak of the doublet eventually attain resonance with the reservoir chemical potential at @xmath117 .
this allows electrons to resonantly tunnel from the lead into the dot , thus generating a backwards current that suppresses the total photocurrent and increases the @xmath93 population .
when the lower peak lies below @xmath5 , the enhancement of @xmath85 tends to depopulate this channel , consequently suppressing @xmath38 , as seen in fig . [
fig3](b ) for @xmath118 . finally , it is valid to point out that the @xmath38 current component plays a role in the transport whenever @xmath119 , which allows electrons in the reservoir to resonant tunnel to the dot .
it is possible to entirely suppress the incoming current by moving @xmath4 high enough above @xmath5 , so that @xmath120 . for this regime electrons can flow only in one direction , i.e. , from the dot to the reservoir , thus reducing the backwards charge flow .
[ fig4 ] shows the photocurrent against time for different @xmath4 values . for @xmath121 and @xmath122
the channel @xmath123 lies below @xmath5 , as illustrated in the energy diagram at the lower part of the panel .
this results in a relatively high @xmath124 component ( see the inset ) . on the contrary , for @xmath125 and @xmath126 we find @xmath123 higher than @xmath5 ( see the upper energy diagram sketched ) , which suppresses @xmath124 and makes the total current larger . 0.5 cm values , with respect to @xmath5 . as @xmath4 increases the photocurrent
is amplified .
this is due to the suppression of the incoming current for larger @xmath4 .
the temperature adopted is @xmath127 . in the inset
we show the incoming current component for all the @xmath4 values used .
as @xmath4 enlarges , the incoming current tends to zero.,title="fig : " ]
in conclusion , via nonequilibrium green function technique we have investigated the dynamics of electron - hole pairs in a quantum dot tunnel coupled to fermionic reservoirs .
we found that the thermal fluctuation of the reservoir and consequent occupation of the conduction band level appears as a new source of decoherence for the optically induced rabi oscillation in qds , which has not been reported yet .
as temperature increases , the thermal excited carriers in the left reservoir acquires enough energy to tunnel into the dot .
this gives rise to an enhancement of electronic dot population , which results in a thermal activated pauli blockade that suppresses slightly the rabi oscillations .
this effect is strongly dependent on the temperature of the reservoirs and on the mismatch between @xmath4 and @xmath5 .
finally , a nonlinearity signature is found in the current against @xmath15 .
this results from a doublet that brings into resonance a transport channel with the chemical potential @xmath5 .
this laser induced resonance generates a competition between outgoing and incoming currents in the quantum dot that yields to the observed nonlinearities . as a final remark , we note that the present study is a fundamental example of the use of nonequilibrium green function for optical processes , which can be applied to more intricate systems , including non - markovian processes . p. w. fry , i. e. itskevich , s. r. parnell , j. j. finley , l. r. wilson , k. l. schumacher , d. j. mowbray , m. s. skolnick , m. al - khafaji , a. g. cullis , m. hopkinson , j. c. clark , and g. hill , phys . rev .
b * 62 * , 16784 ( 2000 ) .
it is valid to point out that one can control the noise levels in nanoscale conductors in the presence of ac - fields , by tunning the amplitude of the field , see for instance , s. camalet , j. lehmann , s. kohler , and p. hnggi , phys .
90 * , 210602 ( 2003 ) .
the state of the art quantum transport experiments allow ultrafast pa - photocurrent measurement in the ps time scale .
for instance , it was recently measured ultrafast photocurrent dynamics in a @xmath0-@xmath1-@xmath2 carbon nanotube based junction , see n. m. gabor _ et al_. , phys .
* 108 * , 087404 ( 2012 ) .
the parameter @xmath129 is in fact much smaller than the realistic values for semiconductor dots that result in excitonic energy typically of @xmath130ev .
however , since @xmath131 is already much larger than any other energy scale in the problem , if its value is increased even further no changes will be observed in the results .
it is valid to note that our formulation would nt be valid in the limit where @xmath101 is comparable to the difference @xmath132 or the laser frequency @xmath133 , since we adopt the rotating wave approximation .
additionally , in the experimental point of view @xmath101 is much smaller than @xmath132 or @xmath133 .
it is valid to point out that in the standard approach of the two - level semiconductor quantum dot systems , based on density matrix formalism , it is adopted the base @xmath134 and @xmath135 for no - exciton and one - exciton in the dot , respectively .
the initial state @xmath134 in our formalism stands to @xmath136 and @xmath137 , which means that the conduction band is empty , while the valence band is filled with one electron .
though , due to the thermal effects reported here , the occupation @xmath93 fluctuates , thus resulting in @xmath138 .
a similar pair of doublets in a two - level system also takes place when , instead of light , a phonon field connects the levels .
see , for instance , e. vernek , e. v. anda , s. e. ulloa , and n. sandler , phys .
b * 76 * , 75320 ( 2007 ) .
|
we theoretically investigate laser induced quantum transport in a single quantum dot attached to electric contacts .
our approach , based on nonequilibrium green function technique , allows to include thermal effects on the photon - induced quantum transport and excitonic dynamics , enabling the study of non - markovian effects . by solving a set of coupled integrodifferential equations ,
involving correlation and propagator functions , we obtain the photocurrent and the dot occupation as a function of time .
two distinct sources of decoherence , namely , incoherent tunneling and thermal fluctuations , are observed in the rabi oscillations . as temperature increases a thermally activated pauli blockade results in a suppression of these oscillations .
additionally , the interplay between photon and thermal induced electron populations results in a switch of the current sign as time evolves and its stationary value can be maximized by tunning the laser intensity .
|
the miniboone experiment @xcite is a @xmath0 oscillation search at fermi national accelerator laboratory designed to confirm or rule out the lsnd signal @xcite .
the fermilab booster accelerates protons to 8 gev ; these protons strike a beryllium target , generating mesons which decay to produce the miniboone @xmath1 beam .
the neutrinos interact in a 12 m diameter sphere filled with mineral oil of food - grade purity , and the erenkov light from charged particles produced in these interactions is detected by 8 inch hamamatsu @xcite photomultiplier tubes ( pmts ) lining the sphere .
there are 1280 pmts in the light - tight inner signal region : 956 are hamamatsu model r1408 , inherited from the lsnd experiment ; the remainder are hamamatsu model r5912 .
light produced in the outer concentric veto region is detected by an additional 240 r5912 pmts .
a typical tube is pictured in figure [ lsnd_pmt ] .
light signatures recorded by the pmts are used to reconstruct events within the detector .
leptons from charged current neutrino interactions are of particular interest , since these leptons tag the incoming neutrino flavor .
it is important to sufficiently understand the operating characteristics of the pmts , since they are the only active detector element in miniboone .
miniboone uses a monte carlo simulation to understand the detector s response to neutrino interactions .
this program simulates photons traveling through the detector from creation to detection , and models all intermediate processes .
the geometry of all detector elements is coded into the simulation , including that of the pmts .
photons striking the photocathode in a pmt produce photoelectrons ( pes ) , whose number is amplified via a dynode chain ; the magnitude of the current so produced is proportional to the number of incident photons whose energy is above a certain threshold .
a specific operating voltage must be found for each pmt so that all tubes have the proper gain .
gain is defined here as observed output charge ( divided by charge per electron ) per input photoelectron . the charge and timing information recorded from each pmt
can then be used to reconstruct the events off - line .
ideally , this information , along with photocathode efficiency as determined by hamamatsu , specifies exactly when and how many photons hit the pmt .
each pmt is tested in several ways to determine : basic functionality , timing resolution , charge resolution , dark rate and operating voltage .
the operating voltage was chosen so that the gain is closest to the desired @xmath2 signal electrons per photoelectron .
in addition to these global pmt tests , specialty tests were done on seven pmts . in order for the monte carlo simulation to properly model the behavior of the pmts ,
it is necessary to correctly include their efficiency as a function of incident photon angle .
the dominant effect comes simply from the solid angle subtended by a pmt as a function of angle , and this is handled by coding an approximate pmt shape into the simulation .
any remaining efficiency function must be measured and entered into the code .
the measurement , however , will itself contain the solid angle effects ; these must be factored out of the measured angular efficiency to avoid double - counting .
a thorough understanding of the angle - dependent response , time and charge resolution , and individual variations of the photomultiplier tubes will help accurately identify neutrino events within the miniboone detector .
this paper discusses the following topics . in section [ sec : technical design of the pmts ] , the technical specifications of the phototubes are described , along with the custom bases that miniboone used . in section [ sec : global testing ] , the tests which were performed on every tube , with their results , are described . in section [ sec : angular testing ] , the angular tests which were performed on 7 tubes are described , along with their results and a comparison of these results to other angular tests .
section [ sec : implosion risk studies ] describes the studies of miniboone s risk of an implosion chain reaction . ) before the pmt is installed into the detector . ]
phototubes in the miniboone detector satisfy the following technical requirements .
their maximum dark rates lie below the response rate expected from cosmic rays .
their gains are high , around @xmath3 , to be adequately sensitive .
their responses are reliably fast , and have a narrow enough pulse width to minimize response pileup .
each of these characteristics was tested ; these tests are described in subsequent sections .
the two models ( r5912 and r1408 ) of phototubes used in miniboone differ primarily in their dynode structures and number of stages , and are similar in all the characteristics described above .
the r5912 is a ten - dynode - stage photomultiplier tube with an 8 inch hemispherical photocathode .
the base supplies the high voltages for the grids , dynodes , and anode , and provides the back termination in the anode circuit .
the voltage divider chain is entirely passive , and is fabricated with carbon - film resistors and ceramic - disk capacitors .
a single coaxial cable is used to provide high voltage ( on the order of 1500v ) and to connect the anode signal to the readout .
the anode circuit is back terminated in 50 @xmath4 and balanced to compensate for the distortion caused by the capacitive coupling in the termination circuit .
the high voltage taper is the one recommended by hamamatsu ( ca .
1996 ) for gain .
a drawing of the tube dome and a diagram of the dynode circuitry from the hamamatsu data sheets are included in the appendix as figure [ hamamatsu_drawing_dynode ] .
figure [ baseart ] shows the printed circuitboard layout for the r5912 custom bases .
the r1408 is a nine - dynode - stage photomultiplier tube .
hamamatsu upgraded it to the r5912 at least 10 years before this publication .
the circuit diagram for the base ( designed for the lsnd experiment ) is shown in figure [ r1408base ] .
the detector design requires that the tubes , bases , and cables be immersed in oil for several years without adversely affecting the phototubes or the oil . to accomplish this
, the bases and necks of all miniboone pmts were coated with the two - component epoxy ep21lv from masterbond @xcite , which is impervious to and insoluble in oil .
for the r1408 tubes recycled from lsnd , the masterbond was applied on top of the existing hysol @xcite coating .
[ pgh : masterbond ] . ]
the method of global testing for the pmts has been described in detail elsewhere @xcite ; a synopsis of that report is provided here .
the purpose of global testing was to select an operating voltage which resulted in a similar response from each pmt , regardless of model ( r1408 or r5912 ) .
the selected gain was @xmath510@xmath6 electrons per photoelectron ( @xmath7 ) .
the testing process served to determine five characteristic data about each pmt : dark rate , time jitter , charge resolution , double - pulsing rate , and pulse shape .
these data were used to sort the tubes into five categories , from best to worst in timing and charge resolution .
categories 4 and 5 had similar resolutions , and were divided between the two categories by dark rate .
these classifications were used to distribute tubes ranked 1 through 4 ( with about 320 in each rank ) evenly throughout the tank .
category 5 tubes were used in the veto , where timing and charge resolution is less crucial but low dark rate is important . [
pgh : categories ] figure [ test_setup ] shows a schematic of a single tube in the test setup .
the full setup accommodated 30 r1408 pmts and 16 r5912 pmts .
tubes were dark adapted for 12 - 24 hours . after this , tube responses were recorded using an automated vxi readout system .
fiber optic cables located about eight inches from the face of each tube transmitted a steady stream of pulses from an led pulser .
a diagram of a signal tube setup is shown in figure [ test_setup ] .
the output pulse was sent to the vxi electronics , where the waveform was digitized ( fig .
[ fig : vxi ] ) .
different delays were added to the led trigger to center the captured waveform in the time windows ( fig .
[ waveform ] ) .
the pulses were digitized over 512 channels at a rate of 5 giga - samples / sec using the tektronix oscilloscope inside the vxi crate .
100 channels or @xmath820 ns .
each time window was 100 ns long.,width=475 ] the testing data were acquired in two modes .
first , the dark rate rates were collected by recording the noise rates measured at a range of operating voltages , with no light source .
second , the pmts were illuminated by an led light source whose wavelength was 450 nm ( corresponding approximately to the erenkov frequency of the oil ) .
the source flashed at a rate of 1 khz ; the width of each pulse was 1 ns .
pmt responses to 600 - 1000 led flashes were used to determine the operating voltage for the required gain .
this procedure also allowed both the charge and time resolutions , and pre- and post - pulsing anomalies to be studied .
each pmt was tested to determine its operating voltage , as well as its gain , dark rate , charge resolution , timing resolution , and double pulsing rate .
the results of the later five measurements are reported below .
pmt gain corresponds to the mean number of electrons produced by the phototube in response to one pe .
this was determined using the equation @xmath9 where @xmath10 is the total charge from the pmt for all pulses with response past threshold , @xmath11 is the number of responses past threshold , and @xmath12 is the light level .
total charge was computed by summing up charge in the main pmt pulse for all responses with in - time pulses that pass threshold , excluding double pulses .
the second factor is the reciprocal of the average number of pes seen by a pmt for a given light level , @xmath12 , ( excluding zero responses ) predicted from poisson statistics .
the voltage was selected in order to obtain a gain of @xmath13 electrons per @xmath7 .
the distribution of operating voltages at which the pmts run in the detector is shown in fig .
[ fig : opvolts ] .
dark rate is defined as the number of pulses larger than 3 mv in one second .
the experiment required that the tubes operate with a dark rate below 8 khz in the main tank and below 4 khz in the veto .
dark rates were measured over a range of voltages , from several hundred volts below suggested operating voltage up to operating voltage and above .
plots of dark rate versus voltage , or plateau plots , serve as a measure of pmt quality .
the pmts are operated on the plateau , where dark rate does not change significantly as the operating voltage increases .
the r1408 pmts were found to have poorly defined plateaus , but they were considered acceptable if the dark rates had a steady , shallow rise when the operating voltage was increased . the charge resolution , @xmath14 is determined from the width of the peak in the pmt response corresponding to one pe . the distribution of charge resolutions for all pmts is shown in fig . [ fig : chargedist2 ] .
the time jitter in the measurements is dominated by the timing resolution of the pmts ; the timing resolution of the oscilloscope was found to be negligible .
the distribution of timing resolutions for all pmts is shown in fig .
[ fig : timedist2 ] .
pre- and post - pulsing , where the main pmt pulse is preceded or followed by additional small pulses , is a known effect in the types of pmts used on miniboone @xcite .
pre - pulsing was observed in only four r1408 tubes , which were rejected .
post - pulsing occurs in two different time intervals : between 8 - 60 ns after the main pulse , and 100 ns-16 @xmath12s after the main pulse .
miniboone is concerned about the first case ( `` early post - pulsing '' ) , because data in the detector are recorded in 100 ns intervals , making it unlikely that the second case will contaminate events .
early post - pulsing can begin when an electron accelerated to the first dynode starts a typical cascade , and causes another electron to be ejected from the first dynode .
this second electron can move around the inside of the pmt dome before settling back to the first dynode and initiating a second cascade , which becomes the post - pulse .
hamamatsu reports that r5912 and r1408 pmts are expected to have early post - pulses in @xmath15 of the responses from every tube .
miniboone observed higher rates .
pmts were required to have a double - pulsing rate of @xmath16 for the r5912 pmts and @xmath15 for the r1408 pmts at the time of installation .
the distribution of pmt double - pulsing rates for all pmts is shown in fig .
[ fig : doubleplot ] .
extensive specialized testing was done on seven phototubes to explore the behavior of other important parameters , especially the dependence of the pmt response on the angle of the incoming light .
the data from r1408 and r5912 tubes were kept separate until the final analyses .
the main objective of the angular tests was to find how the response of the pmts depended on the angle of incident light .
there were three additional tests .
one tested a few tubes at various voltages .
another rotated the tubes about two orthogonal axes arbitrarily labelled yaw " and pitch " to test the assumption that the tubes are rotationally symmetric .
the last tested tubes in air , without the oil present in the other tests .
the apparatus used for these tests is designed to replicate as closely as possible the characteristics of the miniboone detector , on a much smaller and more manageable scale .
the central part of the apparatus is a 40 gallon , stainless steel tank , painted on the inside with the paint used in the detector tank : a primer and black top coat . the primer is hydralon p water - based epoxy primer ( sherwin williams e72ac500/v66vc503 hydralon p water - based epoxy primer ) , and the black topcoat is a flat black moisture cure aliphatic polyurethane chemical agent resistive coating ( f93b102 flat black moisture cure polyurethane ) .
the tank can be filled with miniboone oil when tests require it .
it houses a single pmt , which can be rotated along the polar or azimuthal angle of the tube , using two precision rotary tables , one for pitch and one for yaw with respect to the table .
the tubes were fixed to these tables with an arbitrary rotation , so pitch " and yaw " refer only to the mechanical apparatus and not any characteristic of the tubes .
these tables have externally - coupled mechanical controls which allow the phototube position angle to be changed without opening the tank and exposing the tube to light .
a 10.25 inch diameter window at the front of the tank allows the tests to simulate the light in miniboone events by propagating the light several meters before it strikes the phototube . because of this window
, the entire testing room must be kept dark throughout each test .
the light source is a picoquant pls 450 , a sub - nanosecond pulsed led .
its center wavelength is 460 nm , with 50 @xmath12w average power at 40 mhz , approximately 30 nm spectral width , and typically 800 ps pulse width .
it was used in 10 second sample periods at about @xmath17 hz . the light was filtered to produce a low - intensity beam ( @xmath181 photoelectron per pulse ) .
the light was directed through an optical cable to face the tank from a distance of approximately 3 meters .
the light from this source makes a cone with half - angle about 23@xmath19 , with greatest intensity along the centerline .
the 3 m distance is great enough that one can assume that the entire face of the tube is uniformly illuminated with light in parallel rays .
the signals from the tube in the testing tank can be read out via an oscilloscope / gpib card , or via the miniboone data acquisition system . for the angular tests ,
a specialized data aquisition system was used ( fig .
[ mmboone daq schematic ] ) .
the signal from the tube in the tank is amplified by a factor of 100 , sent to a discriminator with 30 mv threshold , and finally put into coincidence with the delayed sync out signal from the led pulser .
the number of coincidences and the number of input light pulses each go to a scaler ; they are then recorded and used to calculate the relative efficiency .
the tank is magnetically shielded with a @xmath12-metal box .
this was done because ambient magnetic fields ( e.g. , the earth s ) can influence the relative efficiency of the tubes , and thus rotation within these fields may change the shape of the relative efficiency results . to see if ambient magnetic fields could have a significant effect , the tank was rotated ( while holding the configuration of the tube and internal light source constant ) .
the resulting effect was smaller than ( but on the same order as ) the angular dependence . when the tank was shielded and the same tests were performed , this dependence on orientation within ambient fields was not seen .
the apparatus detailed above was used to study relative efficiency ( @xmath20 ) as a function of angle .
using rapid , short - duration pulses of the led , the @xmath20 of the tube was measured at a given position by taking the ratio of the number of responses which coincided with led pulses to the total number of pulses , with about @xmath3 pulses per datum . @xmath21
@xmath20 values were measured at @xmath22 increments through the entire @xmath23 of response , with two repetitions at each point and two rotations of the tube ( from @xmath24 to @xmath25 and back again ) for a total of four measurements at each tube position .
data were recorded directly into a spreadsheet , providing an ongoing real - time check .
this rotational test was performed on four r1408 and three r5912 tubes .
most of these tubes had been previously tested with the global testing ( see section [ sec : global testing ] ) , so the expected dark rate was known , and known to be comparable to that for the other tubes .
the global testing results also specified tube operating voltages .
it was necessary to dark adapt each tube for at least 12 hours to obtain a dark rate comparable to that of the tubes in the detector , which have not been exposed to bright light for over three years .
[ pgh : roll does nt matter ] the phototubes are essentially rotationally symmetric about their central vertical axis , with variation only in the dynode structure below the main dome .
the r1408 tubes have `` venetian blind '' style dynodes ; the r5912 tubes have box and line " style dynodes .
no significant differences were observed when the dynodes at the starting position were oriented horizontally , vertically , or at @xmath26 .
additionally , a test was performed on one tube varying the yaw angle ( similar to the pitch measurements performed on all tubes ) ; again no significant differences were observed .
the equivalence of pitch and yaw implies that that the roll orientation of the tube does not affect the results .
this simplified the measurements greatly .
another test was performed by varying the operating voltages by 50 v above and below the nominal voltages found by previous miniboone tests . the gain variation with voltage , which determines the @xmath27 ,
was already known , so this test checked for variation in the shape of the @xmath20 curve more than the expected variations in amplitude .
none of significance was observed .
plots of relative efficiency as a function of angle were made for each set of measurements ; these are roughly bell - shaped , as one would expect from the shape of the tubes .
typical results from an r1408 and an r5912 ( with the fits as described below ) are included in figures [ pmtgraphold ] and [ pmtgraphnew ] .
these plots are similar from tube to tube ( see figure [ pmtgraphall ] ) .
the parameters for the fits , as well as the average which will be used in the miniboone monte carlo simulation , are included in table [ table : old parameters ] .
the results obtained were compared to expectations from the shape of the pmts and results obtained by the sno @xcite experiment . because the pmt rotator mount inside the tank does not precisely specify the initial angle , @xmath27
is found for each curve ; the angle corresponding to @xmath27 is assumed to correspond to @xmath28 .
each plot s horizontal axis is shifted accordingly before the plots are compared .
each y - axis is scaled to @xmath29 .
an initial quadratic fit is performed on the central points of each plot to determine the horizontal shift .
a 6-degree polynomial fit is then performed on each plot in the region @xmath30 to @xmath31 . outside of this region ,
statistics were significantly lower , making the data less comparable from tube to tube .
the fit polynomial is forced to be symmetric by employing only even powers of the incident angle .
the polynomial coefficients are varied to minimize the @xmath32 . finally , before comparing the functions , each fit is normalized by dividing by its constant term ( which equals @xmath27 ) , forcing @xmath29 .
the results of the fits are shown in table [ table : old parameters ] and superimposed on figures [ pmtgraphall ] and [ pmtgraphairsno ] . .
] at @xmath33 . the average fit is shown . ] the errors on these data are dominated by tube - to - tube variation .
the tests on each tube individually had very high statistics , typically around @xmath34 light pulses at each measured angle , and about 600 000 responses at central angles or 4 000 at extreme angles ( where the tube was facing backwards ) .
thus , to describe the error , it is more important to look at the relatively large differences between data sets from different tubes .
this was done by examining the parameters of the fits .
the average of the seven fitted values of each parameter and their standard deviation from that average are given in table [ table : old parameters ] .
figure [ pmtgraphall ] compares each pmt to the average fit , and deviations from the average fit are shown in fig .
[ pmtgraphdiff ] .
@xmath35 .
normalized parameters from all all tubes tested .
the error is the standard deviation from the average of the seven tubes tested .
these parameters are to be used in the following function for the monte carlo simulation , where @xmath36 is the angle of the incident light from the tube axis measured in degrees . [
cols="<,>,>,>,>",options="header " , ] [ table : old parameters ] ) . ]
the even - polynomial fits are sufficiently accurate and consistent for the r1408 tubes . for the r5912 tubes , the fits are less accurate on one side ( see figure [ pmtgraphnew ] ) . to describe this difference , several other fits of the r5912 tube data were attempted .
the main attempted non - symmetric fits used a different 5-degree polynomial for angles above or below zero , requiring only that they meet at @xmath28 . the same horizontal shift and @xmath27
were used for this fit as for the initial , even polynomial fit .
however , it was found that the variation between individual tubes was larger than the average difference between r5912 tubes and r1408 tubes , so the even polynomial fit was used for all tubes .
miniboone s relative efficiency angular data were compared to several geometrical models , and to data from previous tests from the sudbury neutrino observatory ( sno ) experiment @xcite .
the three geometrical models assumed the photocathode was a flat disk , a hemisphere , and the shape from the hamamatsu technical specifications .
the data from sno were taken from tests on two r1408 pmts in air and water , for a total of four tests .
the pmts in their detector are all hamamatsu r1408 pmts ( the same as the miniboone r1408 tubes ) , making comparisons with them particularly useful .
these seven results are shown together in figure [ pmtpredictions ] .
note that all of the sno results lie between the hemisphere and the ovoid shape predictions on the top and the flat disk prediction on the bottom .
sno used water instead of oil because the sno detector contains water . )
( blue ) , and the results found by sno from testing two pmts each in air ( purple ) and water ( cyan ) . ] function ( black curve ) with the sno results from testing two pmts each in air ( purple ) and water ( cyan ) and the geometrical @xmath20 function based on the hamamatsu drawing ( blue points ) . ]
while developing the testing procedure , one tube was tested without filling the testing tank with oil .
the results from this test , normalized to @xmath29 , were very similar to the air tests done by sno .
this comparison is shown in figure [ pmtgraphairsno ] .
comparisons among the angular response measurements , sno data , and geometrical models reveal overall similarities .
the present data are also observed to be closer to the geometrical model from the hamamatsu ovoid shape description than to the results obtained in either air or water by sno .
this may be the case because these data were taken in oil , where the index of refraction is much closer to that of glass than either air or water , resulting in less reflection at the oil - glass interface .
on november 12 , 2001 , a pmt implosion in the super kamiokande(super - k)@xcite@xcite detector started a chain reaction that resulted in the loss of several thousand tubes . at the time ,
the miniboone detector was not yet filled with oil .
assessment of the possibility of a similar accident occurring at miniboone was immediately undertaken .
the super - kamiokande detector is cylindrical , 36 m tall and 34 m in diameter , and filled with water .
hamamatsu 20-inch pmts line the interior on a support structure .
they are spaced 0.8 m center to center to give 40% photocathode coverage in the tank .
these tubes are rated for a pressure of 608 kpa ( 6 atm ) . when full , the static pressure at the bottom of the detector is 208 kpa ( 2.05 atm ) . at the time of the accident
the tank was being refilled following some maintenance work , and was approximately @xmath37 filled with water .
the failure began with the implosion of a single tube on the detector floor ; the reason for this failure is not known .
this set off a chain reaction which ultimately destroyed most of the tubes more than 5 m below the water line .
all tubes within 5 m of the surface survived .
miniboone has a significantly smaller ( 12 m diameter ) spherical detector , using 8 inch hamamatsu pmts rated for 709 kpa ( 7 atm ) pressure .
the tubes are spaced 0.55 m center to center .
the detector is filled with oil , which is a factor of 0.9 less dense than water .
the static pressure at the bottom of the miniboone tank is 208 kpa ( 4.5 atm ) .
each of these factors suggests that the miniboone detector should be more resistant to a chain reaction of implosions than was the super - k detector .
a computational model to calculate the shock wave pressure resulting from tube implosion was developed , based upon the stored energy in an evacuated tube at a given depth , assuming the pulse length is the tube radius .
it should be noted that the energy stored in an 8 inch miniboone tube is over an order of magnitude less than in a 20 inch super - k tube . the pressure on a neighboring tube when a tube at the bottom of the miniboone detector implodes was compared to the pressure resulting from a super - k implosion at a depth of 5 m , the depth above which all super - k tubes survived .
it was found the pressure in miniboone was 0.45 times the super - k pressure at survival depth , thus indicating a safety factor of two for miniboone if the tubes were equally strong .
in fact , the 8 inch tubes are rated for a higher pressure than the 20 inch tubes .
miniboone is therefore confident that the experiment is operating with a safety factor of greater than two , and that an accidental implosion of one of miniboone s tubes will not result in a chain reaction .
a more rigorous hydrodynamic calculation was done using numerical techniques . from these calculations a somewhat greater safety factor of 2.6 can be inferred .
after the super - k accident , several tests were done at super - k using the 20 inch tubes . personal communication from several super - k collaborators
indicate that at a depth of 30 m of water chain reaction implosions occurred , but at 15 m there were no chain reactions .
since the pressure at 15 m is significantly greater than the maximum pressure in miniboone this supports the conclusion that miniboone tubes are not at risk for this sort of accident . in 1990 , the sno collaboration also performed a series of tests of 8 inch pmts in a water - filled pressure vessel .
no chain reactions occurred at a pressure of 608 kpa ( 6 atm ) .
this test also supports the previous conclusion .
this paper has reported on tests of the r1408 and r5912 photomultiplier tubes used in the miniboone experiment .
we have briefly described the global testing of all tubes , described in detail in reference @xcite .
the main results reported in this paper pertain to the phototube angular tests , and to the calculation of implosion risk for miniboone .
the angular response of the tubes determined a sixth order polynomial function describing the angular response .
implosion risk calculations for the tubes in the miniboone detector were described .
it was found that the miniboone detector has a large safety factor .
the authors would like to thank the following people for their valuable contributions : jesse guerra , andy lathrop , zhijing tang , ryan patterson , t. neil thompson , sally koutsoliotas , and christi bohmbach .
thanks are due also to the following people for helpful discussions and editing : bruce brown , david finley , bill louis , peter meyers , mike shaevitz , ray stefanski , and morgan wascko .
this material is based in part upon work supported by the national science foundation under grants no .
nsf phy-0139464 and nsf phy 00 - 98826 and the u.s .
department of energy no .
de - fg02 - 91er40671 . 00 proposal for the miniboone experiment : http://www-boone.fnal.gov/publicpages/proposal.ps .
c. athanassopoulos _ et al . _
[ lsnd collaboration ] , phys .
lett . * 77 * , 3082 ( 1996 ) [ arxiv : nucl - ex/9605003 ] .
hamamatsu catalog , large photocathode area pmts " , catalog no.tpmh1286e02 .
+ http://usa.hamamatsu.com/ + http://www.hpk.co.jp/eng/main.htm .
master bond inc .
154 hobart street , hackensack , new jersey 07601 .
+ phone : ( 201 ) 343 - 8983 + fax : ( 201 ) 343 - 2132 + web site : http://www.masterbond.com loctite brand product , http://www.loctite.com .
+ in us toll free : 800-loctite ( 800 - 562 - 8483 ) + in canada , call : 800 - 263 - 5043 + henkel corp - industrial , 1001 trout brook crossing , rocky hill , ct 06067 b. t. fleming , l. bugel , e. hawker , v. sandberg , s. koutsoliotas , s. mckenney and d. smith , ieee trans . nucl .
sci . * 49 * , 984 ( 2002 ) .
j. may , m. wysocki , l. bugel , b.t .
fleming , p. nienaber , and d. smith , nuclear science symposium conference record , ieee , * 1 * , 10 - 16 nov . 2002 , pp . 446 - 449 , vol . 1 .
lyon , phd thesis , university of oxford 59 ( 2002 ) .
http://neutrino.phys.washington.edu/@xmath8superk .
fermi news , vol .
24 no.19 , fri nov .
|
the detector for the miniboone @xcite experiment at the fermi national accelerator laboratory employs 1520 8 inch hamamatsu models r1408 and r5912 photomultiplier tubes with custom - designed bases .
tests were performed to determine the dark rate , charge and timing resolutions , double - pulsing rate , and desired operating voltage for each tube , so that the tubes could be sorted for optimal placement in the detector .
seven phototubes were tested to find the angular dependence of their response .
after the super - k phototube implosion accident , an analysis was performed to determine the risk of a similar accident with miniboone .
miniboone , phototube , photomultiplier tube , r1408 , r5912 29.40.ka
|
in recent years , observational advances have made it possible to measure column density profiles in prestellar cores .
various methods are being used in the literature .
@xcite measured mm continuum emission , while @xcite utilized mid - ir absorption , and @xcite measured dust extinction and reddening of the light of background stars in the near - ir .
a fourth method is to use flux measurements in optically thin lines @xcite .
most of the column density profiles measured in these ways show certain prominent common features : a central flat region , followed by a power - law decline .
some cores additionally exhibit signatures of steepening of the profile , while some show a more - or - less smooth merger to some background value of the column density @xcite .
one model often used to fit to such cores ( e.g. , * ? ? ?
* ; * ? ? ?
* ) is the _ bonnor - ebert sphere _ ( hereafter be sphere ) .
this model assumes an isothermal sphere in equilibrium , acted upon only by gravity and thermal pressure , and bound by some external pressure @xcite . in the central region its density profile is flat with density @xmath1 .
the size of this region is proportional to the jeans length , @xmath2 .
it transitions into a power - law decline that approaches the _ singular isothermal sphere _ ( sis ) with @xmath3 for large radii @xcite .
the cloud is finally cut off at some finite radius , where external pressure forces match the internal forces .
the be model invokes the dimensionless radius parameter @xmath4 .
the value @xmath5 marks a dividing line such that clouds truncated at dimensionless radius @xmath6 are in an unstable equilibrium , and clouds with @xmath7 are stable .
the column density profile of the be sphere , obtained by integrating the volume density numerically along a line of sight , matches well with some observations ( b68 ; * ? ? ?
depending on the parameters , the power - law decline @xmath8 in column density can be more or less pronounced , or even almost completely absent .
the profile steepens at the edge because the line of sight through the sphere becomes shorter .
this is a geometric effect which is present in all truncated models , unless the density _ increases _ sufficiently fast with radius . while physically motivated and reproducing features of several observed column density profiles , the be model has shortcomings .
the most important is the key assumption of equilibrium .
in fact , most fits are found to be supercritical @xcite , representing unstable equilibria .
these states are not expected to exist in reality , as any perturbation will send them to immediate collapse . on a more practical side ,
the procedure of fitting the be model to observations is quite involved .
the volume density is only available as a numerical solution , which then needs to be integrated ( again numerically ) to calculate the column density .
sometimes the fit demands temperatures _ well above _ those measured for the centres of prestellar cores @xcite , and predicted by detailed models of the thermodynamics within the core @xcite . finally , most of the observed cores are deeply embedded within their parent clouds , and the source of a bounding external pressure is not obvious .
the often - cited example b68 @xcite is an exception , as it is thought to lie within a hot hii region .
we stress that the generic features of the be density profile are not unique to equilibrium situations ( see * ? ? ?
the flat region with adjacent @xmath9 density profile appears also in solutions of the hydrodynamic equations for gravitationally collapsing objects @xcite , as long as pressure is not completely negligible @xcite .
the pressure gradient then establishes a region where it nearly balances gravity . here
, the density is nearly constant on the scale of the local jeans length , shrinking in size over time as the density increases . inverted
, this requires that @xmath10 in the outer profile that is left behind outside the central region ( see * ? ? ?
there is no reason why the above two features ( flat region with size of jeans length and adjacent @xmath9 density profile ) should be present in a non - self - gravitating cloud .
however , @xcite find that convergent turbulent flows , with and even without self - gravity ( which are expected to have very different volume density profiles ) , nevertheless yield column density profiles that resemble those of a be sphere .
there are three reasons for this seemingly surprising result .
the first one is the effect of smoothing the data , by angle - averaging , and also by integration along the line of sight .
secondly , the be sphere leaves the modeller the freedom to fit the size of the central flat region by varying the temperature . indeed ,
@xcite fit their simulated cores with be temperatures in the range of @xmath11 , despite their models being strictly isothermal at @xmath12 .
finally , the position of the outer radius cutoff is somewhat arbitrary .
this often leaves large parts of profiles unfitted . in this paper
, we argue that fitting a prestellar core profile at a _ set _ temperature _ does _ still allow one to distinguish between different models of internal structure .
we propose an analytic density profile reproducing the characteristics of not only isothermal equilibria @xcite , but also non - equilibrium collapse solutions ( e.g. , * ? ? ?
* ) , and many observed profiles @xcite . within the margins of uncertainty
it fits the observations as well as the be model does .
however , it possesses a closed - form expression for the column density , and is therefore very easy to fit .
furthermore , the temperature can be an input parameter instead of a fitting parameter , so that the model avoids some of the possible inconsistencies of fitting the be model to either observations or simulation results .
we can use our model to make some inference about the dynamical state of the core .
this paper is organized the following way : section [ sec : sphericalgeometry ] describes the spherical model and its parameters , while section [ sec : diskgeometry ] presents a corresponding model for intrinsically flattened objects . in section [ sec : applications ] , we apply our model to b68 and l1689b , and we summarize our results in section [ sec : summary ] .
the characteristics found in observed column density profiles and theoretically both in equilibrium and collapse solutions can be parametrized by a volume density more generic than the be sphere .
we propose using the profile@xmath13characterized by a central volume density @xmath14 and truncated at some radius @xmath15 .
the parameter @xmath16 fits the size of the flat region in terms of the jeans length , and is given by @xmath17where @xmath18 is the gravitational constant and @xmath19 is a constant of proportionality .
this profile is also mentioned in @xcite and @xcite .
the temperature @xmath20 , which can be constrained observationally , is _ not _ used as a fitting parameter .
it enters through the value of the isothermal sound speed @xmath21 . the boltzmann constant
is denoted by @xmath22 , and @xmath23 is the mean mass of a particle , where @xmath24 is the mass of a hydrogen atom .
we assume a @xmath25 number fraction of helium .
the column density is derived by integrating the volume density along a line of sight through the sphere : @xmath26where we have used the transformation @xmath27 and hence @xmath28 . fig .
[ fig : cloudcut ] defines the quantities appearing in this derivation . .
the observer is positioned along the direction of the coordinate @xmath29 , and measures an integrated column density @xmath30 as a function of the offset @xmath31 . ] inserting equation ( [ eq : rho ] ) into equation ( [ eq : integral_rho ] ) , we find that the integral is analytically tractable .
the closed - form expression for the column density is then@xmath32this can be re - written in terms of the central column density by introducing the ratio @xmath33 , called the concentration in @xcite .
with @xmath34 we find @xmath35 .
\label{eq : sigma}\end{aligned}\ ] ] fig .
[ fig : sigmageneric ] demonstrates that this profile possesses generic features that it shares with observations , collapse solutions , and with the be model : a flat central region , a power - law decline , and steepening at the edge .
we note that the effect of the boundary is exclusively contained in the factor in square brackets .
the quantity @xmath36 determines the size of the region described by the power - law .
if it is large , there is a pronounced power - law , whereas if @xmath36 approaches unity , the cut - off already dominates near the flat region and inhibits the power - law . ) for varying @xmath33 .
all cases have identical values for @xmath16 .
the central column densities are chosen to be different in order for the plots not to overlap .
the solid line illustrates the three major characteristics of this function : the flat region in the centre of size @xmath16 , the adjacent power - law decline @xmath37 , a steepening at the edge for geometric reasons ( see text ) , out to the cut - off radius @xmath15 . in this case ,
the dashed - line shows a model dominated by the power - law regime , with @xmath39 .
@xmath15 is chosen so that the influence of the cut - off is negligible .
the long - dashed line finally demonstrates that the power - law can be suppressed entirely if @xmath40 . in this case , @xmath41 . ]
the three parameters to fit are ( i ) the outer radius @xmath15 , ( ii ) the central column density @xmath42 , and ( iii ) the size of the flat region @xmath16 .
the latter contains the product of @xmath19 and @xmath43 , as shown in equation ( [ eq : definitiona ] ) .
if the temperature is pre - determined , @xmath16 only depends on the value of @xmath19 .
this then allows for a stability assessment , as demonstrated in the following section .
the pressure in an isothermal system is given by @xmath44 , and hence the pressure gradient for the spherically symmetric volume density profile of equation ( [ eq : rho ] ) is@xmath45 ^{2}}.\]]in the inner regions , where @xmath46,@xmath47this demonstrates a simple point : the larger the flat region @xmath16 , the smaller the pressure gradient in that region .
the minimal value for @xmath16 is reached in an equilibrium situation . a larger @xmath16 would result in a smaller pressure gradient , gravity would win out , and collapse would ensue .
conversely , for @xmath16 smaller than its equilibrium value the pressure force would dominate , causing expansion . however , that case is not probable since a profile with @xmath48 mandates a strong gravitational influence .
another well - known model besides the sis is the so - called larson - penston solution ( lp solution ) , a self - similar spherical collapse solution , assuming a homogeneous initial density distribution @xcite .
this highly dynamical model does not assume equilibrium , and asymptotically reaches a density profile for which @xmath49everywhere .
we now want to find an expression for the parameter @xmath16 in the sis and the lp models . in the outer regions ,
equation ( [ eq : rho ] ) becomes@xmath50where @xmath51 for the sis ( equilibrium ) and @xmath52 for the asymptotic lp collapse solution .
therefore@xmath53this has a direct application to observed prestellar cores .
if an estimate for the core temperature is available , equation ( [ eq : sigma ] ) can be fit to the column density profile , with @xmath19 as a fitting parameter . if @xmath54 , the system can be considered to be in equilibrium . if @xmath19 is significantly larger than @xmath55 , and closer to @xmath56 , the cloud under scrutiny is collapsing .
, for which the profile approaches the asymptotic lp solution .
the lower line shows @xmath57 , for which the sis profile is reached .
the dashed line represents the be model , which also asymptotically approaches the sis . ]
[ fig : bevssisvslpvsgeneric ] shows our model with @xmath16 chosen to converge to either the sis or the asymptotic lp solution .
we also plot the equilibrium be model . in order to avoid boundary effects ,
the outer radius was moved to @xmath58 .
the be profile does not join onto the sis profile right away .
it overshoots , then steepens and asymptotically approaches the sis profile at large radii .
we may ask which value of @xmath19 is most appropriate to model a critical be sphere .
the overshoot effectively produces a larger flat region if the solution is truncated at @xmath59 .
the resultant @xmath16 has to be larger than for the sis .
having fixed the temperature this can only be achieved by increasing @xmath19 , as we show in fig .
[ fig : bevsgeneric ] . profile ( a ) shows a subcritical be sphere , i.e. a stable equilibrium solution . in this case
, there is no discernible power - law region and the central flat region makes up a significant portion of the total radius . the cut - off becomes dominant just outside the flat region , and the best - fit dimensionless dynamics parameter is @xmath60 .
note that the density contrast is less than an _ e_-folding , which means a mean column density enhancement of @xmath61 over the background .
such an object would not be observationally characterized as a prestellar core @xcite .
profile ( c ) in fig .
[ fig : bevsgeneric ] shows the opposite situation , where the power - law portion of the profile is fit , and @xmath62 is closer to the equilibrium value . as shown in fig .
[ fig : bevssisvslpvsgeneric ] , the be model initially has a power - law index steeper than @xmath63 .
this is the reason why the profiles do not match up as well as for the other cases .
profile ( b ) finally shows a critical be sphere .
a density contrast greater than @xmath64 would characterize an unstable equilibrium , initiating collapse upon the smallest of perturbations .
the two profiles are very similar , only minor deviations are discernible . in order to fit our model to the overshoot of the be profile over the sis , @xmath16 has to be larger , which can only be achieved by @xmath65 .
we demonstrate this effect on the examples of b68 and l1689b in sections [ subsec : b68 ] and [ subsec : l1689b ] : as a consistency check , we insert the derived be temperatures into our model , and indeed find @xmath66 and @xmath67 respectively .
note that the be model _ is _ a valid model in absence of any effects other than gravity and thermal pressure , and thus we conclude that in this case , not the asymptotic value of @xmath68 but rather the critical be value of @xmath65 is relevant for the stability assessment .
we stress that if @xmath19 is significantly larger than that , it is strongly indicative of collapse .
( see text for explanation ) .
( c ) presents a fit to a highly supercritical ( unstable ) be model . ] since the be model effectively has @xmath69 and collapsing clouds have @xmath70 , trying to force a be model fit to such a cloud results in spurious higher temperatures .
this can be seen in @xcite .
a factor of @xmath71 in the dimensionless dynamics parameter @xmath19 needs to be compensated by a factor of @xmath72 in temperature when fitting a be equilibrium state .
both our model and the be sphere have a well - defined outer radius , and the total mass depends on it . for the profile of equation ( [ eq : rho ] ) , the total mass has a closed - form expression . evaluating @xmath73 yields@xmath74 .
\label{eq : mass}\]]there is evidence for some power - law behaviour in most observed profiles ( e.g. , * ? ? ?
in such a situation , the value of @xmath75 must be at least @xmath76 ( cf .
[ fig : sigmageneric ] ) .
then , the contribution of @xmath77 is quite close to its limiting value of @xmath78 , and the total mass is given by@xmath79with @xmath80 error .
the equivalent expression for the be model is ( derived assuming an sis profile at large radii ) @xmath81 both expressions for the masses are linearly dependent on the unknown radius @xmath15 . for a given ( measured ) temperature , our model additionally invokes the dimensionless dynamics parameter @xmath19 , which is well constrained by the size of the flat region .
the be mass instead depends on the temperature of the fit , which we show in sections [ subsec : params_spher ] and [ subsec : l1689b ] to potentially be a very poor value , and prone to overestimation by as much as a factor of @xmath72 .
most observed cores do not appear circular in projection . a corresponding analysis as in section [ subsec : params_spher ] can be done for a disc - like geometry as well . here , a generic face - on column density profile is@xmath82this profile provides a good fit to the column density of collapsing flattened clouds @xcite , even though the models in that paper are not truncated . assuming vertical hydrostatic equilibrium and ignoring the effect of external pressure , the volume density is proportional to the square of the column density@xmath83and the density accordingly is given by@xmath84the assumption of vertical hydrostatic equilibrium
is well justified by simulations when some source of flattening is present , such as rotation @xcite , or magnetic fields @xcite .
abbreviating @xmath85 , the volume density profile has exactly the same form as for the spherical case ( with a different constant ) . integrating @xmath86 through
the disc viewed edge - on , the column density versus the offset @xmath31 has the same functional form as derived above for the spherical case in equation ( [ eq : sigma ] ) .
the only difference is that the central column density is now given by @xmath87 , since for a thin disc , the relation between @xmath16 and the jeans length is @xmath88 this similarity means that both a flattened and a spherical object ( see section [ sec : sphericalgeometry ] ) can be fit with the same formula . note that @xmath89 is the face - on column density , whereas @xmath90 is its edge - on counterpart .
we can compare the ( face - on ) generic column density profile to one of an equilibrium solution , in order to get the minimal value for @xmath16 , as done for the spherical case above . here ,
the appropriate profile is the singular isothermal disc , characterized by@xmath91following the same steps as in section [ subsec : params_spher ] , the large - radius asymptote of equation ( [ eq : disk_sigma ] ) is @xmath92yielding @xmath93 . comparing this expression to equation ( [ eq : jeans_length_disk ] ) , we find @xmath94 for equilibrium .
@xcite show that for an isothermal disc - like cloud during runaway collapse , the self - similar column density profile is given by @xmath95 , analogous to how the lp solution is over - dense compared to the sis .
this provides a good estimate for the value of the dimensionless dynamics parameter for a dynamically collapsing disc - like cloud , @xmath96 .
flattened cloud cores best fit with values for the dimensionless dynamics parameter @xmath97 are therefore close to equilibrium , while @xmath98 is strongly indicative of dynamical collapse .
inclusion of magnetic fields into a spherical model poses a problem : the magnetic field lines can not be arranged in a spherically symmetric way .
there has to be a preferred direction .
however , in disc - like geometry , the axis perpendicular to the disc is preferred already , and can be chosen as the axis of orientation of a large - scale magnetic field . starting with initially straight field lines through a collapsing astrophysical gas sphere , flattening ensues along this preferred direction .
matter can contract significantly along the field lines , but the lorentz force will impede motions perpendicular to the field @xcite . in the limit of a magnetic field much stronger than the field of the ambient cloud , @xcite , @xcite and @xcite showed that the contribution of the magnetic field can be folded into a force calculation of the collapse of a magnetized disc by magnetic pressure modifying the effective sound speed , and magnetic tension changing the effective gravitational constant .
then we can write ( see * ? ? ?
* ) @xmath99we use the standard definition of the mass - to - flux ratio in units of the critical value : @xmath100collapse requires @xmath101 , otherwise the magnetic forces will dominate over gravity and stabilize the cloud , and evolution can only happen on long time scales by ambipolar diffusion ( see , e.g. , * ? ? ?
we can absorb the modification into an effective @xmath19 , writing it as @xmath102 .
the value @xmath103 as derived for the non - magnetic case remains the minimal value , while @xmath104 can exceed it by a factor of @xmath105 to @xmath71 for @xmath106 , which is a reasonable estimate for supercritical cores @xcite .
this means @xmath107 , which leaves it still a factor of @xmath71 smaller than the value for the dimensionless dynamics parameter for dynamical collapse ( @xmath98 , as shown above ) , and thus clearly distinguishable .
similar to the problem for magnetic fields , rotation can not be considered in strict spherical symmetry .
in fact , rotation can be the cause of disc - like geometry , as a collapsing rotating spherical cloud will flatten along the rotation axis before contracting in the radial direction .
there exist self - similar solutions for the collapse of a rotating thin disc @xcite . following the discussion in @xcite
, one can express the additional effect of rotation as an effective acceleration @xmath108 , where @xmath109 is the thermal acceleration , while @xmath110 denotes the centrifugal acceleration .
assuming the column density profile of equation ( [ eq : disk_sigma ] ) and proportionality between the specific angular momentum and enclosed mass , this leads to @xmath111this number is computed for a background rotation rate of the molecular cloud @xmath112 , a central column number density @xmath113 , and a temperature @xmath114 .
hence , we find that the effective radial acceleration opposing gravity exceeds the thermal acceleration by less than @xmath115 .
this shows that , unlike magnetic forces , rotation does not significantly modify the size of the flat region and hence the dimensionless dynamics parameter @xmath19 .
we fit both the be model and our model to observational data by determining the best - fitting parameters using a standard levenberg - marquardt least - squares minimization algorithm based on minpack .
we fit our model to b68 , the prime example in the literature for an extraordinarily good fit of the be model to a ( angular - averaged ) column density profile measured in near - ir dust extinction @xcite .
b68 is an isolated bok globule , a small dark cloud which has been studied extensively .
@xcite assume a distance of @xmath116 and quote a be mass of @xmath117 , a temperature of @xmath118 , an outer radius of @xmath119 , and a dimensionless outer radius @xmath120 . @xcite and @xcite more recently updated some of these values by measuring the temperature to be @xmath121 and @xmath122 , respectively , and estimated the cloud to be closer by about @xmath123 , placing it at the near side of the ophiuchus complex .
this reduces the outer radius and decreases the be mass to @xmath124 , but does not change the value for @xmath125 , as that is determined by the shape of the profile .
we perform a similar analysis as @xcite .
we calculate a be sphere , vary the temperature , the dimensionless and physical outer radii @xmath125 and @xmath15 , and fit the line - of - sight integral to the observational data , assuming a distance of @xmath126 .
this procedure yields a be mass of @xmath127 , a be model temperature of @xmath128 , and a central number density of @xmath129 . the outer radius of the best
be fit is @xmath130 , and the dimensionless outer radius @xmath131 .
the best fit of our model to the same data yields a total mass of @xmath132 , @xmath133 , and @xmath134 .
the size of the flat region is @xmath135 .
assuming a temperature of @xmath136 , as in the observations mentioned above , this corresponds to @xmath66 .
we showed in section [ subsec : params_spher ] that a critical be sphere is fit with @xmath65 , and we conclude that the internal structure of b68 may indeed very closely resemble a critical be sphere .
[ fig : fitb68 ] shows that the best fits to b68 of the be sphere and our model differ by very little over the whole range of data .
the be analysis is much more involved and computationally expensive than our model .
it necessitates a numerical solution of the lane - emden equation .
this ordinary differential equation ( ode ) underlies the be model , and does not have a general analytical solution .
the numerical solution is truncated at the dimensionless radius @xmath125 and converted into physical units . only then can one integrate along lines of sight through the solution numerically , yielding the column density to be compared with observational data .
in contrast , fitting our model to a dataset requires less than a dozen lines of code .
another prestellar core to which we apply our model is l1689b .
its central column density is much greater and it is more extended than b68 .
furthermore , it is not isolated , but rather embedded within the larger complex of the ophiuchus molecular cloud .
its profile was measured by @xcite using mid - ir absorption observations and updated by @xcite .
we use the east - west profile , which shows evidence of steepening in the outer regions .
the be fit to the data in fig .
[ fig : fitl1689b ] yields @xmath137 ( supercritical , unstable core ) and a mass of @xmath138 , similar to what @xcite found . fitting our model results in @xmath139
. however , in contrast to the case of b68 , the temperatures of the two fits are vastly different .
the be model requires a temperature @xmath140 , whereas our model can achieve the same quality of fit with a temperature @xmath114 .
we fit the size of the flat region @xmath141 which is a combination of the temperature and @xmath19 .
a temperature of @xmath114 mandates @xmath142 , which is very far from the value required for equilibrium , and even surpasses the value @xmath143 achieved in the highly dynamical asymptotic lp solution .
as discussed in section [ subsec : params_spher ] , this is a strong indication that l1689b can not be in equilibrium , but must instead be collapsing .
again , our model fits this dynamically evolving object just as well as it fit the presumably almost static b68 .
it can do so without introducing inconsistent temperatures , and with much less computational and coding effort .
, while our model can fit the data with @xmath114 .
the dimensionless dynamics parameter @xmath142 , which means that the cloud is collapsing . ]
we have introduced an analytic profile for the integrated line - of - sight column density of an isothermal spherical or flattened cloud . this cloud can either be in equilibrium or in a state of dynamical collapse .
our model is very simple to calculate compared to the be model , and a few lines of code suffice to find a best - fit set of parameters .
another advantage lies in its ability to also encompass non - equilibrium states .
the dimensionless dynamics parameter @xmath19 allows one to assess whether a cloud is near equilibrium or vigorously collapsing . at the same time
, our model does not produce inconsistencies like the be model regarding the object s temperature ( which can be treated as a constrained quantity and not a free parameter ) .
our model can be applied both to spherical and to flattened clouds , and in both cases yields the same functional form for the column density @xmath90 .
it fits the size of the central flat region and allows the modeller to adjust either @xmath19 or the temperature to match it .
results of fitting our model to b68 show that it is indeed a near - equilibrium cloud , with parameters very similar to the best - fit be model . for l1689b ,
our model avoids the need for a high temperature ( the be model requires @xmath140 ) since it can be interpreted as having a temperature of @xmath114 but being in a state of dynamical collapse .
this finding is confirmed by the detection of infall motions for l1689b by means of the shape of the line profiles in optically - thick molecular transitions @xcite . in the future ,
our model can be applied to many other prestellar cores .
we thank joao alves for giving access to his data on b68 .
we also thank philippe andr for his data on l1689b and for his comments on the manuscript .
sb was supported by a research grant from nserc .
|
we present a new analytical three - parameter formula to fit observed column density profiles of prestellar cores .
it represents a line - of - sight integral through a spherically symmetric or disc - like isothermal cloud .
the underlying model resembles the bonnor - ebert model in that it features a flat central region leading into a power - law decline @xmath0 in density , and a well - defined outer radius .
however , we do not assume that the cloud is in equilibrium , and can instead make qualitative statements about its dynamical state ( expansion , equilibrium , collapse ) using the size of the flat region as a proxy .
instead of having temperature as a fitting parameter , our model includes it as input , and thus avoids possible inconsistencies .
it is significantly easier to fit to observational data than the bonnor - ebert sphere .
we apply this model to l1689b and b68 .
we show that l1689b can not be in equilibrium but instead appears to be collapsing , while our model verifies that b68 is not far from being a hydrostatic object .
[ firstpage ] ism : clouds ism : globules stars : formation
|
based on earlier ideas of wigner @xcite dyson introduced a three - fold classification of quantum systems according to their behavior under time - reversal and spin rotation @xcite .
this symmetry classification turned out to be very useful , for instance in semiclassical , disordered and random - matrix approaches to complex quantum systems .
the success of random - matrix theory is based on universal features in spectra of complex quantum systems . while not capable of predicting single eigenvalues random - matrix theory has become one of the key ingredients in predicting physical features that depend on non - trivial spectral statistics @xcite . in each symmetry class
various universality classes have been identified
each described by some ensemble of random matrices .
most prominent are the three gaussian ensembles gue , goe and gse .
they define the _ ergodic universality classes _ and they have been applied successfully to a wide range of quantum systems ( see the recent review @xcite for an overview and further references ) .
recently the three - fold classification has been extended to a ten - fold classification by incorporating spectral mirror symmetries .
this lead to seven novel symmetry classes @xcite .
they are partly realized for a dirac particle in a random gauge field , and for quasi - particles in disordered superconductors or normalconducting - superconducting hybrid systems . in the presence of a spectral mirror symmetry
the spectrum is symmetric with respect to one point @xmath0 : if @xmath1 is in the spectrum so is @xmath2 .
the invention of the novel classes has become necessary due to the impact such a symmetry has on spectral correlations .
these new universal features appear near the symmetry point @xmath0 and they can be described by defining random - matrix ensembles which incorporate the corresponding spectral mirror symmetry .
it has been conjectured by bohigas , giannoni and schmit that the spectra of classically chaotic systems display the spectral fluctuations described by the three gaussian wigner - dyson ensembles of random - matrix theory @xcite .
though the fidelity to the universal predictions of random - matrix theory have an overwhelming support by both experimental and numerical data the physical basis of universality is not completely understood .
quantum graphs have been introduced by kottos and smilansky @xcite as simple quantum models with an exact semiclassical trace formula for the density of states which is expressed as a sum over periodic orbits on the graph .
they have since become an important tool in the semiclassical approach to universality . in this series of papers we will construct star graphs for all ten symmetry classes and investigate their spectral statistics both numerically and analytically . while the following paper @xcite is devoted to a semiclassical periodic - orbit approach this paper focusses on the construction of appropriate star graphs and some numerical results .
we start with giving a short introduction to the ten symmetry classes in section [ sec : symmetry ] with all details needed for the subsequent construction of star graphs . in the following section [ sec : universality ] on spectral statistics
we introduce the spectral form factors , review the results of gaussian random - matrix theory for the ten symmetry classes and generalize the bohigas - giannoni - schmit conjecture . after a general introduction to quantum graphs in section [ sec : stargraphs ] we construct one ensemble of star graphs for each of the ten symmetry classes .
numerical results then show the fidelity of these ensembles to the predictions of the gaussian random - matrix ensembles .
in quantum mechanics most symmetries are described by some unitary operators @xmath3 that commute with the hamilton operator @xmath4 .
thus the operators @xmath3 ( or its hermitian generators ) describe constants of motion and they lead to a block diagonal form of the hamilton matrix in an eigenbasis of @xmath3 . if enough constants of motion @xmath5 are available such that the corresponding hermitian generators form a complete set of _ commuting _ observables the hamilton operator is eventually diagonalized in the common eigenbasis of the symmetry operators ( or their generators ) .
however , for any hermitian hamilton operator @xmath6 there is always a complete set of commuting hermitian operators @xmath7 which also commute with the hamilton operator ( e.g. projectors on eigenstates ) . in some sense
the notion of symmetry in this wide sense is obsolete for a single quantum system .
however , in most cases such a set of commuting operators will not have any corresponding classical observable and will only apply to a single system .
it is more appropriate for our purposes to consider a family or class of quantum systems .
such a family may arrise by varying some physical parameters ( like the strength of an applied magnetic field ) or , for disordered systems , by an ensemble of random potentials . in the derivation and application of semiclassical methods
one formally considers the asymptotics @xmath8 which is equivalent to a family of operators with fixed @xmath9 but some varying physical parameters .
a unitary operator @xmath5 is a unitary symmetry of a class of systems if it commutes with all hamilton operators in that class .
this notion avoids `` symmetry '' operators that would rely on a diagonalization of the hamilton operator . in semiclassics a unitary symmetry will have a classical correspondence . in the sequel
we will consider each hamilton operator as a member of a class without explicitly referring to it .
if a family of systems has a unitary symmetry all its hamilton operators can be brought to a blockdiagonal form .
each block can be regarded as a new hamilton operator on a reduced hilbert space .
let us assume that the hilbert space is completely reduced such that there are no more unitary symmetries .
what types of symmetry may such a reduced quantum system still have ?
what are the possible structures of the hamilton operator ( or the hamilton matrix ) and what are the consequences on its spectrum and its eigenvectors ?
such questions were for the first time addressed and partially answered by wigner and dyson @xcite .
dyson proposed a symmetry classification based on the behavior of quantum systems under time - reversal and spin rotation .
this lead to three symmetry classes ( the three - fold way ) : _ i. _ systems that are not time - reversal invariant , _ ii . _ time - reversal invariant spin - less particles , and _ iii . _ time - reversal invariant particles with spin @xmath10 .
time - reversal symmetry has immediate consequences on the form of the hamilton operator : spin - less particles can be described by real symmetric hamilton matrices in a time - reversal invariant basis , while systems without time - reversal invariance do not have any canonical basis and the hamilton matrix remains complex .
the influence of the symmetry class on spectral properties such as level repulsion has been investigated extensively within the field of random matrix theory @xcite .
we will give more details on random matrix theory in section [ sec : universality ] .
recently the wigner - dyson symmetry classification has been extended to a ten - fold way by including all different types of mirror symmetries in the spectrum @xcite . in the presence of a spectral mirror symmetry
every eigenvalue @xmath1 has a partner eigenvalue at @xmath2 ( we will set @xmath11 in the sequel without loss of generality ) . below we will describe the various ways
a mirror symmetry may arise and be combined with time - reversal invariance .
this leads to the seven novel symmetry classes . as shown in @xcite
there is a one - to - one correspondence between cartan s ten - fold classification of riemannian symmetric spaces and the ten symmetry classes of quantum systems .
we will use the convention to adopt the names given by cartan to the different classes of symmetric spaces for the according symmetry classes .
the novel symmetry classes are partly realized for dirac fermions in a random potential ( the _ chiral classes _ ) @xcite and partly for quasi - particles in mesoscopic superconductors or superconducting - normalconducting ( sn ) hybrid systems ( the _ andreev classes _ ) .
it is possible to construct much more general systems in the appropriate symmetry classes , e.g. two coupled spins or a generalized version of the pauli equation ( which includes the bogoliubov - de - gennes equation as a special case ) , and quantum graphs .
quantum maps which incorporate the corresponding symmetries have been discussed recently @xcite . due to their simplicity graphs
will be the focus of this work .
the following discussion of symmetry classes is summarized in table [ table1 ] .
quantum systems obey generalized time - reversal symmetry if there is an anti - unitary operator @xmath12 the _ generalized time reversal operator _ that first , commutes with the hamilton operator @xmath13=0 \label{eq : timerevinv}\ ] ] and second , is the condition that any state is invariant upto a phase when it is time - reversed twice .
the anti - unitarity of @xmath12 then leads to @xmath14 . ]
obeys @xmath15 anti - unitarity implies _
i. _ anti - linearity @xmath16 and _ ii . _
@xmath17 . for time - reversal invariant systems
@xmath12 changes the direction of time when applied to the schrdinger equation .
equivalently , when @xmath12 is applied to the retarded green s operator @xmath18 one gets @xmath19 which is just the advanced green s operator . time - reversal symmetry also effects other dynamic operators such as the unitary time evolution operator @xmath20 scattering problems can often be described by some unitary operator @xmath21 that connects incoming and outgoing states of energy @xmath22 .
time - reversal invariance leads to @xmath23 these equations also define time - reversal symmetry for quantum maps .
the transformation of the time development operator follows immediately from the condition on the hamiltonian . in scattering problems
@xmath21 can be related to a unitary combination of green s functions for definiteness consider @xmath24 and equation follows from the transformation of @xmath25 .
we have used the term _ generalized _ time - reversal operator because @xmath12 need not be the well - known conventional time - reversal operator . for a particle in @xmath26 the anti - unitary conventional time - reversal operator obeys @xmath27 where @xmath28 is the particle spin .
this conventional time - reversal operator obeys @xmath29 if the spin quantum number is integer @xmath30 , and @xmath31 if the spin is half - integer @xmath32 .
thus the most relevant and simplest realizations are for spin - less ( @xmath29 ) and spin-@xmath33 ( @xmath34 ) particles .
when a given quantum system is studied one should be aware that a generalized time - reversal operator may still exist which commutes with the hamiltonian while the conventional time - reversal operator may not commute with @xmath6 .
the consideration of time - reversal symmetries leads to three symmetry classes : either a system is not time - reversal invariant , or it is time - reversal invariant in the latter case the time - reversal operator either obeys @xmath35 or @xmath36 .
these classes have been called _ wigner - dyson classes _ and their impact on the form of hamilton matrices and universal spectral features will be discussed further in section [ sec : wigner - dyson ] and [ sec : universality ] .
additional spectral mirror symmetries lead to the novel symmetry classes to be discussed below .
kramers degeneracy occurs in time - reversal invariant quantum systems with @xmath36 .
if @xmath37 is an eigenvector with eigenvalue @xmath22 , then due to time reversal invariance @xmath38 is an eigenvector with the same eigenvalue @xmath22 .
it is straight forward to show that @xmath39 is orthogonal to @xmath37 using the properties of the time reversal symmetry operator .
this degeneracy is well known for spin-@xmath33 particles with conventional time - reversal symmetry .
a quantum system has a spectral mirror symmetry if the spectrum is symmetric : for every eigenvalue @xmath40 there is another eigenvalue @xmath41 . in general
, there may be some vanishing eigenvalues @xmath42 .
we will now discuss the symmetry operators related to such spectral mirror symmetries . according to a theorem by wigner any symmetry operation on hilbert space
is either represented by a unitary operator @xmath43 or an anti - unitary operator @xmath44 .
now take any eigenstate @xmath45 such that @xmath46 it is obvious that spectral mirror symmetry demands that either @xmath47 or @xmath48 .
this condition on an eigenstate leads eventually to the condition that the hamilton operator anti - commutes with either a unitary or an anti - unitary symmetry operator @xmath49_+&=0&\text{or } & & \left[\mathcal{c},\mathcal{h}\right]_+&=0 .
\label{eq : spectralmirror}\end{aligned}\ ] ] as an additional condition any state should be invariant upto a phase factor when acted upon twice with @xmath43 or @xmath44 .
it can be shown that it suffices to consider @xmath50 if a system is not time - reversal invariant one can always choose @xmath51 while in presence of time reversal symmetry either @xmath51 or @xmath52 .
spectral mirror symmetries relate green s operators at energy @xmath22 and @xmath53 @xmath54 for scattering problems this leads to @xmath55 and for the time development operator to @xmath56 the seven novel symmetry classes are obtained by all possible combinations of a spectral mirror symmetry with time - reversal symmetry ( with the additional requirement that @xmath57=0 $ ] or @xmath58=0 $ ] if both symmetries are supposed to hold ) .
first , there are three novel symmetry classes that are not time reversal invariant : either there is a unitary operator with @xmath51 or an anti - unitary with @xmath59 . in time - reversal invariant systems
one has both unitary and anti - unitary spectral mirror symmetry operators : if a unitary operator @xmath43 anti - commutes with the hamilton operator so does the anti - unitary operator @xmath60 . as @xmath61 and @xmath62 this leads to four symmetry classes that combine time - reversal symmetry with spectral mirror symmetry : if @xmath35 either @xmath63 ( @xmath51 ) or @xmath64 ( @xmath52 ) , if @xmath36 either @xmath64 ( @xmath51 ) or @xmath63 ( @xmath65 . for historic reasons
these seven classes have been split into two groups , the first group is given by the three _
chiral classes _ the ones that have a unitary mirror symmetry with @xmath51 .
their importance has first been observed in investigations of dirac fermions in a random gauge field where the spectral symmetry is related to chirality . for this reason we will call @xmath43 a _ chiral symmetry operator _ though in general @xmath43
need not be related to chirality .
the four remaining classes have mainly been discussed in connection to mesoscopic disordered superconductors or superconducting - normalconducting hybrid systems where the anti - unitary mirror symmetry is connected to electron - hole conjugation .
for this reason we call @xmath44 a _ charge conjugation symmetry operator _ , though again , in general @xmath44 need not be related to charge conjugation at all .
since andreev reflection is a main ingredient in the dynamics of superconducting - normalconducting hybrid systems we will call these classes _ andreev classes_. the detailed discussion of these symmetry classes and their impact on universal spectral features will be discussed in sections [ sec : chiral ] , [ sec : andreev ] and [ sec : universality ] . .
[ table1 ] the ten symmetry classes of quantum systems .
if a symmetry class obeys time - reversal symmetry or a spectral mirror symmetry the entry @xmath66 in the corresponding column indicates if the symmetry operator squares to @xmath67 .
the entry @xmath68 indicates that the corresponding symmetry is broken .
the last column gives the corresponding riemannian symmetric space ( of compact type ) . [
cols="^,^,^,^,^ " , ] time - reversal and spectral mirror symmetries restrict the form of hamilton and scattering matrices due to the relations , , and . by choosing an appropriate hilbert space basis for each symmetry class the symmetry operators
are represented by a simple matrix ( combined with the complex conjugation operator for anti - unitary operators ) .
these determine the explicit form of scattering matrices for each symmetry class .
note , that the following derivation of the scattering matrices depends on the choice of the basis .
there are many choices for the hilbert space basis in which the symmetry operators have a simple form . as a consequence many of the following identities are only valid in that special basis . especially the `` complex conjugation operator '' @xmath69 is defined with respect to a given basis . however , one may show that the bases chosen here can always be constructed from the general properties of the time - reversal and spectral mirror symmetries .
our choice of basis is biased by their later application to star graphs in section [ sec : stargraphs ] .
in addition , some symmetry classes have a further division into subclasses .
though we will mention all subclasses we will only give the scattering matrix in one of the subclasses .
quantum systems without spectral mirror symmetries belong to one of the three wigner - dyson classes @xmath70 , @xmath70i or @xmath70ii .
class @xmath70 contains quantum systems that are not time - reversal invariant .
there is no preferred basis in hilbert space and the scattering matrix @xmath21 may be any unitary @xmath71 matrix .
a time - reversal invariant system belongs either to class @xmath70i if @xmath35 or to class @xmath70ii
if @xmath36 . in class @xmath70i
there are time - reversal invariant bases such that @xmath72 for any basis state . in any such basis
the time - reversal symmetry operator is represented by the complex conjugation operator @xmath73 where the complex conjugation operator acts on a general state by complex conjugation of the coefficients @xmath74 .
the condition implies that a scattering matrix is represented by a unitary symmetric @xmath75 matrix @xmath76 for class @xmath70ii there is no time - reversal invariant basis . instead
, there are always bases in which the time - reversal symmetry operator is be represented by @xmath77 where @xmath78 is the @xmath75 identity matrix .
hilbert space has even dimension due to kramers degeneracy .
in such a basis the scattering matrix has the form @xmath79 with complex @xmath71 matrices @xmath80 that satisfy @xmath81 and @xmath82 and are further restricted by unitarity of @xmath21 .
a system with a spectral mirror symmetry connected to a unitary chiral symmetry operator @xmath43 ( with @xmath51 ) falls into one of the three chiral symmetry classes @xmath70iii , @xmath83i or @xmath84ii . since @xmath51 the eigenvalues are either @xmath85 or @xmath86
. in general , there will be @xmath87 positive and @xmath88 negative eigenvalues . the number @xmath89 distinguishes between different subclasses in each of the chiral classes ( @xmath90 is always even for class @xmath84ii ) . the integer @xmath90 has impact on both the form of hamilton ( or scattering ) matrices and on the spectral statistics . because @xmath43 relates states with positive energy to states with negative energy
there are @xmath90 vanishing energy eigenvalues due to the chiral symmetry .
we will focus on the subclasses with @xmath91 and set @xmath92 in classes @xmath70iii and @xmath83i , @xmath93 in class @xmath84ii .
hilbert space has even dimension in all three classes .
there are many bases that can be used as reference basis , for example the one , where @xmath43 is diagonal . here , biased by our following construction of star graphs we choose @xmath94 which can be obtained from the diagonal representation by a simple rotation .
the chiral class @xmath70iii contains systems without additional time - reversal invariance .
the other two chiral symmetry classes are time - reversal invariant with @xmath35 for class @xmath83i and @xmath36 for class @xmath84ii . in class @xmath83i
one may always choose a time - reversal symmetry operator of the form @xmath95 which commutes with the chiral symmetry operator @xmath43 . in class @xmath84ii
one may choose @xmath96 which also commutes @xmath43 .
due to the condition a scattering matrix @xmath21 in class @xmath70iii has the form @xmath97 where ( besides unitarity ) the @xmath71 matrices @xmath80 are further restricted by @xmath98 and @xmath99 . in class @xmath83i , due to time - reversal invariance , @xmath21 is symmetric , thus @xmath100 where @xmath101 and @xmath98 . finally , in class @xmath84ii , @xmath21 is a @xmath102 matrix of the form @xmath103 with additional constraints @xmath81 , @xmath99 , @xmath104 and @xmath105 .
a quantum system with a spectral mirror symmetry that does not belong to any of the chiral symmetry classes belongs to one of the four andreev classes @xmath84 , @xmath84i , @xmath83 or @xmath106iii .
the spectral mirror symmetry for these classes is related to an anti - unitary charge conjugation operator @xmath44 with @xmath64 for @xmath84 and @xmath84i while @xmath63 for @xmath83 and @xmath106iii .
the classes @xmath84 and @xmath83 are not time - reversal invariant while @xmath84i and @xmath106iii are time - reversal invariant with @xmath35 in @xmath84i and @xmath36 in @xmath106iii . the classes @xmath84 and @xmath84i do not split into subclasses . in appropriate @xmath107-dimensional bases the charge conjugation operator can be represented as @xmath108 in contrast the classes @xmath83 and @xmath106iii fall into two subclasses each .
the symmetry class @xmath83 allows for either an even or odd dimensional hilbert space . due to spectral mirror symmetry
there is always an eigenvalue on the symmetry point @xmath42 in an odd - dimensional hilbert space .
the subclass with odd ( even ) dimensional hilbert space may be called @xmath83-odd(even ) . in the following
we will restrict ourselves to the even - dimensional case and will follow the convention to call it symmetry class @xmath106 .
similarly @xmath106iii falls into the two subclasses @xmath106iii - odd and @xmath106iii - even .
the dimension of the corresponding hilbert spaces is twice an odd or twice an even number .
spectral mirror symmetry combined with kramers degeneracy implies two eigenvalues @xmath42 on the spectral symmetry point in class @xmath106iii - odd . in the sequel we will restrict to @xmath106iii - even which is physically more relevant .
an appropriate choice of basis in the hilbert space takes the charge conjugation operator of the symmetry classes @xmath106 and @xmath106iii ( we will not mention the `` even '' further ) to the form @xmath109 where @xmath78 is the @xmath71 identity matrix for class @xmath106 and the @xmath110 identity for class @xmath106iii .
the time - reversal symmetry operators in the classes @xmath84i and @xmath106iii have the representations @xmath111 and @xmath112 in an appropriate basis here
@xmath78 is the @xmath71 identity matrix in both equations .
these representations commute with the corresponding representations of the charge conjugation operators . the conditions and to scattering matrices @xmath21 of the form @xmath113 in classes @xmath84 and @xmath84i .
there are no further restrictions on the complex @xmath71 matrices @xmath80 for class @xmath84 ( apart from unitarity ) .
time - reversal invariance in class @xmath84i requires @xmath21 to be symmetric , thus @xmath101 and @xmath114 . in the symmetry class @xmath106
the scattering matrix has the form @xmath115 without further restrictions on the @xmath75 matrices @xmath80 . for class @xmath106iii
@xmath21 is a complex @xmath116 matrix of the form @xmath117 with @xmath81 , @xmath118 , @xmath104 and @xmath119 .
in the previous chapter we have summarized the symmetry classification of quantum systems .
it is completely general .
we have not yet related it to universal spectral properties .
this will be done in this section . in each symmetry class
there are several universal regimes with respect to their spectral statistics .
a universality class is a subset of a symmetry class which share the same spectral statistics ( or at least some universal spectral correlation functions ) .
the spectral statistics of a given universality class can be described ( and defined ) by some ensemble of random matrices ( usually there will be a lot of different ensembles that share the same universal spectral statistics ) . in this paper
we will focus on the _ ergodic _ universality classes that can be described by gaussian ensembles of hermitian matrices in each of the ten symmetry classes .
note , that three chiral symmetry classes and the symmetry classes @xmath83 and @xmath106iii fall into various subclasses as the universal spectral statistics is different in each of these subclasses they define different ergodic universality classes in the same symmetry class . as in the previous section we will only discuss one subclass in each of these cases . in the symmetry classes
@xmath70 , @xmath70i , @xmath70ii , @xmath84 and @xmath84i there is _ one _ unique ergodic universality class . in the chiral classes
@xmath70iii , @xmath83i and @xmath84ii we restrict to @xmath120 ( see section [ sec : chiral ] ) . finally , the classes @xmath83 and @xmath106iii have two subclasses ( see section [ sec : andreev ] ) and we will restrict ourselves to the subclasses @xmath106 ( @xmath83-even ) and @xmath106iii - even .
let us mention that apart from the ergodic universality classes there are a lot of other physically relevant universality classes within each symmetry class . in random - matrix theory these correspond to ensembles which are not equivalent to the gaussian ensembles .
for instance ensembles of banded or sparse hermitian matrices can describe quantum systems in a localized regime @xcite . in andreev systems more specialized random - matrix ensembles
can describe the so - called hard gap in the quasi - particle excitation spectrum that appears when a small part of the boundary of a normalconducting chaotic billiard is coupled to a superconductor @xcite . if no magnetic field is applied the resulting combined electron - hole dynamics near the fermi level is no longer chaotic and the system does not belong to an ergodic universality class . to reveal universality in the statistics of quantum spectra the system dependent mean density of states
has to be separated .
this is done by writing the density of states as a sum @xmath121 in presence of kramers degeneracy ( symmetry classes @xmath70ii , @xmath84ii , and @xmath106iii ) we define the density of states such that every doubly degenerate energy is counted only once in the sum @xmath122 .
let us also introduce a degeneracy factor @xmath123 , where @xmath124 for systems with kramers degeneracy and else @xmath125 . in equation
the first part @xmath126 is the average density of states which may be obtained by counting all states @xmath127 in an interval @xmath128 , then the number @xmath129 of states in that interval divided by @xmath130 is the average density of states @xmath131 for this to be well - defined it is necessary to choose @xmath130 self - consistently in range such that _ i. _ @xmath132 which is equivalent to taking the energy interval much larger the the mean spacing @xmath133 , and _ ii .
_ @xmath130 is small compared to compared to the scale on which the resulting @xmath134 changes . in systems that allow for
a classical limit and one may consider the semiclassical regime .
the scale @xmath130 is then chosen classically small ( @xmath135 as @xmath136 ) but large compared to the mean level spacing .
thus the average density of states is well - defined in the semiclassical regime .
it is given by weyl s law @xmath137 where @xmath138 is the classical hamilton function , and @xmath139 the number of freedoms .
this equation shows that the average density of states defined by weyl s law is system dependent and universal features can only arise due to the fluctuating part @xmath140 .
note , that weyl s law gives the mean density of states on scales much larger than the mean level spacing . in the presence of mirror symmetries the fluctuating part
@xmath140 may contribute to _ universal _ features in the density of states on the scale of the mean level spacing .
for classically chaotic ( hyperbolic ) systems the fluctuating part of the density of states is given by gutzwiller s trace formula @xcite as a sum over periodic orbits of the classical system @xmath141 here , @xmath142 is the primitive period of the orbit ( the time needed for a single traversal ) , @xmath143 is the stability amplitude of the periodic orbit ( @xmath144 is the _ reduced monodromy matrix _ and @xmath145 the _ maslov index _ ) and @xmath146 is the ( reduced ) action .
note , that hyperbolic chaos is a strong condition on a classical system all periodic orbits are hyperbolically unstable and isolated in the energy shell .
the energy scale for universal features is given by the mean level spacing @xmath147 .
introducing rescaled energies @xmath148 one obtains a density of states @xmath149 for the _ unfolded _ spectrum .
each ergodic universality class can be associated to a gaussian ensemble of random matrices . within one class
the gaussian ensembles differ only by the dimension of their matrices .
the universal features of spectral statistics are extracted in the limit of large matrices . in each gaussian ensemble
the probability for a hamiltonian matrix @xmath6 ( with symmetries according to one of the ten symmetry classes ) has the form @xmath150 where @xmath151 is a normalization constant , @xmath70 is an overall scale that fixes the mean level spacing , and the measure @xmath152 is given by @xmath153 where the product runs over all independent elements of @xmath6
. in general , on may denote the gaussian ensemble for the symmetry class @xmath154 by @xmath154-ge .
we will use this notion for the andreev classes .
note , that for some symmetry classes one should distinguish various ergodic universality classes . as we have restricted our investigations to just one relevant subclass we will use the name of the whole symmetry class for the gaussian ensembles .
let us now define the statistical functions that are in the center of our investigation .
for a physical system the following averages are either performed over some system parameters or over different parts of the spectrum .
we will always use unfolded spectra with unit mean level spacing .
spectral averaging is only possible if the universal results are invariant under shifts of the energy @xmath155 .
there is an important difference between the wigner - dyson classes where the universality was conjectured for a single spectrum of one system and the remaining seven symmetry classes where some universal features near energy @xmath42 can only be obtained by averaging over different spectra . in the classical limit
one naturally obtains many spectra for the same physical system by formally changing @xmath9 . thus even for the seven novel symmetry classes one may average over different spectra for the _ same physical system_. we will be interested in the two simplest correlation functions and their fourier transforms .
we will call the latter _
form factors_. the first correlation function is simply the averaged fluctuating part of the density of states @xmath156 .
if the spectral statistics is invariant under shifts this expectation value must vanish ( if not it would be a constant over scales much larger than the mean level spacing in contradiction to its definition ) .
the spectral statistics near a spectral mirror symmetry is not invariant under energy - shifts .
non - trivial contributions to the mean fluctuating part of the density of states may then arise .
these have to appear on the scale of mean level spacing ( else it would be inconsistent with the separation of the density of states in @xmath157 ) .
the fourier transform of the averaged fluctuating part of the density of states is the _ first - order form factor _
@xmath158 where @xmath159 is the heisenberg time . inverting the fourier
transform one may represent the deviations from weyl s law in the expectation value for the density of states as @xmath160 note , that for @xmath161 the first - order form factor is the trace of the time evolution operator @xmath162 . the second - order correlation function is defined by @xmath163 if the spectral statistics is invariant under energy shifts it only depends on the energy difference @xmath164 averaging over different parts of the spectrum for a given system is an average over @xmath165 .
its fourier transform with respect to @xmath164 is the _ second - order form factor _
@xmath166 where we have suppressed the possible dependency on @xmath165 . for physical spectra
a time average over a small time interval @xmath167 has to be added to the definition of the form factors
. we will now summarize the relevant results from random - matrix theory ( for more details see @xcite ) .
the ergodic universality classes for quantum systems in the wigner - dyson classes are described by the well - known gaussian ensembles of random matrix theory gue ( @xmath70-ge ) , goe ( @xmath70i - ge ) and gse ( @xmath70ii - ge ) .
the universal spectral statistics is invariant under shifts of the energy @xmath168 .
thus the expectation value of the fluctuating part of the density of states vanishes and so does its fourier transform @xmath169 the two - point correlation functions are given by @xmath170 where @xmath171 is the sine integral .
the corresponding second - order form factors are given by @xmath172 the gaussian random - matrix ensembles in the chiral symmetry classes are known as chgue ( @xmath70iii - ge ) , chgoe ( @xmath83i - ge ) and chgse ( @xmath84ii - ge ) .
the andreev ensembles @xmath84-ge , @xmath84i - ge , @xmath106-ge , and @xmath106iii - ge do not have any established name .
the spectral statistics of these ensembles is not invariant under energy - shifts and , as a consequence , deviations from weyl s law need not vanish near @xmath173 . at energies much larger than the mean level spacing @xmath174 wigner - dyson statistics is recovered .
thus , for the two - point correlation function we have @xmath175 the universal features near the symmetry point @xmath173 are most prominent in the density of states .
though there are universal deviations from wigner - dyson statistics in all correlation functions we will focus on the density of states .
the universal deviations from weyl s law for the chiral ensembles are given by @xcite @xmath176 the corresponding first - order form factors can be calculated explicitly in terms of the complete elliptic integrals of first , second and third kind @xmath177 , @xmath178 and @xmath179 ( we use the convention that @xmath180 is real for @xmath181 @xcite ) .
they are given by @xmath182 it has been conjectured by bohigas , giannoni and schmit that quantum systems ( in the semiclassical regime ) with a chaotic classical limit have universal spectral fluctuations that coincide with the predictions of one of the wigner - dyson gaussian ensembles of random - matrix theory gue , goe , or gse .
more precisely in an average over different parts of the unfolded spectrum the @xmath183-point correlation functions for @xmath184 of a _ single _ spectrum are conjectured to coincide with the corresponding correlation functions of the wigner - dyson ensemble .
the mean density of states of a given quantum system is non - universal and can not be described by random - matrix theory .
semiclassically it is given by weyl s law .
a lot of evidence has since been gathered both numerical and analytical that this conjecture is true in generic chaotic systems @xcite ( though a few exceptions are known @xcite ) .
many approaches have been used to understand and proof the fidelity to random - matrix theory in complex quantum systems @xcite .
recently there has been considerable progress in the semiclassical approach using periodic orbit theory @xcite .
bohigas _ et al _ stated their conjecture before the impact of spectral mirror symmetries on spectral statistics has been recognized .
a proper generalization of their statement has to take into account that a spectral average will wipe out all effects of a spectral mirror symmetry .
thus the original conjecture is expected to hold for the novel symmetry classes as well : after averaging over different parts of a single spectrum they will show the universal spectral fluctuations of gue , goe , or gse . the additional universal features in physical systems near the spectral symmetry point can only be observed when an average over various spectra is performed .
this corresponds to an average over some system parameter .
we conjecture that for classically chaotic systems with a spectral symmetry _ all _ correlation functions of the fluctuating part of the unfolded density of states @xmath185 as given by averaged over _
one _ system parameter coincide with those of the corresponding gaussian random - matrix ensemble in the novel symmetry classes .
this includes universal deviations from weyl s law in the density of states itself .
note , that though there are seven symmetry classes there are infinitely many ergodic universal classes due to the different subclasses . though some average is certainly necessary we may still conjecture the fidelity to ergodic random - matrix theories of a single physical system by formally averaging over spectra for different values of an effective planck s constant . in superconducting - normalconducting hybrid structures
this corresponds to an average over fermi energy @xmath186 .
quantum graphs have been introduced by kottos and smilansky @xcite as simple quantum systems with an exact semiclassical trace formula for the density of states .
they consist of @xmath187 vertices and connected by @xmath188 bonds .
each bond @xmath189 connects two vertices and has a length @xmath190 .
a particle propagates freely on the bonds and is scattered at the vertices by prescribed boundary conditions which leads to quantization . in their first approach kottos and smilansky
considered vertex boundary conditions that implied current conservation and continuity of the wave function .
the continuity condition is not always essential and has often been relaxed . in that case
the boundary conditions at a vertex are specified by any unitary scattering matrix that transforms incoming waves to outgoing waves
unitarity of the vertex scattering matrix is equivalent to current conservation . a star graph consists of @xmath188 bonds @xmath191 of length @xmath192 ( @xmath193 ) and @xmath194 vertices @xmath195 ( @xmath196 ) .
each bond @xmath191 emanates from the central vertex @xmath197 and connects it to the peripheral vertex @xmath195 ( see figure [ fig : stargraph ] ) .
we will allow for a multi - component wave function on the graph .
the number of components @xmath186 is assumed to be equal on all bonds .
it may represent different spin components or electron and hole components of a quasi - particle .
the @xmath186-component wave function on the bond @xmath191 is @xmath198 where @xmath199 is the distance from the central vertex and @xmath200 are @xmath186-component vectors of constant coefficients for the incoming ( outgoing ) waves on bond @xmath191 ( `` incoming '' and `` outgoing '' will always be used with respect to the central vertex ) .
it is convenient to combine all the coefficients of incoming ( outgoing ) waves into two vectors of dimension @xmath201 @xmath202 the boundary condition at the center can then be written in the form @xmath203 the diagonal @xmath204 matrix @xmath205 describes the propagation along the bonds . here ( and in the rest of this section ) @xmath206 indicates the component of the wave function and @xmath207 is the bond index . the _ central scattering matrix _
@xmath208 is a fixed unitary @xmath204 matrix that defines the boundary conditions at the center .
for definiteness , we will assume that different components of the wave function do not mix at the center , thus @xmath209 where the @xmath210 matrix @xmath211 describes scattering of the @xmath212-component at the center .
the boundary conditions at the peripheral vertices may be described by one fixed @xmath213 vertex scattering matrices for each peripheral vertex these can be combined to a single unitary @xmath204 peripheral scattering matrix @xmath214 such that @xmath215 we will not allow any dependence of the scattering matrices @xmath208 and @xmath214 on the wave number @xmath218 .
uniqueness of the wave function and the boundary conditions and lead to the quantization condition @xmath219 where we introduced the _ bond scattering matrix _ @xmath220 .
non - trivial solutions of these equations exist only when the wave number belongs to the discrete spectrum @xmath221 given by the zeros of the corresponding determinant @xmath222 the density of states for the graph is defined as @xmath223 where @xmath124 in systems with kramers degeneracy ( else @xmath125 ) .
let us now write the density of states as a sum of its mean @xmath224 and an oscillating part @xmath225 @xmath226 for both contributions one can give an exact semiclassical expression .
the mean density of states is given by weyl s law @xmath227 and the oscillating part obeys the trace formula @xcite @xmath228 in the sequel we will consider star graphs where all bond lengths are equal @xmath229 . in that case the bond scattering matrix is a periodic function of @xmath218 @xmath230 where @xmath231 is the _ reduced bond scattering matrix_. thus , the spectrum is also periodic and the trace formula simplifies to @xmath232 the periodicity of the spectrum will not be relevant here as we are interested in features on the scale of a mean level spacing . for equal bond lengths
the trace formula can be derived in a few lines : let @xmath233 ( @xmath234 ) be the eigenvalues of the unitary reduced bond scattering matrix @xmath235 .
the quantization condition is equivalent to @xmath236 ( @xmath237 ) and the density of states is @xmath238 the mean density of states @xmath239 is just the @xmath240 term in the sum over @xmath183 while the rest gives the trace formula for the oscillating part ( the second line follows from the first by poisson s summation formula @xmath241 ) .
the trace formula can be interpreted as a sum over _ periodic orbits _
@xmath87 on the graph .
a periodic orbit @xmath242 $ ] of length @xmath183 is defined by a sequence of @xmath183 peripheral vertices @xmath243 visited one after the other together with the specification of the wave component @xmath244 between two vertices ( cyclic permutations define the same orbit ) .
a periodic orbit is _ primitive _ if it is not the repetition of a shorter periodic orbit . in terms of primitive periodic orbits
@xmath87 and its repetitions
the trace formula reads @xmath245 where @xmath246 is the length of the primitive periodic orbit @xmath247 is the amplitude of the primitive orbit and @xmath248 its phase ( `` action '' ) .
note , that we set @xmath249 and @xmath250 .
the similarity of the sum over periodic orbits to the semiclassical gutzwiller trace formula is evident .
however , while semiclassics , in general , is an approximation the semiclassical trace formula for quantum graphs is exact . the trace formula will be our main tool in the analysis of universal spectral statistics .
it will lead us to a simple expression for the form factors that can easily be averaged numerically . in the second paper of this series the trace formula will be in the center of an analytic approach to universality .
since universality exists on the scale of the mean level spacing we will write @xmath251 where @xmath252 is the mean level spacing . in terms of the rescaled wave number the trace formula is @xmath253 we have introduced the shorthand @xmath254 for the @xmath183-th trace of the reduced bond scattering matrix .
the first - order form factor is obtained by a fourier transform and a subsequent time average .
it obeys the trace formula @xmath255 where the bar denotes a time average over a small time interval @xmath256 and @xmath257 the brackets @xmath258 denote an average over an ensemble of graphs .
this can be written more compactly as @xmath259 where the continuous time average has been replaced by an average over the discrete time @xmath260 .
the second - order form factor for a graph also obeys a trace formula which , after a spectral average over the central wave number , is given by @xmath261 where @xmath262 if no spectral average is performed additional terms appear .
these are irrelevant for the graphs in the wigner - dyson classes ( they do not survive the subsequent ensemble average ) . here
, we will not consider the second - order form factor for graphs in the novel symmetry classes where the additional terms are relevant near the central energy @xmath263 .
we will now construct ensembles of star graphs for each symmetry class .
the star graphs will be constructed in such a way that spectral fluctuations of the corresponding ergodic universality classes can be expected . though we are not able to prove an equivalent conjecture we will give strong evidence .
the constructions of star graphs for each symmetry class are based on a proper choice of the central and peripheral scattering matrices @xmath208 and @xmath214 .
both have to obey the right symmetry conditions ( see section [ sec : symmetry ] ) . as an example of star graphs in class @xmath70i that _ do not _ belong to the ergodic universality class
let us mention _ neumann _ star graphs .
these have a one - component wave function ( @xmath266 ) on the bonds , dirichlet boundary conditions at the peripheral vertices such that @xmath267 and neumann boundary conditions at the center , thus @xmath268 .
such graphs have been investigated first in @xcite and in more detail in @xcite in contrast to our approach the bond lengths were chosen different for each bond ( and incommensurate ) .
however , neumann boundary conditions at the center favor backscattering and lead to non - universal ( localization ) effects @xcite .
our approach is different in as much as we allow for more general scattering matrices at the center and in as much as we will always consider an ensemble of graphs .
the occurrence or non - occurrence of localization effects can be traced back to a gap condition on the matrix @xmath269 .
this bistochastic matrix describes the corresponding `` classical '' probabilistic dynamics ( equivalent to a frobenius - perron operator ) . in chaotic ( ergodic )
systems the frobenius - perron operator has a finite gap in the spectrum between the ( unique ) eigenvalue one and all other eigenvalues which describe the decay of the probability distribution . in neumann star
graphs this gap is small and vanishes in the limit @xmath270 faster than @xmath271 which leads to non - ergodic spectral statistics .
in general one needs a multi - component wave function to introduce the different symmetries .
the number of components @xmath186 has been chosen minimal under the additional assumptions that the components do not mix at the central vertex and that time - reversal is only broken at the peripheral vertices . the central scattering can be chosen in a very simple way by using the symmetric @xmath210 _ discrete fourier transform matrix _
@xcite @xmath272 or its complex conjugate for each component .
an incoming wave on a given bond is scattered with equal probability to any bond which excludes localization effects .
indeed , the matrix @xmath273 has one eigenvalue @xmath274 while all other @xmath275 eigenvalues vanish ( so the gap is maximal ) .
the bond scattering matrix @xmath276 for each ensemble of graphs is constructed by demanding that the matrices @xmath208 , @xmath214 , and @xmath277 do all have the canonical forms of the desired symmetry class given in section [ sec : symmetry ] .
note , that for star graphs we are interested in the @xmath218 spectrum , so in the canonical forms for scattering matrices the energy @xmath22 has to be replaced by @xmath218 .
the ensemble of graphs is built by introducing some random phases into the peripheral scattering matrix .
let us start with the simplest case : an ensemble of star graphs in class @xmath70i where a one - component wave - function suffices to incorporate the time - reversal symmetry which demands that the unitary matrices @xmath208 , @xmath214 , and @xmath278 are all symmetric .
now , @xmath214 and @xmath279 are diagonal for @xmath266 , and choosing @xmath280 we meet all requirements . at the peripheral vertices we are free to choose one random phase @xmath281 for each peripheral vertex @xmath282 independently such that @xmath283 where @xmath284 is uniformly distributed .
second - order form factor for star graphs in the wigner - dyson ensembles averaged over @xmath285 realizations of with @xmath286 bonds ( additional time average over an interval of length @xmath287 ) .
dashed lines : prediction by gaussian random - matrix theory .
full lines : numerically calculated form factor for graphs .
inlets : discrete time form factor @xmath265 as function of @xmath288 for the class @xmath70 and @xmath70i graphs . for the class
@xmath70ii graphs @xmath265 vanishes for odd @xmath183 the inlet shows @xmath289 for even @xmath183.,title="fig : " ] second - order form factor for star graphs in the wigner - dyson ensembles averaged over @xmath285 realizations of with @xmath286 bonds ( additional time average over an interval of length @xmath287 ) . dashed lines : prediction by gaussian random - matrix theory .
full lines : numerically calculated form factor for graphs .
inlets : discrete time form factor @xmath265 as function of @xmath288 for the class @xmath70 and @xmath70i graphs . for the class
@xmath70ii graphs @xmath265 vanishes for odd @xmath183 the inlet shows @xmath289 for even @xmath183.,title="fig : " ] second - order form factor for star graphs in the wigner - dyson ensembles averaged over @xmath285 realizations of with @xmath286 bonds ( additional time average over an interval of length @xmath287 ) . dashed lines : prediction by gaussian random - matrix theory .
full lines : numerically calculated form factor for graphs .
inlets : discrete time form factor @xmath265 as function of @xmath288 for the class @xmath70 and @xmath70i graphs . for the class
@xmath70ii graphs @xmath265 vanishes for odd @xmath183 the inlet shows @xmath289 for even @xmath183.,title="fig : " ] for class @xmath70 we have to break time - reversal symmetry .
this may be done by choosing a non - symmetric central scattering matrix for a one - component wave - function on the graph .
as we like to keep the simplicity of the discrete fourier transform matrix we choose another simple construction with a two - component wave function , a central @xmath290 scattering matrix @xmath291 and @xmath292 for the peripheral scattering matrix .
the diagonal matrices @xmath293 are @xmath294 where the independent random phases @xmath281 , @xmath295 , and @xmath296 are uniformly distributed . for class
@xmath70ii a four - component wave - function is needed to incorporate time - reversal invariance with @xmath36 into our scheme for star graphs .
indeed , the number of components must be even as discussed above in [ sec : wigner - dyson ] .
in addition , we assumed that components do only mix at the peripheral vertices .
then , a @xmath297 scattering matrix at the peripheral vertices is the minimal matrix dimension that allows for component mixing as can be seen from the canonical form of a @xmath70ii scattering matrix .
the diagonal matrix @xmath277 is a diagonal unitary matrix of the canonical form .
all further requirements are met by choosing @xmath298 for the central @xmath299 scattering matrix , and @xmath300 for the peripheral scattering matrix .
the diagonal matrices @xmath293 are given by @xmath301 where the independent random phases @xmath281 , @xmath295 , and @xmath296 are uniformly distributed .
ergodic spectral statistics may be expected for the three ensembles in the wigner - dyson classes .
this is strongly supported by a numerical calculation of the second - order form factor ( see figure [ fig : numwd ] ) .
first - order form factor for andreev star graphs in the classes @xmath84 and @xmath84i averaged over @xmath285 realizations with @xmath286 bonds ( additional time average over an interval of length @xmath287 ) . dashed lines : prediction by gaussian random - matrix theory . full lines : numerically calculated form factor for graphs .
inlets : discrete time form factor @xmath302 as function of @xmath288 for even @xmath183 ( @xmath264 vanishes by construction for odd @xmath183 ) .
, title="fig : " ] first - order form factor for andreev star graphs in the classes @xmath84 and @xmath84i averaged over @xmath285 realizations with @xmath286 bonds ( additional time average over an interval of length @xmath287 ) . dashed lines : prediction by gaussian random - matrix theory .
full lines : numerically calculated form factor for graphs .
inlets : discrete time form factor @xmath302 as function of @xmath288 for even @xmath183 ( @xmath264 vanishes by construction for odd @xmath183 ) .
, title="fig : " ] let us start with the andreev star graphs for the classes @xmath84 and @xmath84i where the wave function can be chosen in the simplest case to have two components .
the first will be called `` electron '' and the second `` hole '' .
the transfer matrix @xmath277 and the central scattering matrix defined by @xmath303 obey the symmetry condition .
the first - order form factor for andreev star graphs in the classes @xmath106 and @xmath106iii ( see figure [ fig : numandreevc ] for details).,title="fig : " ] the first - order form factor for andreev star graphs in the classes @xmath106 and @xmath106iii ( see figure [ fig : numandreevc ] for details).,title="fig : " ] the peripheral scattering matrix may be chosen such that complete _ andreev scattering _
( electron - hole conversion ) takes place @xmath304 where the diagonal matrix @xmath305 is @xmath306 for class @xmath84 , and @xmath307 for class @xmath84i .
the random phases @xmath281 are uniformly distributed and @xmath308 with equal probability . for the andreev classes @xmath106 and @xmath106iii and as well for the three chiral classes @xmath70iii , @xmath83i , and @xmath84ii a four - component wave function is needed .
we will call the first ( last ) two components `` electron '' ( `` hole '' ) .
the first - order form factor for chiral star graphs ( see figure [ fig : numandreevc ] for details ) .
, title="fig : " ] + the first - order form factor for chiral star graphs ( see figure [ fig : numandreevc ] for details ) .
, title="fig : " ] + the first - order form factor for chiral star graphs ( see figure [ fig : numandreevc ] for details ) .
, title="fig : " ] the symmetry requirements , , , , and are met by the transfer matrix @xmath277 and the central scattering matrix defined by @xmath309 in all five remaining classes we choose the peripheral scattering matrix such that complete andreev scattering takes place . for @xmath106 and @xmath106iii the simplest choice obeying the symmetry requirements are @xmath310 where the diagonal matrices @xmath293 are @xmath311 for class @xmath106 , and @xmath312 for class @xmath106iii . the random phases @xmath281 and
@xmath295 are uniformly distributed and @xmath308 with equal probability .
the simplest choice for peripheral scattering matrices in the chiral classes is @xmath313 the diagonal matrices @xmath293 have to be chosen according to the requirements of each symmetry class . for class @xmath70iii they are @xmath314 where @xmath315 with equal probability and the phases @xmath281 and @xmath295 are uniformly distributed .
the @xmath83i star graphs can be obtained from the @xmath70iii case by the additional restrictions @xmath316 finally , for class @xmath84ii the peripheral scattering matrix is defined by @xmath317 we have checked numerically that the first - order form factor for the constructed andreev and chiral star graphs obeys the corresponding prediction by gaussian random - matrix theory ( see figures [ fig : numandreevc],[fig : numandreevd ] , and [ fig : numchiral ] ) .
we are indebted to felix von oppen and martin zirnbauer for many helpful suggestions , comments and discussions .
we thank for the support of the sonderforschungsbereisch / transregio 12 of the deutsche forschungsgemeinschaft .
|
in a series of two papers we investigate the universal spectral statistics of chaotic quantum systems in the ten known symmetry classes of quantum mechanics . in this first paper
we focus on the construction of appropriate ensembles of star graphs in the ten symmetry classes .
a generalization of the bohigas - giannoni - schmit conjecture is given that covers all these symmetry classes .
the conjecture is supported by numerical results that demonstrate the fidelity of the spectral statistics of star graphs to the corresponding gaussian random - matrix theories .
|
the best known models for the spread of infectious disease in human populations are based on the classical sir model of kermack and mckendrick @xcite .
the same system of ordinary differential equations ( odes ) may be derived as the large population limit of a density - dependent markov jump process using the methods of kurtz @xcite .
this stochastic formulation brings a number of mathematically attractive properties such as explicit likelihood formulas and ease of simulation .
nevertheless , a drawback of the kermack and mckendrick - type models is that they can be unrealistic in describing the interactions of infectives and susceptibles as they are based on assumptions of homogeneous mixing @xcite . in recent years
there has been considerable interest in developing alternatives to the classical sir , for instance , via network - based epidemic models , as reviewed by pellis et al . @xcite and house and keeling @xcite .
pair approximation models have been formulated to account for spatial correlations while maintaining mathematical tractability @xcite .
much of the work on these models has regarded the dynamics of the deterministic system obtained in the large population limit , such as the work of miller and volz on edge - based models @xcite and that of altmann @xcite on a process with dynamic partnerships . as in the work of miller and volz @xcite , the configuration model ( cm ) random graph
is often chosen to dictate the network structure in epidemic models @xcite .
cm networks can be viewed as a generalization of the erds - rnyi random graph , i.e. the @xmath0 model where @xmath1 is the size of the graph and edges are drawn between each pair of nodes independently , with probability @xmath2 . in the limit of a large graph this results in a poisson degree distribution ; however , in a general cm graph model framework poisson may be replaced with any appropriate degree distribution ( see , e.g. , @xcite ) .
miller and volz heuristically formulated the limiting system of odes that govern the dynamics of the sir epidemic on a cm graph .
decreusefond et al .
@xcite and janson et al .
@xcite have recently given formal proofs for the correctness of the miller - volz equations as the law of large numbers ( lln ) for the stochastic system under relatively mild regularity conditions on the degree distribution and on the epidemic initial condition .
in addition , volz and miller have conjectured the limiting equations for generalizations which include dynamic graphs ( edge formation and breakage ) , heterogeneous susceptibility , and multiple types of transmission @xcite . to date , the convergence for the stochastic systems in these cases has not been formally verified but numerical studies done by the authors suggest that they are correct . as in the case of classical sir @xcite the early behavior of an epidemic on a cm graph
can be approximated by a suitable branching process @xcite .
such approximation allows one in turn to use the early epidemic data to statistically ascertain the probability of a major outbreak ( large number of infections among the network nodes ) as well as to estimate the rate of infection spread and changes in the contact network as described below . by and large ,
there have been limited studies of statistical estimation for network - based models @xcite . in one of the early papers on the topic , oneill and britton study bayesian estimation procedures for the @xmath0 model
when the network is assumed to be of reasonable enough size that the network structure may be included as missing data and imputed via a monte carlo scheme @xcite .
more recently , groendyke has extended this approach to non - markovian dynamics by allowing both the infection time and the recovery time to follow an arbitrary gamma distribution @xcite .
however , these methods are generally tailored to networks of small size , where posterior sampling is feasible , and may encounter various difficulties in large networks when the imputation becomes computationally expensive and often impractical .
the framework presented here , on the other hand , assumes the proportion of infectives is small relative to the population size .
this allows us to avoid explicit imputation of the network and makes the numerical complexity of the analysis comparable to that of a small homogenous sir epidemic @xcite .
the main contribution of the current paper is to present a statistical inference method for analyzing the early stages of an epidemic , or a small outbreak , evolving according to sir type dynamics on a random graph .
a novel aspect of our method is that we assume the random graph structure evolves in response to the epidemic progression .
this allows us to account for changing contact patterns in response to infection , for example due to population behavioral changes or health interventions ( e.g. quarantine ) .
the development of our methods is motivated by the devastating @xmath3 outbreak of ebola virus in west africa .
while the west africa outbreak received considerable attention in fall 2014 due to the dramatic rise in the number of cases , there was an independent ebola outbreak in the democratic republic of the congo ( drc ) occurring at the same time .
this much smaller outbreak began july 26 , 2014 in quateur province and lasted until november 2014 .
the index case was a woman living in inkanamongo village that presumably became infected by consuming bushmeat of an infected animal .
several healthcare workers involved in a postmortem cesarean section on the women subsequently became ill and generated further chains of transmission @xcite .
there were a total of @xmath4 confirmed infections in the 2014 drc outbreak .
the time series of cumulative case counts was reported by maganga et al .
however , the analysis method presented here does not require temporal data and instead utilizes the distribution of secondary cases , also given in @xcite .
the index case is believed to have caused @xmath5 secondary cases , presumably due to her funeral acting as a super - spreading event @xcite .
hence , we assume this data point to be an outlier and exclude it from our final analysis , as maganga et al . also did in their estimation of @xmath6 .
the secondary case distribution for other named contacts is shown in figure [ secondcases ] .
one patient caused three subsequent cases , two patients caused two additional cases , @xmath7 caused one additional case and @xmath8 patients caused zero additional cases . these may be viewed as observations from the post - index offspring distribution of the branching process approximation , i.e. the number of infections caused by an individual who is himself not the index case .
although , as already mentioned , the temporal data measurements are not explicitly required for parameter estimation in our approach , the available temporal information can be directly incorporated into the likelihood function , or used to inform the parameterization of the prior distributions , which is what was done for the current drc dataset .
details are given below .
consider a graph @xmath9 where the vertex set , @xmath10 , corresponds to individuals in a population of size @xmath1 and the edge set , @xmath11 , corresponds to potentially infectious contacts occurring between such individuals .
the graph structure is given as a realization of a configuration model random graph with a prescribed degree distribution , @xmath12 , where the probability that a node has degree @xmath13 is denoted by @xmath14 .
each node @xmath15 is assigned @xmath16 half - edges that are drawn at random from @xmath12 .
then , the pool of half - edges is paired off uniformly to form the final network .
this link formation model does not exclude the possibility of self - loops or multiple edges but this has a negligible effect in the large graph limit ( see , e.g. , discussion in @xcite ) .
the general disease framework adopted is the standard compartmental sir model where , for every time @xmath17 , each node @xmath15 is classified as susceptible ( s ) , infectious ( i ) , or recovered ( r ) from the infection . given some number of initially infectious nodes , a node @xmath18 becomes infective via transmission along an edge from one of his infectious neighbors . in the markovian case , we assume that @xmath18 remains infectious for an exponentially distributed amount of time with rate parameter @xmath19 , which we refer to as the recovery rate . more generally , a non - exponential ( e.g. , gamma ) recovery rate leads to the so - called semi - markov sir model @xcite .
while infectious , the infective @xmath18 attempts to transmit the infection to all of his susceptible neighbors according to an exponential distribution with rate @xmath20 . in the case that the realization of the infection clock " for a particular neighbor rings " before the recovery clock " , then this neighbor becomes infected .
the epidemic ends when there are no more infectives .
in addition to these standard sir dynamics , we allow the network structure to change due to infection status .
specifically , we assume that infectious individuals drop each of their contacts according to an exponential distribution with rate @xmath21 .
the dropped contacts , which could account for behavioral changes due to disease such as isolation or decreased mobility , can not be reformed .
for an index case @xmath18 with degree @xmath16 , at most @xmath16 secondary infections can be produced .
conditional on the recovery time of that index case , @xmath22 , the probability that infection has passed to any particular neighbor is given by @xmath23 where the first term in the product on the right - hand side is the probability that an edge transmitted infection prior to being dropped , while the second term represents the probability that either an infection or a drop occurred before recovery .
therefore , the total number of secondary infections caused by node @xmath18 , which we denote by @xmath24 , conditional on the time of recovery @xmath22 , is given by @xmath25 if the recovery time is not known but assumed to follow a distribution @xmath26 , then we may analogously define @xmath27 , the conditional probability of @xmath28 offspring given degree @xmath16 .
the law of total probability implies @xmath29 we will assume identical recovery distributions for all individuals and , thus , the offspring distribution will be the same for all index cases .
the final form for the offspring distribution of an index case may be found by supposing that the degree , @xmath16 , of the index case is unknown .
let @xmath30 denote the probability that the index case has degree @xmath16 .
therefore , the law of total probability gives @xmath31 that is , we sum over all possible degrees that could yield at least @xmath28 secondary infections and weight them according to the degree distribution .
we now consider the offspring distribution for a post - index case . by definition ,
such an individual acquired infection from another individual in the network and , thus , has at least one neighbor .
therefore , some adjustments are needed to account for the fact that post - index cases have a degree distribution which differs from @xmath12 .
let @xmath32 denote the probability that a given neighbor in the cm network has degree @xmath13 .
then it is known @xcite that @xmath33 since at least one of the neighbors of a post - index case has already been infected , he may pass the infection to at most @xmath34 of his neighbors .
let @xmath35 denote the offspring distribution for a post - index infection @xmath18 .
similarly to eq .
@xmath36 we derive @xmath37 for a fixed set of parameters the basic reproductive number , @xmath6 , can be calculated as @xmath38 , i.e. the average number of secondary infections caused by a post - index case .
that is , @xmath6 is given by @xmath39 to illustrate , we assume that the degree distribution is poisson with mean parameter @xmath40 and the recovery distribution is exponential with rate parameter @xmath19 .
this implies @xmath41 and @xmath42 therefore , by eq .
@xmath36 , the offspring distribution for an index case is given by @xmath43 and , by eq .
@xmath44 , the post - index case offspring distribution is given by @xmath45 this expression has no simple analytical form but it is not hard to approximate the integral numerically since it can be written as an expectation against the recovery distribution .
therefore , a simple monte carlo sample from the desired recovery distribution allows for efficient computation of this term . using eq .
@xmath46 , we can calculate the basic reproductive number in this markovian case . for an arbitrary degree distribution , @xmath6
is given by @xmath47 which only differs in the inclusion of @xmath21 from the corresponding formula on a static cm graph @xcite . here
@xmath48 by assumption .
the summation in eq .
@xmath49 represents the expected excess degree , i.e. the degree of a node which is necessarily a neighbor of a node , not counting the known edge . in the particular case here of a poisson degree distribution , @xmath6
is found to be ( cf .
, e.g. , @xcite chapter 6 ) @xmath50 in practice , outbreak data may not arise solely from a single index case and often @xmath51 separate chains of infection are tracked .
suppose the data @xmath52 corresponds to the number of secondary infections for each of @xmath51 independent index cases and the data @xmath53 corresponds to secondary infections caused by each of the @xmath54 post - index cases .
let @xmath55 denote the vector of parameters where @xmath19 and @xmath40 represent the parameters of the recovery time and degree distributions , respectively .
the offspring distributions given in eqs .
@xmath36 and @xmath44 allow for explicit formulation of the likelihood for @xmath56 which is given by @xmath57 with the specification of the likelihood , maximum likelihood estimators ( mles ) for the rate parameters can be found by numerical optimization .
given the mle @xmath58 , the corresponding estimator for the basic reproductive number can be calculated by application of the continuous mapping theorem to the expression for @xmath6 given in eq .
@xmath46 . for example , under the assumptions of our example in section [ sec : ex ] , the estimator following from eq .
@xmath59 would be @xmath60 denote the vector of current parameters as @xmath61 .
if we denote the parameter prior distribution @xmath62 then the likelihood function above may be also used to compute the metropolis - hastings acceptance probability in the markov chain monte carlo ( mcmc ) sampler .
let @xmath63 denote the vector of data counts and @xmath64 be the transition kernel .
the mcmc algorithm for obtaining the posterior distribution of @xmath56 is then as follows 1 .
initiate @xmath65 .
2 . obtain proposal @xmath66 from @xmath67 .
3 . accept ( or not ) this proposal with metropolis - hastings probability given by @xmath68 4 .
return to 2 .
5 . repeat until convergence . in this way , after sampler convergence , we obtain an approximate sample from the posterior distribution of @xmath56 and may subsequently compute its approximate @xmath69 credibility region , given the observed data @xmath63 .
in particular , the credibility interval of the parameter @xmath6 given by eq . may be determined .
note that the likelihood formula is valid in a non - markovian setting such as an arbitrary recovery time distribution and could further be extended to the scenario where the transmission rate varies with time since infection .
note that in the latter case the formula for @xmath6 would differ from eq .
due to a form for @xmath70 that differs from eq .
but would remain calculable as @xmath38 .
if additional data were available , such as the recovery time or number of contacts of each individual , it could be explicitly incorporated into the likelihood function through the joint distribution of recovery times and degrees .
to illustrate our method we perform the bayesian posterior estimation of the model parameters for the 2014 ebola outbreak in the drc based on the data described in section [ data ] .
the specific model considered is as given in section [ model ] , where transmission occurs according to an exponential distribution with rate parameter @xmath20 , and the degree distribution is poisson with parameter @xmath40 .
recovery time is assumed to follow a gamma distribution @xmath71 .
we note that , while incubation periods for ebola range from two to 21 days @xcite , our method does not require consideration of latent exposure since it does not depend on infection timing .
we perform estimation via the mcmc scheme given in section [ mcmc ] .
prior distributions were set to be minimally informative gaussian distributions and hyper - parameters were selected based on previous estimates @xcite . the prior distribution for @xmath40 was taken to be @xmath72 and for @xmath21 was taken to be @xmath73 . to improve mixing of the mcmc scheme ,
@xmath20 was estimated on log - scale under an assumed @xmath74 prior distribution .
lastly , the gamma distribution for the infectious period was chosen to have prior mode ( i.e. , @xmath75 ) distributed as @xmath76 and prior standard deviation ( @xmath77 ) as @xmath78 .
a gaussian transition kernel @xmath64 was used .
central @xmath79 credibility intervals were calculated for the parameters of interest as well as for @xmath6 .
results are summarized in figure [ posteriors ] including histograms of posterior samples . the @xmath79 credibility interval for the basic reproductive number @xmath6 is found to be @xmath80 with the @xmath6 posterior mean of @xmath81 .
the latter value numerically agrees closely with the moment - based estimate given in maganga et al .
@xcite and is also consistent with the relatively small total number of confirmed infections observed during the epidemic .
we further note that our interval estimate compares favorably to the @xmath79 @xmath6 confidence interval of @xmath82 reported by maganga et al .
@xcite indicating that for the drc data the fully parametric model produces a more precise ( shorter ) interval .
the infection rate is found to have a posterior mean of @xmath83 and a corresponding 95% credibility interval of @xmath84 .
the contact drop rate is found to have a posterior mean of @xmath85 and corresponding credible interval of @xmath86 .
these quantities were not estimated directly by the authors in @xcite so comparison here is not possible . the estimated mean for the infectious period
is found to be @xmath87 days with corresponding credibility interval of ( @xmath88 days , @xmath89 days ) .
maganga et al . did not explicitly estimate infectious period but instead gave an estimate for time from symptom onset to death with mean 11.3 days @xcite .
based on a historic 1995 outbreak of ebola in the drc , legrand et al .
estimated the time from onset to end of infectiousness for survivors to be 10 days and time from onset to death to be 9.6 days @xcite . while our estimate of infectious period is somewhat longer than these , we note that our interpretation of the quantity is the total length of time that an individual could cause infection .
this could include several days after death in which the body is being prepared for and undergoing funeral rituals .
lastly , the mean number of contacts is estimated to be @xmath90 with corresponding credibility interval of @xmath91 , which is again in close pointwise agreement with the estimate inferred from the number of contacts traced by maganga et al . @xcite .
overall , the point estimate for @xmath6 and infection and recovery rates are seen to be consistent with a small outbreak behavior observed in the drc dataset and to agree well with the numerical values reported earlier . however , based on the same data , our parametric model is also seen to yield more precise interval estimates .
the final outbreak size of simulated branching processes from the posterior parameter distribution is used as a model diagnostic for fit to the empirical data .
conditioning on the number of infections caused by the index case as the number of independent branches , the posterior parameter samples and corresponding post - index case offspring distributions were used to simulate branching process realizations .
the final outbreak sizes were calculated and the distribution is presented in figure [ fig : finalsize ] .
reasonable agreement is observed between this distribution and the empirical outbreak size of 69 cases @xcite .
we presented here a bayesian parameter estimation method for a class of stochastic epidemic models on configuration model graphs .
the method is based on applying the branching process approximation , and is applicable when the number of infections is small in relation to the size of the population under study . in particular , this includes the case when the total outbreak size is small or when we are at the onset of a large outbreak .
the method are flexible , for example allowing for arbitrary degree and recovery distributions , and in principle requires only a knowledge of the distribution of secondary cases , although additional data can be incorporated into the inference procedure , as the likelihood function under branching approximation remains straightforward to evaluate under a wide range of data collection schemes .
we illustrated our approach with the analysis of data from the 2014 drc ebola outbreak which was originally described and analyzed in maganga et al .
@xcite . our method , under only weakly informative prior distributions , is seen to produce a considerably tighter credibility interval for @xmath6 than the moment - based confidence estimate reported in @xcite .
this demonstrates the utility of the branching process approximation for small epidemics in obtaining more precise estimates of @xmath6 , which is essential in assessing potential risk of a large outbreak and in determining the level of control efforts ( e.g. vaccination or quarantine ) needed to mitigate an outbreak .
the final size comparison indicates that the observed data is within the range of model predictions , although the direct comparison of observed and model predicted offspring distributions indicates some disagreement in the observed frequency of zeros and ones .
the small sample size prohibits definitive conclusions but this may indicate the need to incorporate more complex network dynamics ( e.g. distinguishing between multiple types of infectious contacts ) . as the statistical estimation methods appear essential to inform public health interventions
, we hope that our work here will help in establishing a broader inference framework for epidemic parameters , based on the type of data usually collected in the course of an outbreak .
the bayesian approach is particularly attractive in this context , as it naturally incorporates any prior or historical information .
however , the current approach only addresses the inference problem at the epidemic onset and , in particular , is not appropriate when the number of infected individuals comprises a significant portion of the population .
we plan to address estimation for such large outbreaks , possibly also incorporating more complex network dynamics , in our future work .
we thank the mathematical biosciences institute at osu for its assistance in providing us with space and the necessary computational resources .
l. decreusefond , j .- s .
dhersin , p. moyal , v. c. tran et al . ,
large graph limit for an sir process in random network with heterogeneous connectivity , _ the annals of applied probability _ , * 22 * ( 2012 ) , 541575 . m. keeling , correlation equations for endemic diseases : externally imposed and internally generated heterogeneity , _ proceedings of the royal society of london b : biological sciences _ , * 266 * ( 1999 ) , 953960
. w. o. kermack and a. g. mckendrick , a contribution to the mathematical theory of epidemics , in _ proceedings of the royal society of london a : mathematical , physical and engineering sciences _ , vol .
115 , the royal society , 1927 , 700721 .
g. d. maganga , j. kapetshi , n. berthet , b. kebela ilunga , f. kabange , p. mbala kingebeni , v. mondonge , j .-
t. muyembe , e. bertherat , s. briand et al . , ebola virus disease in the democratic republic of congo , _ new england journal of medicine _ ,
* 371 * ( 2014 ) , 20832091 .
|
we present a method for estimating epidemic parameters in network - based stochastic epidemic models when the total number of infections is assumed to be small .
we illustrate the method by reanalyzing the data from the 2014 democratic republic of the congo ( drc ) ebola outbreak described in maganga et al .
( 2014 ) .
mark g. burch karly a. jacobsen joseph h. tien grzegorz a. rempaa
|
the core of the standard model @xcite of particle physics is easily displayed in a single figure , here figure 1 .
there are gauge groups @xmath0 for the strong , weak , and electromagnetic interactions .
the gauge bosons associated with these groups are minimally coupled to quarks and leptons according to the scheme depicted in the figure .
the non - abelian gauge bosons within each of the @xmath1 and @xmath2 factors also couple , in a canonical minimal form , to one another . the @xmath3 group is spontaneously broken to the @xmath4 of electromagnetism .
this breaking is parameterized in a simple and ( so far ) phenomenologically adequate way by including an @xmath0 @xmath5 scalar ` higgs ' field which condenses , that is has a non - vanishing expectation value in the ground state .
condensation occurs at weak coupling if the bare ( mass)@xmath6 associated with the higgs doublet is negative .
the fermions fall into five separate multiplets under @xmath7 , as depicted in the figure .
the color @xmath1 group acts horizontally ; the weak @xmath2 vertically , and the hypercharges ( equal to the average electric charge ) are as indicated .
note that left- and right - handed fermions of a single type generally transform differently .
this reflects parity violation .
it also implies that fermion masses , which of course connect the left- and right - handed components , only arise upon spontaneous @xmath3 breaking .
only one fermion family has been depicted in figure 1 ; of course in reality there are three repetitions of this scheme . also not represented are all the complications associated with the masses and cabibbo - like mixing angles among the fermions .
these masses and mixing angles are naturally accommodated as parameters within the standard model , but i think it is fair to say that they are not much related to its core ideas more on this below . with all these implicit understandings and discrete choices ,
the core of the standard model is specified by three numbers the universal strengths of the strong , weak , and electromagnetic interactions . the electromagnetic sector , qed , has been established as an extraordinarily accurate and fruitful theory for several decades now .
let me now briefly describe the current status of the remainder of the standard model .
-.50 in 0.75 in some recent stringent tests of the electroweak sector of the standard model are summarized in figure 2 . in general
each entry represents a very different experimental arrangement , and is meant to test a different fundamental aspect of the theory , as described in the caption .
there is precisely one parameter ( the mixing angle ) available within the theory , to describe all these measurements .
as you can see , the comparisons are generally at the level of a per cent accuracy or so . over all
, the agreement appears remarkably good , especially to anyone familiar with the history of weak interactions .
it is so good , that if the small but significant residual discrepancies in the branching ratios @xmath8 and @xmath9 survive further scrutiny , one will certainly prefer to interpret them as an indication of some peculiar additional physics , rather than as a breakdown of the fundamental principles behind the @xmath3 theory . indeed , there are some reasonably attractive concrete possibilities along these lines @xcite .
-.7 in 0.75 in some recent stringent tests of the strong sector of the standard model are summarized in figure 3 @xcite .
again a wide variety of very different measurements are represented , as indicated in the caption .
a central feature of the theory ( qcd ) is that the value of the coupling , as measured in different physical processes , will depend in a calculable way upon the characteristic energy scale of the process .
the coupling was predicted and evidently is now convincingly measured to decrease as the inverse logarithm of the energy scale : asymptotic freedom .
again , all the experimental results must be fit with just one parameter the coupling at any single scale , usually chosen as @xmath10 .
as you can see , the agreement between theory and experiment is remarkably good .
the accuracy of the comparisons is at the few per cent level .
-.65 in = 3.5 in let me emphasize that these figures barely begin to do justice to the evidence for the standard model .
several of the results in them summarize quite a large number of independent measurements , any one of which might have falsified the theory .
for example , the single point labeled ` dis ' in figure 3 describes literally hundreds of measurements in deep inelastic scattering with different projectiles and targets and at various energies and angles , which must if the theory is correct all fit into a tightly constrained pattern .
the central theoretical principles of the standard model have been in place for nearly twenty - five years . over this interval
the quality of the relevant experimental data has become incomparably better , yet no essential modifications of these venerable principles has been required .
let us now praise the standard model : @xmath11 the standard model is here to stay , and describes _ a lot_. since there is quite direct evidence for each of its fundamental ingredients ( _ i.e_. its interaction vertices ) , and since the standard model provides an extremely economical packaging of these ingredients , i think it is a safe conjecture that it will be used , for the foreseeable future , as the working description of the phenomena within its domain .
and this domain includes a very wide range of phenomena indeed not only what dirac called `` all of chemistry and most of physics '' , but also the original problems of radioactivity and nuclear interactions which inspired the birth of particle physics in the 1930s , and much that was unanticipated .
@xmath11 the standard model is a _ principled _ theory .
indeed , its structure embodies a few basic principles : special relativity , locality , and quantum mechanics , which lead one to quantum field theories , local symmetry ( and , for the electro weaksector , its spontaneous breakdown ) , and renormalizability ( minimal coupling ) .
the last of these principles , renormalizability , may appear rather technical and perhaps less compelling than the others ; we shall shortly have occasion to re - examine it in a larger perspective . in any case , the fact that the standard model is principled in this sense is profoundly significant : it means that its predictions are precise and unambiguous , and generally can not be modified ` a little bit ' except in very limited , specific ways .
this feature makes the experimental success especially meaningful , since it becomes hard to imagine that the theory could be approximately right without in some sense being exactly right .
@xmath11 the standard model _ can be extrapolated_. specifically because of the asymptotic freedom property , one can extrapolate using the standard model from the observed domain of experience to much larger energies and shorter distances .
indeed , the theory becomes simpler the fundamental interactions are all effectively weak in these limits .
the whole field of very early universe cosmology depends on this fact , as do the impressive semi - quantitative indications for unification and supersymmetry i shall be emphasizing momentarily .
just because it is so comprehensive and successful , we should judge the standard model by demanding criteria .
it is clearly an important part of the truth ; the question becomes : how big a part ?
critical scrutiny reveals several important shortcomings of the standard model : @xmath11 the standard model contains scattered multiplets with peculiar hypercharge assignments . while little doubt can remain that the standard model is essentially correct , a glance at figure 1 is enough to reveal that it is not a complete or final theory .
the fermions fall apart into five lopsided pieces with peculiar hypercharge assignments ; this pattern needs to be explained .
also the separate gauge theories , which are quite mathematically and conceptually similar , are fairly begging to be unified .
@xmath11 the standard model supports the possibility of strong p and t violation @xcite .
there is a near - perfect match between the necessary ` accidental ' symmetries of the standard model , dictated by its basic principles as enumerated above , and the observed symmetries of the world .
the glaring exception is that there is an allowed gauge invariant , renormalizable interaction which , if present , would induce significant violation of the discrete symmetries @xmath12 and @xmath13 in the strong interaction .
this is the notorious @xmath14 term .
@xmath14 is an angle which _ a priori _ one might expect to be of order unity , but in fact is constrained by experimental limits on the neutron electric dipole moment to be @xmath15 . this problem can be addressed by postulating a sort of quasi - symmetry ( peccei - quinn @xcite symmetry ) , which roughly speaking corresponds to promoting @xmath14 to a dynamical variable a quantum field .
the quanta of this field @xcite , _ axions _ , provide an interesting dark matter candidate @xcite .
other possibilities for explaining the absence of strong @xmath12 and @xmath13 violation have been proposed , but they require towers of hypotheses which seem ( to me ) quite fragile .
i would like to emphasize , since it came up at the conference , that in no way should the absence of strong @xmath12 and @xmath13 violation be taken as evidence against qcd itself . for practical purposes
, one can take the @xmath14 parameter to be fixed phenomenologically .
one finds it to be very small and is done with it !
@xmath11 the standard model does not address family problems .
there are several distinct ` family problems ' , ranging from the extremely qualitative ( digital ) to the semi - qualitative ( some very small numbers and distinctive patterns ) to the straightforward but most challenging goal of doing full justice to experience by computing ( analog ) experimental numbers with controlled , small fractional errors : why are there three repeat families ?
rabi s famous question regarding the muon
`` who ordered _ that _ ? '' still has no convincing answer .
how does one explain the very small electron mass ?
the dimensionless coupling associated with the electron mass , that is its strength of yukawa coupling to the higgs field , is about @xmath16 .
it is almost as small as the limits on the @xmath14 angle ( suggesting , perhaps , that strong p and t violation is just around the corner ? ) .
this question can of course be generalized all the fermion masses , with one exception , are sufficiently small to beg qualitative explanation .
the exception , of course , is the @xmath17 quark .
it is a fascinating possibility that roughly the observed value of @xmath17 quark mass at low energies might result after running from a wide range of fundamental couplings at a high scale @xcite .
if so , one would have a satisfactory qualitative explanation of the value of this parameter .
why do the weak currents couple approximately in the order of masses ?
that is , light with light , heavy with heavy , and intermediate with intermediate .
why are the mixings what they are small , but not miniscule ? the same , for the cp violating phase in the weak currents ( parameterized invariantly by jarlskog s @xmath18 ) @xcite ? and , by the way , are we sure that @xmath19 ? and so on ... @xmath11 the standard model does not allow non - vanishing neutrino masses .
this is the only entry on my list that has a primarily _ experimental _ motivation . at present
there are three quite different experimental hints for non - vanishing neutrino masses : the solar neutrino problem @xcite , the atmospheric neutrino problem @xcite , and the los alamos oscillation experiment @xcite .
the standard model in its conventional form does not allow non - zero neutrino masses .
however i would like to mention that only a very minimal extension of the theory is necessary to accommodate such masses .
one can add a complete @xmath20 singlet fermion @xmath21 to the model .
@xmath21 can be given , consistent with all symmetries and with the requirement of renormalizability , a majorana mass @xmath22 .
note especially that such a mass does not violate @xmath3 .
likewise , @xmath21 can have a yukawa coupling to the to the ordinary lepton doublets through the higgs field .
then condensation of the higgs field activates the `` see - saw '' mechanism @xcite to give a small mass for the observed neutrinos ; with @xmath23 a range of experimentally and perhaps cosmologically interesting neutrino masses can be accommodated .
@xmath11 gravity is not included in the standard model .
this really represents ( at least ) two logically separate problems .
first there is the ultraviolet problem , the notorious non - renormalizability of quantum gravity .
this provides an appropriate context , in which to introduce the modern perspective toward the whole concept of renormalizability .
suppose that one were naive , and simply added the einstein lagrangian for general relativity to the standard model , of course coupling the matter fields appropriately ( minimally ) . following feynman and
many others one could derive , formally , the perturbation series for any physical process
. one would find , however , that , the integrals over closed loops generally diverges at the high - energy ( ultraviolet ) end .
indeed the graviton coupling has , in units where @xmath24 , dimensions of @xmath25 so that in expanding about flat space one must take @xmath26 in order to obtain a properly normalized quadratic kinetic term for @xmath27 . ] . here
@xmath28 is a measure of the stiffness of space - time .
thus higher and higher order terms will , on dimensional grounds , introduce higher and higher factors of the ultraviolet cutoff to compensate .
however if we fix ( by notional experiments ) the couplings at a given scale @xmath29 and calculate corrections by only including energy - momenta between the scale @xmath30 of interest and @xmath31 , the successive terms in perturbation theory will be accompanied by positive powers of @xmath32 and will be small .
thus we can , for example , consistently set all non - minimal couplings to zero at any chosen energy - momentum scale well below the planck scale .
they will then be negligible for all practically accessible scales . for different choices of the scale they will be different , but since they are negligible this hardly matters .
this procedure is , in practice , the one we always adopt and the standard model peacefully coexists with gravity , so long as we refuse to consider @xmath33 .
however from this perspective a second problem looms larger than ever .
the energy ( and negative pressure ) density of matter - free space , the notorious cosmological term , occurs as the coefficient @xmath34 of the identity term in the action : @xmath35 .
it has dimensions of ( mass)@xmath36 , and on phenomenological grounds we must suppose @xmath37 .
the question is : where does such a tiny scale come from ?
what is so special about the present state of the universe , that the value of the effective @xmath34 for it , which one might naively expect to reflect contributions from much higher scales , is so effectively zeroed ?
there may also be problems with forming a fully consistent quantum theory even of low - energy processes involving black holes @xcite .
finally i will mention a question that i think has a rather different status from the foregoing , although many of my colleagues would put it on the same list , and maybe near the top : the standard model begs the question `` _ why _ does @xmath3 symmetry break ? '' .
to me , this is an example of the sort of metaphysical question that could easily fail to have a meaningful answer .
there is absolutely nothing wrong , logically , with the classic implementation of the higgs sector as described earlier .
nevertheless , one might well hunger for a wider context in which to view the existence of the higgs doublet and its negative ( mass)@xmath6 or a suitable alternative .
each of the deficiencies of the standard model mentioned in the previous section has provoked an enormous literature , literally hundreds or thousands of papers .
obviously i can not begin to do justice to all this work . here
i shall concentrate on the first question , that of deciphering the message of the scattered multiplets and peculiar hypercharges . among our questions
this one has so far inspired the most concrete and compelling response a story whose implications range far beyond the question which inspired it .
given that the strong interactions are governed by transformations among three color charges say rwb for red , white , and blue while the weak interactions are governed by transformations between two others say gp for green and purple what could be more natural than to embed both theories into a larger theory of transformations among all five colors ? this idea has the additional attraction that an extra u(1 ) symmetry commuting with the strong su(3 ) and weak su(2 ) symmetries automatically appears , which we can attempt to identify with the remaining gauge symmetry of the standard model , that is hypercharge . for while in the separate su(3 ) and
su(2 ) theories we must throw out the two gauge bosons which couple respectively to the color combinations r+w+b and g+p , in the su(5 ) theory we only project out r+w+b+g+p , while the orthogonal combination ( r+w+b)-@xmath38(g+p ) remains .
georgi and glashow @xcite originated this line of thought , and showed how it could be used to bring some order to the quark and lepton sector , and in particular to supply a satisfying explanation of the weird hypercharge assignments in the standard model .
as shown in figure 4 , the five scattered su(3)@xmath39su(2)@xmath39u(1 ) multiplets get organized into just two representations of @xmath40 .
it is an extremely non - trivial fact that the known fermions fit so smoothly into @xmath40 multiplets .
@xmath41 : 2 different color labels ( antisymmetric tensor ) @xmath42 @xmath43 : 1 anticolor label @xmath44 # 1 in all the most promising unification schemes , what we ordinarily think of as matter and anti - matter appear on a common footing . since the fundamental gauge transformations do not alter the chirality of fermions , in order to represent the most general transformation possibilities one should use fields of one chirality , say left , to represent the fermion degrees of freedom . to do this , for a given fermion , may require a charge conjugation operation .
also , of course , once we contemplate changing strong into weak colors it will be difficult to prevent quarks and leptons from appearing together in the same multiplets .
generically , then , one expects that in unified theories it will not be possible to make a global distinction between matter and anti - matter and that both baryon number @xmath45 and lepton number @xmath46 will be violated , as they definitely are in @xmath40 and its extensions . as shown in figure 4
, there is one group of ten left - handed fermions that have all possible combinations of one unit of each of two different colors , and another group of five left - handed fermions that each carry just one negative unit of some color .
( these are the ten - dimensional antisymmetric tensor and the complex conjugate of the five - dimensional vector representation , commonly referred to as the `` five - bar '' . )
what is important for you to take away from this discussion is not so much the precise details of the scheme , but the idea that _ the structure of the standard model , with the particle assignments gleaned from decades of experimental effort and theoretical interpretation , is perfectly reproduced by a simple abstract set of rules for manipulating symmetrical symbols_. thus for example the object rb in this figure has just the strong , electromagnetic , and weak interactions we expect of the complex conjugate of the right - handed up - quark , without our having to instruct the theory further .
if you ve never done it i heartily recommend to you the simple exercise of working out the hypercharges of the objects in figure 4 and checking against what you need in the standard model after doing it , you ll find it s impossible ever to look at the standard model in quite the same way again . although it would be inappropriate to elaborate the necessary group theory here , i ll mention that there is a beautiful extension of @xmath40 to the slightly larger group @xmath47 , which permits one to unite all the fermions of a family into a single multiplet @xcite .
in fact the relevant representation for the fermions is a 16-dimensional spinor representation .
some of its features are depicted in figure 5 .
the 16th member of a family in @xmath47 , beyond the 15 familiar degrees of freedom with a standard model family , has the quantum numbers of the right - handed neutrino @xmath21 as discussed above .
this emphasizes again how easy and natural is the extension of the standard model to include neutrino masses using the see - saw mechanism . 0.75 in -.15 in @xmath48 @xmath49 @xmath50 @xmath51
we have seen that simple unification schemes are successful at the level of _ classification _ ; but new questions arise when we consider the dynamics which underlies them .
part of the power of gauge symmetry is that it fully dictates the interactions of the gauge bosons , once an overall coupling constant is specified .
thus if su(5 ) or some higher symmetry were exact , then the fundamental strengths of the different color - changing interactions would have to be equal , as would the ( properly normalized ) hypercharge coupling strength .
in reality the coupling strengths of the gauge bosons in su(3)@xmath39su(2)@xmath39u(1 ) are observed not to be equal , but rather to follow the pattern @xmath52 .
fortunately , experience with qcd emphasizes that couplings `` run '' .
the physical mechanism of this effect is that in quantum field theory the vacuum must be regarded as a polarizable medium , since virtual particle - anti - particle pairs can screen charge .
thus one might expect that effective charges measured at shorter distances , or equivalently at larger energy - momentum or mass scales , could be different from what they appear at longer distances .
if one had only screening then the effective couplings would grow at shorter distances , as one penetrated deeper insider the screening cloud .
however it is a famous fact @xcite that due to paramagnetic spin - spin attraction of like charge vector gluons @xcite , these particles tend to _ antiscreen _ color charge , thus giving rise to the opposite effect
asymptotic freedom that the effective coupling tends to shrink at short distances .
this effect is the basis of all perturbative qcd phenomenology , which is a vast and vastly successful enterprise , as we saw in figure 3 . for our present purpose of understanding the disparity of the observed couplings ,
it is just what the doctor ordered .
as was first pointed out by georgi , quinn , and weinberg @xcite , if a gauge symmetry such as su(5 ) is spontaneously broken at some very short distance then we should not expect that the effective couplings probed at much larger distances , such as are actually measured at practical accelerators , will be equal .
rather they will all have have been affected to a greater or lesser extent by vacuum screening and anti - screening , starting from a common value at the unification scale but then diverging from one another at accessible accelerator scales .
the pattern @xmath52 is just what one should expect , since the antiscreening or asymptotic freedom effect is more pronounced for larger gauge groups , which have more types of virtual gluons .
the marvelous thing is that the running of the couplings gives us a truly quantitative handle on the ideas of unification , for the following reason . to fix the relevant aspects of unification ,
one basically needs only to fix two parameters : the scale at which the couplings unite , which is essentially the scale at which the unified symmetry breaks ; and their value when then unite . given these ,
one calculates three outputs : the three _ a priori _ independent couplings for the gauge groups su(3)@xmath39su(2)@xmath39u(1 ) of the standard model .
thus the framework is eminently falsifiable .
the miraculous thing is , how close it comes to working ( figure 6 ) .
= 3.5 in -0.75 in 0.50 in the unification of couplings occurs at a very large mass scale , @xmath53 . in the simplest version
, this is the magnitude of the scalar field vacuum expectation value that spontaneously breaks su(5 ) down to the standard model symmetry su(3)@xmath39su(2)@xmath39u(1 ) , and is analogous to the scale @xmath54 for electroweak symmetry breaking .
the largeness of this large scale mass scale is important in several ways .
@xmath55 it explains why the exchange of gauge bosons that are in su(5 ) but not in su(3)@xmath39su(2)@xmath39u(1 ) , which re - shuffles strong into weak colors and generically violates the conservation of baryon number , does not lead to a catastrophically quick decay of nucleons .
the rate of decay goes as the inverse fourth power of the mass of the exchanged gauge particle , so the baryon - number violating processes are predicted to be far slower than ordinary weak processes , as they had better be .
@xmath55 @xmath56 is significantly smaller than the planck scale @xmath57 at which exchange of gravitons competes quantitatively with the other interactions , but not ridiculously so .
this indicates that while the unification of couplings calculation itself is probably safe from gravitational corrections , the unavoidable logical next step in unification must be to bring gravity into the mix .
@xmath55 finally one must ask how the tiny ratio of symmetry - breaking mass scales @xmath58 required arises dynamically , and whether it is stable .
this is the so - called gauge hierarchy problem , which we shall discuss in a more concrete form a little later . the success of the gqw calculation in explaining the observed hierarchy @xmath52 of couplings and the approximate stability of the proton is quite striking .
in performing it , we assumed that the known and confidently expected particles of the standard model exhaust the spectrum up to the unification scale , and that the rules of quantum field theory could be extrapolated without alteration up to this mass scale thirteen orders of magnitude beyond the domain they were designed to describe .
it is a triumph for minimalism , both existential and conceptual .
however , on further examination it is not quite good enough .
accurate modern measurements of the couplings show a small but definite discrepancy between the couplings , as appears in figure 6 . and heroic dedicated experiments to search for proton decay did not find it @xcite ; they currently exclude the minimal su(5 ) prediction @xmath59 by about two orders of magnitude . given the scope of the extrapolation involved ,
perhaps we should not have hoped for more .
there are several perfectly plausible bits of physics that could upset the calculation , such as the existence of particles with masses much higher than the electroweak but much smaller than the unification scale . as virtual particles
these would affect the running of the couplings , and yet one certainly can not exclude their existence on direct experimental grounds .
if we just add particles in some haphazard way things will only get worse : minimal su(5 ) nearly works , so the generic perturbation from it will be deleterious .
this is a major difficulty for so - called technicolor models , which postulate many new particles in complicated patterns .
even if some _ ad hoc _ prescription could be made to work , that would be a disappointing outcome from what appeared to be one of our most precious , elegantly straightforward clues regarding physics well beyond the standard model .
fortunately , there is a theoretical idea which is attractive in many other ways , and seems to point a way out from this impasse .
that is the idea of supersymmetry @xcite .
supersymmetry is a symmetry that extends the poincare symmetry of special relativity ( there is also a general relativistic version ) . in a supersymmetric theory
one has not only transformations among particle states with different energy - momentum but also between particle states of different _
spin_. thus spin 0 particles can be put in multiplets together with spin @xmath60 particles , or spin @xmath60 with spin 1 , and so forth .
supersymmetry is certainly not a symmetry in nature : for example , there is certainly no bosonic particle with the mass and charge of the electron . more generally if one defines the @xmath61-parity quantum number @xmath62 which should be accurate to the extent that baryon and lepton number
are conserved , then one finds that all currently known particles are @xmath61 even whereas their supersymmetric partners would be @xmath61 odd .
nevertheless there are many reasons to be interested in supersymmetry , and especially in the hypothesis that supersymmetry is effectively broken at a relatively low scale , say @xmath63 1 tev . anticipating this for the moment ,
let us consider the consequences for running of the couplings .
the effect of low - energy supersymmetry on the running of the couplings was first considered long ago @xcite , well before the discrepancy described above was evident experimentally .
one might have feared that such a huge expansion of the theory , which essentially doubles the spectrum , would utterly destroy the approximate success of the minimal su(5 ) calculation .
this is not true , however .
to a first approximation since supersymmetry is a space - time rather than an internal symmetry it does not affect the group - theoretic structure of the calculation .
thus to a first approximation the absolute rate at which the couplings run with momentum is affected , but not the relative rates .
the main effect is that the supersymmetric partners of the color gluons , the gluinos , weaken the asymptotic freedom of the strong interaction .
thus they tend to make its effective coupling decrease and approach the others more slowly .
thus their merger requires a longer lever arm , and the scale at which the couplings meet increases by an order of magnitude or so , to about 10@xmath64 gev .
also the common value of the effective couplings at unification is slightly larger than in conventional unification ( @xmath65 _ versus _
@xmath66 ) .
this increase in unification scale significantly reduces the predicted rate for proton decay through exchange of the dangerous color - changing gauge bosons , so that it no longer conflicts with existing experimental limits . upon more careful examination
there is another effect of low - energy supersymmetry on the running of the couplings , which although quantitatively small has become of prime interest .
there is an important exception to the general rule that adding supersymmetric partners does not immediately ( at the one loop level ) affect the relative rates at which the couplings run .
this rule works for particles that come in complete su(5 ) multiplets , such as the quarks and leptons ( which , since they do nt upset the full su(5 ) symmetry , have basically no effect ) or for the supersymmetric partners of the gauge bosons , because they just renormalize the existing , dominant effect of the gauge bosons themselves .
however there is one peculiar additional contribution , from the supersymmetric partner of the higgs doublet .
it affects only the weak su(2 ) and hypercharge u(1 ) couplings .
( on phenomenological grounds the su(5 ) color triplet partner of the higgs doublet must be extremely massive , so its virtual exchange is not important below the unification scale .
_ why _ that should be so , is another aspect of the hierarchy problem . )
moreover , for slightly technical reasons even in the minimal supersymmetric model it is necessary to have two different higgs doublets with opposite hypercharges .
the net effect of doubling the number of higgs fields and including their supersymmetric partners is a sixfold enhancement of the asymmetric higgs field contribution to the running of weak and hypercharge couplings .
this causes a small , accurately calculable change in the calculation . from figure 7
you see that it is a most welcome one . indeed , in the minimal implementation of supersymmetric unification
, it puts the running of couplings calculation right back on the money @xcite .
= 3.5 in -.75 in 0.65 in since the running of the couplings with scale is logarithmic the unification of couplings calculation is not terribly sensitive to the exact scale at which supersymmetry is broken , say between 100 gev and 10 tev .
( to avoid confusion later let me specify that at this point what i mean by the scale at which supersymmetry is broken is the characteristic mass splitting between supersymmetric partners in a minimal extension of the standard model .
later we shall use the term in a different sense , as referring to the largest splitting between supersymmetric partners in the entire world - spectrum ; this could be much larger , and indeed in popular models it almost invariably is .
this ambiguous terminology is endemic in the literature ; fortunately , the meaning is usually clear from the context . )
there have been attempts to push the calculation further , in order to address this question of the supersymmetry breaking scale , but they are controversial . for example , comparable uncertainties arise from the splittings among the very large number of particles with masses of order the unification scale , whose theory is poorly developed and unreliable .
superstring theory suggests @xcite many possible ways in which the simple calculation described here might go wrong ; if we take the favorable result of this calculation at face value , we must conclude that none of them happen . in any case
, if we are not too greedy the main points still shine through : @xmath11 if supersymmetry is to fulfill its destiny of elucidating the hierarchy problem in a straightforward way , then the supersymmetric partners of the known particles can not be much heavier than the su(2)@xmath39u(1 ) electroweak breaking scale , they should not be beyond the expected reach of lhc .
@xmath11 if we assume this to be the case then the meeting of the couplings takes place in the simplest minimal models of unification , to adequate accuracy , without further assumption .
this is a most remarkable and non - trivial fact .
the preceding result , taken at face value , has extremely profound implications : @xmath11 quantum field theory , and specifically its characteristic vacuum polarization effects leading to asymptotic freedom and running of the couplings , continue to work quantitatively up to energy scales many orders of magnitude beyond where they were discovered and established .
i would like to emphasize also some negative implications of this : there are things that might have , but do _ not _ , happen .
it might have happened that the known particles are some complicated composites of more elementary objects , or that many additional strong couplings appeared at higher energies ( technicolor ) , or that additional dimensions became dynamically active , or that particle physics simply dissolved into some amorphous mess .
unless figure 7 is a cruel joke on the part of mother nature , none of this happens , or at least the complications are in a strong , precise sense walled off from the standard model and the dynamical evolution of its couplings .
@xmath11 supersymmetry , in its virtual form , has already been discovered .
this provides , in my opinion , a very good specific reason to be hopeful about the future of experimental particle physics at the high - energy frontier .
@xmath11 you will notice that we have made progress in uniting the gauge bosons with each other , and the various quarks and leptons with each other , but not the gauge bosons with the quarks and leptons .
it takes supersymmetry perhaps spontaneously broken to make this feasible .
@xmath11 supersymmetry was invented in the context of string theory , and seems to be necessary for constructing consistent string theories containing gravity ( critical string theories ) that are at all realistic .
@xmath11 most important for our purposes , supersymmetry can help us to understand the vast disparity between weak and unified symmetry breaking scales mentioned above .
this disparity is known as the gauge hierarchy problem .
it actually raises several distinct problems , including the following . in calculating radiative corrections to
the ( mass)@xmath6 of the higgs particle from diagrams of the type shown in figure 8 one finds an infinite , and also large , contribution . by this
i mean that the divergence is quadratic in the ultraviolet cutoff .
no ordinary symmetry will make its coefficient vanish .
if we imagine that the unification scale provides the cutoff , we find that the radiative correction to the ( mass)@xmath6 is much larger than the final value we want .
( if the higgs field were composite , with a soft form factor , this problem might be ameliorated . following
that road leads to technicolor , which as mentioned before seems to lead us far away from our best source of inspiration . ) as a formal matter one can simply cancel the radiative correction against a large bare contribution of the opposite sign , but in the absence of some deeper motivating principle this seems to be a horribly ugly procedure .
now in a supersymmetric theory for any set of virtual particles circulating in the loop there will also be another graph with their supersymmetric partners circulating .
if the partners were accurately degenerate , the contributions would cancel .
otherwise , the threatened quadratic divergence will be cut off only at virtual momenta such that the difference in ( mass)@xmath6 between the virtual particle and its supersymmetric partner is relatively negligible .
thus we will be assured adequate cancelation if and only if supersymmetric partners are not too far split in mass in the present context , if the splitting is not much greater than the weak scale .
this is ( a crude version of ) the most important _ quantitative _ argument which suggests the relevance of `` low - energy '' supersymmetry .
-.75 in = 3.5 in @xmath11 supersymmetric field theories have many special features , which make them especially interesting , and perhaps promising , phenomenologically .
i can not be very specific about this here , both because there are as yet no canonical models and because the subject is excessively technical , but let me just mention some appropriate concepts : radiative @xmath3 breaking associated with the heavy top quark ; doublet - triplet splitting mechanisms ; approximate flat directions for generating large mass hierarchies .
supersymmetric models also have additional mechanisms for neutral flavor - changing processes and cp violation , which in generic models are dangerously large , but in appropriate models can be suppressed down to a level which is interesting but not _ too _ interesting experimentally .
theoretical control over the many new parameters that arise even in minimal supersymmetric extensions of the standard model is not adequate to guide one through a bewildering number of choices that substantially affect the predictions for observable phenomena .
recently a class of models that potentially enjoy much greater predictive power has been identified @xcite .
the leading idea of these models is that the fundamental breaking of supersymmetry occurs in a `` hidden sector '' and is characterized by a scale @xmath67 .
this breaking is supposed to be conveyed to our observable sector , including the minimal supersymmetric standard model , by a specific , highly symmetric interaction term ) having the quantum numbers @xmath68 under @xmath40 .
these fields are assumed to acquire supersymmetry breaking masses through their couplings to a standard model singlet field @xmath69 which develops vacuum expectation values along its scalar as well as its @xmath70components .
explicit models where this occurs dynamically have been constructed @xcite .
typically these models contain the superpotential couplings @xmath71 which generates a non supersymmetric spectrum for the @xmath72 fields through @xmath73 .
the @xmath72 fields , through their standard model gauge interactions , induce susy breaking masses for the r - odd scalars @xmath74 ) and gauginos ( @xmath75 . ] .
this is to be contrasted with the situation in a class of models , commonly referred to as supergravity models , that has been much more popular . in supergravity models
the fundamental symmetry breaking scale is @xmath76 , and the hidden sector communicates with the observable sector by gravitational - strength interactions .
a decisive phenomenological difference between these classes of models arises because of the different role played by the gravitino @xmath77 , which is essentially the nambu - goldstone fermion of spontaneous supersymmetry breaking .
its mass is generically given as @xmath78 , whereas its couplings to ordinary matter scale as @xmath79 @xcite .
in supergravity models the gravitino is massive and very weakly coupled , and although it could be significant cosmologically , it plays no very direct role in ordinary phenomenology . in the messenger models with a ( relatively ) much lower supersymmetry breaking scale the gravitino is exceedingly light , and although its couplings are very feeble they can be such as to cause the next lightest , photino - like particle @xmath80 to decay via @xmath81 within a distance that might be difficult to resolve experimentally @xcite . generically , then , in these models one will find that pair - production of supersymmetric ( r - odd ) particles will lead eventually , perhaps after a cascade of decays , to final states containing @xmath82 and missing energy @xcite .
the recent observation @xcite of a dramatic event with a final state containing hard @xmath83 and missing transverse energy by the cdf collaboration therefore lends special interest to these models .
one can work out the consequences of a minimal implementation of a low - energy messenger model for supersymmetry breaking @xcite .
outstanding features of this model include : generic production of final states including @xmath82 and missing energy ; clean separation of scales in the r - parity odd spectrum , with the squarks and gluinos much heavier than the electroweak gauginos and sleptons ; and very small mixing of the @xmath3 neutralinos with higgsinos and with each other .
many specific features are predicted in terms of very few parameters , so that the model could be rigorously tested , or eliminated , with quite a small amount of relevant data .
if the cdf @xmath83 event represents a first indication , such data is not far beyond current reach . of course in insisting upon an absolutely minimal model we may be over reaching .
some apparent difficulties with the minimal model are removed upon careful treatment of coupling renormalization and proper inclusion of d - terms ; nevertheless , one must remain open to such possibilities as different overall mass scales for the sleptons as against the gauginos , or additional contributions to the parameters of the higgs sector .
finally let me briefly mention another possibility for physics capable of generating the cdf @xmath83 event , which if nothing else might serve as a useful foil .
if there were heavy pion - like particles whose decay modes were dominantly @xmath84 and @xmath85 , their production and decay could induce many of the same final states as we have analyzed above , including the cdf event .
the missing energy in this event would be carried off by neutrinos ; but in other cases one might have @xmath86 without significant missing energy , so that this alternative will be readily distinguishable @xcite .
these considerations are meant mainly to illustrate how very important it is , in our present state of ignorance , to obtain some guidance from experiment .
a few informative events could re - shape our ideas on particle physics and cosmology in fundamental ways .
for example , if something like the minimal messenger model of supersymmetry breaking were established , we would learn that the lightest r - odd particle is the gravitino , and that it is extremely light . in that case
planned experiments to detect a cosmological background of weakly interacting massive particles ( wimp ) would be doomed to frustration .
now i would like to draw out from this discussion of particle physics some morals for cosmology .
@xmath11 it makes good sense to extrapolate toward the _ very _ early universe . as figure 3 shows us in hard data , the strong coupling runs toward a small effective value at high energy .
figure 6 , and especially figure 7 , emphasize why the seductive assumption that in crucial respects particle physics remains simple and weakly coupled up to extraordinarily high energies is very hard to resist , because it leads to a strikingly successful account of the unification of couplings .
thus fundamental particle physics becomes profoundly simpler ( though superficially more complex ) toward the earliest moments of the big bang .
the scale for the running of inverse couplings , once they are large , is logarithmic .
this feature naturally connects mass scales identifiable in particle physics experiments with exponentially ( in the inverse couplings ) larger scales .
numerically , as we have seen , one is lead in this way close to the planck scale .
thus there is direct , though of course extremely limited , evidence that quantum field theory at weak coupling governs the interactions of the known particles up to energies ( temperatures , densities ) nearly as high as one could reasonably hope .
@xmath11 cosmic phase transitions happened .
since qcd and asymptotic freedom are firmly established , we can say with complete confidence that the effective low - energy degrees of freedom in the strong interaction hadrons with confined color and spontaneously broken chiral symmetry are quite different from the fundamental colored quark and gluon degrees of freedom which manifest themselves at high temperature .
the transition between them must be accompanied by a rapid crossover and perhaps by a phase transition .
indeed , the existence of a phase transition can be rigorously established in models closely related to real - world qcd , such as the variant of qcd where the @xmath87 or the pure - glue theory .
the nature of the transition , perhaps surprisingly , seems to depend sensitively upon the spectrum of light quarks .
the behavior of qcd at high temperatures is in principle , and to some extent in practice , calculable . the qcd crossover or transition that occurred during the big bang is even in some rough sense _
reproducible _ , and will be approximated in future relativistic heavy ion collisions .
similarly , since the electroweak @xmath88 breaking pattern is firmly established , there is an excellent chance that another phase transition , associated with @xmath3 restoration at high temperature , occurs .
it is not guaranteed that there is a strict phase transition , since there is no gauge - invariant order parameter for the higgs phase , though at weak coupling one certainly expects a sharp transition .
one of the most interesting projects for future accelerators , which has have important implications for cosmology , will be to map out the relevant parameters so that we can characterize this crossover or transition .
these ` established ' examples encourage one to speculate about the possibility of phase transitions associated with unification , or perhaps occurring in some hidden sector .
cosmological phase transitions have many possible consequences for the history of the universe and for observational cosmology , including : _ defects _ of all sorts , including textures , strings , and monopoles that could persist even to the present day .
truly stable domain walls must be avoided or inflated away , but appropriately long - lived ones could be important in the prior evolution of the universe .
_ inflation _ , if one gets trapped in a metastable condition by a barrier or by weakness of the relevant coupling that drives the transition .
_ gravity waves _ , if the scale of the transition is high or if the transition is sufficiently violent .
_ baryogenesis _ , under the same qualitative conditions . of course
, baryogenesis does not require a phase transition .
it will be fascinating to discover which , if any , of these possibilities is realized by the electroweak transition .
there is also a great opportunity , if one can establish compelling , sufficiently detailed models for unification , or a hidden sector , or ... , to examine their cosmological implications . in particular , as was repeatedly emphasized in these dialogues , we need better models of , or alternatives to , inflation .
@xmath11 we have specific , credible dark matter candidates , notably the lightest r - odd particle ( lsp ) and the axion .
@xmath11 we have rich material for fantasies .
i will leave to each of you the joy of choosing his or her own favorite fantasies from among the many we have heard , or of inventing new ones .
i would however like briefly to conclude by alluding a circle of ideas which though it can be made to sound quite fantastic i think is actually deeply implicit in much current thinking about particle physics and cosmology .
the axion is perhaps the most well - motivated and studied exemplar of a family of related fields including familons , dilatons , and moduli fields that in one way or another embody the idea that what we ordinarily consider ` constants ' might not be fundamental parameters fixed in the very formulation of the laws of nature , but rather can be considered usefully as dynamical entities . in a theory with only one fundamental , dimensional , parameter , such as superstring theory appears to be , there is clearly a sense in which all dimensionless couplings are dynamical variables .
they might nevertheless behave effectively as constants either if it costs a very large energy - density to excite them at all ( _ e.g_. massive fields ) or if ordinary matter couples very weakly to long - wavelength fluctuations of the field ( _ e.g_. stiff light fields with derivative couplings ) . in the latter case
one might anticipate long - range forces mediated by the exchange of the field @xcite .
i believe that experiments to look for such forces are among the most fundamental which can presently be attempted .
they address in a concrete way the question : are the constants of nature uniquely determined to be what we observe by dynamical laws , or are they ` frozen accidents ' imprinted at the big bang ? or are they presently relaxing towards some more favorable value to about 1 gev , then relaxes towards the dynamically favored , approximately p and t conserving , value . ] ?
note that although a rapidly ( on cosmological time scales ) oscillating field represents non - relativistic matter , a sufficiently slowly varying field might appear as a contribution to the effective cosmological term . if these ideas are along the right lines , it could be misguided to seek a unique lorentz - invariant ` vacuum ' state as a model for our present world .
one might instead be required , at a fundamental level , to seek the boundary conditions for particle physics from cosmology ( and _ vice versa _ ) .
one could then hope that some single simple , attractive _ ansatz _ would form the basis of both cosmology and particle physics , and truly unify these fundamental sciences .
is it plausible ?
i think that recent developments in string theory , especially in its application to the theory of black holes , are quite encouraging in this regard . in that context one primarily deals with spatially inhomogeneous but essentially stationary ( in time ) configurations .
great insight has been achieved by modeling extremal black holes and other solitons as extended singular surfaces ( d - branes ) @xcite where the conditions for string propagation are modified . a central point worth emphasizing that there is no obvious singularity in directly physical quantities , because there are no pointlike physical objects within the theory capable of probing and resolving the singular mathematical structure directly . at large distances
one has physics which appears to a local observer to be building toward a field - theoretic soliton and spatial geometry with a singular core , but the crisis dissolves into something that can not be described in terms of ordinary fields and spatial geometry .
it ought to be possible to examine the analogous situation for pseudo - singularities occuring at an initial ` time ' which of course would wind up _ not _ being physical time . in this way
, it might be possible to retain the advantages of simplicity and uniqueness inherent in a mathematically singular initial condition , while avoiding any associated physical pathology .
it is conceivable to me that even a very simple initial condition might suffice to generate a region of the universe of sufficient complexity to be identified with our own .
indeed , by analogy with the black hole case one might expect that the ( mathematically ) singular time - origin would look , at sufficient distance so that familiar space - time - matter notions apply , like a sort of universal heat pulse which perhaps could be identified as the trigger for a hot big bang .
as guth and linde have emphasized to us , such a quasi - thermal starting point can lead , through inflation , to an extremely rich and structured dynamics , including generically the possibility that different spatial regions are governed by different ` laws of physics ' specifically differing with respect to the values of the sorts of moduli fields mentioned above , but also conceivably differing even with respect to effective spatial dimension , gauge groups and family structures , supersymmetry breaking ( or not ) mechanism , and so forth .
physics would then give an essentially historical but not vague
account of the world we observe : an account highly constrained but perhaps far from determined by fundamental symmetry and consistency principles .
this would be similar to how biology is constrained , but not determined , by physics .
of course , this meta - proposal for a wave function or rather initial condition of the universe is based on quite different physical considerations from that of hartle - hawking @xcite , and probably contradicts it : in particular there is no appeal to euclidean procedures , and the arrow of time is manifest .
my overarching perception , coming out of this set of dialogues , is that the doors to several new worlds are ajar . on the cosmological side
, we can look forward especially to new observational results on the microwave anisotropy .
these will heavily impact the classic questions of homogeneous cosmology ( including the values of hubble parameter @xmath27 , the spatial curvature of the universe , and the cosmological term @xmath34 ) as well as theoretical ideas about primordial fluctuations .
the semi - phenomenological approach to particle physics , emphasized in the body of this talk , leads us to anticipate that there is a wealth of accessible new phenomena whose broad outlines ( i believe ) we can confidently foresee , and whose details will be richly informative on fundamental questions . and there is today rapid ferment in the ` high theory ' of particle physics and quantum gravity , with results of striking novelty and mathematical depth emerging .
the challenge remains , to connect these results with concrete phenomena .
perhaps cosmology offers our best hope for this .
we see narrow shafts of light emerging from behind each door .
in each case , we do not know what is on the other side whether it is a beautiful habitable world , a labyrinth , or a jungle .
the doors may be heavy , and require elaborate machinery to move them , but we must , and will , push through .
i would like to thank keith dienes for supplying figures 6 and 7 , and especially chris kolda for valuable assistance in the preparation of this manuscript .
after several partial and tentative proposals , the @xmath3 electroweak theory took on in its essentially modern form in : s. weinberg , phys .
* 19 * , 1264 ( 1967 ) ; a. salam , in _ elementary particle physics _ , ed .
n. svartholm ( almqvist and wiksells , stockholm , 1968 ) , p. 367 ; s. glashow , j. iliopoulos , and l. maiani , phys .
rev . d*2 * , 1285 ( 1970 ) .
after several partial and tentative proposals , the @xmath1 strong interaction theory took on its essentially modern form in : d. gross and f. wilczek , phys rev .
d * 8 * , 3633 ( 1973 ) ; s. weinberg , phys .
lett . * 31 * , 494 ( 1973 ) ; h. fritzsch , m. gell - mann , and h. leutwyler , phys . lett . * 47b * , 365 ( 1973 ) .
not coincidentally , the key discovery that allowed one to connect the abstract gauge theory to experiments is asymptotic freedom was first demonstrated just prior to these papers , in : d. gross and f. wilczek , phys . rev . lett . * 30 * , 1343 ( 1973 ) ; h. d. politzer , phys .
lett . * 30 * , 1346 ( 1973 ) .
the existence of the infrared fixed point was first discussed in : c. hill , phys .
d * 24 * , 691 ( 1981 ) .
more recent examinations including supersymmetry appear in : v. barger , m. berger and p. ohmann , phys .
d * 47 * , 1093 ( 1993 ) ; p. langacker and n. polonsky , phys .
d * 47 * , 4028 ( 1993 ) ; m. carena , s. pokorski and c. wagner , nuc .
b * 406 * , 59 ( 1993 ) .
m. gell - mann , p. ramond , and r. slansky , in _
supergravity _ , ed .
p. van neiuwenhuizen and d. freedman ( north holland , amsterdam , 1979 ) , p. 315 ; t. yanagida , proc . of the workshop on unified theory and baryon number in the universe , eds . o. sawada and a. sugamoto ( kek , 1979 ) .
a very useful introduction and collection of basic papers on supersymmetry is s. ferrara , _ supersymmetry _ ( 2 vols . )
( world scientific , singapore 1986 ) .
another excellent standard reference is n .- p .
nilles , phys .
reports * 110 * , 1 ( 1984 ) .
j. ellis , s. kelley , and d. nanopoulos , phys .
b260 * , 131 ( 1991 ) ; u. amaldi , w. de boer , and h. furstenau , phys . lett . *
b260 * , 447 ( 1991 ) ; for more recent analysis see p. langacker and n. polonsky , phys .
d * 49 * , 1454 ( 1994 ) .
s. dimopoulos , m. dine , s. raby and s. thomas , phys . rev . lett . *
76 * , 3494 ( 1996 ) ; s. ambrosanio , g. kane , g. kribs , s. martin and s. mrenna , phys . rev . lett .
* 76 * , 3498 ( 1996 ) ; s. dimopoulos , s. thomas , and j. wells , report slac - pub-7148 , hep - ph/9604452 ( april 1996 ) ; g. dvali , g. giudice and a. pomarol , report cern - th/96 - 61 , hep - ph/9603238 ( march 1996 ) .
s. park , representing the cdf collaboration , `` search for new phenomena in cdf , '' in _
10th topical workshop on proton - antiproton collider physics _
, ed . by r.
raha and j. yoh ( aip press , new york , 1995 ) , report fermilab - conf-95/155-e .
|
i discuss some central issues in particle physics which are potentially relevant to cosmology .
i first briefly review the present ( glorious ) experimental status of the standard model , emphasizing that it provides a firm foundation both for early universe cosmology and for further exploration toward the basic laws of nature .
i then provide a critique , arguing that while there are no clear discrepancies , there are several major , specific deficiencies of the standard model which clearly point up its provisional character .
i elaborate on the story theorists have made up to address one of these problems , the problem of scattered multiplets , and show how upon following it out one finds , within existing experiments , encouragement bordering on evidence for certain ambitious ideas regarding unification and supersymmetry .
i briefly describe and contrast two paradigms of supersymmetry breaking , which have markedly different experimental and cosmological consequences .
i call attention to specific connections with cosmology where appropriate throughout ; and near the end i make some more global remarks .
finally i venture a speculation suggesting , in a fairly concrete way , the possibility that the laws of physics can not , in principle , be disentangled from cosmology . .7ex
.7ex
gev tev _ i.e. _
@=11 caption#1[#2]#3
@=12
|
symmetry occurs in many combinatorial problems . for example , when coloring a graph , we can permute the colors in any proper coloring .
symmetry can also be introduced by modelling decisions ( e.g. using a set of finite domain variables to model a set of objects will introduce the symmetries that permute these variables ) . a common method to deal with symmetry
is to add constraints which eliminate symmetric solutions ( e.g. @xcite ) .
unfortunately , breaking symmetry by adding constraints to eliminate symmetric solutions is intractable in general @xcite . more specifically , deciding if an assignment is the smallest in its symmetry class for a matrix with row and column symmetries is np - hard , supposing rows are appended together and compared lexicographically .
there is , however , nothing special about appending rows together or comparing solutions lexicographically .
we could use _ any _ total ordering over assignments .
for example , we could break symmetry with the gray code ordering .
that is , we add constraints that eliminate symmetric solutions within each symmetry class that are not smallest in the gray code ordering .
this is a total ordering over assignments used in error correcting codes .
such an ordering may pick out different solutions in each symmetry class , reducing the conflict between symmetry breaking , problem constraints , objective function and the branching heuristic .
the gray code ordering has some properties that may be useful for symmetry breaking .
in particular , neighbouring assignments in the ordering only differ at one position , and flipping one bit reverses the ordering of the subsequent bits . as a second example , we can break row and column symmetry with the snake - lex ordering @xcite .
this orders assignments by lexicographically comparing vectors constructed by appending the variables in the matrix in a `` snake like '' manner .
the first row is appended to the reverse of the second row , and this is then appended to the third row , and then the reverse of the fourth row and so on . as a third example , we can break row and column symmetry by ordering the rows lexicographically and the columns with a multiset ordering @xcite .
this is incomparable to the lex - leader method .
we will argue theoretically that breaking symmetry with a different ordering over assignments can not improve the worst case complexity .
however , we also show that other orderings can be useful in practice as they pick out different solutions in each symmetry class .
our argument has two parts .
we first argue that , under modest assumptions which are satisfied by the gray code and snake - lex orderings , we can not reduce the computational complexity from that of breaking symmetry with the lexicographical ordering which considers variables in a matrix row - wise .
we then prove that for the particular case of row and column symmetries , breaking symmetry with the gray code or snake - lex ordering is intractable ( as it was with the lexicographical ordering ) .
many dynamic methods for dealing with symmetry are equivalent to posting symmetry breaking constraints `` on the fly '' ( e.g. @xcite ) .
hence , our results have implications for such dynamic methods too .
a symmetry of a set of constraints @xmath0 is a bijection @xmath1 on complete assignments that maps solutions of @xmath0 onto other solutions of @xmath0 .
many of our results apply to the more restrictive definition of symmetry which considers just those bijections which map individual variable - value pairs @xcite .
however , this more general definition captures also conditional symmetries @xcite .
in addition , a few of our results require this more general definition .
in particular , theorem 3 only holds for this more general definition .
the set of symmetries form a group under composition . given a symmetry group @xmath2 , a subset @xmath3 generates @xmath2 iff any @xmath4 is a composition of elements from @xmath3 . a symmetry group @xmath2 partitions the solutions into symmetry classes ( or orbits ) .
we write @xmath5_{\sigma}$ ] for the symmetry class of solutions symmetric to the solution @xmath6 . where @xmath2 is clear from the context , we write @xmath5 $ ] .
a set of symmetry breaking constraints is _ sound _ iff it leaves at least one solution in each symmetry class , and _ complete _ iff it leaves at most one solution in each symmetry class .
we will study what happens to symmetries when problems are reformulated onto equivalent problems .
for example , we might consider the boolean form of a problem in which @xmath7 maps onto @xmath8 .
two sets of constraints , @xmath0 and @xmath9 over possibly different variables are _ equivalent _ iff there is a bijection between their solutions .
suppose @xmath10 and @xmath11 for @xmath12 $ ] are partitions of the sets @xmath13 and @xmath14 into @xmath15 subsets .
then the two partitions are _
isomorphic _ iff there are bijections @xmath16 and @xmath17\mapsto [ 1,k]$ ] such that @xmath18 for @xmath12 $ ] where @xmath19 .
two symmetry groups @xmath2 and @xmath3 of constraints @xmath0 and @xmath9 respectively are _ isomorphic _ iff @xmath0 and @xmath9 are equivalent , and their symmetry classes of solutions are isomorphic .
the lex - leader method @xcite picks out the lexicographically smallest solution in each symmetry class . for every symmetry @xmath1 , it posts a lexicographical ordering constraint : @xmath20 where @xmath21 to @xmath22 is some ordering on the variables in the problem .
many static symmetry breaking constraints can be derived from such lex - leader constraints .
for example , doublelex constraints to break row and column symmetry can be derived from them @xcite . as a second example
, precedence constraints to break the symmetry due to interchangeable values can also be derived from them @xcite .
efficient algorithms exist to propagate such lexicographical constraints ( e.g. @xcite ) .
we could , however , break symmetry by using another ordering on assignments like the gray code ordering .
we define the gray code ordering on boolean variables . for each symmetry
@xmath1 , we could post an ordering constraint : @xmath23 where the @xmath15-bit gray code ordering is defined recursively as follows : 0 is before 1 , and to construct the @xmath24-bit ordering , we append 0 to the front of the @xmath15-bit ordering , and concatenate it with the reversed @xmath15-bit ordering with 1 appended to the front .
for instance , the 4-bit gray code orders assignments as follows : @xmath25 the gray code ordering is well founded .
hence , every set of complete assignments will have a smallest member under this ordering .
this is the unique complete assignment in each symmetry class selected by posting such gray code ordering constraints . thus breaking symmetry with gray
code ordering constraints is sound and complete .
breaking symmetry with gray code ordering constraints is sound and complete . in section 6 , we propose a propagator for the gray code ordering constraint .
we can not enforce the gray code ordering by ordering variables and values , and using a lexicographical ordering constraint .
for example , we can not map the 2-bit gray code onto the lexicographical ordering by simply re - ordering variables and values . to put it another way ,
no reversal and/or inversion of the bits in the 2-bit gray code will map it onto the lexicographical ordering .
the 2-bit gray code orders 00 , 01 , 11 and then 10 .
we can invert the first bit to give : 10 , 11 , 01 and then 00 . or we can invert the second bit to give : 01 , 00 , 10 , and then 11 . or we can invert both bits to give : 11 , 10 , 00 , and then 01 .
we can also reverse the bits to give : 00 , 10 , 11 , and then 01 . and we can then invert one or both bits to give : 10 , 00 , 01 , and then 10 ; or 01 , 11 , 10 , and then 00 ; or 11 , 01 , 00 , and then 10 . note that none of these re - orderings and inversions is the 2-bit lexicographical ordering : 00 , 01 , 10 , and then 11 .
we will show that , under some modest assumptions , we can not make breaking symmetry computationally easier by using a new ordering like the gray code ordering .
our argument breaks into two parts .
first , we observe how the symmetry of a problem changes when we reformulate onto an equivalent problem .
second , we argue that we can map onto an equivalent problem on which symmetry breaking is easier . if a set of constraints @xmath0 has a symmetry group @xmath2 , @xmath0 and @xmath9 are equivalent sets of constraints , @xmath26 is any bijection between solutions of @xmath0 and @xmath9 , and @xmath27 then : 1 .
@xmath28 is a symmetry group of @xmath9 ; 2 .
@xmath2 and @xmath28 are isomorphic symmetry groups ; 3 . if @xmath3 generates @xmath2 then @xmath29 generates @xmath28 .
we will use this proposition to argue that symmetry breaking with any ordering besides the lexicographical ordering is intractable .
we consider only _ simple _ orderings . in a simple ordering , we can compute the position of any assignment in the ordering in polynomial time , and given any position in the ordering we can compute the assignment at this position in polynomial time .
we now give our main result . given any simple ordering @xmath30 , there exists a symmetry group such that deciding if an assignment is smallest in its symmetry class according to @xmath30 is np - hard .
* proof : * deciding if an assignment is smallest in its symmetry class according to @xmath31 is np - hard @xcite . since @xmath30 and @xmath31 are both simple orderings , there exist polynomial functions @xmath32 to map assignments onto positions in the @xmath31 ordering , and @xmath33 to map positions in the @xmath30 ordering to assignments .
consider the mapping @xmath26 defined by @xmath34 . now
@xmath26 is a permutation that is polynomial to compute which maps the total ordering of assignments of @xmath31 onto that for @xmath30 .
similarly , @xmath35 is a permutation that is polynomial to compute which maps the total ordering of assignments of @xmath30 onto that for @xmath31 .
let @xmath36 be the row and column symmetry group . by theorem 2 ,
the problem of finding the lexicographical least element of each symmetry class for @xmath36 is equivalent to the problem of finding the least element of each symmetry class according to @xmath30 for @xmath37 . thus , for the symmetry group @xmath37 deciding if an assignment is smallest in its symmetry class according to @xmath30 is np - hard .
it follows that there exists an infinite family of symmetry groups such that checking a constraint which is only satisfied by the smallest member of each symmetry class is np - hard .
note that the gray code and snake - lex orderings are simple .
hence , breaking symmetry with either ordering is np - hard for some symmetry groups .
note that we are not claiming that deciding if an assignment is smallest in its symmetry class is np - complete .
first , we would need to worry about the size of the input ( since we are considering the much larger class of symmetries that act on complete assignments rather than on literals ) .
second , to decide that an assignment is the smallest , we are also answering a complement problem ( there is _ no _ smaller symmetric assignment ) .
this will take us to dp - completeness or above .
we next consider a common type of symmetry . in many models
, we have a matrix of decision variables in which the rows and columns are interchangeable @xcite .
we will show that breaking row and column symmetry specifically is intractable with the gray code and the snake - lex orderings , as it is with the lexicographical ordering that considers the variables in a row - wise order .
[ l : l1 ] finding the smallest solution up to row and column symmetry for the snake - lex ordering is np - hard .
* proof : * we reduce from the problem of finding the lex - leader solution of a matrix @xmath38 .
let @xmath38 be an @xmath39 matrix of boolean values .
we assume @xmath38 does not contain a row of only ones since any such row can be placed at the bottom of the matrix .
we embed @xmath38 in the matrix @xmath40 such that finding @xmath41 , denoted @xmath42 , the smallest row and column symmetry of @xmath40 in the snake - lex ordering is equivalent to finding the lex - leader of @xmath38 .
we ensure that even rows in the snake - lex smallest symmetric solution of @xmath40 are taken by dummy identical rows .
then in odd rows , where snake - lex moves from the left to the right along a row like lex does , we embed the lex - leader solution of @xmath38 .
let @xmath43 be the maximum number of zeros in any row of @xmath38 .
we construct @xmath40 with @xmath44 rows and @xmath45 columns so that it contains three sets of rows .
the first set consists of a single row that contains @xmath46 zeros followed by @xmath47 ones .
the second set contains @xmath48 identical rows with @xmath49 ones followed by @xmath50 zeros in each row .
the third set of rows contains @xmath48 rows such that at the @xmath51th row the first @xmath52 positions are ones , the next @xmath53 positions are the @xmath51th row from @xmath38 and the last @xmath50 positions are ones again .
schematic representation of @xmath40 is shown at figure [ f : construction0](a ) .
we determine positions of rows and columns that must be fixed in @xmath42 up to permutation of identical rows and columns .
the first row of @xmath40 has to be the first row of @xmath42 as no other row contains @xmath46 zeros .
note that this also fixes the position of columns from 1 to @xmath46 in @xmath40 to be the first columns in the @xmath42 .
note also that these columns are identical and each of them contains the zero in the first row only .
one of the rows in the second set has to be the second row of @xmath42 , as none of the rows that embed rows from @xmath38 contains @xmath50 zeros .
as we move from the right to the left on even rows , this also makes sure that last @xmath50 columns from @xmath40 must be the last columns in @xmath42 .
we summarise that at this point the first two rows are fixed and the first @xmath46 columns and the last @xmath50 columns in @xmath42 must be equal to a permutation of the first @xmath46 identical columns and the last @xmath50 identical columns in @xmath40 , respectively . by assumption , @xmath38 does not contain rows with all ones .
moreover , only rows that embed rows from @xmath38 can have the value zero at columns from @xmath54 to @xmath55 in @xmath42 .
hence , a row from the third set that embeds a row from @xmath38 has to be the third row in @xmath42 .
we do not specify which row it is at this point .
the fourth row has to be again a row from the second set as any of remaining rows from the second set has @xmath50 zeros in the last @xmath50 columns in @xmath42 while any row that embeds @xmath38 has at most @xmath43 zeros .
we can repeat this argument for the remaining rows .
a schematic representation of the positions of rows from the first and second sets are shown in figure [ f : construction0](b ) .
note that the first and all even rows in @xmath42 are fixed .
the only part of @xmath42 not yet specified is the ordering of odd rows of @xmath53 columns from @xmath54 to @xmath55 .
these are exactly all rows from @xmath38 .
hence , finding @xmath42 is reduced to ordering of this set of rows and columns that embed @xmath38 .
now , all columns from @xmath54 to @xmath55 are interchangeable , all odd rows except the first are interchangeable , and all elements of @xmath42 except elements of @xmath38 are fixed by construction .
as the snake - lex ordering goes from the left to the right on odd rows like the lex ordering , finding @xmath42 is equivalent to finding the lex - leader of @xmath38 to show that finding the smallest row and column symmetry in the gray code ordering is np - hard , we need a technical lemma about cloning columns in a matrix .
we use rowwise ordering in a matrix .
suppose we clone each column in a @xmath56 boolean matrix @xmath38 to give the matrix @xmath57 .
let @xmath58 be the smallest row and column symmetry of @xmath57 in the gray code ordering .
[ l : l2 ] any original column of @xmath38 is followed by its clone in @xmath58 ignoring permutation of identical original columns .
* proof : * by contradiction .
suppose there exists an element @xmath59 $ ] such that the original column @xmath51 and the next column @xmath60 are different at the @xmath61th row .
we denote by @xmath15 the @xmath62 $ ] element of @xmath58 in its row - wise linearization .
we ignore the rows from @xmath63 to @xmath48 at this point as they are not relevant to this discrepancy .
each pair of columns coincide on the first @xmath61 rows for the first @xmath64 columns and on the first @xmath65 rows for the columns from @xmath51 to @xmath53 .
we conclude that ( 1 ) @xmath51 is odd and @xmath60 is even ; ( 2 ) the number of ones between the first and the @xmath66th positions in the linearization of @xmath58 is even as each value is duplicated ; ( 3 ) the clone of the @xmath51th column can not be among the first @xmath64 columns as each such column is followed by its clone by assumption .
hence , the clone of the @xmath51th column is among columns from @xmath67 to @xmath53 .
suppose the clone of the @xmath51th column is the @xmath68th column .
note that the @xmath68th column must coincide with the @xmath60th column at the first @xmath65 rows .
we consider two cases . in the first case ,
@xmath69 = 1 $ ] and @xmath59=0 $ ] .
note that the total number of ones at the positions from @xmath70 to @xmath71 is odd as we have one in the position @xmath71 and the number of ones in the first @xmath72 positions is even .
next we swap the @xmath73th and @xmath68th columns in @xmath58 .
this will not change the first @xmath71 elements in the linearization as the @xmath68th column must coincide with the @xmath60th column at the first @xmath65 rows
. moreover , this swap puts @xmath70 in position @xmath15 .
as the number of ones up to the @xmath74th position is odd then @xmath70 goes before @xmath75 at position @xmath15 in the gray code ordering .
hence , by swapping the @xmath73th and @xmath68th columns we obtain a matrix that is smaller than @xmath58 in the gray code ordering .
this is a contradiction . in the second case ,
@xmath69 = 0 $ ] and @xmath59=1 $ ] .
note that the total number of ones at positions 1 to @xmath71 in the linearization is even as we have zero at the position @xmath71 and the number of ones in the first @xmath72 positions is even .
therefore , @xmath75 precedes @xmath70 at position @xmath15 in the gray code ordering . by swapping the @xmath73th and @xmath68th columns we obtain a matrix that is smaller than @xmath58 in the gray code ordering as @xmath75 appears at the position @xmath15 instead of @xmath70
this is a contradiction .
[ l : l3 ] finding the smallest solution up to row and column symmetry for the gray code ordering is np - hard .
* proof : * we again reduce from the problem of finding the lex - leader solution of a matrix @xmath38 .
we clone every column of @xmath38 and obtain a new matrix @xmath57 .
let @xmath58 be the smallest row and column symmetry of @xmath57 in the gray code ordering .
lemma [ l : l2 ] shows that each original column is followed by its clone in @xmath58 .
next we delete all clones by removing every second column .
we call the resulting matrix @xmath76 .
we prove that @xmath76 is the lex - leader of @xmath38 by contradiction .
suppose there exists a matrix @xmath40 which is the lex - leader of @xmath38 that is different from @xmath77 .
hence , @xmath40 is also the lex - leader of @xmath77 .
we find the first element @xmath78 $ ] where @xmath79 \neq m[j , i]$ ] in the row - wise linearization of @xmath40 and @xmath77 , so that @xmath79 = 1 $ ] and @xmath78 = 0 $ ] .
we denote by @xmath15 the position of the @xmath80 $ ] element of @xmath40 in its row - wise linearization .
we clone each column of @xmath40 once and put each cloned column right after its original column .
we obtain a new matrix @xmath81 .
we show that @xmath81 is smaller than @xmath58 in the gray code ordering to obtain a contradiction . as @xmath79 = 1 $ ] and @xmath78 = 0 $ ] then @xmath82 = 1 $ ] and @xmath83 = 0 $ ] because the matrices @xmath58 and @xmath81 are obtained from @xmath77 and @xmath40 by cloning each column and putting each clone right after its original column . as @xmath77 and @xmath40
coincide on the first @xmath71 positions then @xmath58 and @xmath81 coincide in the first @xmath84 positions . by transforming @xmath77 and @xmath40 to @xmath58 and @xmath81 , we duplicated each value in positions from @xmath70 to @xmath71 .
hence , the total number of ones in positions from @xmath70 to @xmath84 in @xmath69 $ ] and @xmath85 $ ] is even .
therefore , the value zero precedes the value one at position @xmath86 in the gray code ordering . by assumption , the value in the position @xmath86 in @xmath58 , which is @xmath82 $ ] , is 1 , and the position @xmath86 in @xmath81 , which is @xmath83 $ ] , is @xmath75 .
hence , @xmath81 is smaller than @xmath58 in the gray code ordering .
we conjecture that row and column symmetry will be intractable to break for other simple orderings .
however , each such ordering may require a new proof .
despite these negative theoretical results , there is still the possibility for other orderings on assignments to be useful when breaking symmetry in practice .
it is interesting therefore to develop propagation algorithms for different orderings .
propagation algorithms are used to prune the search space by enforcing properties like domain consistency .
a constraint is _ domain consistent _
( _ dc _ ) iff when a variable is assigned any value in its domain , there exist compatible values in the domains of the other variables .
we give an efficient encoding for the new global constraint @xmath87 , [ y_1,\ldots , y_n])$ ] that ensures @xmath88 is before or equal in position to @xmath89 in the gray code ordering where @xmath90 and @xmath91 are 0/1 variables .
we encode the transition relation of an automaton with 0/1/-1 state variables , @xmath92 to @xmath93 that reads a sequence @xmath94 and ensures that the two sequences are ordered appropriately .
we consider the following decomposition where @xmath95 : @xmath96 we can show that this decomposition not only preserves the semantics of the constraint but also does not hinder propagation .
unit propagation on this decomposition enforces domain consistency on @xmath87,[y_1,\ldots , y_n])$ ] in @xmath97 time .
* proof : * ( correctness ) @xmath98 as soon as the two vectors are ordered correctly .
@xmath99 iff @xmath90 and @xmath100 are ordered in the gray code ordering with 0 before 1 .
@xmath101 iff the @xmath51th bits , @xmath90 and @xmath100 are ordered in the gray code ordering with 1 before 0 .
@xmath102 stays the same polarity as @xmath103 iff @xmath104 and flips polarity iff @xmath105 .
( completeness ) this follows from the completeness of cnf encoding of the corresponding automaton @xcite and the fact that unit propagation on this set of constraints enforces dc on a table constraint that encodes the transition relation .
( complexity ) there are @xmath97 disjuncts in the decomposition .
hence unit propagation takes @xmath97 time .
in fact , it is possible to show that the _ total _ time to enforce dc down a branch of the search tree is @xmath97 .
note that this decomposition can be used to break symmetry with the gray code ordering in a sat solver . for row and column symmetry
, we can break symmetry with the @xmath106 constraint that lexicographically orders rows and columns , or the @xmath107 constraint .
this is based on the smallest row and column permutation of the matrix according to an ordering on assignments that linearizes the matrix in a snake - like manner @xcite .
the ( columnwise ) @xmath107 constraint can be enforced by a conjunction of @xmath108 lexicographical ordering constraints on pairs of columns and @xmath109 lexicographical constraints on pairs of intertwined rows . to obtain the rowwise @xmath107 constraint ,
we transpose the matrix and then order as in the columnwise @xmath107 .
note that @xmath106 and @xmath107 _ only break a subset _ of the row and colum symmetries .
however , they are very useful in practice . it was shown in @xcite , that enforcing _
dc _ on the @xmath106 constraint is np - hard .
hence we typically decompose it into separate row and column constraints . here
, we show that enforcing _
dc _ on the @xmath107 constraint is also np - hard .
it is therefore also reasonable to propagate @xmath107 by decomposition .
[ t : snake_double ] enforcing dc on the @xmath107 constraint is np - hard .
* proof : * let @xmath38 be a @xmath48 by @xmath53 matrix of boolean variables , @xmath110 . the main idea is to embed @xmath38 in to a specially constructed matrix @xmath40 in such a way that enforcing _
dc _ on the @xmath106 constraint on @xmath38 ( which we already know is np - hard ) is equivalent to enforcing _
dc _ on the @xmath107 constraint on @xmath40 .
we transform the matrix of variables @xmath38 into a new matrix of variables @xmath40 with @xmath111 rows , @xmath112 and @xmath113 columns , @xmath114 .
@xmath115 = \left\ { \begin{array}{ll } x_{v-1,(k+1)/2 } & \textrm{if $ v \in [ 2,n+1]$ and $ k$ is odd } \\ 1 & \textrm{if $ v \in [ 2,n+1]$ and $ k$ is even } \\ 0 & \textrm{if $ v=1 $ and $ k$ is odd or if $ v = n+2 $ and $ k$ is even } \\ 1 & \textrm{if $ v=1 $ and $ k$ is even or if $ v = n+2 $ and $ k$ is odd } \\
\end{array } \right.\ ] ] figure [ f : snake_dlex ] shows the transformation for a matrix @xmath38 with @xmath116 and @xmath117 . _ row constraints on @xmath40 .
_ the first constraint @xmath118 in @xmath40 is @xmath119 which can be simplified to an entailed constraint @xmath120 .
similarly , the constraint @xmath121 is entailed .
l6 cm consider the @xmath122 constraint , @xmath123 $ ] in @xmath40 : @xmath124 @xmath125 . as
all even elements coincide we can simplify the constraint to @xmath126 , @xmath127 so that @xmath128 $ ] .
hence , all row constraints are simplified to the row ordering constraints of the @xmath106 constraint . _ column constraints on @xmath40 . _
the constraints @xmath129 , @xmath130 , are entailed as @xmath131 . the constraints @xmath129 and @xmath132 , @xmath133 , are entailed
as @xmath129 is @xmath134 and @xmath132 is @xmath135 .
therefore , the only not entailed constraints are @xmath132 , @xmath130 : @xmath136 .
these constraints can be simplified to @xmath137 , @xmath138 so that @xmath139 which gives the column ordering of the @xmath106 constraint . hence , enforcing _
dc _ on the @xmath107 constraint on @xmath40 enforces _ dc _ on the @xmath106 constraint on @xmath38 .
we tested two hypotheses that provide advice to the modeller when breaking symmetry .
1 . other orderings besides the lexicographical ordered can be effective when breaking symmetry in practice ; 2 .
symmetry breaking should align with the branching heuristic , and with the objective function .
all our experiments report the time to find an optimal solution _ and _ prove it optimal .
we believe that optimisation is often a more realistic setting in which to illustrate the practical benefits of symmetry breaking , than satisfaction experiments which either find one or all solutions . breaking symmetry in optimisation problems
is important as we must traverse the whole search space when proving optimality .
all our experiments used the bprolog 7.7 constraint solver .
this solver took second place in the asp 2011 solver competition .
the three sets of experiments took around one cpu month on a macbook pro with an intel core i5 2 core 2.53 ghz processor , with 4 gb of memory .
the three domains were chosen as representative of optimisation problems previously studied in symmetry breaking .
we observed similar results in these as well as other domains .
this is prob032 in csplib @xcite .
this problem arises in conway s game of life , and was popularized by martin gardner . given a @xmath48 by @xmath48 submatrix of the infinite plane
, we want to find and prove optimal the pattern of maximum density which does not change from generation to generation .
for example , an optimal solution for @xmath140 is : @xmath141 this is a still life as every live square has between 2 and 3 live neighbours , and every dead square does not have 3 live neighbours .
we use the simple 0/1 constraint model from @xcite .
this problem has the 8 symmetries of the square as we can rotate or reflect any still life to obtain a new one .
bosch and trick argued that _ `` the symmetry embedded in this problem is very strong , leading both to algorithmic insights and algorithmic difficulties '' .
_ our first experiment used the default search strategy to find and prove optimal the still life of maximum density for a given @xmath48 .
the default strategy instantiates variables row - wise across the matrix . our goal here is to compare the different symmetry breaking methods with an `` out of the box '' solver .
we then compare the impact of the branching heuristic on symmetry breaking .
we broke symmetry with either the lexicographical or gray code orderings , finding the smallest ( lex , gray ) or largest ( anti - lex , anti - gray ) solution in each symmetry class .
in addition , we linearized the matrix either row - wise ( row ) , column - wise ( col ) , snake - wise along rows ( snake ) , snake - wise along columns ( col - snake ) , or in a clockwise spiral ( spiral ) .
table 1 gives results for the 20 different symmetry breaking methods constructed by using 1 of the 4 possible solution orderings and the 5 different linearizations , as well with no symmetry breaking ( none ) .
.backtracks required to find and prove optimal the maximum density still life of size @xmath48 by @xmath48 using the default branching heuristic .
column winner is in _ emphasis_. [ cols="^,>,>,>,>,>",options="header " , ] we make some observations about these results .
first , the best symmetry breaking constraint with the default branching heuristic ( anti - gray col + row heuristic ) was either very good or very bad with the other branching heuristics .
it offers the best overall performance in this experiment ( viz .
anti - gray col + spiral - in heuristic ) , and is the best of all the symmetry breaking methods for 5 other heuristics .
however , it also the worst of all the symmetry breaking methods with 4 other heuristics .
second , aligning the branching heuristic with the symmetry breaking constraint at best offers middle of the road performance ( e.g. lex row + row heuristic ) but can also be counter - productive ( e.g. anti - gray col + col heuristic ) .
third , the spiral - in heuristic offer some of the best performance .
this heuristic provided the best overall result , and was always in the top 2 for every symmetry breaking method . recall that the spiral - in heuristic was one of the worst heuristics on the maximum density still life problem .
we conjecture that this is because it delays constraint propagation on the still life problem constraints but not on the constraints in the peaceable armies of queens problem .
fourth , a bad combination of branching heuristic and symmetry breaking constraints is worse than not breaking symmetry if we have a good branching heuristic ( e.g. none + constr heuristic beats anti - gray col + col - snake heuristic ) .
these results support both our hypotheses .
other orderings besides the simple lexicographical ordering can be effective for breaking symmetry , and symmetry breaking should align with both the branching heuristic and the objective function .
unfortunately , as the last example demonstrated , the interaction between problem constraints , symmetry breaking and branching heuristic can be complex and difficult to predict .
overall , the gray code ordering appears useful .
whilst it is conceptually similar to the lexicographical ordering , it looks at more than one bit at a time . this is reflected in the automaton for the gray code ordering which has more states than that required for the lexicographical ordering .
we have argued that in general breaking symmetry with a different ordering over assignments than the usual lexicographical ordering does not improve the computational complexity of breaking with symmetry .
our argument had two parts .
first , we argued that under modest assumptions we can not reduce the worst case complexity from that of breaking symmetry with a lexicographical ordering .
these assumptions are satisfied by the gray code and snake - lex orderings .
second , we proved that for the particular case of row and column symmetries , breaking symmetry with the gray code or snake - lex ordering is intractable ( as it was with the lexicographical ordering ) .
we then explored algorithms to break symmetry with other orderings .
in particular , we gave a linear time propagator for the gray code ordering constraint , and proved that enforcing domain consistency on the @xmath107 constraint , like on the @xmath106 constraint , is np - hard .
finally , we demonstrated that other orderings have promise in practice .
we ran experiments on three standard benchmark domains where breaking symmetry with the gray code ordering was often better than with the lex - leader or snake - lex methods .
crawford , j. , luks , g. , ginsberg , m. , roy , a. : symmetry breaking predicates for search problems . in : proceedings of the 5th international conference on knowledge representation and reasoning , ( kr 96 ) .
( 1996 ) 148159 aloul , f. , sakallah , k. , markov , i. : efficient symmetry breaking for boolean satisfiability . in : proceedings of the 18th international joint conference on ai , international joint conference on artificial intelligence ( 2003 )
271276 law , y. , lee , j. : global constraints for integer and set value precedence . in : proceedings of 10th international conference on principles and practice of constraint programming ( cp2004 ) , springer ( 2004 ) 362376 law , y.c . , lee , j. , walsh , t. , yip , j. : breaking symmetry of interchangeable variables and values . in : 13th international conference on principles and practices of constraint programming ( cp-2007 ) , springer - verlag ( 2007 ) katsirelos , g. , narodytska , n. , walsh , t. : on the complexity and completeness of static constraints for breaking row and column symmetry . in cohen ,
d. , ed . : proceedings of the 16th international conference on the principles and practice of constraint programming ( cp 2010 ) .
volume 6308 of lecture notes in computer science . , springer ( 2010 ) 305320 grayland , a. , miguel , i. , roney - dougal , c. : snake lex : an alternative to double lex . in gent , i.p . ,
ed . : proceedings of 15th international conference on principles and practice of constraint programming .
volume 5732 of lecture notes in computer science . , springer ( 2009 ) 391399 frisch , a. , hnich , b. , kiziltan , z. , miguel , i. , walsh , t. : multiset ordering constraints . in : proceedings of the 18th international joint conference on artificial intelligence ( ijcai-2003 ) , international joint conference on artificial intelligence ( 2003 ) benhamou , b. , sais , l.
: theoretical study of symmetries in propositional calculus and applications . in : proceedings of 11th international conference on automated deduction .
volume 607 of lecture notes in computer science . , springer ( 1992 ) 281294 backofen , r. , will , s. : excluding symmetries in constraint - based search . in jaffar , j. , ed . :
proceedings of the 5th international conference on principles and practice of constraint programming .
number 1713 in lecture notes in computer science , springer - verlag ( 1999 ) 7387 fahle , t. , schamberger , s. , sellmann , m. : symmetry breaking . in walsh ,
t. , ed . : proceedings of 7th international conference on principles and practice of constraint programming ( cp2001 ) , springer ( 2001 ) 93107 sellmann , m. , hentenryck , p.v . : structural symmetry breaking . in : proceedings of 19th internatinal joint conference on ai ( ijcai 2005 ) , international joint conference on artificial intelligence ( 2005 ) 298303 puget , j.f . : dynamic lex constraints . in benhamou , f. , ed . :
12th international conference on the principles and practice of constraint programming ( cp 2006 ) .
volume 4204 of lecture notes in computer science . , springer ( 2006 ) 453467 katsirelos , g. , walsh , t. : symmetries of symmetry breaking constraints . in : proc . of the 19th european conference on artificial intelligence ( ecai-2010
) , european conference on artificial intelligence , ios press ( 2010 ) narodytska , n. , walsh , t. : an adaptive model restarts heuristic . in : proceedings of 10th international conference on integration of ai and or techniques in constraint programming for combinatorial optimization problems ( cpaior 2013 ) .
volume 7874 of lecture notes in computer science . , springer ( 2013 ) 369377 gent , i. , kelsey , t. , linton , s. , mcdonald , i. , miguel , i. , smith , b. : conditional symmetry breaking . in van beek ,
: proceedings of 11 international conference on principles and practice of constraint programming ( cp 2005 ) .
volume 3709 of lecture notes in computer science . , springer ( 2005 ) 256270 flener , p. , frisch , a. , hnich , b. , kiziltan , z. , miguel , i. , pearson , j. , walsh , t. : symmetry in matrix models . in : proceedings of the cp-01 workshop on symmetry in constraints ( symcon01 ) .
( 2001 ) held alongside cp 2001 .
also apes-30 - 2001 technical report .
frisch , a. , hnich , b. , kiziltan , z. , miguel , i. , walsh , t. : global constraints for lexicographic orderings . in : 8th international conference on principles and practices of constraint programming ( cp-2002 ) , springer ( 2002 ) katsirelos , g. , narodytska , n. , walsh , t. : combining symmetry breaking and global constraints .
in oddi , a. , fages , f. , rossi , f. , eds . : recent advances in constraints , 13th annual ercim international workshop on constraint solving and constraint logic programming ( csclp 2008 ) .
volume 5655 of lecture notes in computer science .
, springer ( 2009 ) 8498 flener , p. , frisch , a. , hnich , b. , kiziltan , z. , miguel , i. , walsh , t. : matrix modelling : exploiting common patterns in constraint programming . in : proceedings of the international workshop on reformulating constraint satisfaction problems .
( 2002 ) held alongside cp-2002 .
flener , p. , frisch , a. , hnich , b. , kiziltan , z. , miguel , i. , pearson , j. , walsh , t. : breaking row and column symmetry in matrix models . in : 8th international conference on principles and practices of constraint programming ( cp-2002 ) , springer ( 2002 ) gent , i. , walsh , t.
: : a benchmark library for constraints . technical report , technical report apes-09 - 1999 ( 1999 ) a shorter version appears in the proceedings of the 5th international conference on principles and practices of constraint programming ( cp-99 ) .
smith , b. , petrie , k. , gent , i. : models and symmetry breaking for peacable armies of queens . in : proceedings of the first international conference on integration of ai and or techniques in constraint programming ( cp - ai - or ) , springer - verlag ( 2004 ) 271286 lncs 3011 .
|
we can break symmetry by eliminating solutions within each symmetry class . for instance , the lex - leader method eliminates all but the smallest solution in the lexicographical ordering .
unfortunately , the lex - leader method is intractable in general .
we prove that , under modest assumptions , we can not reduce the worst case complexity of breaking symmetry by using other orderings on solutions .
we also prove that a common type of symmetry , where rows and columns in a matrix of decision variables are interchangeable , is intractable to break when we use two promising alternatives to the lexicographical ordering : the gray code ordering ( which uses a different ordering on solutions ) , and the snake - lex ordering ( which is a variant of the lexicographical ordering that re - orders the variables ) .
nevertheless , we show experimentally that using other orderings like the gray code to break symmetry can be beneficial in practice as they may better align with the objective function and branching heuristic .
|
the masses of galaxies , and in fact of any larger dynamically bound structure in the universe , inferred from dynamics are found to exceed the masses of the observed baryons in these structures @xcite .
this phenomenon is usually explained by postulating ( non - baryonic ) dark matter , which is also the backbone of structure formation in cosmological simulations ( e.g. * ? ? ?
* ; * ? ? ?
* ) and serves well in accounting for the characteristics of the cosmic microwave background radiation ( e.g. * ? ? ?
* ; * ? ? ?
+ demonstrated a surprising characteristic of this discrepancy between dynamical and baryonic mass : it occurs below a characteristic acceleration level .
@xcite reproduced such a characteristic acceleration scale in the @xmath3cdm context using semi - analytic models of galaxy formation , which were tuned to reproduce the tully - fisher relation .
however , also found that the amplitude of this discrepancy correlates with the acceleration , the so - called mass discrepancy acceleration ( mda ) relation .
moreover , there is only small scatter about this relation , as confirmed by @xcite .
this correlation , and even more its tightness , is surprising in the context of hierarchical structure formation and the large variety of possible merger trees for individual galaxies in @xmath3cdm cosmology ( see also @xcite ) .
+ so far tests of the mda relation have concentrated mainly on spiral galaxies . in these systems
the hi gas makes it relatively easy to trace the dynamical mass far from the centre of the galaxy
. however , spiral galaxies also possess large amounts of baryonic mass in the form of gas .
the mass of this component is rather difficult to estimate accurately from observations and introduces uncertainties in the mda relation .
early - type galaxies ( ellipticals and lenticulars ) , due to their lower hi gas content , do not suffer from this problem , however they lack an easy to measure tracer .
for this reason , studies addressing the mda relation in early - type galaxies have had to largely rely on other tracers such as hot x - ray emitting gas or discrete tracers like planetary nebulae ( pne ) , globular clusters ( gc ) , and satellite galaxies ( e.g. @xcite ; see however @xcite ) .
while theoretical efforts trying to reproduce the mda relation in the @xmath3cdm context are progressing ( e.g. @xcite ) , it is desirable to push the observational side further , and probe the mda relation systematically for a sample of early - type galaxies down to low accelerations .
+ alternatively , the mda relation , and its tightness , have been used to argue in favour of a modification of newtonian dynamics ( mond ; see , e.g. @xcite ; @xcite ) .
a number of previous studies have discussed whether early - type galaxies can be used to falsify the theory .
recently , ( * ? ? ?
* hereafter ) carried out dynamical modelling of a sample of 14 early - type galaxies , empowered by the combination of the inner stellar kinematics from atlas@xmath0 @xcite and stellar kinematics reaching out to a median radius of about 4 half - light radii ( @xmath4 ; @xcite ) from the sluggs survey @xcite .
this represents the first homogeneous , statistically meaningful sample of stellar kinematics of early - type galaxies ( all of which are fast rotators as defined in @xcite ) reaching radii where dark matter is expected to dominate the mass budget ( i.e. the dark matter fraction increases from @xmath11/3 to @xmath12/3 when measured within 2 and 4 @xmath5 , respectively ) .
here we use s mass modelling results to test whether or not early - type galaxies follow the mda relation .
furthermore , we test whether or not mond , which was originally designed to reproduce the rotation curves of spiral galaxies @xcite , fails in early - type galaxies .
finally , we also consider results from mass modelling based on gc kinematics for an extended sample , including slow rotators and reaching even larger radii , also from the sluggs survey @xcite .
our primary sample comprises 14 early - type galaxies in a stellar mass range of @xmath6 @xcite .
mass models for these galaxies were built by using axisymmetric jeans anisotropic modelling ( jam ; @xcite ) with a stellar and a dark matter component . to ensure the method to be applicable , and to achieve a homogenous sample of nearly axisymmetric galaxies , restricted the sample to fast rotators .
one input to the jam modelling is the light distributions in these galaxies .
those were parametrized using multi - gaussian expansions ( mge ; ) and were taken from various studies ( @xcite ; ) . here
, we use the same characterizations of the distributions of stars within the galaxies .
other galaxy parameters , such as half - light radius and distance , are taken from , and the velocity dispersion within @xmath7 from @xcite and for ngc 3115 .
the combination of data from atlas@xmath0 @xcite and sluggs @xcite allowed to probe the stellar dynamics of the galaxies from their inner parts out to a median radius of @xmath14@xmath8 .
the stellar kinematics were extracted in the optical ( atlas@xmath0 ; @xcite ) and ca triplet spectral region ( sluggs ; @xcite ) , in both cases with ppxf @xcite .
for the dynamical modelling the data were symmetrized , outliers were removed , and the two data sets combined .
we refer the reader to for a description of the process .
however , we do note the key characteristic of was the use of a very general parametrization for the dark halo in the modelling process , to be able to focus on the _ total _ density profile alone .
the jam models are very accurate and robust , which was confirmed by comparison with the results of @xcite based on hi data , but despite the increase in radial extent when compared to similar earlier studies , they probe the dynamics to accelerations an order of magnitude higher than those typically probed in spiral galaxies .
@xcite recently applied the tracer mass estimator of @xcite to the kinematics of the gc systems for a superset of early - type galaxies , mostly from the sluggs survey @xcite .
this method estimates the enclosed mass from the line - of - sight velocities of the tracer population based on assumptions for the power - law slope of the gravitational potential , the power - law slope of the tracer density profile , and the orbital anisotropy . while being less reliable
, these models allow us to get indications about trends beyond 5@xmath8 , and also for slow rotators .
@xcite analysed the dependence of their mass estimates on these assumptions , took into account corrections for non - sphericity and substructures in the tracer population , and compared their results to previous studies .
their total mass estimates with the assumption of isotropy agree with similar earlier studies within a factor of 1.6 .
we refer the reader for the details to @xcite and adopt their estimates for isotropic orbits .
the sluggs galaxies were selected to be representative early - type galaxies across various environments ( see @xcite ) and span a wide mass range up to ngc 4486 , i.e. m87 . for 23 of the 25 sluggs galaxies @xcite derived estimates for the total mass within @xmath9 and beyond .
most of the galaxies have rich enough gc data sets to derive not only a single dynamical mass estimate , but also radial mass density profiles .
this applies for 10 of the 14 galaxies of our primary sample . beyond the galaxies common to both samples
, the study of @xcite comprises ngc 1400 , ngc 3607 , ngc 4564 , ngc 5866 and also the slow rotators ngc 720 , ngc 1407 , ngc 3608 , ngc 4365 , ngc 4374 , ngc 4486 , and ngc 5846 .
before analysing the mda relation , we need to convert the stellar light to stellar mass , in order to estimate the expected accelerations .
several options are available .
the stellar populations can be fitted with models to infer a stellar mass - to - light ratio .
this requires knowledge of the stellar populations and needs to assume a stellar initial mass function ( imf ) , which is suspected to vary from galaxy to galaxy ( e.g. * ? ? ?
* ; * ? ? ?
* ) , even in the framework of mond @xcite .
the stellar mass - to - light ratio @xmath10 can also be obtained from the dynamical model .
for atlas@xmath11 , the ratio is then based on the assumption of newtonian gravity ( * ? ? ?
* denoted as @xmath12 therein ) . here , we also compute the stellar mass - to - light ratio @xmath13 by fitting the dynamics expected from the light profile based on mond to the observed dynamical profile , and with a spatially constant mass - to - light ratio as fitting parameter .
this fit is dominated by the inner regions where density and flux are highest .
we compare the various mass - to - light ratios in fig .
[ fig : ml ] .
since the mge used in the dynamical modelling came from observations with different photometric filters , we convert all mass - to - light ratios to the @xmath14-band , using the photometric predictions of @xcite and @xcite . the various ways of determining the mass - to - light ratio all show the same trend of an increase with increasing velocity dispersion @xmath7 within @xmath5 ( see also , e.g. , * ? ? ?
* ; * ? ? ?
this trend is due to underlying changes of the stellar population characteristics , e.g. increasing age and metallicity ( and increasing @xmath15 due to changes in the imf ) . in the following we use the mass - to - light ratios from the mond fitting @xmath13 for consistency with the comparisons in section [ sec : mond ] . in the next step
the enclosed mass is calculated , both for the dynamical mass from jam , which is calculated from the density profiles of , and for the stellar mass .
the latter is obtained from the published mges of the galaxy light distribution and the spatially constant stellar mass - to - light ratio @xmath13 as follows .
the mass of an axisymmetric mge model , enclosed within a spherical shell of radius @xmath14 is given by @xmath16 \,\,\
, - \vphantom { { { \exp\left[-r^2/(2\sigma_i^2)\right ] \textrm{erf}\left[r \sqrt{1-q^2_i}/(\sqrt{2}q_i\sigma_i ) \right ] } \bigg/ { \sqrt{1-q^2_i } } } } \right . \\ & \left . { { \exp\left[-r^2/(2\sigma_i^2)\right ] \textrm{erf}\left[r \sqrt{1-q^2_i}/(\sqrt{2}q_i\sigma_i ) \right ] } \bigg/ { \sqrt{1-q^2_i } } } \right \ } , \end{split}\ ] ] with the mge parameters for width , flattening , and total mass ( @xmath17 , @xmath18 , and @xmath19 ) , for each gaussian component respectively .
this equation was obtained by integration of the density profile in terms of the mge as given in footnote 11 of .
in addition to stellar and dynamical mass , the newtonian acceleration due to the baryons is needed for the mda relation , and it is given in the spherical limit by @xmath20 we calculate uncertainties in a monte carlo fashion , by randomly perturbing the density profiles of within their errors ( see sec .
[ sec : mond ] ) .
however , rather than using only formal uncertainties , we try to be conservative and account for systematic uncertainties as follows .
we adopt a 6% uncertainty in the overall mass normalization as inferred by @xcite and an upper limit of 0.11 in the profile slope , derived from the observed scatter in .
therefore , we perturb the profiles by adding a random constant with @xmath21 dex to log@xmath22 , and by adding a linear trend in @xmath23 versus @xmath24 with a random slope ( @xmath25 ) .
we calculate and plot for each galaxy 100 realizations of this process .
the resulting mda relation is plotted for all 14 galaxies in the left - hand panel of fig .
[ fig : mda ] .
it shows that our fast rotators follow a relation similar to that found for spiral galaxies , with the dynamical to stellar mass discrepancy systematically increasing with decreasing acceleration at that radius .
@xcite introduced a modification to newtonian dynamics as an alternative to dark matter in explaining flat rotation curves in spiral galaxies ( for recent reviews see @xcite ; @xcite , and the entire special issue in which it was published ) . in newtonian dynamics
the acceleration caused by a spherical matter distribution within a radius @xmath14 is given by equation ( [ eq : newton ] ) .
in mond the acceleration @xmath26 felt by matter is modified . under the assumption of spherical symmetry the two quantities
are related by @xmath27 with the interpolating function @xmath28 and a characteristic acceleration scale @xmath29 m s@xmath30 . for high accelerations ( @xmath31 ) @xmath32 , i.e. the modification vanishes and the newtonian limit applies .
equation ( [ eq : mu ] ) can be easily inverted @xmath33 with @xmath34 $ ] .
one common choice for the interpolating function @xmath35 , especially on galactic scales , is the _ simple _ interpolating function @xmath36 , so that @xmath37 . the dynamically inferred mass @xmath38 , when assuming newtonian dynamics , is then given by ( see , e.g. , * ? ?
* ) @xmath39 m_*(r ) .
\label{eq : massr}\ ] ] this offers a natural explanation for flat rotation curves in disc galaxies and for the baryonic tully - fisher relation ( see , e.g. , @xcite ) , and it means a newtonian observer concludes the presence of from the mond point of view ` phantom ' dark matter given by @xmath40 . in terms of the mda
this leads to a simple prediction , which is plotted as the black solid curve in fig .
[ fig : mda ] ( see also , e.g. , @xcite for a more theoretical account ) . given the uncertainties the mond curve looks like a rather good representation for our sample .
found the galaxies to have nearly isothermal mass density profiles ( @xmath41 ) , with only small amounts of scatter , which was confirmed by @xcite for a sample of early - type galaxies based on hi data . in fig .
[ fig : profiles ] the stellar density profile , the dynamical mass density profile from , and that predicted from the stellar density profile using mond are shown .
it can be seen that the mond profiles reproduce the slopes of the dynamically determined profiles well .
this is not a surprise , since galaxies have asymptotically isothermal profiles in mond .
however , when compared in more detail , for each galaxy individually , inconsistencies become noticeable .
there are galaxies like ngc 3115 that start to diverge strongly from the mond prediction at the outermost radii , but also galaxies like ngc 4473 and ngc 4494 for which features at smaller radii are not well reproduced .
this becomes especially evident when considering the residual profiles in the right panel , which are calculated individually for each monte carlo representation .
the general reproduction of the mda and the slopes of the density profile by mond is intriguing at first .
we note that this is achieved without fitting for the mond acceleration scale , but using for all galaxies a value of @xmath42 m s@xmath30 , which was found by @xcite for spiral galaxies , and which is still used in more recent studies ( e.g. * ? ? ?
the sensitivity to the exact value is not very critical , as can be seen in the comparison curve in the mda relation for another value @xmath43 m s@xmath30 , which was also used for spiral galaxies ( e.g. @xcite ) .
however , as showed , the dynamical profiles can also be expected to be simple power - laws in the @xmath3cdm context out to about 10 @xmath8 .
furthermore , the profile for individual cases are less consistent as mentioned above .
our analysis makes a few potentially limiting assumptions , such as a spatially constant mass - to - light ratio and anisotropy .
accounted for two individual anisotropies in the inner and outer part . also , for the calculation of the mondian predictions we assumed spherical symmetry , which may seem to be a rather strong assumption given that the galaxies in our sample are often flattened fast rotators . @xcite and @xcite demonstrated that the expected deviations due to non - sphericity effects on the non - linear mond equivalent of poisson equation are however small even in flat spiral galaxies .
the modelling uncertainties are dominated by systematics , which can be important in individual cases .
for this reason significant progress can best be made in a statistical manner using large galaxy samples . with this in mind ,
the following could be considered as a more serious challenge for mond than the mismatches for individual galaxies .
we showed our comparisons with the mond predictions when using the simple interpolating function , since it gives more consistent results for our sample than the _ standard _ one ( cf .
also * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
while changing the mond constant @xmath44 within the range of values used for spiral galaxies only leads to minor changes , switching to the standard interpolating function has more severe effects .
the number of density profiles that are still consistent with mond given our assumptions and estimates of uncertainties is roughly halved as compared to that when using the simple interpolating function . in the mda relation
this is even more evident .
the calculation with the standard interpolating function largely underpredicts the mass discrepancy for our sample ( fig .
[ fig : mda ] ) and marks essentially the lower edge of the trend of observed mda curves , while the mda relation calculated with the simple interpolating function runs through the middle of this trend
. however , the opposite is true for the comparison sample of spirals from @xcite , for which the standard interpolating function provides a superior representation .
this is at odds with mond , where there should be one universal interpolating function and mda relation .
the above is consistent with @xcite s ( @xcite ) results from stellar dynamics .
they concluded that , from their analysis of the mda for a sample of early - type galaxies , the upturn of the mda relation occurs at higher accelerations than for spiral galaxies , in disagreement with mond ( when using accelerations @xmath45 based on the dynamics as in their fig .
19 our fig .
[ fig : mda ] looks similar ) .
their sample included two of our galaxies and also slow rotators .
at that time , @xcite doubted that the change in the mass - to - light ratios , which is what @xcite actually plotted , marked the onset of the mass discrepancy , and pointed out discrepancies between the results of @xcite and the profiles of @xcite for the galaxies common to both studies .
even with the combination of stellar kinematics from atlas@xmath11 and sluggs surveys we only reach radii at which dark matter just starts to be dominant and where the accelerations just decrease to values of the order of the mond constant @xmath44 . using gc kinematics from the sluggs survey , @xcite studied dynamical models for a superset of galaxies reaching beyond @xmath46 .
the conclusions based on these should be considered more tentative , since they are less robust than those from the jam modelling . here , we are interested in the comparison to the mond expectations .
these were obtained by fitting the mond profile to the jam density profile via the ( spatially constant ) mass - to - light ratio as a fitting parameter . in fig .
[ fig : profiles ] , we vertically shift the density profiles of @xcite so that they match the jam profiles in the radial overlap region for comparison to the same mond profiles .
the @xcite profiles are less smooth due to the discrete nature of the tracers , but give nonetheless some indication how the dynamically inferred density profiles in fig .
[ fig : profiles ] continue at larger radii . in some cases , e.g. ngc 821 , the deviations from mond increase , while in other cases , e.g. ngc 3377 , the onset of the deviation appears less critical , since the continuation is close to consistent with mond .
previous studies with dynamical tracers concluded that ngc 821 is mondian to @xmath13.5 @xmath4 , using pne @xcite , while could not reconcile ngc 3115 with mond , similar to our analysis .
the same applies to ngc 4278 , which is in our analysis only marginally inconsistent . for the mda relation , adding the @xcite data is problematic , since their profiles are non - monotonic and ` jumpy ' , again due to the discrete nature of the tracers .
instead , we use the total masses and dark matter fractions at 5 @xmath4 and at @xmath47 from their table a4 .
the resulting dynamical - to - stellar mass ratios and accelerations are shown in the right - hand panel of fig .
[ fig : mda ] , including their 11 additional galaxies .
the sample of fast rotators , which overlaps with our sample of galaxies with stellar dynamics , generally appears to follow the same mda .
the even larger scatter prevents conclusions as to which interpolating function performs better .
however , the gc system data seem also to suggest that the slow rotators have systematically higher dynamical - to - stellar mass ratios , despite the large uncertainties .
this is qualitatively consistent with , who generally found more severe inconsistencies with mond for slow rotators based on binned kinematics of gc systems , also using sluggs data ( see also @xcite for an account of pressure supported dwarf spheroidals within mond ) .
while these tentative gc based conclusions do not share the robustness of those based on the jam models ( which were reinforced by @xcite ) , they suggest an interesting trend of increasing offsets in the mda relation from spiral galaxies to fast rotators to slow rotators . in the framework of mond
, these findings may be explained by unseen ( normal ) matter , which is known to be required in this context on the larger scales of galaxy groups and clusters @xcite . in the above context
it is noteworthy that found for the atlas@xmath11 galaxies with hi data that the baryonic tully - fisher relation has little scatter which would be consistent with the mond framework .
however , @xcite previously found an offset between the baryonic tully - fisher relation for spiral and early - type galaxies .
we analysed the mass discrepancy acceleration ( mda ) relation for early - type galaxies by using the dynamical models for the stellar dynamics of 14 fast rotators of @xcite .
the range , robustness , and accuracy of these models allowed us to determine that , while the galaxies broadly follow such a relation , they are systematically offset from the comparison sample of spiral galaxies @xcite .
this adds to the challenges found for mond when comparing the dynamically determined profiles to the mond predictions for individual galaxies .
meanwhile , the simulations of @xcite demonstrated that the mda for spiral galaxies could arise in @xmath3cdm from variations of the dark matter profile shape with galaxy mass , instead of a universal nfw profile .
our analysis predicts the corresponding models for fast - rotator early types to be offset from the mda for spiral galaxies .
we are grateful for the referee s suggestions , which improved the presentation of our results .
jj and daf thank the arc for financial support via dp130100388 .
mc acknowledges support from a royal society university research fellowship .
lc was supported by the miur grant prin 2010 - 2011 , project ` the chemical and dynamical evolution of the milky way and local group galaxies ' , prot .
2010ly5n2 t .
|
the dark matter ( dm ) haloes around spiral galaxies appear to conspire with their baryonic content : empirically , significant amounts of dm are inferred only below a universal characteristic acceleration scale .
moreover , the discrepancy between the baryonic and dynamical mass , which is usually interpreted as the presence of dm , follows a very tight mass discrepancy acceleration ( mda ) relation .
its universality , and its tightness in spiral galaxies , poses a challenge for the dm interpretation and was used to argue in favour of modified newtonian dynamics ( mond ) . here , we test whether or not this applies to early - type galaxies .
we use the dynamical models of fast - rotator early - type galaxies by cappellari et al . based on atlas@xmath0 and sluggs data , which was the first homogenous study of this kind , reaching @xmath1@xmath2 , where dm begins to dominate the total mass budget .
we find the early - type galaxies to follow an mda relation similar to spiral galaxies , but systematically offset .
also , while the slopes of the mass density profiles inferred from galaxy dynamics show consistency with those expected from their stellar content assuming mond , some profiles of individual galaxies show discrepancies .
[ firstpage ] galaxies : elliptical and lenticular , cd gravitation dark matter
|
relativistic models built with hadronic degrees of freedom have been very successful in describing different properties of nuclei and hadrons at low and intermediate energies ( for comprehensive reviews see refs .
@xcite ) . in some of these models ,
the hadronic lagrangian has symmetries which are inspired by the underlying qcd theory .
this allows one to have fewer parameters , thereby reducing ambiguities in the hadronic models .
one of the first models which incorporated the @xmath7 symmetry was the gauged linear @xmath8 model ( glsm ) developed in ref .
this model was an extension of the linear @xmath8 model and included , in addition to the pion and scalar meson , the vector @xmath9 and axial - vector @xmath10 mesons as gauge bosons of the local @xmath7 symmetry .
the local symmetry was explicitly broken by the vector - meson mass terms , and spontaneous symmetry breaking ( ssb ) by the scalar field led to the mass splitting between the @xmath9 and @xmath11 this model was elaborated in @xcite , where the current - field identities were established .
later the model was applied @xcite in a description of the meson properties . because of some difficulties additional terms are often included .
these terms break further the local symmetry and introduce additional parameters , which allow for a better description of the meson observables @xcite .
the qhd - ii model , which respects the local @xmath12 isospin symmetry , was developed in refs .
it was extended in ref .
@xcite by adding the chiral @xmath13 symmetry .
this model , called qhd - iii , is a chiral invariant theory based on the local symmetry @xmath14 .
the @xmath9 and @xmath10 mesons are included as the gauge bosons which are initially massless .
the masses are generated through ssb and the higgs mechanism .
the lagrangian of qhd - iii includes the lagrangian of glsm and the lagrangian of the higgs fields .
the need for the latter sector of the model was clearly explained in ref .
@xcite : the higgs mechanism in glsm with local symmetry leads to the disappearance of the pion which plays the role of a would - be goldstone boson giving its degree of freedom to the massive @xmath10 meson . therefore the higgs sector serves to generate the masses of the @xmath15 and @xmath16 mesons and to preserve the pion as a physical degree of freedom .
due to the local gauge symmetry the model is renormalizable and does not require the introduction of cut - off parameters .
it is also parity conserving by construction .
a subtle aspect of qhd - iii ( but also of glsm and other hadronic models including the axial - vector meson ) is the presence of a bilinear term mixing the @xmath10 and pion fields .
this considerably complicates the interpretation of the physical particles in the theory as well as calculations with this lagrangian .
one way to get rid of the mixing was considered in @xcite and later used in other papers @xcite .
it consists of a re - definition of the @xmath10 field and subsequent wave - function renormalization of the pion field .
the final lagrangian takes a complicated form with strongly momentum - dependent vertices .
this has undesirable implications in low - energy meson phenomenology . as examples
, we mention the vanishing of the @xmath17 and @xmath18 vertices at some values of the invariant masses of the @xmath10 and @xmath19 , and the difficulties with the @xmath20 vertex @xcite .
the authors of @xcite , instead of re - defining the fields , preferred to sum the self - energy generated by the @xmath21 mixing to all orders .
an alternative method in the framework of qhd - iii was recently suggested in @xcite .
this method exploits the freedom in choosing the gauge due to the local gauge invariance .
originally @xcite the so - called unitary gauge was chosen from the beginning , and the two massless isovector goldstone bosons ( we will call them @xmath22 and @xmath23 ) were gauged away .
this choice leads to the above - mentioned complication with the mixing .
note that the pseudoscalar boson @xmath23 has the same quantum numbers @xmath24 as the pion @xmath25 originating from glsm sector .
therefore the physical pion field can be chosen as a linear combination of @xmath23 and @xmath25 , while the orthogonal combination is decoupled in the unitary gauge @xcite .
the lagrangian then takes a simpler form , without complicated momentum - dependent vertices . in the present paper
we apply this new representation of qhd - iii in the calculation of some meson properties .
first , we include the electromagnetic ( em ) interaction in this model .
this is done via an extension of the symmetry of qhd - iii to the local @xmath26 group .
we use an arbitrary gauge where all eight higgs fields are initially present . the em interaction in both sectors of the model
is obtained .
after an appropriate diagonalization of the @xmath23 and @xmath27 fields and fixing the gauge along the lines of ref .
@xcite , we obtain the em interaction in terms of the physical pion field .
the final lagrangian includes the minimal em interaction , as well as the ( nonminimal ) em interaction with the intrinsic magnetic moment of the @xmath16 and @xmath15 mesons .
we then study the strong and em decays of the vector and axial - vector mesons .
some of the decays can be calculated on the tree - graph " level , while others require a calculation beyond the tree level .
in particular , we calculate the width of the following decays : @xmath28 @xmath29 and @xmath30 we also address the issue of the width of the scalar @xmath8 meson , in view of the interest @xcite in this subject .
the matrix elements for the above decays are given directly by the corresponding vertices in the lagrangian . in order to calculate the decays @xmath31 and @xmath32 we need to include loop diagrams .
in particular , the @xmath33 process is described by a large number of one - loop diagrams , which can be grouped according to the intermediate state in the diagram .
the diagrams where only vector or axial vector mesons in the loop are present are not yet included in this exploratory study .
all matrix elements of the em processes turn out to be finite , due to a cancellation of divergencies between different amplitudes .
the fulfillment of em gauge invariance serves as a check of the calculation .
the decay widths of these processes are listed in the pdg reviews , and we compare the model predictions with experimental values . to determine the parameters of the model we fix the strong - coupling constant @xmath34 from the @xmath35 decay .
all other parameters are strongly correlated , once the masses of the particles are taken equal to their experimental values .
only the mass @xmath36 of the @xmath8 meson and the mass @xmath37 of the higgs particles remain unconstrained .
the mass @xmath37 is taken to infinity in the calculation .
as an additional test of the model we calculate the low - energy parameters for @xmath38 scattering . because of some unusual features of the model , such as the presence of higgs mesons and a suppression of the @xmath39 interaction , it is not _ a priori _ clear whether the model can reasonably describe the experiment .
the scattering lengths and effective ranges for the @xmath40 and @xmath41waves are calculated on the tree level and compared with the data and other approaches .
the paper is organized as follows . in sect .
[ sec : em ] the lagrangians of the em and strong interactions are obtained .
we start with the higgs sector in sect .
[ subsec : em - higgs ] .
the glsm is briefly discussed in sect .
[ subsec : sigma ] . in sect .
[ subsec : removing ] the procedure for removing the mixing terms in the lagrangian is described .
the final em and strong - interaction lagrangians in terms of the physical pion field are presented in sect .
[ subsec : em - physical ] . in sect .
[ subsec : strong decays ] the widths of the strong decay of the mesons are calculated , and in sect .
[ subsec : em decays ] we consider the em decay of the mesons .
results are compared with experiment .
pion - pion scattering at low energies is studied in sect .
[ sec : pi - pi ] . in sect .
[ sec : discussion ] we discuss the results and prospects , and draw conclusions . in appendix
[ app : a ] we outline the derivation of the lagrangian , originating from the higgs sector .
explicit expressions are given for the lagrangian in glsm .
finally , appendix [ app : b ] contains details of the calculation of the one - loop integrals .
in this section we discuss the em interaction in the framework of chiral quantum hadrodynamics ( qhd - iii ) .
the strong lagrangian @xcite consists of the glsm lagrangian @xmath42 and the higgs part , @xmath43 where @xmath42 will be discussed in the next section , and @xmath44 with the potential @xmath45-\mu _
{ h}^{2}(\phi _ { r}{}^{\dagger } \phi _ { r}+\phi _ { l}{}^{\dagger } \phi _ { l } ) .
\label{eq.3}\ ] ] the complex doublets of spinless fields , @xmath46 transform as the spinor representation of @xmath47 .
the covariant derivatives @xmath48are expressed in terms of the right and left isovector gauge fields @xmath49 and @xmath50 the lagrangian is symmetrical under the local @xmath7 gauge transformations ( for more details see @xcite ) . to include the em interaction we extend the model by adding the gauge @xmath51 symmetry of hypercharge .
the method is formally equivalent to that in the theory of glashow , weinberg and salam ( gws ) of electroweak interactions ( see , e.g. , @xcite ch.20.2 , and also @xcite ) .
the hypercharge @xmath52 is assigned to the scalar fields , and the covariant derivatives acting on @xmath53 and @xmath54 take the form @xmath55\phi _ { r / l } , \label{eq.4}\ ] ] where @xmath56 is the em field , @xmath57 and @xmath58 are the strong and em coupling constants .
the fields @xmath59and@xmath60 are associated with the vector ( isovector ) meson @xmath61 , and axial - vector ( isovector ) meson @xmath62 we should also add the free lagrangians of all vector fields @xmath63 where @xmath64 in these expressions we included primes on @xmath65 and @xmath66 , anticipating that these are not yet the fields of the physical photon and @xmath9 meson , but will be redefined . due to the chosen form of the potential @xmath67 in eq.([eq.3 ] ) , the masses of the vector and axial - vector mesons are generated via the higgs mechanism , as suggested in ref .
the fields @xmath53 and @xmath68 acquire a nonzero vacuum expectation value ( vev ) @xmath69 where the value of @xmath70 will be specified later .
we now define the eight higgs fields via @xmath71\left ( \begin{array}{c } 0 \\ 1 \end{array } \right ) .
\label{eq.8}\ ] ] the fields @xmath72 and @xmath22 are scalars , whereas @xmath73 and @xmath74 are pseudoscalars under the parity transformation , @xmath75 so that the fields @xmath53 and @xmath54 satisfy the relations @xmath76 eq.([eq.10 ] ) is the condition that the model is parity conserving @xcite
. the lagrangian can now be rewritten in terms of the fields @xmath77 and @xmath78 we insert eq.([eq.4 ] ) in the lagrangian ( [ eq.2 ] ) and use the representation of eq.([eq.8 ] ) . in the derivation
there appears a mixing between @xmath56 and the 3d component of @xmath79 , which requires a re - definition of these fields .
one can introduce new physical fields ( without primes ) , @xmath80 with the mixing angle defined through @xmath81 and rewrite the lagrangian in terms of the physical fields .
the covariant derivatives now read @xmath82\phi _ { r / l}\ , , \label{eq.13}\ ] ] where we introduced the coupling of the neutral @xmath16 meson , @xmath83 , and the electric charge of the proton , @xmath84 .
the charge operator is given by @xmath85 ; it is seen that it yields zero when acting on the vacuum .
the latter condition is crucial to ensure that the photon does not acquire mass due to ssb . for the physical values of the couplings we have @xmath86 and , up to @xmath87
, we use the substitutions @xmath88 without distinguishing @xmath89 from @xmath34 , and @xmath58 from @xmath90 .
eq.([eq.13 ] ) simplifies correspondingly , to this order .
the derivation of the lagrangian is tedious , and some details are collected in appendix [ app : a ] .
the result can be written as a sum of the em and strong - interaction parts , @xmath91 where the em lagrangian is @xmath92 @xmath93_{3 } + ( \mbox{\boldmath $ \rho$\unboldmath}^{\mu \nu } \times \mbox{\boldmath $ \rho$\unboldmath}_{\nu } + \mathbf{a}^{\mu \nu } \times \mathbf{a}_{\nu } ) _ { 3 } , \label{eq.17}\end{aligned}\ ] ] @xmath94 it is seen that in the arbitrary gauge there is a contribution originating from the fields @xmath95 and @xmath23 .
if they are omitted from the beginning then only the last term in eq.([eq.17 ] ) would remain .
in fact the field @xmath23 may survive even in the unitary gauge ( see sect.[subsec : removing ] ) and contribute to the em current . to clarify this point we need to consider explicitly the second sector of the model
glsm ( sect .
[ subsec : sigma ] ) .
it is also worthwhile to notice the nonminimal em interaction in eq.([eq.18 ] ) , which comes from the free @xmath9-meson lagrangian after making the substitutions of eq.([eq.12 ] ) .
the strong - interaction lagrangian in eq.([eq.14 ] ) has the following structure @xmath96 where @xmath97 and @xmath98 are the free lagrangians of the massless goldstone bosons@xmath99 , the massive higgs bosons @xmath100 , and the gauge bosons @xmath101 .
we have respectively , @xmath102+\frac{1}{2 } [ ( \partial _ { \mu } \zeta ) ^{2}-m_{h}^{2}\zeta ^{2 } ] , \ \label{eq.20 } \\ \mathcal{l}_{\rho
a}^{(0 ) } & = & -\frac{1}{4}\mbox{\boldmath $ \rho$\unboldmath } _ { \mu \nu } ^{2}+\frac{1}{2}m_{\rho } ^{2}\mbox{\boldmath $ \rho$\unboldmath } _ { \mu } ^{2}-\frac{1}{4}\mathbf{a}_{\mu \nu } ^{2}+\frac{1}{2}m_{\rho } ^{2 } \mathbf{a}_{\mu } ^{2}. \label{eq.21}\end{aligned}\ ] ] the expression for the interaction term @xmath103 is complicated and given in eqs.([eq.a - higgs - strong ] ) and ( [ eq.a - v - higgs ] ) of appendix [ app : a ] .
the last term in eq.([eq.19 ] ) describes the ssb induced mixing of @xmath104 with @xmath105 and of @xmath106 with @xmath23 .
this term will be dealt with in sect .
[ subsec : removing ] .
the mass of the @xmath9 and @xmath10 mesons , there would appear mass splitting between the neutral and charged @xmath9 mesons , @xmath107 the mass of the @xmath72 and @xmath73 , the vev @xmath70 , and the parameters of the potential are related via @xmath108 the lagrangian of glsm @xcite can be written in terms of the fields of the nucleon ( @xmath109 ) , pion ( @xmath110 ) , scalar meson ( @xmath111 ) , and vector mesons ( @xmath112 and @xmath106 ) as follows @xmath113n+\frac{1}{2}[(\delta _
{ \mu } \mbox{\boldmath $ \pi$\unboldmath})^{2}+(\delta _ { \mu } \phi ) ^{2}]+\frac{1}{2}m_{\omega } ^{2}\omega _ { \mu } ^{2}-\frac{1}{4}\omega _ { \mu \nu } ^{2}-v(\phi , \mbox{\boldmath $ \pi$\unboldmath})+\mathcal{l}_{sb } , \label{eq.23}\ ] ] where the covariant derivatives acting on the nucleon , pion and scalar fields are defined respectively as @xmath114n , \label{eq.24 } \\
\delta _ { \mu } \mbox{\boldmath $ \pi$\unboldmath } & = & \partial _ { \mu }
\mbox{\boldmath $ \pi$\unboldmath}+g^{\prime}_{\rho } \mbox{\boldmath $ \pi$\unboldmath } \times \mbox{\boldmath $ \rho$\unboldmath}_{\mu } ^{\prime } -g^{\prime}_{\rho } \phi \mathbf{a}_{\mu } , \,\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \delta _ { \mu } \phi = \partial _ { \mu } \phi + g^{\prime}_{\rho } \mbox{\boldmath $ \pi$\unboldmath } \mathbf{a}_{\mu } , \;\;\ \ \ \ \ \ \ \;\;\ \ \label{eq.25}\end{aligned}\ ] ] and the kinetic energy of the @xmath115 meson is expressed through the tensor @xmath116the potential energy term is @xmath117 note that there are no mass terms for the nucleon , @xmath9 and @xmath10 mesons , whereas a mass term is present for the isoscalar @xmath118 .
this lagrangian is invariant under local @xmath7 transformations , apart from a possible explicit symmetry - breaking term @xmath119 generating the pion mass .
the em interaction is included by changing @xmath120 @xmath121to @xmath122 , where the nucleon hypercharge @xmath123 is taken equal to unity .
we also have to make the substitutions of eq.([eq.12 ] ) .
the covariant derivative for the nucleon , in the order @xmath87 , takes the form @xmath124n\ , , \label{eq.27}\ ] ] and the electric charge of the nucleon is @xmath125 in accordance with the gell - mann - nishijima relation .
the nucleon mass is generated via the ssb , if @xmath126 and @xmath127 in eq.([eq.26 ] ) .
the scalar field acquires a nonzero vev @xmath128 and after redefining the sigma field via @xmath129 we obtain the following lagrangian of the em interaction @xmath130 with the em current @xmath131_{3}. \label{eq.29}\ ] ] the strong - interaction lagrangian can be written in the following form @xmath132 where the free lagrangian @xmath133 of the nucleon , pion , sigma and omega reads @xmath134
+ \frac{1}{2}[(\partial _ { \mu }
\sigma ) ^{2}-m_{\sigma } ^{2}\sigma ^{2}]-\frac{1 } { 4}\omega _ { \mu \nu } ^{2}+\frac{1}{2}m_{\omega } ^{2}\omega _ { \mu } ^{2}. \label{eq.31}\ ] ] the interaction @xmath135 is not needed in this section , and is given in eq.([eq.a - n - strong ] ) of appendix [ app : a ] .
the third term on the right in eq.([eq.30 ] ) arises due to the nonzero vev of the scalar field @xmath111 .
it gives an additional contribution to the mass of the @xmath10 meson @xmath136 the last term in eq.([eq.30 ] ) mixes the pion field with the axial - meson field .
what remains to be specified are the relations between the masses of the nucleon , sigma , pion , and the parameters of the potential .
they read as follows @xmath137 the lagrangians obtained so far are still not complete .
they contain bilinear terms which mix different fields , namely @xmath106 and @xmath138 @xmath106 and @xmath139 @xmath140 and @xmath141 to remove these terms we will follow the method of @xcite , with some variations . collecting the mixing terms from eq.([eq.19 ] ) and eq.([eq.30 ] )
one gets @xmath142 where we dropped a full divergence in the first line and used the following definition @xmath143 of the new fields @xmath144 and @xmath145 the mixing angle @xmath146 is determined by @xmath147 _
i.e. _ by the ratio of the vev s of the scalar fields in the two sectors of the lagrangian
. the transformation ( [ eq.c34 ] ) leaves the sum of the kinetic terms invariant , ( @xmath148(@xmath149 the mixing terms in eq.([eq.c33 ] ) can now be removed by adding the gauge - fixing term @xmath150 similarly to the procedure fixing the so - called @xmath151 gauge in gauge theories ( @xcite , ch.21 ) . for the sum we obtain @xmath152 which shows that @xmath144 and @xmath22 are fictitious fields with masses @xmath153 and @xmath154 respectively .
these fields do not contribute to physical processes because their contribution is always canceled by the @xmath155-dependent part of the @xmath156 or @xmath157 propagator @xcite ( ch.21.1 ) .
we will choose the unitary gauge @xmath158 in which @xmath144 and @xmath22 completely decouple and the propagator of the vector meson takes the form @xmath159 .
in this gauge @xmath144 ( @xmath22 ) provides a longitudinal degree of freedom to the massive @xmath10 ( @xmath9 ) meson .
, otherwise the pion mass term in eq.([eq.31 ] ) also contributes to the mass of the @xmath160 the pion can be given mass after the transformation to the new fields is done and the gauge is chosen . ] setting @xmath161 in eq.([eq.c34 ] ) gives @xmath162 and @xmath163 .
the latter formulas have been obtained in @xcite in a slightly different way .
it is convenient to use the notation @xmath164 then in all formulas of the previous sections we just have to set @xmath165 and make the replacements @xmath166 in terms of the physical pion field @xmath167 .
it is seen , in particular , that choosing @xmath168 from the beginning leads to a different lagrangian .
now we are in a position to write the total em interaction .
for the sake of brevity we omit from now on the `` tilde '' on the pion field and , after the substitutions of eq.([eq.c36 ] ) are made , use the notation @xmath169 for the physical pion .
the current arising from the higgs sector reads @xmath170_{3}+ ( \mbox{\boldmath $ \rho$\unboldmath}^{\mu \nu } \times \mbox{\boldmath $ \rho$\unboldmath}_{\nu } + \mathbf{a}^{\mu \nu } \times \mathbf{a}_{\nu } ) _ { 3 } , \label{eq.c38}\end{aligned}\ ] ] while the contribution from the @xmath171 sector reads @xmath172_{3 } .
\label{eq.c39}\ ] ] adding these currents , and noticing that @xmath173 we obtain the total em current @xmath174_{3}. \label{eq.c40}\end{aligned}\ ] ] the nonminimal em interaction remains the same as in eq.([eq.18 ] ) .
it describes the interaction with the intrinsic magnetic moment of the @xmath9 and @xmath10 mesons , which is equal to one in this model .
the gyromagnetic ratio for the @xmath9 ( @xmath10 ) turns out to be @xmath175 in units of @xmath176 ( @xmath177 ) .
this is analogous to the nonminimal em interaction in gws theory and in qhd - ii @xcite .
eq.([eq.c40 ] ) is one of the important results of the paper .
it shows the following features .
the pion em current is restored to its original form ( the current of the free pion ) . due to a cancellation between the currents , the term proportional to ( @xmath178 disappears .
therefore there is no @xmath179 interaction on the tree level . as a result of the diagonalization in eq.([eq.11 ] )
the @xmath180 meson does not couple directly to the photon , so there is no explicit vector - meson dominance of the em interaction .
the em lagrangian includes the 3-field interactions @xmath181 as well as the 4-field interaction pieces @xmath182 and @xmath183 the latter vertices are important for the em gauge invariance of the amplitudes which will be calculated in sect .
[ sec : meson decays ] . for completeness
we present the strong - interaction lagrangian which follows from eqs.([eq.19 ] ) and ( [ eq.30 ] ) , @xmath184 @xmath185n \nonumber \\ & & -\lambda ( \sigma ^{2}+x_{\pi } \mbox{\boldmath $ \pi$\unboldmath } ^{2})[v\sigma + \frac{1}{4}(\sigma ^{2 } + x_{\pi } \mbox{\boldmath$\pi$\unboldmath}^{2 } ) ] -\frac{1}{2}g_{\rho } ( 1+x_{\pi } ) \mbox{\boldmath $ \rho$\unboldmath}_{\mu } ( \mbox{\boldmath $ \pi$\unboldmath}\times \partial ^{\mu } \mbox{\boldmath $ \pi$\unboldmath } ) \nonumber \\ & & + g_{\rho } \sqrt{x_{\pi } } \mathbf{a}_{\mu } ( \mbox{\boldmath $ \pi$\unboldmath}\partial ^{\mu } \sigma -\sigma \partial ^{\mu } \mbox{\boldmath $ \pi$\unboldmath } ) + \frac{1}{2}g_{\rho } ^{2}[x_{\pi } ( \mbox{\boldmath $ \pi$\unboldmath } \mathbf{a}_{\mu } ) ^{2}+(\sqrt{x_{\pi } } \mbox{\boldmath $ \pi$\unboldmath } \times \mbox{\boldmath $ \rho$\unboldmath } _ { \mu } -\sigma \mathbf{a}_{\mu})^{2 } \nonumber \\ & & -2v\mathbf{a}_{\mu } ( \sqrt{x_{\pi } } \mbox{\boldmath $ \pi$\unboldmath } \times \mbox{\boldmath $ \rho$\unboldmath}^{\mu } -\sigma \mathbf{a}^{\mu } ) ] , \label{eq.c42}\end{aligned}\ ] ] @xmath186 + 4 \mbox{\boldmath $ \rho$\unboldmath}_{\mu } \mathbf{a}^{\mu } ( u+\eta ) \zeta \ } \nonumber \\ & & + \frac{1}{2}g_{\rho } \sqrt{1-x_{\pi } } [ \mbox{\boldmath $ \rho$\unboldmath } _
{ \mu } ( \zeta \partial ^{\mu } \mbox{\boldmath $ \pi $ \unboldmath}- \mbox{\boldmath $ \pi $ \unboldmath}\partial ^{\mu } \zeta ) + \mathbf{a}_{\mu } ( \eta \partial ^{\mu } \mbox{\boldmath $ \pi $ \unboldmath}- \mbox{\boldmath$\pi $ \unboldmath}\partial ^{\mu } \eta ) ] \nonumber \\ & & -\frac{\lambda _ { h}}{32}[\eta ^{4}+\zeta ^{4}+4u\eta ^{3}+6\eta \zeta ^{2}(2u+\eta ) + ( 1-x_{\pi } ) ^{2}\mbox{\boldmath $ \pi $ \unboldmath}^{4 } \nonumber \\ & & + 2(1-x_{\pi } ) \mbox{\boldmath $ \pi$\unboldmath}^{2 } ( \eta ^{2}+\zeta ^{2}+2u\eta ) ] .
\label{eq.c43}\end{aligned}\ ] ] the lagrangian @xmath187 is given in eq.([eq.21 ] ) , where now we have to take the mass of the @xmath10 meson from eq.([eq.31bis ] ) .
although the em current and strong - interaction lagrangian look somewhat complicated , they contain only simple vertices with at most one derivative .
this greatly simplifies practical calculations .
it is seen from eq.([eq.c42 ] ) that , apart from the @xmath20 coupling , the strength of the coupling to the pion is scaled down by a factor @xmath188 for each pion field operator . at the same time the coupling to the pion from the higgs lagrangian in eq.([eq.c43 ] ) is scaled by a factor @xmath189 .
it also follows that the @xmath16-meson coupling to the pion , @xmath190 , is not equal to the @xmath16 coupling to the nucleon , @xmath191 .
so in this model the @xmath9 does not couple universally to the hadrons . the presence of the higgs fields @xmath192 and @xmath73 may seem as an obstacle .
however , as was argued in ref .
@xcite , these fields serve as regulators in the calculation of loop integrals . by taking the mass @xmath37 very large
the higgs contributions can be suppressed in many cases .
we will study this aspect while calculating meson decays and low - energy pion - pion scattering in the next sections .
it is interesting to note that the em current in eq.([eq.c40 ] ) can be derived in a simpler way , directly from the strong - interaction lagrangian in eq.([eq.c41 ] ) .
indeed , the minimal substitution @xmath193 ( @xmath194 is the charge operator ) applied to all charged fields leads to eq.([eq.c40 ] ) . at the same time the nonminimal em interaction in eq.([eq.18 ] ) can not be obtained in this way .
the em current satisfies the relation @xmath195 where @xmath196 is the 3d component of the conserved isospin current @xmath197 , and @xmath198 is the conserved baryon current .
the latter is @xmath199 , while the expression for the former is given by @xmath200 conservation of the isospin current is a consequence of the symmetry of the strong - interaction lagrangian in eqs.([eq.c41]-[eq.c43 ] ) with respect to the _ global _
@xmath12 isospin transformations .
indeed , it can be readily verified that @xmath201 in eq.([eq.c45 ] ) is the corresponding noether current .
it also follows from eqs.([eq.c41]-[eq.c43 ] ) that the @xmath9 meson is coupled to a source current @xmath202 which is in general different from @xmath203 the source current is closely related to a current corresponding to the underlying _ local _
@xmath204 symmetry , which is `` hidden '' .
this symmetry may be classified according to ref .
@xcite as a symmetry of the 2nd kind .
a more detailed study of this aspect is beyond the scope of the present paper . for a related discussion in qhd - ii
see appendix d of ref .
first we consider the decay of the mesons which can be obtained on the tree level .
these are the decays @xmath205 @xmath206 and @xmath207 we will need the general expression for the width of the decay @xmath208 in the rest frame of the decaying particle with the mass @xmath209 and spin @xmath210 @xmath211 where @xmath212 and @xmath213 are the corresponding 4-momenta , such that @xmath214 the 3-momentum of the particles in the final state is @xmath215 ( @xmath216 ) and the sum runs over the polarizations of all particles .
the decay width for the @xmath217 will be discussed below .
we first fix the parameters of the model . in general ,
the coupling constants ( @xmath218 the parameters of the potentials ( @xmath219 and all the masses can be considered as parameters .
there are however many relations between these parameters : eq.([eq.22 ] ) , eq.([eq.31bis ] ) and eq.([eq.32 ] ) .
simple considerations show that if we choose the masses of the nucleon , pion , rho , @xmath220 and omega equal to their experimental values , then there remain only four free parameters : @xmath221 the poorly known sigma mass @xmath222 and the unknown mass @xmath223 of the higgs particles .
we will fix the coupling @xmath34 from the @xmath35 decay width .
it is seen from eq.([eq.c42 ] ) that the @xmath35 decay is determined by the matrix element @xmath224 .
the polarization vector of the @xmath9 is denoted by @xmath225 , and latin indices label the charge states of the mesons . from the experimental width
150.2 mev one finds @xmath226 taking @xmath2271.23 gev @xcite , we obtain @xmath228 @xmath229 8.49 , @xmath230 mev , @xmath231 mev , and @xmath232 curiously enough , the ratio @xmath233 appears to be 1.437 , which is close to a factor @xmath234 ( with a deviation of 1.6% ) .
it follows that @xmath235 and @xmath236 the decay @xmath237 is governed by the matrix element @xmath238 where @xmath239
[ @xmath240 is the polarization vector of the @xmath10 @xmath241 meson .
the @xmath242 vertex is simpler than that in glsm @xcite , or in the massive yang - mills approach @xcite .
moreover it does not vanish for any invariant mass of the @xmath243 the calculated width of 272 mev can be compared with the experimental estimate 150 to 361 mev @xcite . in general , this decay is characterized by the two amplitudes , @xmath244 and @xmath245defined through @xmath246 .
those in turn define the s- and d - wave amplitudes @xcite @xmath247,\;\ \ \ \ \ \ \ \ f_{d}=-\frac{\sqrt{8\pi } } { 3m_{\rho } } [ ( \varepsilon _ { \rho } -m_{\rho } ) f+\mathbf{p}^{2}m_{a}g]\ , , \label{eq.47}\ ] ] where @xmath248 is the energy of the @xmath9 meson in the final state@xmath249 since @xmath250 in the present model we obtain the d / s ratio @xmath251% , which agrees in sign and order of magnitude with the experimental ratio @xmath2521.6% @xcite .
the @xmath29 matrix element , according to eq.([eq.c42 ] ) , is @xmath253 the corresponding calculated width comes out 46 mev and the branching ratio is @xmath254 \approx$ ] 14% . from the branching ratio given by pdg we can estimate the corresponding width as 32 to 147 mev .
of course this process is not very well defined , in view of the uncertain status of the @xmath8 meson .
we used the value 770 mev for the mass of the sigma .
next we address the issue of the width of the @xmath8 meson .
this subject has been discussed extensively in @xcite , mainly because in the linear @xmath8 model the width is too large , and sometimes larger than its mass , which makes it difficult to identify the @xmath8 with a particle state . in our model
the matrix element of the @xmath255 decay is given by @xmath256 calculation yields @xmath257 mev for @xmath258 mev@xmath259 and @xmath260 mev for @xmath261 gev .
if we had used - as is appropriate in the linear @xmath8 model - the values @xmath262 and @xmath263 , where @xmath264 mev is the pion weak - decay constant , then we would indeed have obtained a very large width of @xmath265 gev ( @xmath266 gev ) for @xmath267 mev ( @xmath261 gev ) .
the width , however , is reduced considerably due to the factor @xmath268 in the @xmath18 vertex , and to a lesser extent due to the difference between @xmath269 and @xmath270 we should also mention that the vertex , because of its simple structure , does not vanish for any values of the invariant mass of the @xmath271 the vanishing of the vertex is an undesirable feature of glsm , as was pointed out in ref .
@xcite .
let us consider the em decay @xmath273 , which can be described by the tree - level amplitude shown on fig .
[ fig:1 ] .
the matrix element can be written , using eq.([eq.c40 ] ) and eq.([eq.c42 ] ) , as @xmath274 where @xmath275 the last term comes from the @xmath276 vertex in eq.([eq.c40 ] ) .
the total amplitude is gauge invariant , @xmath277 calculation of the decay width involves integration over invariant masses of the pairs of particles in the final state , @xmath278 the limits in the integral over @xmath279 are @xmath280 where @xmath281 and @xmath282 are respectively the photon momentum , pion energy and pion momentum in the rest frame of the @xmath283 system , @xmath284 the sum and average over polarizations is most easily evaluated in the system where the oz axis is along the photon 3-momentum @xmath285 .
we obtain @xmath286 \right\ } , \label{eq.51}\end{aligned}\ ] ] where @xmath287 are the components of the pion momenta orthogonal to the oz axis , _
e.g. _ @xmath288 .
the decay @xmath289 includes the bremsstrahlung from the charged pions and is infrared divergent at small photon energies .
experimentally a cut - off is introduced while measuring the decay width , namely @xmath290 this means that in the integral in eq.([eq.49 ] ) the invariant mass ( squared ) of the @xmath283 pair has an upper limit @xmath291 .
the value of the integral in eq.([eq.49 ] ) depends strongly on @xmath292 for @xmath293 mev we obtain @xmath294 mev , while for @xmath295 mev we get @xmath296 mev .
the pdg review @xcite gives the value 1.487@xmath297 mev for the photon energies above 50 mev .
diagrams for the @xmath298 decay.,width=566 ] next we consider the em decay @xmath300 there is no contribution to the matrix element on the tree level , and we have to include at least the one - loop processes shown in fig .
[ fig:2 ] ( see also table [ tab : diagrams ] ) .
diagrams for @xmath301 decay at the one - loop level .
the intermediate propagators labeled @xmath302 , @xmath303 , @xmath304 refer to particles as defined in table [ tab : diagrams ] .
depending on the intermediate state @xmath305 , the contact diagrams have the structure @xmath306 for @xmath307 and the structure @xmath308 for @xmath309 . for @xmath310
there is no contact diagram.,width=566 ] .[tab : diagrams ] intermediate particles in the loop diagrams indicated in fig .
[ fig:2 ] . [ cols="^,^,^,^,^,^,^,^,^,^,^",options="header " , ] it is seen from table [ tab : pi - pi ] that the scattering length @xmath311 is described fairly well .
the other parameters , however , are overpredicted by a factor of 1.5@xmath3122 .
the @xmath9 meson gives the dominant contribution ( compare the 2nd and 4th rows ) , while the contribution of the higgs meson is small ( see the 2nd and 3d rows ) . in the 6th row
we show parameters obtained with the soft - pion amplitude @xmath313 from ref .
@xcite , which is based on pcac and current commutation relations .
our amplitude will reduce to @xmath314 if @xmath315 and @xmath316 we also show results of the chpt calculations : @xcite , where only pions are included and @xcite , where resonances are added .
comparison with the experiment indicates that the effect of the @xmath9 meson is overemphasized in the present model . in this connection
we compare the @xmath9-exchange amplitude in eq.([eq.78 ] ) with the corresponding amplitude by bernard _
[ eq.(3.14 ) in ref .
one of the differences is that in @xcite the propagator of the vector meson ( to be exact , the part contributing for on - shell pions ) is modified according to @xmath317 which coincides with weinberg s suggestion @xcite based on chiral - symmetry arguments .
such a modification strongly reduces the effect of the @xmath9 meson at low energies . in the present model
however there is no compensating term of the form @xcite @xmath318 which would lead to eq.([eq.79 ] ) . using the @xmath319 interaction with more derivatives , as is advocated in ref .
@xcite , would also be inconsistent with the lagrangian ( [ eq.c42 ] ) .
besides , the calculation shows that the effect of the @xmath9 meson can not be eliminated completely as would follow from eq.([eq.79 ] ) , since the @xmath8 contribution in itself is by far too small .
one of the mechanisms which can partly reduce the @xmath9 contribution is the dependence of the @xmath20 vertex on the invariant mass of the @xmath320 one may assume that @xmath321 with a form factor normalized to unity at @xmath322 the typical form often used in phenomenological models is @xmath323 where @xmath324 is a cut - off parameter . choosing the value @xmath325 gev @xcite we obtain a considerable reduction of the low - energy parameters ( see table [ tab : pi - pi ] , the 5th row ) .
of course this is only one plausible argument which may explain the unsatisfactory description of low - energy @xmath283 scattering on the tree level . besides the form factor
, other higher - order corrections to the amplitude would have to be consistently included .
this issue will be studied elsewhere .
in the previous sections we have presented the results of calculations in the framework of a new representation of qhd - iii @xcite .
an advantage of this new form , compared to the original formulation @xcite , is that only simple vertices with at most one derivative appear in the lagrangian . as a result
the calculations are simpler , and more importantly , the strong - interaction vertices describing the decay of the mesons do not vanish at any values of the meson invariant mass .
a comparison with experiment ( table [ tab : widths ] ) shows that most of the decay widths are reasonably described .
this was not anticipated in view of the fact that practically no free parameters were used in the calculations .
in fact , only the coupling @xmath34 was fixed from the @xmath35 decay .
the masses of the nucleon , pion , rho and @xmath10 mesons were taken from the latest pdg review @xcite .
the mass of the @xmath8 meson was chosen equal to the mass of the @xmath326 _ i.e. _ @xmath327 mev , in line with ref .
@xcite where the @xmath8 is supposed to be degenerate with the @xmath16 . with this mass
the calculated width for the @xmath328 decay , 46 mev , is surprisingly close to the value predicted in @xcite : @xmath329 mev . in some calculations we use the value @xmath330 1 gev . as a result
the widths of the @xmath328 and @xmath331 decays change considerably due to a change of the available phase space .
the other observables are not sensitive to the @xmath8 mass .
calculations were also performed with the value @xmath332 gev , which is based on different arguments @xcite . however , agreement with experiment is better with the pdg value @xmath333 gev . as a curious observation
we mention that this value approximately obeys the relation @xmath334 within a 3% accuracy .
this relation also implies that @xmath335 _
i.e. _ the ratio of the @xmath9 couplings to the nucleon and the pion , is very close to @xmath336 the em decay @xmath289 is described by the two - step tree - level amplitude supplemented by a contact diagram .
the latter comes from the @xmath276 vertex in the em current ( [ eq.c39 ] ) and guarantees the em gauge invariance .
the decay is mainly determined by the @xmath20 coupling and is not sensitive to other ingredients of the model .
the decay @xmath337 is a more informative process . as there is no direct @xmath338 coupling the process
is described by many one - loop diagrams which strongly interfere .
the total amplitude of @xmath339 comes out finite despite the divergence of the separate diagrams .
since the lagrangian was obtained in the unitary gauge , the propagator of the vector mesons includes the longitudinal component @xmath340 the latter gives rise to a quadratic divergence , @xmath341 , but due to the em gauge invariance these divergences from different diagrams cancel .
furthermore , the logarithmic divergences , @xmath342 , cancel as well .
the higgs mesons , @xmath72 and @xmath343 , play the role of auxiliary particles in this model . according to ref .
@xcite these mesons serve as regulators and should have minimal effect on the low - energy predictions .
they do not appear in the initial and final states , but may contribute as intermediate particles .
for instance , the @xmath72 and @xmath73 appear in one - loop diagrams for the @xmath33 decay . in the diagrams with intermediate @xmath344 state , the @xmath345 coupling in eq.([eq.c43 ] )
increases proportionally to @xmath346 however , due to the presence of the @xmath347 propagator this contribution stays finite .
the situation is different in the diagrams with an intermediate @xmath348 and @xmath349 , where the vertices @xmath350 and @xmath351 are independent of @xmath352 .
nevertheless the amplitudes do not vanish in the limit @xmath353 , as one would naively expect by taking the limit of the propagator in the loop integrand .
this is because the longitudinal component of the vector meson propagator leads to divergent integrals ; the correct procedure is to take the limit @xmath354 after the loop integration , and leads to a nonzero contribution . at @xmath353
the total amplitude corresponding to the above processes involving higgs mesons takes a form equivalent to an effective @xmath355 lagrangian .
numerically its contribution to the @xmath356 decay turns out to be relatively small . in processes where @xmath72 and @xmath343 appear on the tree level
the amplitude may diverge at @xmath357 , if there are more @xmath345 vertices than the @xmath347 propagators . using the @xmath283 scattering as an example we demonstrated that this is not the case . in this reaction
there is the @xmath347-exchange diagram which at first sight behaves as @xmath358 .
however there is a compensating @xmath359 vertex in the lagrangian ( [ eq.c43 ] ) .
the couplings in the @xmath72 exchange and @xmath360 contact diagrams are tuned in such a way that the sum remains finite .
this mechanism of cancellation is similar to that in the linear sigma model , in which the amplitude given by the @xmath8 exchange and the @xmath359 vertex does not diverge in the limit @xmath361 , but rather takes the value dictated by the soft - pion amplitude @xcite .
the contribution of the @xmath72 exchange and the associated @xmath359 term at low energies is small . the pion interaction with hadrons in the model
is scaled down by the factor @xmath362 ( with the exception of the @xmath319 , @xmath363 and @xmath364 vertices ) .
this influences the matrix elements of the meson decays , and in most cases improves agreement with experiment .
the factor @xmath365 also leads to a reduction of the @xmath366 decay width .
for example , compared to the linear sigma model the width is reduced by almost an order of magnitude .
therefore the width comes out smaller than the @xmath8 mass [ of the order of 150 ( 350 ) mev for @xmath3670.77 ( 1.0 ) gev ] which may be helpful in identifying the @xmath8 with the scalar - isoscalar state around 1 gev @xcite . at the same time
the effect of the @xmath8 in @xmath283 scattering is also diminished .
our calculation shows that the @xmath8 contribution to the low - energy parameters becomes very small , and the dominant contribution comes from the @xmath9 exchange .
the agreement between the calculation and experiment is not very impressive compared to , for example , chpt calculations . the scattering length in the @xmath368 channel
is described quite well , but the other calculated parameters overestimate the experiment .
this deficiency may be related to the tree - level approximation .
we included a form factor in the @xmath369 vertex , similarly to what is often done in phenomenological models .
this is one of the effects which contribute beyond the tree level . in this way
the @xmath9 contribution is reduced and the agreement is improved , but other higher - order corrections need to be consistently taken into account before definite conclusions can be drawn . in this paper we focused on the meson properties and left out the nucleon sector .
inclusion of the latter can also be an important test of the model , especially because the nucleon - pion vertex is reduced . at this point
we would like to mention that we did not include an additional baryon , the cascade @xmath370 with the hypercharge @xmath371 .
this isodoublet was added in @xcite to cure the problem with the chiral anomaly and render the theory renormalizable . the chiral anomaly in qhd - iii
will show up when the isoscalar @xmath115 meson couples to the baryon loop with other two isovector vertices , one of which contains @xmath372 .
such processes were not considered here .
the anomaly may however be important in the em decay @xmath373 considered in sect .
[ subsubsec : rho - pi - gam ] . if we added the loop with the cascade then the calculated width would change , though we did not include this effect .
it seems logical first to extend the @xmath7 model to the strange sector , as for example , was done in ref .
@xcite for glsm .
the extension to @xmath374 would allow one to include along with @xmath375 the other strange hadrons , like @xmath376 and @xmath377 in conclusion , we applied chiral quantum hadrodynamics ( qhd - iii ) @xcite in the calculation of some properties of the mesons .
first we included electromagnetic interaction by extending the symmetry to the local @xmath26 group .
this allowed us to obtain consistently the minimal and nonminimal contributions to the electromagnetic interaction in an arbitrary gauge .
after an appropriate diagonalization and fixing the gauge the lagrangian is obtained in terms of the physical pion field .
we calculated the strong and em decays of the vector and axial - vector mesons , @xmath237 , @xmath378 , @xmath379 , @xmath339 , @xmath373 , and addressed the issue of the width of the @xmath8 meson . for the @xmath380 decay
some loop diagrams are not yet included , and a more complete analysis is reserved for future work .
most of the calculated decay widths are in reasonable agreement with experiment @xcite .
the only free parameter used in calculations is the mass of the @xmath8 meson , although most of the results were not sensitive to @xmath381 we studied the effect of auxiliary higgs bosons of qhd - iii in the @xmath33 decay and @xmath283 scattering .
the contribution of these particles to the amplitude , both on the tree level and in the one - loop diagrams , turns out to be finite and small in the limit @xmath382 .
this goes in line with the viewpoint of ref .
@xcite on the role of @xmath72 and @xmath73 mesons in low - energy hadron physics .
our exploratory study shows that qhd - iii in the representation of ref.@xcite can describe some features of meson phenomenology in the non - strange sector .
there are many interesting issues which can be further addressed , such as an extension of calculations to the baryon sector ( @xmath383 scattering and nucleon form factors ) , a clarification of the role of the cascade @xmath375 and inclusion of the other strange hadrons .
finally , applications to the many - body sector , i.e. nuclear matter and finite nuclei , may be considered . as was shown in ref .
@xcite ( see also @xcite , sect .
3a ) , the linear sigma model , when applied to finite nuclei on the mean - field level , has some deficiencies .
the present model is much richer and differs in several respects , such as the presence of additional particles and vertices , and a considerable suppression of the pseudo - scalar pion - nucleon coupling .
detailed calculation will help to assess the applicability of the model to the many - body sector .
we would like to thank gary przeau for clarifying communication regarding ref .
@xcite , and olaf scholten for useful discussions .
this work was supported by the fund for scientific research - flanders ( fwo - vlaanderen ) and the research board of ghent university .
the most complicated part of the higgs lagrangian are the terms with the covariant derivatives . for the right field
, for example , such a term can be written in the form @xmath384\left ( \begin{array}{c } 0 \\ 1 \end{array } \right ) \
, , \label{eq.a1}\ ] ] where the operators @xmath385 and @xmath386 are defined as @xmath387 and the notation @xmath388 is used .
the similar term for the field @xmath54 can be obtained from the above formulas by changing @xmath389 @xmath390 and @xmath391 the potential @xmath392 in eq.([eq.3 ] ) is chosen such that @xmath393 and @xmath394 thus allowing for ssb of the global gauge invariance @xcite . from the condition that the term linear in @xmath72 must be absent , the vev @xmath70 can be found .
the fields @xmath72 and @xmath73 acquire a mass @xmath37 , whereas @xmath22 and @xmath23 remain massless , indicating that these are the goldstone bosons .
calculation of the matrix elements in eq.([eq.a1 ] ) leads to the em current in eq.([eq.16 ] ) and ( [ eq.17 ] ) .
the strong lagrangian consists of the free lagrangians in eqs.([eq.20]-[eq.21 ] ) , the mixing term in eq.([eq.19 ] ) , and the interaction lagrangian @xmath395 \ } \nonumber \\ & & + \frac{1}{2}g_{\rho } [ \mbox{\boldmath $ \rho$\unboldmath}_{\mu } ( \eta \partial ^{\mu } \mathbf{h}-\mathbf{h}\partial ^{\mu } \eta + \zeta \partial ^{\mu } \mathbf{z}-\mathbf{z}\partial ^{\mu } \zeta -\mathbf{h}\times \partial ^{\mu } \mathbf{h}-\mathbf{z}\times \partial ^{\mu } \mathbf{z})+\mathbf{a } _ { \mu } ( \eta \partial ^{\mu } \mathbf{z}-\mathbf{z}\partial ^{\mu } \eta \nonumber \\ & & + \zeta \partial ^{\mu } \mathbf{h}-\mathbf{h}\partial ^{\mu } \zeta -\mathbf { h}\times \partial ^{\mu } \mathbf{z}-\mathbf{z}\times \partial ^{\mu } \mathbf { h})]-v_{h}(\eta , \zeta , \mathbf{h , z})\ , , \label{eq.a - higgs - strong}\end{aligned}\ ] ] where the remaining piece of the potential is @xmath396\ , .
\label{eq.a - v - higgs}\end{aligned}\ ] ] eqs.([eq.a - higgs - strong]-[eq.a - v - higgs ] ) will reduce to the lagrangian of qhd - iii @xcite if we take @xmath397 the interaction lagrangian of glsm from sect .
[ subsec : sigma ] follows from eqs.([eq.23 ] ) - ( [ eq.26 ] ) after we define the @xmath19 field through @xmath398 , where the vev @xmath269 can be fixed from the minimum of the potential , @xmath399 .
the lagrangian has the form @xmath400n
\nonumber \\ & & + g_{\rho } [ \mathbf{a}_{\mu } ( \mbox{\boldmath $ \pi$\unboldmath}\partial ^{\mu } \sigma -\sigma \partial ^{\mu } \mbox{\boldmath $ \pi$\unboldmath})- % \mbox{\boldmath $ \rho$\unboldmath}_{\mu } ( \mbox{\boldmath $ \pi$\unboldmath } \times \partial ^{\mu } \mbox{\boldmath $ \pi$\unboldmath } ) -g_{\rho } va_{\mu } ( \mbox{\boldmath $ \pi$\unboldmath}\times \mbox{\boldmath $ \rho$\unboldmath}^{\mu } -\sigma \mathbf{a}^{\mu } ) ] \nonumber \\ & & + \frac{1}{2}g_{\rho}^{2}[(\mathbf{a}_{\mu } \mbox{\boldmath $ \pi$\unboldmath})^{2 } + ( \mbox{\boldmath $ \pi$\unboldmath}\times \mbox{\boldmath $ \rho$\unboldmath}_{\mu } -\sigma \mathbf{a}_{\mu } ) ^{2 } ] -\lambda ( \sigma ^{2}+%
\mbox{\boldmath $ \pi$\unboldmath}^{2})[v\sigma + \frac{1}{4}(\sigma ^{2}+% \mbox{\boldmath $ \pi$\unboldmath}^{2})]\ , .
\label{eq.a - n - strong}\end{aligned}\ ] ]
let us consider the amplitudes for @xmath33 decay with an intermediate @xmath401 state , corresponding to diagrams in fig .
[ fig:2 ] with @xmath402 .
we introduce the following notation : @xmath403 with @xmath404 and @xmath405 the amplitude labeled @xmath406 in fig .
[ fig:2 ] has an integrand proportional to @xmath407 \ , .
\label{eq.b2}\end{aligned}\ ] ] the terms proportional to @xmath408 and @xmath409 are omitted hereafter because of the relations @xmath410 and @xmath411 the diagram @xmath306 in fig .
[ fig:2 ] has an integrand @xmath412 for the diagram @xmath413 with photon bremsstrahlung off the @xmath10 meson we obtain after some algebra , @xmath414 .
\label{eq.b4}\end{aligned}\ ] ] similarly we find for the diagram @xmath415 in fig . [ fig:2 ] with bremsstrahlung off the pion , @xmath416 in the chosen unitary gauge the propagator of the vector meson includes three space - like polarization states .
we notice that the most divergent terms proportional to @xmath417 , which come from the longitudinal component of the @xmath9 propagator , cancel .
the term @xmath418 that remains in eq.([eq.b4 ] ) is equal to zero after integration over @xmath419 .
further , it is straightforward to check that the sum of the four diagrams does not depend on the photon gauge , _
i.e. _ @xmath420 to evaluate the integrals we apply the method of dimensional regularization ( see , e.g. @xcite , app.a ) . using the feynman parametrization and integrating over @xmath419 in @xmath421-dimensional space - time
we obtain @xmath422\right\ } \ , , \label{eq.b6}\end{aligned}\ ] ] where the constant @xmath423 reads @xmath424 .
the expressions for @xmath425 will be specified below .
we kept in eq.([eq.b6 ] ) only the @xmath426 part because of the condition @xmath427 ( in the rest frame of @xmath10 ) .
since the gamma - function @xmath428 has a pole at @xmath429 we have to verify that the expression in the curly brackets vanishes at @xmath430 using the expansions near @xmath429 @xcite : @xmath431 and @xmath432 , we find that the pole term drops out because of the relation @xmath433=0\ , .
\label{eq.b7}\ ] ] the residue gives the amplitude @xmath434\mbox{\rm d}% x\right\ } , \label{eq.b8}\ ] ] where the integration variables @xmath435 have been changed to @xmath436 defined such that @xmath437 in the above formulas @xmath438 is the euler - mascheroni constant and @xmath439 the argument of the logarithm can formally be made dimensionless by changing @xmath440 where @xmath324 is a mass - scale parameter .
this will not affect the amplitude , due to ( [ eq.b7 ] ) .
expression ( [ eq.b8 ] ) can be further simplified by carrying out the integration over @xmath441 the @xmath442-loop amplitudes [ @xmath310 in fig .
[ fig:2 ] ] require a trace calculation , for example , @xmath443}{d_{k}\ d_{k+q}d_{k+p } } \nonumber \\ & \longrightarrow & 4m_{n}\left [ \frac{g^{\mu \nu } q\cdot ( 2k+q)-(2k+q)^{\mu } q^{\nu } } { d_{k}\ d_{k+q}d_{k+p}}-\frac{g^{\mu \nu } } { d_{k}\ d_{k+p}}\right ] , \label{eq.b10}\end{aligned}\ ] ] where @xmath444and@xmath445 .
the @xmath413 amplitude in fig .
[ fig:2 ] reads @xmath446 this amplitude cancels the divergent and em gauge noninvariant piece of @xmath447 finally , the amplitude @xmath448 ( radiation from the pion line ) is zero , being proportional to @xmath449 an important observation is the cancellation of divergent terms between different diagrams .
this , together with the em gauge invariance , helps in evaluating the other diagrams in fig .
[ fig:2 ] .
most of these diagrams are calculated similarly to @xmath450 while others , where the photon couples to a vector meson in the loop , are more algebraically involved . after this general consideration
we present expressions for the amplitudes corresponding to the diagrams in fig .
[ fig:2 ] with @xmath451 . for the @xmath452 loops ( @xmath453 )
we obtain @xmath454\ \mbox{\rm d}x\ } , \label{eq.b12 } \\
t_{2 } & = & c_{2}\{\frac{1}{2}+\int_{0}^{1}\log \frac{\delta _ { 2}(x)}{\delta _ { 3}(x)}[x-3+\frac{1}{x}(\frac{\delta _ { 3}(x)}{m^{2}}+2)]\ \mbox{\rm d}x\ } , \label{eq.b13 } \\
\delta _ { 2}(x ) & = & m_{\pi } ^{2}x+m_{\rho } ^{2}(1-x)-m_{a}^{2}x(1-x),\,\ \ \ \ \ \ \ \ \ \ \delta _ { 3}(x)=m_{\pi } ^{2}x^{2}+m_{\rho } ^{2}(1-x ) , \nonumber \\ c_{1 } & = & c_{2}=(4\pi ) ^{-2}\frac{1}{2}eg_{\rho } ^{3}v\sqrt{x_{\pi } } ( 1+x_{\pi } ) , \nonumber\end{aligned}\ ] ] and we introduced the notation @xmath455 since @xmath456 these amplitudes acquire an imaginary part if @xmath457 and we have to select the proper branch of the logarithm . recalling the prescription @xmath458 for the masses of the particles in the propagators , we have correspondingly @xmath459 and we can use the substitution @xmath460^{1/2}\}/2m_{a}^{2 } , \nonumber\end{aligned}\ ] ] in eq.([eq.b12 ] ) and ( [ eq.b13 ] ) to calculate the real and imaginary parts . here
@xmath461 if@xmath462 and @xmath463 otherwise . for the @xmath464 contribution
we obtain @xmath465\ \mbox{\rm d}x\ } , \label{eq.b15}\\ \delta _ { 2}(x ) & = & m_{\pi } ^{2}x+m_{\sigma } ^{2}(1-x)-m_{a}^{2}x(1-x),\,\ \ \ \ \ \ \ \ \ \ \ \ \ \ \delta _ { 3}(x)=m_{\pi } ^{2}x^{2}+m_{\sigma } ^{2}(1-x ) , \nonumber \\ c_{3 } & = & ( 4\pi ) ^{-2}2eg_{\rho } \lambda v\sqrt{x_{\pi } } x_{\pi } , \,\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ 2\lambda = ( m_{\sigma } ^{2}-m_{\pi } ^{2})/v^{2}\ , .
\nonumber\end{aligned}\ ] ] a substitution similar to eq.([eq.b14 ] ) is applied to calculate the real and imaginary part of @xmath466 next , for the diagrams with @xmath467 in fig .
[ fig:2 ] the amplitude is real and reads , @xmath468\ \mbox{\rm d}x\ } , \label{eq.b16}\\ \delta _ { 2}(x ) & = & m_{\sigma } ^{2}x+m_{a}^{2}(1-x)^{2},\,\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \delta _ { 3}(x)=-m_{\pi } ^{2}x(1-x)+m_{\sigma } ^{2}x+m_{a}^{2}(1-x ) , \nonumber \\ \ c_{4 } & = & ( 4\pi ) ^{-2}2eg_{\rho } ^{3}v\sqrt{x_{\pi } } , \,\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \delta = ( m_{\sigma } ^{2}-m_{\pi } ^{2})/m_{a}^{2}\ . \nonumber\end{aligned}\ ] ] the contribution @xmath469 in fig .
[ fig:2 ] with an intermediate higgs meson @xmath72 ( @xmath470 state ) is @xmath471\ \mbox{\rm d}x\},\ , \label{eq.b17 } \\
\delta _ { 2}(x ) & = & m_{\pi } ^{2}x+m_{h}^{2}(1-x)-m_{a}^{2}x(1-x),\,\ \ \ \ \ \ \ \ \ \ \ \ \delta _ { 3}(x)=m_{\pi } ^{2}x^{2}+m_{h}^{2}(1-x ) , \nonumber \\ c_{5 } & = & -(4\pi ) ^{-2}\frac{1}{4}e\lambda _ { h}m_{\rho } \sqrt{1-x_{\pi } } ( 1-x_{\pi } ) , \ , \nonumber\end{aligned}\ ] ] the amplitudes for @xmath472 ( @xmath473 state ) and @xmath474 ( @xmath475 state ) can be written in the form @xmath476\ \mbox{\rm d}x\ } , \label{eq.b18}\ ] ] where @xmath477 for @xmath478 and @xmath479 for @xmath480 in the limit of large higgs mass , @xmath481 the sum of the amplitudes with the intermediate @xmath72 and @xmath73 mesons can be written as @xmath482 this amplitude corresponds to a contact - like lagrangian in eq.([eq.54 ] ) of sect .
[ subsubsec : a - pi - gam ] . finally , the amplitude for @xmath310 ( @xmath483 state ) on fig . [ fig:2 ] has the form @xmath484 \frac{(1 - 2x)\mbox{\rm d}x}{1-x}\ , \label{eq.b20 } \\
c_{8 } & = & ( 4\pi ) ^{-2}2eg_{\rho } g_{\pi } m_{n}\sqrt{x_{\pi } } \ . \nonumber\end{aligned}\ ] ] the amplitude on fig .
[ fig:3 ] describing the decay @xmath485 is calculated similarly .
evaluating the trace we obtain @xmath486}{d_{k}\ d_{k+q}d_{k+p } } \nonumber \\ & \longrightarrow & 4m_{n}\,\varepsilon ^{\mu \nu \lambda \sigma } q_{\lambda } q_{\sigma } \frac{1}{d_{k}\ d_{k+q}d_{k+p}}. \label{eq.b21}\end{aligned}\ ] ] the result of the integration over @xmath419 is given in eq.([eq.70 ] ) of sect .
[ subsubsec : rho - pi - gam ] .
|
the chiral invariant qhd - iii model of serot and walecka is applied in the calculation of some meson properties . the electromagnetic interaction is included by extending the symmetry of the model to the local @xmath0 group .
the minimal and nonminimal contributions to the electromagnetic lagrangian are obtained in a new representation of qhd - iii .
strong decays of the axial - vector meson , @xmath1 , and the electromagnetic decays @xmath2 and @xmath3 are calculated .
the low - energy parameters for the @xmath4 scattering are calculated in the tree - level approximation .
the effect of the auxiliary higgs bosons , introduced in qhd - iii in order to generate masses of the vector and axial - vector mesons via the higgs mechanism , is studied as well .
this is done on the tree level for @xmath5 scattering and on the level of one - loop diagrams for the @xmath6 decay .
it is demonstrated that the model successfully describes some features of meson phenomenology in the non - strange sector .
|
in optical pumping , absorption of the near - resonance circularly polarized light could lead to large population imbalances in atomic ground states and induce the polarization of the atoms@xcite .
a variety of applications , from traditional studies on nuclear resonance gyroscope@xcite , atomic clock , optical pumped magnetometer to the recently developed experiments in atomic spin gyroscope , have use the electron - spin polarization for precise measurements . among them
, the recently developed serf magnetometer features a higher sensitivity and a much more portable package , compared to the superconducting quantum interference devices ( squids ) , thus provides a better way to measure magnetocardiography ( mcg ) and magnetoencephalography ( meg ) signals . with a sufficient high cell temperature and weakly shielded environment , such that @xmath0 ( @xmath1 is the time between spin - exchange collisions and @xmath2 is the spin precession rate ) , the spin - exchange collisions would not broaden the magnetic resonance linewidth and it indicates the realization of the serf regime@xcite . as the operation principle of the sensors mentioned above relies on the spin polarization , the value of the polarization would affect the performance of these sensors .
thus , it is necessary to monitor the polarization of the alkali - metal atoms . a general experimental technique by turning the pump laser beam on and off has been used to calculate the spin polarization@xcite . by applying an oscillating magnetic field ,
the transverse electronic paramagnetic resonance ( epr ) spectroscopy is another efficient method to obtain the population status .
however , both of these processes would introduce an extra perturbation into the sensor proper and affect its normal operation .
the motivation of this article is to develop a non - perturbative scheme to acquire the polarization of the electron - spin while minimizing the introduction of perturbation to the system . by investigating the relationship between the electron - spin polarization and the transmission of the circularly polarized pump laser beam ,
here we propose a new scheme , which only relies on the pump laser intensity , to extract the electron - spin polarization .
compared to the previous methods , this technique has the advantage that no extra disturbance is introduced into the sensor . besides , this scheme could achieve a real - time observation and requires less components .
we experimentally validate the method in a serf magnetometer and the results are in excellent agrement with the theory .
the model could deep our understanding of the absorption process of the pump laser and facilitate the optimization of the polarization in the atomic sensors .
the electron - spin polarization of the alkali - metal atoms along the @xmath3 direction ( pumping axis ) @xmath4 is defined by@xcite @xmath5 where @xmath6 is the optical pumping rate due to the circularly polarized pump laser , @xmath7 is the spin - destruction rate due to collisions with other alkali atoms , buffer gas atoms and quenching gas molecules@xcite .
while the atoms illuminated by the pump laser are polarized , those atoms outside the illumination region still could be polarized by the diffusion effects .
and the diffusion length in one relaxation time is @xmath8 , where @xmath9 is the diffusion coefficient@xcite . under typical experimental condition of the cs magnetometer , where @xmath10 1@xmath11s and @xmath12 mm@xmath13s , the corresponding diffusion length is about 0.36 mm @xcite . for a monochromatic laser
, the optical pumping rate could be described by @xmath14 where @xmath15 determined by the photo absorption cross - section as well as the waist of the laser beam is defined as the pumping rate per unit laser intensity , @xmath3 is the position in the cell and @xmath16 is the corresponding laser intensity . inserting eq.([eq : pumpingrate ] ) into eq.([eq : electronpolrization ] ) , we arrive at the following expression for the spin polarizatioin as a function of the pump laser beam : @xmath17 where the coefficient @xmath18 is defined as @xmath19 .
note that eq.([eq : electronpolrizationreplace ] ) gives us the relationship between the electron - spin polarization and the laser intensity .
if the value of @xmath18 is known , then eq.([eq : electronpolrizationreplace ] ) provides a useful method to measure the atomic spin polarization . as the pump beam with circular polarization propagates through the vapor cell , the the populations of the atoms are redistributed and the reduction in the laser intensity @xmath16
is given by@xcite @xmath20 where @xmath21 is the density of the alkali vapor , @xmath22 is the absorption cross - section determined by the atomic frequency response about the resonance frequency @xmath23 .
eq.([eq : reductioninintensity ] ) could be solved via the lambert @xmath24-function@xcite , and the solution is expressed as @xmath25 where @xmath26 satisfies the function @xmath27 , @xmath28 and @xmath29 represent the incident laser intensity and the corresponding pumping rate . to facilitate the data fitting , we rewrite eq.([eq : transmittedintensity ] ) in a much more simplified form , @xmath30 where
@xmath31 is replaced by @xmath18 and @xmath32 is the optical depth of the cell . the transmitted laser intensity , @xmath16 , the incident laser intensity , @xmath28 , could be measured directly .
the value of @xmath18 as well as the optical depth of the cell , @xmath33 , are the free parameters and could be obtained from the fit based on eq.([eq : transmittedintensitysimple ] ) . substituting the values of @xmath18 and the transmitted pump beam intensity @xmath16 , which could be monitored continuously , into eq.([eq : electronpolrizationreplace ] )
, we would obtain the corresponding electron - spin polarization in real time .
note that eqs.([eq : electronpolrizationreplace ] ) and ( [ eq : transmittedintensitysimple ] ) only rely on the intensity of the pump laser , which is the essential part of the devices mentioned above , thus this scheme could be conducted without adding any additional component . meanwhile ,
no perturbation has been introduced into the sensor proper .
a magnetic field @xmath34 , which is applied perpendicular to the magnetic moment arising from the polarized atoms , will cause a torque on the atom and induce the larmor precession of the atom .
the precession of the spin polarization @xmath4 will result in a projection of on the @xmath35 axis , indicated as @xmath36 .
for the ultra - sensitive serf magnetometry , the component @xmath36 is generally detected by a linear polarized probe beam and the magnetic field response could be expressed as @xcite , @xmath37 where @xmath36 is the @xmath35 projection of the spin polarization @xmath4 and @xmath38 is the gyromagnetic ratio of a bare electron .
@xmath36 could be detected via the optical rotation of a linearly polarized probe beam , given by @xcite @xmath39 } \label{eq : rotationangle}\ ] ] where @xmath40 is the path length of the probe beam through the cell , @xmath41 m is the classical electron radius ( @xmath42 and @xmath43 are the electric charge and the mass of the electron ) , @xmath44 is the speed of the light , @xmath45 is the oscillation strength ( roughly @xmath46 for @xmath47 transition and @xmath48 for @xmath49 transition ) , @xmath50 is optical resonance lineshape of the probe beam with voigt profile , @xmath23 is the frequency of the probe light and @xmath51 is the resonance frequency of the transition @xmath52 , @xmath53 is the strength of the optical rotation arising from the individual hyperfine resonance @xmath52 and @xmath54 .
a step in the @xmath34 field with an amplitude of 0.05 nt has been applied at each pump laser intensity and the response should be maximized when the spin polarization is equal to 50@xmath55 according to eq.([eq : magsignal ] ) .
the configuration of the experimental setup used is depicted in fig.[fig : experimentalsetup ] , and is centered around a spherical vapor cell with a diameter of 12 mm .
the vapor cell contains a droplet of cs metal , 50 torr of n@xmath56 gas and 500 torr of @xmath57he gas .
the n@xmath56 prevents radiation trapping , while @xmath57he was included to reduce the diffusion effect to the cell walls .
it was held in an oven heated to 100 @xmath58c with twisted - pair wires driven by ac current at 50 khz .
a set of four cylindrical layers of @xmath59-metal magnetic shield enclosed the cell and provides a shielding factor of @xmath60 to quasi - static magnetic fields .
the vapor cell was optically pumped by a circularly polarized pump beam originating from a distributed feedback ( dfb ) laser .
the pump laser , with a spectral width of about 2 mhz , was tuned to the cs @xmath61 resonance line and directed along the @xmath3 axis . according to eq.([eq : transmittedintensity ] ) , the transmitted laser intensity @xmath16 would vary with the different incident laser intensity . the incident and the transmitted laser intensities
were recorded by two photodiodes ( pd1 and pd2 ) , simultaneously .
the probe laser beam , formed by another dfb system , is detuned by about 0.3 nm to the red side of the @xmath62 resonance line and propagates in the @xmath35 axis .
the first calcite polarizer ( lp2 ) determines the initial linear polarization axis at 0@xmath58 .
faraday rotation from the atoms rotates the polarization plane of the probe beam by an angle @xmath63 according to eq.([eq : rotationangle ] ) .
then , a quarter waveplate ( qw2 ) with one of its axes nominally aligned with the calcite polarizer contributes a circular component proportional to @xmath63 .
a final calcite polarizer ( lp3 ) , which acts as an analyzer and is crossed with the original polarizer , selects the time varying intensity of the transmitted light .
a photo - elastic modulator ( pem ) following the quarter waveplate is used to modulate the polarization of probe laser in order to improve the s / n ( the ratio of signal to noise ) . and the modulation axis of the pem is oriented at 45@xmath58 with respect to the transmission axis of the polarizer .
finally , the probe beam transmitted through the cell is detected by a photodiode ( pd3 ) and the time - varying intensity on pd3 is given by@xcite @xmath64 where @xmath65 is the incident intensity of the probe beam , @xmath66 rad and @xmath67hz are the depth and the frequency of the pem modulation .
note that the time - varying intensity of the probe beam @xmath68 is definitely different from the distance - dependent intensity of the pump beam @xmath16 .
the photocurrent is amplified by a low - noise transimpedance amplifier and demodulated by a commercial computer - controlled digital lock - in amplifier ( zurich instruments , model hf2li ) referenced to the pem . via detecting the optical rotation of the polarization plane
, the residual magnetic fields could be further canceled actively by three orthogonal coils inside the shields in order to realize the serf regime@xcite . under the serf regime , the spin relaxation arising from the spin - exchange collisions is completely eliminated and the spin - destruction collisions dominate the absorption of the pump laser .
c and the red line overlapping the data points is the fitting curve according to eq.([eq : transmittedintensitysimple ] ) , from which the value of @xmath18 is deduced .
the inset shows magnification of the transmission curves .
note that the linearly polarized components of the pump laser beam would be completely absorbed by the hot atoms and thus contribute to the further reduction of the pump laser intensity . in our experiment , the degree of the pump beam polarization is approximately 0.9 and this factor should be included in the curve fitting process based on eq .
( [ eq : transmittedintensitysimple ] ) . ] ) , while the theoretical prediction according to eq.([eq : electronpolrization ] ) is marked by the red line . ]
axis ( @xmath69 ) at different pump laser intensity .
the blue line shows the unmodified model based on eq.([eq : rotationangle ] ) while the red dashed line is the fitting curve according to the modified version of eq.([eq : rotationangle ] ) , where the diffusion term described by eq.([eq : diffusionsig ] ) should be included . ] in fig.[fig : incidenttransmittedintensity ] , we plot the transmitted laser intensity as a function of the incident laser intensity at room temperature and 100 @xmath58c .
the density of the vapor cell at room temperature ( @xmath70 ) is more than two orders less than that at 100 @xmath58c ( @xmath71 ) , and the absorption due to the atoms at room temperature is ignorable .
as the blue triangles shown in fig.[fig : incidenttransmittedintensity ] , the transmitted laser intensity is proportional to the incident laser intensity and the reduction of the laser intensity reflects the transmission of the system components , including the uncoated optical glass windows , focus lenses as well as the vapor cell made from aluminosilicate glass . compared to the results obtained at room temperature , the data points at 100 @xmath58c show nonlinear absorption when the incident intensity is less than about 1 mw ( see inset in fig.[fig : incidenttransmittedintensity ] ) , in which the value of @xmath18 could be extracted form the fitting according to eq.([eq : transmittedintensitysimple ] ) . however , at high laser intensity the transmitted laser intensity @xmath16 scales linearly with @xmath28 , which implies that the atoms are almost fully polarized . utilizing the coefficient @xmath18 obtained above and the incident laser intensity
, we could achieve the polarization measurements of the electron - spin using eq.([eq : electronpolrizationreplace ] ) .
the electron - spin polarization of cs atoms as a function of the incident laser intensity is shown in fig.[fig : polarizationsignal ] .
the measured values based on eq.([eq : electronpolrizationreplace ] ) agree very well with the prediction described by eq.([eq : electronpolrization ] ) . the dependence of the magnetic response on the electron polarization has also been studied .
the magnetometer responses contained in eq.([eq : rotationmeasure ] ) are extracted by the lock - in amplifier and are shown in fig.[fig : magneticresponsesignal ] .
the magnetometer signal increase with the incident laser intensity in the weak intensity range of @xmath72 mw and the system achieves the maximum value when the polarization is nearly @xmath73 , shown in fig.[fig : magneticresponsesignal ] .
however , the values at hight intensity deviate from the theory , which we suspect it is the result of the diffusion effects and the model described by eq.([eq : magsignal ] ) need to be modified .
the polarization of the atoms outside the illumination region degrades as the distance @xmath35 from the illumination region increases , @xmath74 where the initial polarization @xmath75 .
the polarization of the atoms outside the illumination region also contributes to the final response of the magnetic field . to get the corresponding contribution from @xmath76 to the rotation signal similar to eq.([eq : magsignal ] ) , we need to know the @xmath35 projection arising from the polarized atoms in the diffusion region .
it could be obtained by integrating @xmath76 over the diffusion length , given by @xmath77 where the term associated with the optical pumping rate has been dropped out compared to eq.([eq : magsignal ] ) .
the rotation angle arising from eq.([eq : diffusionsig ] ) should be accounted for and added to eq.([eq : rotationangle ] ) .
as the red dashed line shown in fig.[fig : magneticresponsesignal ] , the fitting curve based on the modified model fit the data better . as the contribution from eq.([eq : diffusionsig ] ) depends on the species of the alkali - metal atoms and the noble gases as well as the cell pressure , thus the total optimal polarization would deviate from 50@xmath55 and tends to become greater .
note that the specific optimal value would not be a constant and changes with the cell parameters of the system .
in summary , we have studied the relationship between the electron - spin polarization and the intensity of the circularly polarized pump laser , based on which a scheme has been proposed to extract the electron - spin polarization . compared to the traditional methods to obtain the polarization , this technique only relying on the transmitted pump laser intensity operates in a non - perturbative mode and has the advantage that no extra disturbance is introduced into the sensor .
we therefore anticipate that the scheme described here could detect real - time polarization without disturbing the normal operation of the sensor .
a comparison of the experimental results with theoretical calculations is also presented .
the experimental results are in excellent agreement with the theoretical prediction and verify the validity of the scheme proposed here .
we also investigated the corresponding magnetic field response of the serf magnetometer as a function of the electron - spin polarization . and
a modified model , where we has added the diffusion term to the original model , has been reported to explain the discrepancy of the response between the experimental data and the theory . as the atomic spin polarization can be derived from the intensity of transmitted pump laser beam
, the transmitted intensity could be used to maintain the stability of the pump beam intensity and thus keep the electron spin polarization stable .
this strategy allows to reduce the number of components in the pumping systems and results in a simplification of the system structure . from our perspective ,
our results give new insight into the laser absorption progress for many optical pumping systems , and especially useful for the magnetometer or comagnetometer operated in the serf regime .
s. knappe , p. schwindt , v. shah , l. hollberg , j. kitching , l. liew , and j. moreland , opt .
express * 13 * , 1249 ( 2005 ) .
j. c. allred , r. n. lyman , t. w. kornack , and m. v. romalis , phys .
lett . * 89 * , 130801 ( 2002 ) .
k. kim , s. begus , h. xia , s .- k .
lee , v. jazbinsek , z. trontelj , and m. v. romalis , `` multi - channel atomic magnetometer for magnetoencephalography : a configuration study , '' neuroimage * 89 * , 143 - 151 ( 2014 ) .
|
in the optical pumping systems based on the pump - probe arrangement , the spin polarization of the atoms is generally monitored utilizing the probe laser beam , in which way an extra perturbation must be introduced and thus affects the normal operation of the sensors . by investigating the absorption rate of the circularly polarized pump laser , here we demonstrate the feasibility of extracting the electron - spin polarization from the transmitted pump laser intensity .
we experimentally validate the method in a spin - exchange relaxation free ( serf ) magnetometer and the results are in excellent agreement with the theory .
the scheme operates in a non - perturbative mode and features a real - time observation .
we also study the corresponding magnetic field response of the serf magnetometer and a term arising from the diffusion effects has been added to the original model to explain the discrepancy of the response .
|
the m81 group contains three major closely interacting galaxies : m81 , m82 and ngc 3077 .
studies of this triple system at radio wavelengths have shown an extremely disturbed hi distribution with tidal bridges connecting the three galaxies ( e.g. gottesman & weliachew 1975 ; van der hulst 1979 ; appleton , davies & stephenson 1981 ; yun , ho & lo 1994 ) . within the tidal bridges ,
hi knots with optical counterparts are seen .
the most prominent of these to the east / north - east of m81 are holmberg ix and arp s loop or a0952@xmath569 ( arp 1965 ) .
it has been suggested ( e.g. boyce et al . 2001 ) that these are forming tidal dwarf galaxies .
numerical simulations of the system by yun ( 1999 ) can successfully reproduce the the hi tidal debris if the closest approaches of m82 and ngc 3077 to m81 took place 220 and 280 myr ago respectively .
arp s loop , the topic of this paper , occurs at the intersection of the three tidal streamers in the simulation of yun ( 1999 ) who gives an hi mass of @xmath6 m@xmath7 .
sun et al .
( 2005 ) present very deep wide - field optical continuum images of the m81/m82 galaxy group , and show that groups of young stars are distributed across the entire region between the two galaxies .
although tidal tails usually have blue optical colors ( e.g. schombert et al .
1990 ) , it is not clear whether these colors are due to tidally - disrupted stellar disk material or active star formation in the tails .
recently , ultraviolet ( uv ) imaging obtained with the galaxy evolution explorer ( galex ) satellite has shown that regions of bright uv emission , containing recently formed stars , correlate well with regions of high hi column density in tidal tails ( e.g. hibbard et al . 2005 ; neff et al .
these findings suggest that the combination of uv and hi data provides a powerful technique for identifying and studying star formation in the intergalactic medium ( igm ) between interacting galaxies .
this topic is of importance for earlier epochs in the universe because galaxy interactions , and thus tidal tails were more common , leading to a higher star formation rate in the igm .
this may well have implications for the enrichment of the igm ; ryan - weber et al .
( 2004 ) have shown that stars forming in the igm offer a viable alternative to galactic wind enrichment on large scales .
we have thus started a study of star formation in the igm of the m81 galaxy group with the overall aim of elucidating the importance of this mode of star formation .
we present results for arp s loop in this paper and for holmberg ix in sabbi et al .
arp s loop is located 17 arcmin north - east of m81 and its composition was first discussed by efremov et al .
they resolved the ring - like structure into blue diffuse and star - like objects which they interpreted as young star clusters and associations .
makarova et al . ( 2002 ) investigated the stellar population of arp s loop using photometry obtained with the wide field and planetary camera 2 ( wfpc2 ) on board the hubble space telescope ( hst ) .
they were unable to clarify the evolutionary nature of arp s loop because of photometric limitations and wfpc2 s small field of view .
they find evidence for : enhanced star formation between 20 and 200 myr ago , coincident with the likely epoch of tidal interactions in the m81 group ; intermediate age stars which were either formed prior to the interaction in the arp s loop or were torn out of the massive galaxies during the interaction ; and also faint stars with ages @xmath8 8 gyr but which were too close to the detection limit for detailed analysis . in this paper , we use multi - wavelength observations , including deep photometry obtained with the advanced camera for surveys ( acs ) , to resolve the stellar population , and thus determine the evolutionary status of arp s loop .
the m81 group was observed with the galaxy evolution explorer ( galex ) mission in the far ( fuv : @xmath9@xmath10=1528 ; @xmath11@xmath9@xmath12=269 ) and near ultraviolet ( nuv : @xmath9@xmath10=2271 ; @xmath11@xmath9@xmath13=616 ) .
we obtained the data from the multimission archive at stsci ( mast ) as part of the team general releases .
the galex field of view is [email protected] and [email protected] in the fuv and nuv bands , and the pixel scale is 1.5@xmath15/pixel .
the exposure times were 3093s for each band .
fluxes were calculated using the equations of morrissey et al .
( 2005 ) : m@xmath16 log[f@xmath17/a@xmath17 ] + b@xmath17 , where a@xmath12 = 1.4 @xmath19 10@xmath20 erg s@xmath21 @xmath22 @xmath21 , a@xmath13=2.06@xmath19 10@xmath23 erg s@xmath21 @xmath22 @xmath21 , b@xmath12=18.82 and b@xmath13=20.08 for fuv and nuv , respectively .
fluxes were multiplied by the effective filter bandpass ( @xmath11@xmath9@xmath12=269 and @xmath11@xmath9@xmath13=616 ) to give units of erg s@xmath21 @xmath22 and luminosities were calculated for a distance of 3.6 mpc ( freedman et al .
1994 ) . figure [ m81m82hi ] shows the distribution of the atomic hydrogen in the m81 group of galaxies ( yun , ho & lo 1994 ) on top of the galex fuv image . we have identified 8 fuv objects ( fig .
[ nuv_feet ] , table 1 ) inside the hi cloud located on the northeast side of the bridge between m81 and m82 and north of holmberg ix .
this region , called concentration ii by yun et al . , is the brightest hi complex in the tidal bridge .
it includes arp s loop , and may not be a single cohesive feature , containing tidally stripped gas from m81 and m82 .
makarova et al . ( 2002 ) found 250 stars in the hst / wfpc2 f606w / f814w images of arp s loop .
however , the wfpc2 observations were centered on the hi concentration and cover only objects # 7 and # 8 , missing the other fuv objects .
we searched the hst archive at mast and found new pointings of the same area but taken with the wide field channel ( wfc ) of acs using the f475w and f814w filters and covering a larger field of view ( go 10915 : pi dalcanton ) .
[ acs_feet ] shows the acs f475w image where we have plotted the wfpc2 footprint , the galex contours , and marked the 8 fuv objects .
the high resolution of acs ( 0.05@xmath15 ) resolves the fuv objects into hundreds of starlike objects as can be seen in fig .
[ blobszoom ] .
however , the acs pointing misses two of the fuv objects ( # 3 and # 4 ) and part of another object ( # 1 ) .
we have also obtained an h@xmath0 image of arp s loop with the optic imager at the wiyn 3.5 m telescope .
the observations were taken with wiyn filter # 13 ( @xmath11@xmath9 = 47 ) and a standard johnson @xmath24 filter for continuum subtraction .
optic is a mosaic of two 2kx4k orthorgonal transfer ccds , resulting in a field of view of 9.5@[email protected]@xmath25 . in fig .
[ wiyn ] we present only an image of chip # 2 which was bias - corrected and flat - fielded using iraf the two images taken in each filter were combined with cosmic - ray rejection , and the resulting scaled @xmath24-band image was subtracted from the combined h@xmath0 image to remove the continuum signal .
our data confirm karachentsev & kaisin s ( 2007 ) detection of objects # 1 , # 8 , # 7 , and # 3 , with # 1 being the brightest and # 3 the faintest object ( see fig.1 in karachentsev & kaisin ) . as shown in fig .
[ wiyn ] , although we have detected the faintest object , we have not detected two of their h@xmath0 sources which are marked as squares in fig .
this could be due to the different bandwidths , since karachentsev & kaisin used a broader filter ( @xmath11@xmath9 = 75 ) than ours .
detections of the fuv sources suggest that these are small hii regions where hot , blue ob stars are emitting ultraviolet radiation .
in order to investigate the stellar content of the fuv objects , we have performed photometry on the two acs / wfc images . the total exposure times for the fully reduced and multidrizzled images are 2250 s in the f475w ( @xmath26 ) band and 2265 s in the f814w ( @xmath27 ) band respectively .
each image covers an area of @xmath28 , which corresponds to @xmath29 for the distance of m81 .
the images were reduced using the standard stsci acs pipeline calacs ( pavlovsky et al .
2004 ) to remove bias , dark current and flat field response signatures .
the dithered images were combined using the multidrizzle package , which removes cosmic rays , and corrects the images for geometric distortion .
photometry was performed on the drizzled images using the daophot package within iraf .
stars were automatically detected with daofind in the f475w band , with the detection threshold set at 4 @xmath30 above the local background level ; the detections were then forced on the @xmath31 band image .
the instrumental magnitude of each object in each filter was estimated via a psf fitting technique .
a spatially variable psf was computed for both the f475w and f814w images , using @xmath32 isolated stars in different positions on the images .
selection criteria based on the shape of the objects have been applied to the photometric catalog to remove as many spurious detections , blended and/or extended objects as possible .
we have considered the daophot `` sharpness '' parameter , which sets the intrinsic angular size of the objects , and selected only those with @xmath33 in both filters .
these sharpness values allow us to reject spurious and extended objects without eliminating the bright stars .
we emphasize that the fuv objects are only partially resolved in the acs data : one wfc / acs pixel corresponds to 0.05@xmath15 , which is @xmath41.5 pc at the distance of m81 .
our final photometry is calibrated on the acs vegamag photometric system , with zero points adopted from sirianni et al .
( 2005 ) .
we have also measured the optical integrated flux at @xmath34 of the fuv objects detected by galex by averaging the f475w and f814w images , and using the polyphot routine in iraf with the galex fuv contours , to determine @xmath34 magnitudes in the ab system .
although the @xmath34 magnitudes were measured in a pseudo @xmath34 image , we included it in table 1 since it is a standard magnitude provided in the literature .
figure [ cmd ] shows the @xmath35 versus @xmath36 ( @xmath37 ) color magnitude diagram ( cmd ) for all stars measured in the acs field of view .
the cmd shows a young and blue main sequence ( ms ) evolutionary phase , located at @xmath38 , with the brightest stars at @xmath39 .
a scarcely populated red plume is visible at @xmath40 and extends up to @xmath41 .
it is populated by red supergiants ( rsgs ) at the brighter magnitudes , and asymptotic giant branch stars ( agbs ) at fainter luminosities .
one of the most interesting features in the cmd is the diagonal strip of stars extending through @xmath42 , @xmath43 and @xmath44 , @xmath45 .
this strip is populated by stars at the hot edge of the helium core - burning phase .
the concentration of stars in the range of colors @xmath46 and magnitudes @xmath47 corresponds to low mass ( @xmath48 ) , older ( age @xmath49 ) stars in the red giant branch ( rgb ) evolutionary phase .
we used padova isochrones for metallicity @xmath50 and @xmath51 ( fagotto et al .
1994a , b ) to enable us to interpret the features visible in the cmd . superimposed on fig .
7 is the set of padova isochrones for a metallicity @xmath52 .
we obtain this value for @xmath53 from the following considerations . for the old population
, we can rule out metallicities as high as solar from the shape and broadening of the rgb .
for the same reasons , we find that @xmath53 is unlikely to be @xmath54 .
however we can not exclude a slightly higher metallicity .
the color and morphology of the blue edge of the blue loop evolutionary phase are sensitive to @xmath53 and age .
we find that the colors of this structure can only be reproduced for @xmath55 , assuming that star formation began @xmath56 myr ago in arp s loop and is continuing to the present .
we can not put strong constraints on @xmath53 for the young stellar population because there are not enough stars , and some of these stars may be multiple .
the youngest stars are concentrated where galex detected the fuv regions .
we used galex isophotal contours to select stars within the fuv regions .
cmds corresponding to fuv objects # 1 & # 2 , # 5 , # 6 , and # 7 & # 8 are shown in fig .
[ cmdblobs ] .
a comparison with padova isochrones indicates that the regions where the fuv objects are detected are mostly dominated by o and b stars with ages younger than 30 myr .
individual o stars are difficult to distinguish in cmds based on optical filters because of their well known color degeneracy at these wavelengths .
we also have the additional problem that any o stars in arp s loop may be multiple objects ( one wfc / acs pixel corresponds to @xmath41.5 pc ) . to estimate the number of o stars in arp s loop , we can use both the galex fuv fluxes and the nebular h@xmath0 luminosity of karachentsev & kaisin ( 2007 ) .
we measure a total fuv flux of @xmath57 erg s@xmath21 ( table 1 ) and , using standard numbers , this is equivalent to @xmath4 50 o8v stars .
this is an upper limit because we have neglected the contribution of b stars to the fuv flux .
karachentsev & kaisin ( 2007 ) measure a total h@xmath0 flux of @xmath58 erg s@xmath21 @xmath22 giving @xmath59(h@xmath60 erg s@xmath21 .
thus , assuming one o8v star has a photon ionizing luminosity @xmath61 of @xmath62 s@xmath21 in the lyman continuum ( smith , norris & crowther 2002 ) , the observed h@xmath0 flux is equivalent to 13 o8v stars .
these rough estimates will increase if there is any internal extinction .
arp s loop is then equivalent to @xmath63 orion nebulae , if we assume that the orion nebula has an h@xmath0 luminosity of @xmath64 erg s@xmath21 .
fig.[spatial ] shows the spatial distribution of the main sequence and the rgb stars in the acs image .
it is clear that stars of both types are located in the same general areas of the image with only two exceptions , at ( 1400,1600 ) and ( 3400 , 3400 ) .
these regions of the image have many ms stars and only a couple of rgbs are seen .
the first region is not among the galex sources , it is close to the bright star on the south - east side of fig.[acs_feet ] and the other region corresponds to object # 6 . in summary , we find both an old and a young stellar population in arp s loop with a similar spatial distribution and a metallicity @xmath65 .
it is likely that the old stellar population was formed in the m82 and/or m81 galaxies and ejected into the intergalactic medium during a tidal passage ( @xmath4 200300 myr ago ) , whereas the young stars detected by galex have formed since the encounter in the tidal debris .
observational evidence is accumulating that star formation often occurs in low gas density environments .
examples include the outer disks of galaxies ( e.g. ferguson et al .
1998 , thilker et al . 2005 ) and tidal tails ( e.g. saviane , hibbard , & rich 2004 , hibbard et al . 2005 ) .
the nearest example is the hi bridge between the magellanic clouds ( see demers & battinelli 1998 ) , located at a distance of 5060 kpc which allows very detailed investigations of both its interstellar gas and stars .
the bridge has a complex structure in hi with typical column densities of 10@xmath6610@xmath67 @xmath22 ( muller et al .
2003 , 2004 ) and somewhat higher in smaller regions .
the harris ( 2007 ) optical survey confirms that stars along the magellanic bridge not only formed in place but also formed in material that contains very few old red giant stars .
molecular line surveys ( mizuno et al .
2006 ) and follow - up searches for pre - main - sequence stars ( nishiyama et al . 2007 ) demonstrate that despite overall low hi column densities , the conditions necessary for star formation and young stellar candidates exist in some places along the western part of the bridge that extends from the wing of the small magellanic cloud . the more diffuse magellanic stream with @xmath68(hi ) @xmath69@xmath70 @xmath22 ( e.g. brns et al . 2005 ) , however , is not known to contain any stars ( majewski et al .
1999 ) . at the other extreme ,
relatively intense star formation is detected in fuv images of more massive gas rich tidal tails ( e.g. hibbard et al . 2005 for the antennae , and neff et al .
2005 for ngc 7769/71 , ngc 5713/19 , arp 295 , and the ngc 520 system ) . in these cases , large hii regions , ob associations , and even super star clusters
can form , especially in situations where the hi column densities are @xmath68(hi ) @xmath71 @xmath22 over large regions ( maybhate et al . 2007 ) .
thus the tidal bridge between m81m82 where arp s loop is located ( @xmath68(hi ) @xmath72@xmath73 @xmath22 ; yun et al .
1994 ) appears to support star formation at rates which are intermediate between the extremely low levels seen in an object like the magellanic bridge and very active tidal tails like those found in , for example , ngc4676 ( `` the mice '' ; de grijs et al .
this suggests that the tidal hi streams in the m81 system sit near the low end of a continuum of levels of star formation in gas - rich tidal matter .
star formation requires the production of gravitationally bound molecular clouds .
since tidal streams normally are at most only mildly gravitationally unstable on large spatial scales , star formation is not expected to behave in the same way as in the inner disks of galaxies which exist on the edge of large scale gravitational instability ( kennicutt 1989 ) .
the possibility that collisions between gas clouds could lead to star formation in low density but gas - rich environments such as the magellanic bridge was explored by christodoulou et al .
( 1997 ) . in their model , compression to the densities required to make bound molecular structures was accomplished through post - shock compression .
this picture gained support as observations improved ; e.g. muller & parker ( 2007 ) make a case for the magellanic bridge being a prime example of star formation resulting from turbulence .
more recent investigations , such as that of bergin et al .
( 2004 ) and heitsch et al .
( 2006 ) , include the effects of turbulence which can enhance density perturbations more readily leading to molecular regions that can collapse to make stars in lower velocity cloud collisions . in these models sufficient columns of hi must interact with speeds of @xmath7410 km s@xmath21 to produce local regions with @xmath68(hi ) @xmath74 10@xmath75 @xmath22 that are required to readily shield molecular gas from radiative dissociation .
the basic idea of triggering star - formation by cloud collision in arp s loop is consistent with the yun ( 1999 ) model for this feature being a stream crossing zone and with the hi observations showing comparatively high @xmath68(hi ) in a region with a substantial @xmath430 km s@xmath21 kpc@xmath21 and complex velocity field ( rots & shane 1975 ; rots 1975 ; yun et al .
1994 ) .
ongoing star formation in relatively diffuse structures , such as arp s loop , is consistent with expectations for gas collisions and subsequent turbulence to lead to inefficient conversion of gas into stars in low density systems such as tidal tails . on the other hand ,
the younger star forming sites seem to be relatively normal objects with properties similar to the orion nebula . here
we find support for the expectation that once a gravitationally bound molecular cloud forms in a low density environment , its subsequent evolution is largely controlled by factors that only weakly depend on how the cloud was formed .
the ages and locations of objects containing young stars indicate that self - gravitating substructures form within hi tidal material in a variety of interacting systems .
this tidal matter from interacting galaxies is expected to remain bound to one of the galaxies .
much of it will relatively quickly fall back into the galaxies , although some matter may remain detached for as long as a hubble time .
for example , according to hibbard & mihos ( 1995 ) @xmath420% of the tidal material of ngc7252 , a well - known merger remnant , is likely to remain in the local intergalactic medium .
tidal optical features are also known to host tidal dwarf galaxies " ( tdgs ) ( e.g. duc et al .
1997 ; weilbacher et al .
2000 ; knierman , et al . 2003 ; bornaud et al .
the eventual fate of the tdgs and the stars and metals that they produce remains unclear .
for instance , the tdg - candidate found in the compact group hcg100 ( de mello et al .
2007 ) observed with galex has young ( @xmath15 myr ) stellar populations and is located @xmath76100 kpc away from the main galaxies .
these systems therefore have the possibility to remain as independent objects over time scales of @xmath741 gyr .
the fuv objects in the m81 group are much closer to a large spiral than the hcg100 tdg , as they are only @xmath417 kpc in projected distance from m81 ( nucleus ) and @xmath415 kpc from the central region of the tdg - candidate , holmberg ix .
besides being located closer to the parent galaxies than a long - lived tdg - candidate , the 8 fuv objects have much lower uv luminosities than the objects studied by de mello et al .
( 2007 ) and neff et al .
( 2005 ) . as discussed in sect .
3 , their fuv luminosities are equivalent to @xmath4 50 late o stars .
the fact that the m81 group is so close and that galex is sensitive to very low star - formation levels , 10@xmath77 m@xmath78 yr@xmath21 ( salim et al . 2005 ) , makes it possible to detect the arp s loop fuv objects .
are we then seeing a more diffuse example of star formation , as may be the case in the magellanic bridge , or is the arp s loop region in an earlier evolutionary phase where star formation has not progressed far enough to produce a denser region ?
increasing the stellar density is critical for this object to become a tdg ( mizuno et al .
better kinematic and baryonic mass density information are required to address this question .
it is also possible that the arp s loop fuv regions are analogs to the so - called ` intergalactic hii ' regions of mendes de oliveira et al .
( 2004 ) , ryan - weber et al .
( 2004 ) and oosterloo et al .
these objects seem to be similar to the hii regions in our milky way , but are located in regions with no optical connections to nearby galaxies .
the fate of these type of objects and their importance in galaxy evolution and in the enrichment of the intergalactic medium are still debatable .
these star- forming clumps can ( 1 ) remain as independent entities and grow to become tdgs by accreting more gas and forming more stars , ( 2 ) make star clusters that survive to live in the distant halos of their hosts or even become intergalactic objects , or ( 3 ) dissolve and not remain gravitationally bound yielding only very sparse star streams .
we have analyzed the uv and optical light within the hi bridge between m81 and m82 system and identified 8 fuv regions in an area known as arp s loop .
four of these objects were also detected in h@xmath0 .
high resolution acs images show six of the regions in great detail and allow us to trace their star formation history .
the synthetic cmd modeling shows the presence of a young population ( @xmath1 10 myr ) together with an older component with ages @xmath79 1 gyr .
it is possible that these two populations are not related . in any case
the young population must have formed in situ " whereas the old population probably was formed in the m82 and/or m81 galaxies , and removed from their disks during a tidal passage @xmath4 200300 myr ago which also triggered star formation in the hi debris .
the fuv luminosities of the eight objects we detect are modest and typical of small clusters or associations of o and b stars .
it is likely that these young stars have condensed recently from the tidally stripped gas from m81 and m82 .
the tidal bridge between m81m82 appears to be intermediate between the very low levels of star formation seen in an object like the magellanic bridge and very active tidal tails like those found in interacting galaxies such as the mice and the antennae .
ongoing star formation in relatively diffuse structures , such as arp s loop , is consistent with expectations for gas collisions and subsequent turbulence to lead to inefficient conversion of gas into stars in low density systems such as tidal tails .
the fate of these star - forming clumps is not clear .
they can grow to become tdgs , or make star clusters that survive to live in the distant halos of their hosts , or even become intergalactic objects .
finally , they may dissolve and not remain gravitationally bound , yielding only very sparse star streams . in any case , they will have a composite population of old and young stars formed in very different environments such as the m81/m82 disks and the tidal bridge .
more observations of interacting systems similar to the m81 group will help establish the importance of these systems in galaxy evolution and in the enrichment of the intergalactic medium .
galex is a nasa small explorer , launched in 2003 april .
we gratefully acknowledge nasa s support for construction , operation , and science analysis for the galex mission , developed in cooperation with the centre national detudes spatiales of france and the korean ministry of science and technology . this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
cccccccc 1 + 2 & 149.313562 & 69.278637 & 20.25@xmath80 0.07 & 18.17 @xmath80 0.02 & 1.09e+039 & 19.46 @xmath80 0.02 & 6.48e+038 + 3 & 149.304548 & 69.273482 & & 19.77 @xmath80 0.05 & 2.48e+038 & 21.17 @xmath80 0.06 & 1.34e+038 + 4 & 149.317475 & 69.273476 & & 20.03 @xmath80 0.06 & 1.94e+038 & 21.13 @xmath80 0.04 & 1.40e+038 + 5 & 149.328273 & 69.273482 & [email protected] & 20.53 @xmath80 0.07 & 1.23e+038 & 20.78 @xmath80 0.05 & 1.92e+038 + 6 & 149.374412 & 69.266872 & [email protected] & 19.60 @xmath80 0.04 & 2.90e+038 & 20.45 @xmath80 0.04 & 2.61e+038 + 7 & 149.393938 & 69.294389 & [email protected] & 21.33 @xmath80 0.1 & 5.91e+037 & 22.23 @xmath80 0.09 & 5.06e+037 + 8 & 149.390473 & 69.296530 & [email protected] & 21.03 @xmath80 0.09 & 7.80e+037 & 21.95 @xmath80 0.08 & 6.52e+037 +
|
we present multi - wavelength observations of stellar features in the hi tidal bridge connecting m81 and m82 in the region called arp s loop .
we identify eight young star - forming regions from galaxy evolution explorer ultraviolet observations .
four of these objects are also detected at h@xmath0 .
we determine the basic star formation history of arp s loop using f475w and f814w images obtained with the advanced camera for surveys on board the hubble space telescope .
we find both a young ( @xmath1 10 myr ) and an old ( @xmath2 gyr ) stellar population with a similar spatial distribution and a metallicity @xmath3 .
we suggest that the old stellar population was formed in the stellar disk of m82 and/or m81 and ejected into the intergalactic medium during a tidal passage ( @xmath4 200300 myr ago ) , whereas the young uv - bright stars have formed in the tidal debris .
the uv luminosities of the eight objects are modest and typical of small clusters or ob associations .
the tidal bridge between m81m82 therefore appears to be intermediate between the very low levels of star formation seen in the magellanic bridge and actively star - forming tidal tails associated with major galaxy mergers .
|
the dbq4 star lds749b ( wd2129 + 00 ) has long been considered for a flux standard ( e.g. , bohlin et al .
1990 ) . to establish the flux on the _ hubble space telescope _
( _ hst _ ) white dwarf ( wd ) flux scale , stis spectrophotometry was obtained in 20012002 .
the virtues of lds749b as a flux standard include an equatorial declination and a significantly cooler flux distribution than the 3300061000 k primary da standards gd71 , gd153 , and g191b2b .
full stis wavelength coverage is provided from 0.1151.02 @xmath0 m , and the peak in the sed is near 1900 . at @xmath8 ( landolt & uomoto 2007 ) , lds749b is among the faintest _ hst _ standards and is suitable for use with larger ground - based telescopes and with the more sensitive _
hst _ instrumentation , such as the acs / sbc and cos .
the bulk of the stis data was obtained as part of the fastex ( faint astronomical sources extention ) program .
finding charts appear in turnshek et al .
( 1990 ) and in landolt & uomoto ( 2007 ) ; but there is a large proper motion of 0.416 and 0.034 arcsec / yr in right ascension and declination , respectively . the absolute flux calibration of _ hst _ instrumentation is based on models of three pure hydrogen wd stars gd71 ,
gd153 , and g191b2b ( bohlin 2000 ; bohlin , dickenson , & calzetti 2001 ; bohlin 2003 ) . in particular , the nlte model fluxes produced by by the tlusty code ( hubeny & lanz 1995 ) determine the shape of the flux distributions using the known physics of the hydrogen atom and of stellar atmospheres .
if there are no errors in the basic physics used to determine the stellar temperatures and gravities from the balmer line profiles , then the uncertainty of 3000 k for the effective temperature of g191b2b means that the relative flux should be correct to better than 2.5% from 0.13 to 1 @xmath0 m and to better than 1% from 0.35 to 1 @xmath0 m . a model that matches the observations serves as a noise free surrogate for the observational flux distribution and provides a reliable extrapolation beyond the limits of the observations for use as a calibration standard for _ jwst _ , _ spitzer _ , and other ir instrumentation .
currently , the best ir absolute flux distributions are found in a series of papers from the epic and pioneering work of m. cohen and collaborators , i.e. , the cohen - walker - witteborn ( cww ) network of absolute flux standard ( e.g. , cohen , wheaton , & megeath 2003 ; cohen 2007 ) .
the cww ir standard star fluxes are all ultimately based on models for vega and sirius ( cohen et al .
more recently , bohlin & gilliland ( 2004 ) observed vega and published fluxes on the _
hst_/stis wd flux scale .
a small revision in the stis calibration resulted in excellent agreement of the stis flux distribution with a custom made kurucz model with @xmath9 k ( bohlin 2007 ) , which is the same @xmath10 used for the cohen et al.(1992 ) vega model .
the model presented here for lds749b and archived in the calspec database should have a better precision than the kuruzc @xmath9 k model for vega , especially beyond @xmath112 @xmath0 m , where the vega s dust disk becomes important ( engleke , price , & kraemer 2006 ) .
vega is also a pole - on rapid rotator , which may also cause ir deviations from the flux for a single temperature model . our modeled flux distribution for lds749b should have an accuracy comparable to the pure hydrogen model flux distributions for the primary wd standards gd71 , gd153 , and g191b2b .
a helium model atmosphere flux distribution for lds749b is calculated with the lte code of koester ( e.g. , castanheira et al .
2006 ) for @xmath11 k and @xmath3 .
at such a cool temperature , the differences between lte and nlte in the continuum flux distributions should be @xmath40.1% from the far - uv to the ir .
for example for a pure hydrogen da , the difference between the continua of a hot 40,000 k lte / nlte pair of models is 1% between 0.12.5 @xmath0 m .
the same maximum difference at 20,000 k is only 0.3% .
napiwotzki ( 1997 ) did not discuss pure he models but concludes that nlte effects tend to become smaller with lower effective temperature .
for cool da wds , koester et al .
( 1998 ) show that the only nlte effect that approaches 1% is a deeper line core of h@xmath12 . the matter densities in helium - rich white dwarfs are significantly higher , leading to a higher ratio of collisional versus radiative transitions between atomic levels
. the larger importance of collisions increases the tendency towards lte occupation numbers because of the robust maxwell distribution of particle velocities .
the @xmath11 k is higher than the @xmath13 k published for lds749b ( alias g26@xmath1410 ) in castanheira et al .
( 2006 ) , because only uv spectra of lower precision ( _ iue _ heritage ) were used in that analysis .
voss et al .
( 2007 ) found @xmath15 k with large uncertainty , because only line profiles in the optical range were used and @xmath16 had to be assumed . a trace of carbon at @xmath17 of the solar c
/ he ratio is included , i.e. the c / he number ratio is 3.715@xmath18 .
the model mass is 0.614 @xmath19 and the stellar radius is 0.01224 @xmath20 , which corresponds to a distance of 41 pc for the measured stis flux .
the line broadening theory for the he lines combines van der waals , stark , and doppler broadening to make a voigt profile .
however , the stark broadening uses a simple lorentz profile with width and shift determined from the broadening data in griem ( 1964 ) , instead of the elaborate calculations of beauchamp et al.(1997 ) .
the griem method is computationally much faster , and data are available for more lines than are calculated by beauchamp et al .
the fit of the higher series he lines is much improved , if the neutral - neutral interaction is decreased in comparison to the original formalism of the hummer - mihalas occupation probabilities .
a similar effect was noticed by koester et al .
( 2005 ) , and our model uses the same value of the quenching parameter that koester et al . derived ( @xmath21 ) .
the model wavelengths are all on a vacuum scale .
the sensitivities of the five stis low dispersion spectrophotometric modes have been carefully tracked since the stis commissioning in 1997 .
after correcting for changing sensitivity with time ( stys , bohlin , & goudfrooij 2004 ) and for charge transfer efficiency ( cte ) losses for the three stis ccd spectral modes ( bohlin & goudfrooij 2003 ; goudfrooij et al .
2006 ) , stis internal repeatability is often better than 0.5% ( bohlin 2003 ) .
thus , _ hst_/stis observations of lds749b provide absolute spectrophotometry with a precision that is superior to ground based flux measurements , which require problematic corrections for atmospheric extinction .
observations with a resolution @xmath221500 in four stis modes from 11501710 ( g140l ) , 15903170 ( g230l ) , 29005690 ( g430l ) , and 530010200 ( g750l ) were obtained in 20012002 .
earlier observations of lds749b in 1997 to test the time - tagged mode were unsuccessful .
two observations in the ccd g230 lb mode overlap the wavelength coverage of the mama g230l but are too noisy to include in the final combined absolute flux measurement from the other four modes .
table 1 summarizes the individual observations used for the final combined average , along with the unused g230 lb data for completeness .
figure 1 shows the ratios of the three individual g230l and the two g230 lb observations to the model fluxes , which are normalized to the stis flux in the 53005600 range .
the excellent repeatability of stis spectrophotometry over broad bands is illustrated and the global average ratio over the 17503000 band pass is written in each panel .
this ratio is unity to within 0.3% , even for the shorter ccd g230 lb exposures despite their almost @xmath23 higher noise level and cte corrections .
the other ccd modes g430l and g750l also require cte corrections .
repeatability for all the stis spectral modes is comparable , i.e. , the global ratio deviates rarely from unity by more than 0.6% .
the observations in each of the four spectral modes are averaged and the four segments are combined .
this composite standard star spectrum extends from 115010226 and can be obtained at http://www.stsci.edu/hst/observatory/cdbs/calspec.html/ along with the remainder of the _ hst _ standard star library ( bohlin , dickenson , & calzetti 2001 ) .
this binary fits table named @xmath24 has 3666 wavelength points and seven columns .
an ascii file of the flux distribution in table 2 is available via the electronic version of this paper .
table 2 contains the wavelength in and the flux in erg @xmath25 s@xmath26 @xmath26 in the first two columns , while columns 34 are the poisson and systematic uncertainty estimates in flux units , respectively .
column 5 is the fwhm of the resolution in .
the @xmath27 version has two more columns than the ascii version : column 6 is a data - quality flag , where one is good and zero may be poor quality . the seventh column is the exposure time in seconds .
the fluxes at the shortest wavelengths below 1160 are unreliable because of the steepness of the sensitivity drop - off .
to compare the model and observations , a convenient method of removing the slope of the spectral energy distribution ( sed ) is to divide both fluxes by the same theoretical model continuum .
small differences between the observations and the model , either in the lines or in the actual continuum , are easily illustrated in such plots .
the theoretical continuum contains only continuum opacities with an extrapolation across the he i opacity edges at 2601 , 3122 , and 3422 , in order to avoid discontinuities .
figure 2 shows an overview of the comparison of the stis fluxes with the model after division of both seds by this same smooth line - and - edge - free continuum .
the mean continuum level of the data between the absorption lines agrees with the model within @xmath11% almost everywhere .
the most significant deviation of the data from the model is in the broad 14001550 region , where each of the three spectra comprising the g140l average have 350,000 photo - electron events in this 150 band .
the background level is @xmath40.1% of the net signal , so that neither counting statistics nor background subtraction error could cause the observed @xmath11.5% average disparity .
of the five low dispersion modes , g140l shows the worst photometric repeatability of individual spectra in broad bands of @xmath280.6% .
the three individual spectra comprising the g140l average do show occasional 23@xmath29 broadband dips within their 550 coverage region ; but the probability of such a large excursion as 1.5% in their average is extremely unlikely at any particular wavelength .
however , the probability is much greater that such a large excursion could occur in some 150 band .
individual g140l spectra of the monitoring standard grw+70@xmath305824 often show a broad region differing by 12% from the average
. the cause of such excursions could be flat field errors , temporal instabilities in the flat field , or other detector effects that might make the flat field inapplicable to a narrow spectral trace .
the uncertainty in the model @xmath10 is determined by the uncertainty in the slope of the uv flux distribution . for a constant @xmath31 model that is cooler or hotter by 50 k and normalized to the measured 53005600 flux ,
there are increasing differences with the data from 1% near 2000 to 2% at the shorter wavelengths .
such a large change in the modeled continuum level in figure 2 ( red line ) is inconsistent with the stis flux ( black line ) .
this 50 k uncertainty of the model @xmath10 is an internal uncertainty relative to the temperatures of the primary wd standards gd71 , gd153 , and g191b2b .
if a re - analysis of the balmer lines in these primary da standards produces a systematic shift in the temperature scale , this shift would be reflected in a revised @xmath10 for lds749b that is independent of the 50 k internal uncertainty .
a 50 k temperature difference causes a @xmath40.5% flux change in the ir longward of 1 @xmath0 m . to estimate the uncertainty in @xmath31 ,
models are computed at the 13575 k baseline temperature but with an increment in @xmath31 .
positive and negative increments produce nearly mirror image changes in the flux distribution . for a decrease of 0.7 in @xmath31 ,
the flux decreases by a nearly uniform 1.5% below 3600 after normalizing to unity in the 53005600 range . increasing the @xmath10 by the full 50 k uncertainty to 13625
k can compensate for this flux decrease below @xmath12000 .
however , the + 50 k increase compensates little in the 25003600 range , leaving a disparity of @xmath11% . because this 25003600 range includes some of the best s / n stis data , a 1% disparity establishes the uncertainty of 0.7 dex in @xmath31 as barely compatible with the stis flux distribution . the ir uncertainty corresponding to this
limiting case of @xmath32 and @xmath33 is @xmath11% longward of 1 @xmath0 m , because the fractional percent changes in the ir from the higher temperature and from the lower @xmath31 are both in the same direction .
another source of error in the model @xmath10 is interstellar reddening .
the standard galactic reddening curve has a strong broad feature around 2200 ; and a tiny limit to the extinction @xmath34 is set by the precise agreement of stis with the model in this region of figure 2 .
for the upper temperature limit of @xmath35 k , an @xmath36 brings the reddened model into satisfactory agreement with stis .
however , for a temperature increment of 100 k and @xmath37 , the model is @xmath11% high at 1300 and @xmath11% low at 2200 .
thus , for standard galactic reddening , @xmath34 must be less than 0.004 , and @xmath10 is less than 13675 k. in this case of @xmath10 and @xmath34 at these allowed limits , the ir flux beyond 1 @xmath0 m is still the same as for the unreddened baseline @xmath11 k within 0.5% . however , bohlin ( 2007 ) presented arguments for reddening with a weak 2200 bump for other lines of sight with tiny amounts of extinction . reddening curves measured in the smc ( e.g. , witt & gordon 2000 ) are missing the 2200 feature and can cause larger uncertainty in @xmath10
. additional evidence for extinction curves more like those in the magellanic clouds is presented by clayton et al .
( 2000 ) for the local warm intercloud medium , where the reddening is low .
changes in the shape of the flux distribution after reddening with the smc curve of witt and gordon are similar to the change in shape with @xmath10 .
for example , reddening a model with @xmath38 k by smc extinction of 0.015 is required to make an equally unacceptable fit as for 13675 k and galactic extinction of @xmath39 . in this extreme limiting case of smc extinction ,
the ir flux beyond 1 @xmath0 m is still the same within @xmath11% as for @xmath11 k and @xmath40 . despite small uncertainties in the interstellar reddening and consequent uncertainty in @xmath10 ,
our modeling technique still predicts the continuum ir fluxes to 1% from 1 @xmath0 m to 30 @xmath0 m . discounting the most pathological case of smc reddening ,
the worst far - ir uncertainty is from the combined 50 k temperature and 0.7 @xmath31 uncertainties , because the changes in the slope from the visual band normalization region into the ir due to higher temperature and lower @xmath31 are both in the same direction . in the absence of modeling errors or other physical complications like
ir excesses from dust rings , the measured fluxes of lds749b relative to the three primary wds should be the same as predicted by the relative fluxes of the respective models to a precision of 1% in the ir . an upper limit on the equivalent width for h@xmath12 of @xmath10.1
constrains the fraction by number of hydrogen in the atmosphere of lds749b to @xmath41@xmath5 of helium .
however , a stricter limit of @xmath41@xmath6 is provided by the weak ly@xmath12 line .
because interstellar absorption at ly@xmath12 could be significant , zero hydrogen is consistent with the observations and is adopted for the final best model for lds749b .
after normalization in the @xmath41 band , the continuum of a model with @xmath42 hydrogen composition and the baseline @xmath11 k and @xmath3 agrees with the zero hydrogen baseline continuum to @xmath10.5% from ly@xmath12 to 30 @xmath0 m .
figure 3 compares the observed hei lines with the baseline model after correcting the model wavelengths by the radial velocity of @xmath1481 km s@xmath26 ( greenstein & trimble 1967 ) .
the model is smoothed with a triangular profile of fwhm corresponding to a resolution @xmath43 for the mama spectra shortward of 3065 and to @xmath22 for the ccd spectra longward of 3065 . in general
, the model underestimates the line strengths , even for the quenching of the neutral - neutral interactions with @xmath21 .
there is a suggestion of some systematic asymmetry with stronger absorption in the short wavelength side of the line profile .
this asymmetry could be in the stis line spread function ( lsf ) ; or perhaps , a more exact treatment of the stark line broadening theory would reproduce the observed asymmetries . with a c /
he ratio of @xmath41@xmath5 solar , i.e. a c / he number ratio of 3.715@xmath18 , the modeled c i and c ii lines reproduce the observations within the observational noise , as shown in figure 4 . in particular , the agreement of the modeled c i(1329)/c ii(1335 ) line ratio with the observed ratio means that the carbon ionization ratio corresponds to the photospheric temperature of the star . with this small amount of carbon ,
the spectral clasification of lds749b should more properly be dbq ( wesemael et al .
1993 ) .
the oxygen triplet at 1302.17 , 1304.87 , and 1306.04 constrains the fraction of oxygen in the lds749b atmosphere .
this triplet absorption feature extends over 4 or about seven stis pixels ; but no obvious absorption feature appears above the noise level .
after binning the stis data by seven pixels , the rms noise in the 1300 region is 0.8% .
the corresponding 3@xmath29 upper limit to the equivalent width is 0.10 , which implies an upper limit to the atmospheric oxygen fraction by number of 7@xmath7 of helium .
in the absence of any interstellar reddening , a helium model with @xmath44 , @xmath45 , and a trace of carbon at @xmath41@xmath5 of solar fits the measured stis flux distribution for lds749b .
the noise - free , absolute flux distribution from the model after normalization to the observed broadband visual flux is preferred for most purposes .
this normalized model sed is a high fidelity far - uv to far - ir calibration source ; and the flux distribution is available via table 3 in the electronic version of the _ journal_. both the observed flux distribution and the modeled fluxes are also available from the calspec database .
primary support for this work was provided by nasa through the space telescope science institute , which is operated by aura , inc .
, under nasa contract nas5 - 26555 .
additional support came from doe through contract number c3691 from the university of california / lawrence berkeley national laboratory .
this research has made use of the simbad database , operated at cds , strasbourg , france .
beauchamp , a. , wesemael , f. , & bergeron , p. 1997
, apjs , 108 , 559 bohlin , r. c. 2000 , aj , 120 , 437 bohlin , r. c. 2003 , in 2002 hst calibration workshop , eds .
s. arribas , a. koekemoer , & b. whitmore , ( baltimore : stsci ) , p. 115
bohlin , r. c. 2007 , in the future of photometric , spectrophotometric , and polarimetric standardization , asp conf .
series , vol .
364 , p. 315 ed . c. sterken ; also astro - ph 0608715 bohlin , r. c. , dickinson , m. e. , & calzetti , d. 2001 , aj , 122 , 2118 bohlin , r. c. , & gilliland , r. l. 2004 , aj , 127 , 3508 bohlin , r. , & goudfrooij , p. 2003
, instrument science report , stis 03 - 03 , ( baltimore : stsci ) bohlin , r. c. , harris , a. w. , holm , a. v. , & gry , c. 1990 , apjs , 73 , 413 castanheira , b. , kepler , s. , handler , g. , & koester , d. 2006 , a&a , 450 , 331 clayton , g. , gordon , k. , & wolff , m. 2000 , apjs , 129 , 147 cohen , m. 2007 , in the future of photometric , spectrophotometric , and polarimetric standardization , asp conf . series , vol . 364 , p. 333
, ed . c. sterken cohen , m. , walker , r. g. , barlow , m. j. , & deacon , j. r. 1992 , aj , 104 , 1650 cohen , m. , wheaton , w. , & megeath , s. 2003 , aj , 126 , 1090 engleke , c. , price , s. , & kraemer , k. 2006 , aj , 132 , 1445 goudfrooij , p. , bohlin , r. , maz - apellniz , j. , & kimble , r. 2006 , pasp , 118 , 1455 griem , h. r. 1964 , plasma spectroscopy , ( new york : mcgraw - hill ) greenstein j. l. , & trimble v. l. 1967 , apj , 149 , 283 hubeny , i. , & lanz , t. 1995 , apj , 439 , 875 koester , d. , dreizler , s. , weidemann , v. , & allard , n. f. 1998 , a&a , 338 , 612 koester , d. , napiwotzki , r. , voss , b. , homeier , d. , & reimers , d. 2005 , a&a , 439 , 317 landolt , a. u. , & uomoto , a. k. 2007 , aj , 133 , 768 napiwotzki , r. 1997 , a&a , 322 , 256 stys , d. j. , bohlin , r. c. & goudfrooij , p. 2004 , instrument science report , stis 2004 - 04 , ( baltimore : stsci ) turnshek , d. a. , bohlin , r. c. , williamson , r. l. , lupie , o. l. , koornneef , j. , & morgan d. h. 1990 , aj , 99 , 1243 voss , b. , koester , d. , napiwotzki , r. , christlieb , n. , & reimers , d. 2007 , a&a , 470 , 1079 wesemael , f , et al .
1993 , pasp , 105 , 761 witt , a. , & gordon , k. 2000 , apj , 528 , 799 2 pixel resolution of stis before dividing . the global average and
rms for the range between the vertical dotted lines are written in each panel along with the identifying information for each observation : spectral mode , root name , aperture , and exposure time in seconds .
the noisier , short - exposure ccd mode g230 lb data shown in the bottom two panels are not used in the final average of the observations.,height=672 ] lllcc o6il01020 & g230 lb & 52@xmath462 & 2001 - 07 - 13 & 576 + o6il01030 & g430l & 52@xmath462 & 2001 - 07 - 13 & 619 + o6il01040 & g750l & 52@xmath462 & 2001 - 07 - 13 & 1380 + o6il01060 & g230 lb & 52@xmath462 & 2001 - 07 - 13 & 576 + o6il01070 & g430l & 52@xmath462 & 2001 - 07 - 13 & 576 + o6il01080 & g750l & 52@xmath462 & 2001 - 07 - 13 & 1584 + o8h107010 & g140l & 52@xmath462 & 2002 - 09 - 01 & 2266 + o8h107020 & g430l & 52@xmath462 & 2002 - 09 - 01 & 400 + o8h107030 & g430l & 52@xmath462e1 & 2002 - 09 - 01 & 400 + o8h107040 & g230l & 52@xmath462 & 2002 - 09 - 01 & 1469 + o8h108010 & g140l & 52@xmath462 & 2002 - 10 - 16 & 2266 + o8h108020 & g430l & 52@xmath462 & 2002 - 10 - 16 & 400 + o8h108030 & g430l & 52@xmath462e1 & 2002 - 10 - 16 & 400 + o8h108040 & g230l & 52@xmath462 & 2002 - 10 - 16 & 1469 + o8h109010 & g140l & 52@xmath462 & 2002 - 12 - 27 & 2266 + o8h109020 & g430l & 52@xmath462 & 2002 - 12 - 27 & 400 + o8h109030 & g430l & 52@xmath462e1 & 2002 - 12 - 27 & 400 + o8h109040 & g230l & 52@xmath462 & 2002 - 12 - 27 & 1469 ccccc 1150.31 & 1.68860e-14 & 8.09170e-16 & 1.68860e-16 & 1.16699 + 1150.89 & 1.59170e-14 & 7.54240e-16 & 1.59170e-16 & 1.16687 + 1151.47 & 1.51450e-14 & 7.07050e-16 & 1.51450e-16 & 1.16687 + 1152.06 & 1.57620e-14 & 7.02910e-16 & 1.57620e-16 & 1.16699 + 1152.64 & 1.54940e-14 & 6.71560e-16 & 1.54940e-16 & 1.16699 + ccc 900.000 & 8.7713311e-15 & 8.7863803e-15 + 900.250 & 8.7785207e-15 & 8.7950878e-15 + 900.500 & 8.7853486e-15 & 8.8037809e-15 + 901.000 & 8.7975857e-15 & 8.8212095e-15 + 901.500 & 8.8064389e-15 & 8.8386236e-15 +
|
observations from the space telescope imaging spectrograph define the flux of the dbq4 star lds749b from 0.121.0 @xmath0 m with an uncertainty of @xmath11% relative to the three pure hydrogen wd primary _ hst _ standards . with @xmath2 , @xmath3 , and a trace of carbon at @xmath41@xmath5 of solar , a he model atmosphere fits the measured stis fluxes within the observational noise , except in a few spectral lines with uncertain physics of the line broadening theory .
upper limit to the atmospheric hydrogen and oxygen fractions by number are 1@xmath6 and 7@xmath7 , respectively .
the excellent agreement of the model flux distribution with the observations lends confidence to the accuracy of the modeled ir fluxes beyond the limits of the stis spectrophotometry .
the estimated precision of @xmath11% in the predicted ir absolute fluxes at 30 @xmath0 m should be better than the model predictions for vega and should be comparable to the absolute accuracy of the three primary wd models .
|
this article describes lindbladians associated with _ controlled _
stochastic hamiltonians in weak coupling . controlled stochastic hamiltonians arise in the context of `` dynamical decoupling '' and `` coherent control '' and are used to examine protocols for extending the coherence time of qubits @xcite .
lindbladians in the weak coupling limit have been rigorously studied in @xcite in the _ time independent _ setting .
recently , controls aimed at extending the coherence of qubits have been suggested in @xcite and periodically controlled lindbladians have been studied in @xcite .
however a careful derivation of the lindbladians for the controlled stochastic evolutions and in particular eq .
( [ main - o ] ) for _ general _ and in eq .
( [ fcg ] ) for _ stationary control _ are new .
this is also the case for the notion of effective control introduced in appendix [ effective ] and some of the examples in section [ s : ex ] .
consider the stochastic controlled hamiltonian .
[ hlab ] _ weak noises+_control , @xmath0 are fixed hermitian matrices representing independent and in general non - commuting sources of noise .
@xmath1 are stationary gaussian random processes [ jj ] ( _ ( t))=0 , ( _ ( t)_(u))=j_(|t - u| ) . with @xmath2 rapidly decreasing on a time scale @xmath3 .
we shall sometimes assume , w.l.o.g .
, that @xmath2 is a diagonal matrix ( this can be achieved by a redefinition of @xmath0 ) .
a spin in a magnetic field having fixed direction but noisy amplitude , often a good approximation @xcite , is represented by a single term @xmath4 .
the case where the direction of the field is also stochastic is modeled by several @xmath4 s and gives rise to noise that is non - commutative ( not been treated before . )
@xmath5 , a _ time - dependent _ ( hermitian ) matrix , represents the control .
it is convenient to reformulate the problem in the interaction picture .
let [ model ] h^i_(t)=__(t)h^i _ ( t ) , h^i _
( t)= v^ * ( t ) h_v ( t ) where @xmath6 is the unitary generated by the control @xmath7 , [ h ] h_c= i v ( t)v^ * ( t),v(0)= weak coupling parameter in the present context is defined by [ e ] ^2=jh_^21j()=_-^e^it j(t ) dt 0 there are several ways to think about weak coupling : if we think of @xmath8 then weak coupling means what its names suggests , namely , that the noise is weak in the sense that @xmath9 .
an alternate approach , which is also insightful , is to take @xmath10 and then weak coupling means short correlation time @xmath11 .
the noise affects the ( average ) state on the _ coarse grained _ time scale lead to effects of @xmath12 . ]
[ s ] s=^2 t/ control problems are characterized by the rate of rotation of @xmath13 .
for example ,
when the control @xmath5 is time independent , ( constant control ) , @xmath14 while for periodic bang - bang , where @xmath7 is a ( periodic ) sequence of delta pulses , @xmath15 is the frequency of the bangs .
this gives rise to a second dimensionless parameter @xmath16 .
our analysis of the weak coupling limit holds independently of @xmath17 .
dynamical decoupling requires however @xmath18 where the time scale of the control , @xmath19 , is not resolved on the coarse grained time scale @xmath20 . by _ stationary controls _
we shall mean that @xmath13 has a finite number of fourier coefficients .
it is convenient to factor @xmath21 so that the fourier coefficients are @xmath22 are [ stationary ] h^i_(t)=_fh _ ( ) e^it , h_()=h^*_(- ) @xmath23 a finite set .
when @xmath24 we shall show that the evolution is governed by ( complete ) positivity preserving lindbladian [ repar ] = _ moreover , we shall show that , in the case of stationary control , @xmath25 has a limit as @xmath26 given by : = _ _ , _
= h_-d _ with @xmath27,\rho\big],\nonumber \\
{ \mathbfcal d}_\alpha\rho&=\frac \tau 8\sum_{\omega\in f } \tilde j_\alpha ( \omega)\big [ \tilde h_\alpha(\omega ) , [ \tilde h_\alpha^*(\omega),\rho]\big ] \end{aligned}\ ] ] @xmath28 denotes fourier transform and @xmath29 is the anti - symmetric partner of @xmath2 : k_(u)=i ( u)j_(u ) , the special form of @xmath30 reflects the fact that stochastic evolutions are unital : the fully mixed state @xmath31 is stationary .
( [ fcg ] ) can be used to determine @xmath28 from the measured rates @xmath32 @xcite .
see the examples in section [ s : ex ] .
moreover , it implies that the optimal measurement time is @xmath33 . to see this
observe that repeated measurements of a projection @xmath34 in the state @xmath35 generates a poisson process with an average @xmath36 given total allotted time @xmath37 , an optimal estimator minimizes the standard deviation in @xmath38 .
this fixes @xmath39 to be the minimizer of the sensitivity @xmath40 eq . ( [ fcg ] ) determines @xmath35 for @xmath41 . for a depolarizing qubit
@xmath42 so @xmath43 takes its minimum at the left edge of the interval , @xmath33 .
the hamiltonian @xmath44 generates a stochastic unitary evolution @xmath45 given by we can avoid ito s calculus . ]
@xmath46 the time ordering , denoted by the subscript @xmath37 in the first line is defined explicitly in the second .
more crucial to us is the super - operator @xmath47 acting on states _ 0_(t)= _ _ 0= u_(t ) _ 0 u_^*(t ) the super - operator can be written similarly @xmath48 where the super - operators @xmath49 acts by the adjoint action [ ad ] = ( ad[h ] ) note that @xmath50 ad[h_2]\neq ad[h_1h_2]$ ] . rather , [ prod ] ( ad[h_1 ] ad[h_2])()= ( _ 1 _ 2)()= _ 1 ( _ 2)= [ h_1 , [ h_2 , ] ] we also need the fact that [ jacobi ] ad= which follows from jacobi s identity .
the key object of this study is the ( stochastic ) averaged evolution [ average ] _ 0 ( _ ( t))=((t))_0 the super - operator @xmath51 is trace preserving , ( completely ) positivity preserving and unital ( i.e. @xmath52 ) , but , in general , not unitary or markovian . recall that for gaussian averages [ gaussian ] ( e^i)= e^- ( ^2)/2 it follows that for @xmath53 a stationary gaussian process , @xmath54 where @xmath55 so far , no approximation has been made .
however , the time ordering remains a major complication may be viewed as the grand canonical partition function of a 1-d quantum gas with short range interaction . ] . for its precise meaning
one can either go back to eq .
( [ uxi ] ) , or alternatively , see the discussion and graphical representation in appendix [ s : lin ] .
there is no issue with time ordering in two cases : when @xmath53 is white noise and when the ( interaction picture ) hamiltonian commute at different times .
we examine these cases first .
white noise is the limit @xmath56 with @xmath57 . by eq .
( [ e ] ) this corresponds to @xmath58 .
not surprisingly , the reduction of white noise to lindblad evolution is exact .
since @xmath59 we have @xmath60 with @xmath61 since @xmath62 and @xmath63 have the same time argument @xmath39 , we may use the definition of time - ordering in eq .
( [ uxi ] ) with @xmath64 , to conclude that @xmath62 is the generator of @xmath65 .
the lindbladian reduces to : [ lwn ] _
t=-1 2 _ j_the result is exact .
since @xmath66 are unitarily related for different @xmath39 it follows that the family @xmath67 is unitarily related and hence isospectral .
in particular , the instantaneous dephasing rates are independent of the control @xmath68 .
this could be anticipated since to affect the dephasing rates , the control must be at least as fast as the noise correlations . in general
, it is difficult to extract a generator of the evolution from eq .
( [ u2 ] ) because of the time ordering .
in the commutative case this is not an issue and the generator of the evolution follows from eq .
( [ k1 ] ) .
let us denote @xmath69 we then have @xmath70 of course , in the commutative case the index @xmath37 is redundant .
now although @xmath71 is an exact generator , it is not in general of lindblad form .
more precisely , it may fail to satisfy positivity at short times as the following example shows .
[ ex : com ] the commutative case arises , for example , when the ( interaction picture ) noise has a stochastic amplitude but a fixed `` direction '' , i.e. when @xmath72 since @xmath73 is a commuting family , eq . (
[ gu ] ) is exact and the generator of the evolution is [ stoc - lin ] ( t)=- 2 _ 0_0=- 2 [ h_0,[h_0 , ] ] , , the `` dephasing rate '' @xmath74 is given by [ dote ] ( t)= 2_0^t du j(t - u)=2 _ 0^t du j(u ) although @xmath75 for very short times ( since @xmath76 ) , @xmath74 may be negative for @xmath77 as in fig .
[ fig : gammat ] . in these cases
@xmath78 does not generate a contraction for all times .
this reflects the fact that the evolution is not strictly markovian . at longer times , @xmath79 ,
one always has @xmath80 ( since @xmath81 ) .
positivity is regained in the weak coupling limit .
here it is convenient to consider the limit in the sense of short correlation time so @xmath82 .
we get from eqs .
( [ s],[repar ] ) = , ( = ^2 ) in the limit @xmath26 , we get for @xmath83 , [ dir - noise ] = - [ h_0 , [ h_0 , ] ] , with time independent positive dephasing rate : 0<j(0)=_-^ j(u ) du=_02_0^s/^2 j(u ) du for @xmath84 with @xmath85 near the first minimum .
[ fig : gammat ] , scaledwidth=40.0% ]
our aim is to obtain an approximate generator that is valid for small nonzero @xmath21 .
the first step is to show that @xmath71 defined in eq .
( [ k ] ) remains an approximate generator in the noncommutative case .
more precisely , moving the time ordering @xmath37 in the exact formula eq .
( [ u2 ] ) into the exponential , comes with the penalty : @xmath86 here @xmath87 denotes time ordering with respect to the integration variable @xmath88 .
this differs from the usual time ordering defined with respect to the argument of the hamiltonian .
the error term results from the inequivalence of the two types of ordering .
it is proportional to @xmath39 and hence is cumulative .
it reflects the non - commutativity of the hamiltonian at different times ( there is no error in the commmutative case as we have seen in section [ s : commute ] ) . in particular , on the coarse grained time scale @xmath89
the error is @xmath90 .
we justify this estimate in appendix [ s : lin ] . as we have seen ( again in section [ s : commute ] )
@xmath71 may not be a lindbladian .
our next step is to show that within the framework of weak coupling , @xmath71 is close to a generator that is of the lindbladian form up to an @xmath90 error . to show this we introduce a useful representation of the noise @xmath53 in terms of white noises @xmath91 : [ jj ] _
( t)= _ -^du j_(t - u ) w_(u ) ( summation implied ) where ( w_(t))=0 , ( w_(t)w_(u))=_(t - u ) there is freedom in defining @xmath92 which allows us to assume , w.l.o.g .
, that its fourier transform is non - negative , @xmath93 .
@xmath2 is then the convolution of @xmath92 with itself : [ convolution ] j_(u - v)=_-^dw j_(u - w)j_(v - w ) and the corresponding @xmath94 , a bessel function .
@xmath92 is narrower than @xmath2.,scaledwidth=40.0% ] with this notation in place we first note the identity @xmath95 which follows from eq .
( [ convolution ] ) .
the second step is the claim that we can interchange the limits of the @xmath96 and @xmath97 integration in eq .
( [ error ] ) up to a small error , i.e. @xmath98 where [ lin ] _ ( w)=1 2 _ _ -^ du j_(w - u)^i_(u)this follows from the fact that @xmath99 is localized near the origin on a time scale @xmath100 .
it is clear that the main contribution to the integral comes from the region where both @xmath101 $ ] .
the error corresponds to contributions where @xmath102 are in an @xmath100 neighborhood of the interval endpoints ( see fig .
( [ err ] ) ) . as this region has volume @xmath103 and
the integrand is @xmath104 we conclude that the error is of magnitude @xmath105 , whereas the first term is of order @xmath106 which dominates the error .
this proves eq .
( [ error2 ] ) with an error uniform in time .
( 4,0 ) ; ( -1/2,-1/2 + 1/4)(3 - 1/4,3 ) ; ( -1/2,-1/2 - 1/4)(3,3 - 1/4 ) ; ( -1/2,2 ) (3,2 ) ; ( 2,0 ) (2,3 ) ; at ( 4,0 ) v ; at ( 0,3 ) w ; at ( 0,2 ) t ; at ( 2,0 ) t ; ( 2,2 )
( 2 + 1/4,2 )
( 2,2 - 1/4 )
( 2,2 ) ; ( 0,0 ) ( 0 - 1/4,0 )
( 0,0 + 1/4 )
( 0,0 ) ; ( 2,2 )
( 2,2 + 1/4 ) ( 2 - 1/4,2 ) ( 2,2 ) ; ( 0,0 ) ( 0 + 1/4,0 )
( 0,0 - 1/4 )
( 0,0 ) ; it follows that for @xmath21 small the ( time - dependent ) super - operator [ main - o ] _
= - 1 2 _ ( _ ( t)^2)_t generates a cp map which is @xmath90 close to @xmath65 . to rephrase @xmath107 in terms of operators , rather than super - operators , use [ order ] 2((u ) ( v))_t= \ { ( u),(v)}+(u - v ) [ ( u),(v ) ] and the dictionary in eq .
( [ prod],[jacobi ] ) gives
a time dependent generator @xmath108-{1\over2 } \sum_\alpha [ { d}_\alpha(t ) , [ { d}_\alpha(t),\rho]]\\ h^{ren}(t)&=\frac i 4 \sum_\alpha\int j_\alpha(u ) j_\alpha(v ) \operatorname{sgn}(u - v)~[h^i_\alpha(t+u),h^i_\alpha(t+v ) ] \ , dudv\nonumber \end{aligned}\ ] ] since the operators @xmath109 and @xmath110 are self - adjoint @xmath111 is a bona - fide time dependent generator of a cp map .
so far we kept @xmath21 small but finite and allowed arbitrary time dependence of @xmath112 .
this gives the time dependent generator of the previous section . to properly define the limit @xmath26
, one should also specify the limiting behavior of the dimensionless parameter @xmath17 .
if @xmath113 then @xmath3 is the smallest time scale and @xmath53 becomes effectively equivalent to white noise discussed in section [ white ] .
the interesting case and the one relevant to dynamic decoupling is when @xmath114 ( where @xmath115 ) . in this limit
( [ f ] ) reduces to eq .
( [ fcg ] ) . to see this note : * weak coupling may be interpreted as @xmath116 while @xmath9 .
( [ stationary ] ) then implies that @xmath117 .
* the ansatz eq .
( [ stationary ] ) says that the integral in eq .
( [ lin ] ) reduces to the sum : _ ( t)= 2 _ f j ( ) e^it ad(h _ ( ) ) * the limit @xmath26 means that [ weak ] _
0 e^is/^2= + 0 & 0 + 1 & = 0 + in the sense of distributions . *
the limiting lindbladian generates the evolution on the time scale @xmath118 , it is related to @xmath107 by @xmath119 .
it follows that for the second term in eq .
( [ f ] ) we get @xmath120 which is @xmath121 of eq .
( [ fcg ] ) .
similarly , for the first term in eq .
( [ f ] ) we get @xmath122\big ) \ , dudv\nonumber\end{aligned}\ ] ] the @xmath123 integration an be carried out explicitly to give j_(u ) j_(v ) ( v - u ) e^i(u - v ) dudv = j_(u ) ( u ) e^iu= k _ ( ) [ oded ] the figure shows @xmath124 as a function of coarse grained time @xmath20 for constant control .
the parameters are the same as in fig .
[ gammat ] , i.e. @xmath125 and @xmath126 .
the figure compares the average lindblad evolution on the course grained time scale , eq .
( [ fcg ] ) , and the time dependent lindblad evolution , eq .
( [ f ] ) .
note the different scales in the two figures .
the oscillations are tiny .
, title="fig:",scaledwidth=45.0% ]
consider the ( commutative ) stochastic hamiltonian for spin @xmath127 [ spin ] h= s_z eq .
( [ dir - noise ] ) gives the dephasing lindbladian [ dephasing ] = - 2 [ s_z,[s_z,]],=j(0 ) with a single rate parameter @xmath38 .
the coherences can be computed using the spectral properties of the super - operator of angular momentum given in appendix [ a : spin ] [ spectrum ] spectrum((ad ( s_z))^2)=\ { m^2 | m\{0 , , 2s } } the coherence decreases quadratically with the polarization @xmath128 : @xmath129 = { \left| m_1 \right\rangle}{\left\langle m_2 \right|}(m_1-m_2)^2 $ ] .
bang - bang " makes @xmath38 smaller and improves the coherence : a sequence of rapid @xmath130 rotations about an axis perpendicular to the magnetic field self - average the noise .
the @xmath130 rotations are given by the unitary : @xmath131 the controlled stochastic hamiltonian corresponding to eq .
( [ spin ] ) then takes the form : [ hbb(t ) ] h^i_(t)=(t ) s_z w(t ) where @xmath132 is the square wave @xmath133 using the fourier expansion @xmath134 and eq .
( [ fcg ] ) , we obtain the functional form of the dephasing lindbladian of eq .
( [ dephasing ] ) but with a renormalized one has @xmath135 . ] @xmath136 : [ bb ] _ b()= _
n0 since , @xmath137 as @xmath138 , in the limit @xmath139 , and there is no loss of coherence .
consider the stochastic hamiltonian with time - independent control [ zx ] h= _ control+s_x the control is effective in the sense of appendix [ effective ] . in the interaction picture
the stochastic hamiltonian has the form [ h(t ) ] h^i_(t)=(t ) ( s_x_c t + s_y _ c t ) the frequency set @xmath23 in eq .
( [ stationary ] ) has two elements , @xmath140 and @xmath141 we find , from eqs .
( [ fcg ] ) the lindbladian [ acc ] = - [ s_z,]- 2 _ jx , y [ s_j,[s_j , ] ] the two terms in @xmath62 commute .
this follows from the fact that @xmath142 give an @xmath143 representation . as in any such representaion @xmath144
is invariant under rotation around the @xmath145-axis , one has @xmath146$]=0 .
this may also be verified directly by calculating the commutators .
it follows that the first term in @xmath62 determines the imaginary part of the eigenvalues while the second term determines the real part .
the coherence is then determined by the spectrum of @xmath147 and the index denotes multiplicities . in the case of general @xmath43
the spectrum is @xmath148 as computed in appendix [ a : spin ] .
in particular using schur s lemma implies that @xmath149 is always a simple eigenvalue .
it follows that the lindbladian is depolarizing : the unique equilibrium state is the fully mixed state . for @xmath150 for a spin qubit in a stochastic magnetic field with a fixed direction as function of the rate of control @xmath151 .
the lowest line corresponds to bang - bang of eq .
( [ bb ] ) .
the two upper lines show the two nonzero eigenvalue of constant control .
( note that setting @xmath152 in the graph is not meaningful since eq .
( [ weak ] ) fails.),scaledwidth=40.0% ] the simplest case of non - commutative noise is `` planar '' noise @xmath153 eq .
( [ fcg ] ) gives the depolarizing lindbladian [ gyg7 ] = - 2 _ jx , y_j [ s_j[s_j , ] ] , _ j = j_j(0 ) @xmath154 is an effective control .
moreover , the results of sections [ s : bb ] , [ s : cc ] carry over to this case , mutatis mutandis : bang - bang leads to eq .
( [ gyg7 ] ) with @xmath155 as in eq .
( [ bb ] ) .
constant control leads to an equation similar to eq .
( [ acc ] ) up to an extra factor of @xmath156 .
isotropic noise is represented by the hamiltonian @xmath157 leading to the isotropic depolarizing lindbaldian @xmath158,\ ] ] for @xmath159 constant control is not effective .
a possible effective constant control is @xmath160 .
] one can , however , find an effective bang - bang .
matrices in eqs .
( [ uioger],[oickm ] ) by the appropriate rotation operator @xmath161 . ] the simplest version of bang - bang about all three axes is associated with the unitary @xmath162 [ uioger ] v(t)= _ 1 & t2 + _ 2 & t2 + _ 3 & t2 + & t2 this control self - averages the hamiltonian in the interaction picture to zero but leads to a non - isotropic lindblad equation ( with @xmath163 ) . in order to retain isotropy
, we choose a somewhat more complicated @xmath68 corresponding to dividing @xmath164 $ ] into 12 equal parts .
we demand @xmath165 for @xmath166 $ ] where \{v_j}_j=1 ^ 12=\{_1,_2,_3,,_2,_3,_1,,_3,_1,_2 , } [ oickm]this gives the stochastic hamiltonian h_=_s _ w_(t ) where @xmath167 takes on @xmath168\mod2\pi$ ] the values w_1(t)\{+1,-1,-1,+1,-1,-1,+1,+1,-1,+1,-1,+1}_j as function of time .
it self - averages the noise hamiltonian to zero while preserving the isotropy of the lindbladian . , scaledwidth=40.0% ] by symmetry considerations , @xmath169 ( since @xmath170\propto [ s_\alpha , s_\alpha]=0 $ ] ) .
and the depolarizing lindbldian has renormalized rates : @xmath171,\ ] ] @xmath172 is a more complicated version of eq .
( [ bb ] ) [ iso ] ( ) = 8 ^2_n0 ^ 4 ( ) p(n ) @xmath173 the stochastic harmonic oscillator provides a good model for trapped atoms , mechanical oscillators and trapped ions @xcite . since @xmath174 is not a state in an infinite dimensional hilbert space , the lindbladian associated with stochastic evolution may have no stationary state .
there are various types of noises one may consider .
the first is @xmath175 with @xmath176 and @xmath177 gaussian ( possibly correlated ) processes .
this is known as ` linear noise ' since it does not affect the frequency of the oscillator .
the interaction hamiltonian is @xmath178 @xmath179 is an effective control since @xmath44 has vanishing time average .
observe that @xmath180 and @xmath181 commute since [ xp ] [ ad(x),ad(p)]=ad([x , p])=iad()=0 it follows from eq .
( [ fcg ] ) that the lindbladian is real ( has no hamiltonian piece ) and has the form @xmath182 with matrix @xmath183 at the oscillator frequency . since @xmath180 and @xmath181 commute , and
@xmath184 and @xmath185 is a positive matrix , we have @xmath186\ ] ] @xmath149 is in the spectrum but is not associated with an eigenvalue : there is no stationary equilibrium state . in the case of noise in the frequency of the harmonic oscillator
the hamiltonian is : @xmath187 in the interaction picture one has @xmath188 hence @xmath189=\frac 1 2 ( p^2+x^2 ) , \quad \tilde h_\alpha(\pm 2 \omega_c)=\frac 1 4(-1)^{\alpha } ( p\pm i x)^2\ ] ] and the lindbladian is : = _ unitary+_dephasing+ _ parametric drive where @xmath190 , @xmath191 and @xmath192 .
while all the terms @xmath193 are eigenstates of the dephasing part with eigenvalues @xmath194 the parametric drive part does not have a steady state and drives the system towards the infinite temperatures .
numerical algorithm for solving stochastic evolution equations have two advantages : they can work also beyond weak coupling and evolve states rather than density matrices .
they also have several disadvantage : they tend to be slow because of the necessity to accumulating enough statistics ; they are prone to long time drifts , and can be adversely affected by a poor random number generator and finally are prone to bugs . our results on the lindbland evolutioon can be used to test numerical algorithms for stochastic evolutions in those cases that both apply . a comparison between lindbladian evolutions of sections [ s : nc],[s : bb],[s : cc ] and stochastic evolutions with orenstein - uhlenbeck process is shown in fig . [ gammat ] .
three cases have been studied : no control , control by constant @xmath179 and bang - bang .
the weak coupling parameter is @xmath125 and the agreement is satisfactory .
the numerical code is available upon request . as a function of the coarse grained time @xmath20 , eq .
( [ s ] ) for stochastic evolutions and the corresponding lindbladians . the weak coupling , eq .
( [ e ] ) , is @xmath195 .
the numerical grid @xmath196 and correlation time @xmath197 .
the control parameter is @xmath198 and the stochastic averaging is done on an ensemble of @xmath199 runs .
, scaledwidth=90.0% ]
we derived the lindbladian for controlled weakly stochastic evolutions both for small but finite @xmath21 and in the limit @xmath26 for stationary control .
our results can be used to measure the power spectrum of the noise and to test numerical algorithms for solving stochastic evolution .
several examples are studied in detail .
we thank nir bar - gil for drawing our attention to ref .
@xcite , amos nevo and especially ori hirschberg and martin fraas for useful discussions .
the research is supported by isf , the eu project diadems , the marie curie career integration grant ( cig ) and the eu strep project equam .
the purpose of this appendix is to justify the estimate of eq .
( [ u1 ] ) .
this requires a comparison of two different time orderings of the same exponent .
let us first ignore the ordering and consider e^x = , x=-_0^t ds ( s ) the taylor series for the exponent @xmath200 is dominated by terms of order @xmath201 . in our case
this gives @xmath202 .
writing the n - th term in the expansion @xmath203 we conclude that typically @xmath204 .
( 12,0 ) ; ( 1,0 ) to [ out=60 , in=120 ] ( 2,0 ) ; ( 5 - 1.,0 ) to [ out=60 , in=120 ] ( 8 - 1.,0 ) ; ( 6 - 1.,0 ) to [ out=60 , in=120 ] ( 7 - 1.,0 ) ; ( 5 + 3.5,0 ) to [ out=60 , in=120 ] ( 7 + 3.5,0 ) ; ( 6 + 3.5,0 ) to [ out=60 , in=120 ] ( 8 + 3.5,0 ) ; ( 1,0 ) circle [ radius=0.1 ] ; ( 2,0 ) circle [ radius=0.1 ] ; ( 5 - 1.,0 ) circle [ radius=0.1 ] ; ( 6 - 1.,0 ) circle [ radius=0.1 ] ; ( 7 - 1.,0 ) circle [ radius=0.1 ] ; ( 8 - 1.,0 ) circle [ radius=0.1 ] ; ( 5 + 3.5,0 ) circle [ radius=0.1 ] ; ( 6 + 3.5,0 ) circle [ radius=0.1 ] ; ( 7 + 3.5,0 ) circle [ radius=0.1 ] ; ( 8 + 3.5,0 ) circle [ radius=0.1 ] ; at ( 1,0 ) @xmath205 ; at ( 5 - 1.,0 ) @xmath206 ; at ( 6 - 1.,0 ) @xmath207 ; at ( 2,0 ) @xmath208 ; at ( 7 - 1.,0 ) @xmath209 ; at ( 8 - 1.,0 ) @xmath210 ; at ( 5 + 3.5,0 ) @xmath211 ; at ( 6 + 3.5,0 ) @xmath212 ; at ( 7 + 3.5,0 ) @xmath213 ; at ( 8 + 3.5,0 ) @xmath214 ; ( 5 - 1.,1 ) ( 8 - 1.,1 ) ; at ( 6,1 ) @xmath3 ; ( 3,-1 ) ( 5,-1 ) ; at ( 4,-1 ) @xmath39 ; next let us consider the possible time orderings .
the naive @xmath215 time ordering with respect to the argument of @xmath216 implied by eq .
( [ xx ] ) differs from the correct @xmath37-ordering because the relation between @xmath71 and @xmath49 [ yy ] ( s_i)=-12_u_i > v_i
du_i dv_i j(u_i - s_i)j(v_i - s_i)((u_i)(v_i ) ) is non -local in time : the ordering of @xmath217 does not guarantee the ordering of @xmath218 .
the fact that @xmath219 is fast decaying implies , however , that the nonlocality in time is rather small @xmath220 . when @xmath221 the wrong ordering is almost the same as the correct one . in order to estimate the error generated by using the @xmath215 ordering consider more closely the two orderings .
each contribution to the exponent is given as in eq .
( [ xx ] ) by some choice of @xmath222 and we associate a choice of @xmath223 to each @xmath217 as in eq .
. typically @xmath224 and hence @xmath225 which implies that the two ordering are equivalent .
if however there exists some @xmath226 for which @xmath227 then the two expressions do not coincide .
consider for example the case where @xmath228 while all other points are at typical positions .
this will lead to an error term of the type @xmath229{{\cal { h } } } ( v_{j})\\ \times j(s_{j}-u_{j})j(s_{j}-v_{j})j(s_{j+1}-u_{j+1})j(s_{j+1}-v_{j+1 } ) \times { { \cal { u } } } _ { j-1, ... 1 } \ ] ] here @xmath230 correspond to the ( unitary ) evolution before @xmath231 and after @xmath232 .
the integrand in eq([dd])is clearly fast decaying whenever its six integration variables are at inter - distance large compared to @xmath3 .
it thus follows that the main contribution to the integral comes from a region of volume @xmath233 .
the integral is thus at most changes slowly in time , then a tighter bound on the commutator is possible . ] of order of @xmath234 .
other nontypical cases ( e.g. @xmath235 ) lead to error terms of a similar general form which again scale as @xmath236 .
the error terms we found are of the form @xmath237 for some @xmath238 which is quartic in @xmath49 .
this suggests defining an improved generator as @xmath239 .
we however did not pursue this direction here .
the adjoint representation @xmath240 of a representation @xmath43 is constructed as the tensor product of @xmath43 with its dual ( contragredient ) representation @xmath241 . since @xmath143 has a single representaions in each dimension , it is obvious that @xmath242 .
it thus follows that @xmath243 the spectrum ( including multiplicities ) of various operators such as @xmath244 and @xmath245 is then easily deduced @xmath246 @xmath247 @xmath248 in particular the eigenvalue zero appears in @xmath249 with trivial multilicity 1 . this last fact
could also be deduced from schur s lemma since by positivity @xmath250 imply @xmath251 and hence also @xmath252\rho=0 $ ] .
in dynamical decoupling one is interested in making @xmath62 small at the price of strong control , @xmath253 . since @xmath254 is small for large argument and since the terms @xmath255 in eq .
( [ fcg ] ) tend to be of order @xmath256 the bad term " in @xmath62 is the one with @xmath257 .
we say that the control is effective " if @xmath258 .
the notion is independent of @xmath259 , which is often not known .
consider first strong continuous controls .
let @xmath260 be the ( instantaneous ) spectral projections of @xmath5 : @xmath261 and suppose that @xmath260 vary smoothly with @xmath39 and that the @xmath262 do not cross .
then , by the adiabatic theorem , for @xmath15 large @xmath263 ( in the sense of distributions . )
it follows that the control is effective if , for all @xmath39 , [ effective ] _ jp_j(t)h_p_j(t)=0 bang - bang at times @xmath264 is effective if @xmath258 , which is the case if @xmath265 has zero average , i.e. @xmath266 ido almog , yoav sagi , goren gordon , guy bensky , gershon kurizki , and nir davidson .
direct measurement of the system - environment coupling as a tool for understanding decoherence and dynamical decoupling .
, 44:154006 , 2011 .
berthold - georg englert and giovanna morigi .
five lectures on dissipative master equations . in andreas buchleitner and klaus hornberger , editors ,
_ coherent evolution in noisy environments _ , volume 611 of _ lecture notes in physics _ , pages 55106 .
springer berlin heidelberg , 2002 .
carlos meriles , liang jiang , garry goldstein , jonathan hodges , jeronimo maze , mikhail lukin , and paola cappellaro . imaging mesoscopic nuclear spin noise with a diamond magnetometer .
, 133:124105 , 2010 .
y romach , c mller , t unden , l. j. rogers , t isoda , k.m .
itoh , m. a. markham , stacey , j meijer , s pezzagna , b naydenov , l.p .
mcguinness , n. bar - gill , and f jelezko .
nuclear magnetic resonance spectroscopy on a ( 5-nanometer)3 sample volume . , 1404:3879 , 2014 .
|
we construct lindbladians associated with _ controlled _
stochastic hamiltonians in weak coupling .
this allows to determine the power spectrum of the noise from measurements of dephasing rates ; to optimize the control and to test numerical algorithms that solve controlled stochastic schrdineger equations .
a few examples are worked out in detail .
|
quantum entanglement is a key resource for quantum information processing and quantum communication .
one now has a fairly good understanding of how to produce entanglement among microscopic systems , such as atoms @xcite , ions @xcite , or artificial qubits @xcite . in recent years
there has been considerable interest in studying macroscopic entanglement in the context of cavity optomechanics @xcite , where the great experimental advance has made it possible to study the quantum effects and the preparation of nonclassical states for the optomechanical systems , including the realization of the squeezed light @xcite , squeezed mechanical resonator @xcite , macroscopic superposition state @xcite , and mechanical entanglement @xcite .
macroscopic entanglement of mechanical resonators plays a key role in testing the fundamental principles of quantum mechanics , quantum information processing , and ultrahigh - precision measurements @xcite .
the generation of stationary entanglement with two or several mechanical objects in an optical cavity has been widely discussed @xcite .
very recently , the coupling scheme between two cascaded optomechanical setups has also been proposed .
the quantum entanglement between two macroscopic mechanical resonators can be generated with the systems being connected by an optical fiber or direct photon tunneling @xcite , but the mechanical entanglement is small and limited to the enviromental bath of low temperature .
chen et al . have found that the entanglement can be enhanced by using a periodically modulated pump laser @xcite , where the chirped frequency needs to be subtlely controlled and is therefore sensitive to fluctuation of experimental parameters . further increasing
the mechanical entanglement may require the strong single - photon coupling , and the mechanical resonators should be firstly cooled down to the quantum ground state @xcite .
recent studies have found that including an optical parametric amplifier inside a cavity could affect the normal - mode splitting behavior between the coupled movable mirror and the cavity field @xcite and significantly enhance the optomechanical effective coupling @xcite .
moreover , squeezing of cavity modes can considerably improve the cooling of the micromechanical mirror @xcite and the squeezing properties can be transferred from photons to phonons with high efficiency @xcite . here
, we propose a scheme to enhance the mechanical stationary entanglement for two distant movable mirrors by placing two opas inside two coupled optomechanical cavities .
the cavity fields are coupled through a photon - hopping interaction .
it is shown that the limits for the stability condition for the coupled optomechanical system can be relaxed by controlling the parametric gain , and therefore , the stable region of the corresponding parameter space expands .
the enhancement of mechanical entanglement can be achieved by appropriately tuning the parametric gain and the effective laser - cavity detuning , and is even more pronounced with laser blue - detuning from cavity resonance .
in addition , since the difficulty of preserving quantum entanglement of macroscopic systems is often attributed to environment - induced decoherence , we also show that the opas inside the cavities can help to witness the steady - state entanglement between the mechanical resonators suffered from certain thermal noise .
the paper is organized as follows . in section
ii , we introduce the setup with two coupled cavity optomechanical systems and build the corresponding theoretical model . in section iii , we give the covariance matrix for the system and use it to calculate the logarithmic negativity , which quantifies the steady - state mechanical entanglement . in section iv , we present the effect of the opas on the generated mechanical entanglement by numerical simulation .
finally , we summarize the results in section v.
we consider a physical setup consisting of two identical fabry - perot cavities , and each cavity contains a degenerate optical parametric amplifier(opa ) and has one movable and one fixed mirror , as plotted in fig .
the fixed mirror is partially transmitting , but the movable mirror is totally reflecting .
we assumed that each cavity only has one resonant mode , and the cavity modes are coupled via a photon - hopping interaction .
the single mode assumption is valid when the cavity coupling @xmath0 is small compared with the free spectral range ( fsr ) of each uncoupled cavity .
the cavity modes with the same frequency @xmath1 are driven equally by external lasers with frequency @xmath2 and amplitude e .
the movable mirrors are considered as quantum mechanical harmonic oscillators , with the same effective mass m , frequency @xmath3 , and damping rate @xmath4 . under the action of cavity - photon - induced radiation pressure
, the movable mirrors will make oscillations around their equilibrium positions . for the degenerate opa
, we assume that the pump fields interact with the second - order nonlinear optical crystals at frequency @xmath5 , with the gain of the opas being @xmath6 , which depends on the pumping power .
the phase of the pump field is assumed to be @xmath7 . in the rotating frame at the laser frequency @xmath2
, the hamiltonian of the system reads @xmath8 ( color online ) schematic diagram of the cascaded cavity optomechanical system .
each cavity contains a nonlinear crystal that is pumped by an external laser .
the two optomechanical subsystems couple to each other via a photon - hopping interaction . ] @xmath9+\lambda({\textstyle a_{1}^{\dagger}a_{2}+a_{1}{\textstyle a_{2}^{\dagger}})},\end{array}\ ] ] where @xmath10 is the laser detuning from the cavity resonance frequency @xmath1 ; @xmath11 and @xmath12 are the annihilation and creation operator of the cavity modes , and @xmath13 and @xmath14 are the dimensionless position and momentum operators for the movable mirrors , satisfying the standard canonical commutation relation @xmath15=i$ ] .
the parameter @xmath16 is the single - photon coupling strength between the cavity and the mirror due to radiation pressure , where @xmath17 is the zero - point fluctuation of the mirror and l is the rest length of the cavity .
@xmath18 is related to the input laser power p , with @xmath19 being the cavity decay rate due to the leakage of photons through the partially transmitting mirror .
a proper analysis of the system must take the cavity decay and the mechanical damping into account , according to the heisenberg equation of motion , the dissipative dynamics of the full system is described by the following set of nonlinear quantum langevin equations ( qles ) @xcite : @xmath20 here @xmath21 are the independent input quantum noise operators with zero mean value , and their nonzero correlation function are given by @xcite @xmath22{cll } \left\langle { \textstyle a_{j}^{in\dagger}(t)a_{j}^{in}(t')}\right\rangle & = & n_{a}\delta(t - t'),\\ { \textstyle \left\langle a_{j}^{in}(t)a_{j}^{in\dagger}(t')\right\rangle } & = & ( n_{a}+1)\delta(t - t ' ) , \end{array}\ ] ] with @xmath23^{-1}$ ] being the mean thermal excitation number in the optical mode , and @xmath24 being the boltzmann constant . the hermitian brownian noise operators @xmath25 and @xmath26 have the quantum statistical properties @xcite @xmath22{rll } \left\langle \xi_{1}(t)\right\rangle & = & \left\langle \xi_{2}(t)\right\rangle = 0,\\ \left\langle { \textstyle \xi_{i}(t)\xi_{j}(t')}\right\rangle & = & \frac{\gamma_{m}}{\omega_{m}}\int\frac{d\omega}{2\pi}e^{-i\omega(t - t')}\omega[1+coth(\frac{\hbar\omega}{2k_{b}t})]\delta_{ij } , \end{array}\ ] ] where i , j @xmath27 and @xmath28 is the kronecker delta function . by assuming the mechanical quality factor @xmath29 , the correlation function of the noise @xmath30 becomes @xmath31 where @xmath32^{-1}$ ] is the mean thermal excitation number in the movable mirror . in the presence of strong laser driving , we can always rewrite each heisenberg operator as linear sum of the classical mean@xmath33 and the quantum fluctuation @xmath34 with @xmath35 , i.e. @xmath36 by inserting eq .
( [ eq : fluctuation ope ] ) into eq .
( [ eq : qles_initial ] ) , it is easy to find the steady - state momentum @xmath37 and displacement @xmath38 of the movable mirrors , and the steady - state amplitude @xmath39 of the cavity fields given by @xmath40 with @xmath41 being the effective input - laser detuning from the cavity resonance including the frequency shifts caused by the optomechanical interaction @xcite , and the linearized quantum langevin equation for the quantum fluctuation operators as following @xmath22{lll } \delta\dot{q}_{1 } & = & \omega_{m}\delta p_{1},\\ \delta\dot{p}_{1 } & = & -\omega_{m}\delta q_{1}-\gamma_{m}\delta p_{1}+g{\textstyle ( a_{s}^{*}\delta a_{1}+a_{s}\delta{\textstyle a_{1}^{\dagger}})}+\xi_{1}(t),\\ \delta\dot{q}_{2 } & = & \omega_{m}\delta p_{2},\\ \delta\dot{p}_{2 } & = & -\omega_{m}\delta q_{2}-\gamma_{m}\delta p_{2}+g{\textstyle ( a_{s}^{*}\delta a_{2}+a_{s}\delta{\textstyle a_{2}^{\dagger}})}+\xi_{2}(t),\\ \delta\dot{a}_{1 } & = & -(\kappa+i\delta_{0})\delta a_{1}+iga_{s}\delta q_{1}+igq_{s}\delta a_{1}+2\lambda e^{i\theta}\delta{\textstyle a_{1}^{\dagger}}-i\lambda\delta a_{2}+\sqrt{2\kappa}{\textstyle a_{1}^{in}(t),}\\ \delta\dot{a}_{2 } & = & -(\kappa+i\delta_{0})\delta a_{2}+iga_{s}\delta q_{2}+igq_{s}\delta
a_{2}+2\lambda e^{i\theta}\delta{\textstyle a_{2}^{\dagger}}-i\lambda\delta a_{1}+\sqrt{2\kappa}{\textstyle a_{2}^{in}(t ) , } \end{array}\label{eq : operator quantum}\ ] ] where the nonlinear terms @xmath42 and @xmath43 have been safely neglected @xcite .
now , we start looking into the measurement of the steady - state entanglement between the two mechanical modes . by introducing the amplitude and phase fluctuations @xmath44 and @xmath45 for the cavity modes , and @xmath46 and @xmath47 for the input quantum noises , the linearized qles for the quantum fluctuation in eq .
( [ eq : operator quantum ] ) can be rewritten as the matrix form @xmath48 where@xmath49 is the column vector of continuous variables(cv ) fluctuations , @xmath50 is the column vector of the noise sources , and m is the drift matrix given by @xmath51 with @xmath52 @xmath53 @xmath54 @xmath55 here , @xmath56 and @xmath57 are the real and imaginary parts of the effective coupling @xmath58 , respectively .
the stability of the system demands that all the eigenvalues of the drift matrix m has negative real parts by applying the routh - hurwitz criterion @xcite . for the later discussion of the mechanical entanglement in the present work , the stability condition for all situations are carefully verified .
as the stability condition is fulfilled , the system will finally evolve into a unique steady state independent of the initial condition .
since the dynamics of the fluctuations is linearized , and @xmath30 and @xmath59 are zero - mean quantum gaussian noises , the steady state for the quantum fluctuations must be a gaussian state @xcite , for which all the quantum correlations can then be characterized by the @xmath60 covariance matrix(cm ) v with @xmath61 , where @xmath62 denotes the vector of the fluctuation operators at the steady state @xmath63 ) . under the stable condition , the cm elements take the form @xmath64 where @xmath65 , and @xmath66 is the matrix of the stationary noise correlation functions .
for the mechanical oscillators with large quality factor @xmath29 , the quantum brownian noise becomes delta - correlated @xmath67{ccl } \phi_{kl}(t - t ' ) & = & d_{kl}\delta(t - t')\\ & = & diag[0,\gamma_{m}(2n_{m}+1),0,\gamma_{m}(2n_{m}+1),\kappa(2n_{a}+1),\kappa(2n_{a}+1),\kappa(2n_{a}+1),\kappa(2n_{a}+1)]\delta(t - t ' ) , \end{array}\ ] ] then the eq .
( [ eq : cm initial ] ) can be simplified to @xmath68 finally , the cm v satisfies the lyapunov equation @xcite @xmath69 the mechanical entanglement can be measured by the logarithmic negativity @xmath70 @xcite . to calculate @xmath70 , we firstly numerically solve the lyapunov equation above and find the solution for v , and then exact the reduced @xmath71 cm @xmath72 for the two mechanical modes from the full @xmath60 cm v by just keeping the first four rows and columns . by rewritting cm @xmath72 in the following form @xmath73 with a , b and c being the @xmath74 sub - block matrices of @xmath72 ,
the logarithmic negativity @xmath70 is finally given by @xmath75,\ ] ] with @xmath76 where @xmath77 .
if the two mechanical objects are entangled , the value of @xmath78 must be less than 1/2 , which is equivalent to simon s necessary and sufficient entanglement nonpositive partial transpose criterion for gaussian states @xcite .
( color online ) ( a ) the mechanical entanglement @xmath70 versus the parametric amplifier phase @xmath7 and parametric gain @xmath6 .
( b ) @xmath70 versus the parametric amplifier phase @xmath7 for different parametric gain , @xmath79(red curve ) , @xmath80(blue curve ) , @xmath81(black curve ) .
other parameters are ( in units of @xmath3 ) : @xmath82 , @xmath83 , @xmath84 , @xmath85 , @xmath86 , @xmath87 , @xmath88 , @xmath89 @xcite . ]
firstly , we examine the effect of the parametric amplifier phase @xmath7 on the mechanical entanglement @xmath70 , as shown in fig . [ fig2 ] . in all cases
, the blue region represents that the system is unstable . for weak parametric gain @xmath6 ,
the mechanical system remains unstable independent of the amplifier phase .
the steady - state mechanical entanglement can only be achieved for @xmath90 and @xmath7 being limited between @xmath91 and @xmath92 .
we can also see from the fig .
[ fig2](b ) that the maximal entanglement @xmath70 always appears at @xmath93 for varied parametric gain ensuring the stability of the system .
this fact is true as well for negative laser detuning @xmath94 , which is checked numerically .
therefore , in the following , the parametric amplifier phase will be taken as @xmath93 in all cases .
( color online ) the mechanical entanglement @xmath70 versus driving amplitude @xmath95 and cavity - cavity coupling strength @xmath0 with the parametric gain ( a ) @xmath96 and ( b ) @xmath97 other parameters are identical to those in fig .
[ fig2 ] . ] figs . [ fig3](a ) and ( b ) show the mechanical entanglement @xmath70 as a function of the driving amplitude @xmath95 and the coupling strength @xmath0 with opas and without opas inside the cavities , respectively . in both cases
, it can be seen that the mechanical entanglement @xmath70 increases with increasing driving amplitude @xmath95 and cavity - cavity coupling strength @xmath0 . for fixed @xmath0 ,
increasing laser driving strength does not implies that the mechanical entanglement will be enhanced .
the system becomes unstable when the driving amplitude @xmath95 reaches certain limit .
however , we find that in the presence of the opas @xmath98 , the stable region for obtaining the steady - state mechanical entanglement expands in the parameter space ( @xmath0 , @xmath95 ) , that is the reasonable high steady - state entanglement can be achieved for smaller cavity - cavity coupling @xmath99 .
( color online ) the mechanical entanglement @xmath70 for cavity - cavity coupling strength @xmath100 versus the parametric gain @xmath6 and the driving amplitude @xmath95 .
other parameters are same as those in fig .
[ fig2 ] . ]
the limit for the laser driving strength @xmath95 can be enlarged with the assistance of opas , as can be seen in fig .
[ fig4 ] . while for parametric gain @xmath96
, the optomechanical system becomes unstable for @xmath101 , the steady - state entanglement @xmath102 is accessible even for @xmath103with parametric gain @xmath104 .
the maximal entanglement in this case can be found at @xmath105 and @xmath106 .
( color online ) the steady - state mechanical entanglement @xmath70 versus parametric gain @xmath6 and cavity - cavity coupling strength @xmath0 , and the effective cavity detuning ( a ) @xmath89 and ( b ) @xmath107 .
other parameters are the same as in fig .
[ fig2 ] . ] as we have seen in the fig .
[ fig3 ] , the steady - state mechanical entanglement is also limited by the cavity - cavity coupling @xmath0 , namely , the larger the cavity - cavity coupling is , the stronger laser driving strength required . however
, this limit can also be released by using the opas , as shown in fig .
[ fig5 ] , where the steady - state mechanical entanglement @xmath70 for positive laser detuning @xmath89 [ fig .
[ fig5](a ) ] and negative laser detuning @xmath107 [ fig .
[ fig5](b ) ] as a function of the parametric gain @xmath6 and the coupling strength @xmath0 are plotted . setting the driving amplitude @xmath88 , in the absence of opas , the fulfillment of the stability condition is impossible for @xmath108 with @xmath89 , and for @xmath109 with @xmath107 .
the involvement of the opas with finite parametric gain can help to stably entangle the two mechanical oscillators under the condition of less cavity - cavity coupling @xmath0 , which may be significant for relieving experimental constraints .
moreover , the steady - state mechanical entanglement @xmath70 are enhanced by the parametric gain in both cases for the same cavity - cavity coupling .
( color online ) the mechanical entanglement @xmath70 versus parametric gain @xmath6 and the effective cavity detuning @xmath94 .
other parameters are same as the fig .
the coordinate data at @xmath110 , @xmath111 @xmath112 @xmath113 and @xmath114 are indicated , respectively , by right triangle , upper triangle , circular , square , and diamond . ]
the effect of the opas on the steady - state mechanical entanglement can be further seen in fig .
[ fig6 ] , where we show the mechanical entanglement @xmath70 versus the parametric gain @xmath6 and the effective cavity detuning @xmath94 .
when the opas are absent @xmath115 , the maximal mechanical entanglement can only reach @xmath116 for the effective detuning around @xmath117 .
then , it becomes nearly double @xmath118 at @xmath119 with the parametric gain @xmath120 . on the other hand ,
the stability of the system is confined in a band area with the increasing the parametric gain @xmath6 . for @xmath121
, the stability can be found even at negative detuning @xmath122 , where the input laser may be blue - detuned from the cavity resonance ( i.e. @xmath123 ) .
this intriguing effect is very different from the situation without opas @xcite . the mechanical entanglement @xmath70 as a function of the average thermal occupancy of the two mirrors @xmath124 is shown in fig .
[ fig7 ] , where we find that @xmath70 almost decreases linearly as @xmath124 increases .
thus , for a non - zero thermal reservoir for the mechanical oscillators , the well selected laser detuning and parametric gain can help to preserve the mechanical entanglement .
( color online ) the mechanical entanglement @xmath70 versus the thermal phonon number @xmath124 .
the curves are corresponding to the different coordinates indicated in fig .
other parameters are same as the fig .
[ fig6 ] . ]
in conclusion , we have discussed the stability of the coupled optomechanical systems and the enhancement of mechanical entanglement for the two movable mirrors by using the opas .
we have found that the stable region in the parameter space consisting of photon - hopping rate and laser driving strength can be enlarged due to the parametric amplification .
comparing with the case without opas , a well - selected parametric gain can help to stabilize the system for increasing laser intensity and negative laser detuning , and enhance the steady - state mechanical entanglement .
the application of opas can be good for preserving the mechanical entanglement suffered from the decoherence induced by certain temperature limit of the environment .
the scheme discussed here provides an alternative method for improving the quantum entanglement of two distant mechanical oscillators in addition to quantum reservoir engineering and periodic modulation with chirped driving laser .
this work is supported by the major state basic research development program of china under grant no .
2012cb921601 ; the national natural science foundation of china under grants nos . 11305037 , 11347114 , 11374054 and 11422437 .
the fund from fuzhou university csh thanks ze - lin zhang and du ran for useful discussions .
paternostro , m. , vitali , d. , gigan , s. , kim , m. s. , brukner , c. , eisert , j. , aspelmeyer , m. : creating and probing multipartite macroscopic entanglement with light .
lett . * 99 * , 250401(4 ) ( 2007 ) .
chen , r. x. , shen , l. t. , yang , z. b. , wu , h. z. , zheng , s. b. : enhancement of entanglement in distant mechanical vibrations via modulation in a coupled optomechanical system .
rev . a * 89 * , 023843(9 ) ( 2014 ) .
vitali , d. , gigan , s. , ferreira , a. , b?hm , h. r. , tombesi , p. , guerreiro , a. , aspelmeyer , m. : optomechanical entanglement between a movable mirror and a cavity field .
lett . * 98 * , 030405(4 ) ( 2007 ) .
|
we theoretically investigate the stability of a two cascaded cavity optomechanical system with optical parametric amplifiers ( opas ) inside the two coupled cavities , and study the steady - state entanglement between two distant mechanical resonators .
we show that the parameter regime where the system is unstable without opas , such as relatively high laser intensity and blue detuning , can be exploited to build the steady - state mechanical entanglement by modulating the parametric gain .
the application of opas is helpful to preserve the mechanical entanglement suffered from the dissipation at some finite temperature .
the scheme provides an alternative way for improving and engineering the quantum entanglement of two distant mechanical oscillators .
|
-5 mm the organisers requested a talk which would both be a colloquium style talk understandable to a wide spectrum of mathematicians and one which would survey the recent developments in the subject .
i have found it hard to meet both desiderata , and have opted to concentrate on the former .
thus the first three sections of this paper contain a simple presentation of a web of deep conjectures connecting galois representations to algebraic geometry , complex analysis and discrete subgroups of lie groups .
this will be of no interest to the specialist .
my hope is that the result is not too banal and that it will give the non - specialist some idea of what motivates work in this area .
i should stress that nothing i write here is original . in the final section
i briefly review some of what is known about these conjectures and _ very briefly _ mention some of the available techniques .
i also mention two questions which lie outside the topic we are discussing , but which would have important implications for it .
maybe someone can make progress on them ?
we will let @xmath1 denote the field of rational numbers and @xmath2 denote the field of algebraic numbers , the algebraic closure of @xmath1 .
we will also let @xmath3 denote the group of automorphisms of @xmath2 , that is @xmath4 , the absolute galois group of @xmath1 .
although it is not the simplest it is arguably the most natural galois group to study .
an important technical point is that @xmath3 is naturally a ( profinite ) topological group , a basis of open neighbourhoods of the identity being given by the subgroups @xmath5 as @xmath6 runs over subextensions of @xmath7 which are finite over @xmath1 . to my mind
the galois theory of @xmath1 is most interesting when one looks not only at @xmath3 as an abstract ( topological ) group , but as a group with certain additional structures associated to the prime numbers .
i will now briefly describe these structures . for each prime number @xmath8
we may define an absolute value @xmath9 if @xmath10 with @xmath11 and @xmath12 integers coprime to @xmath8 .
if we complete @xmath1 with respect to this absolute value we obtain the field of @xmath8-adic numbers @xmath13 , a totally disconnected , locally compact topological field .
we will write @xmath14 for its absolute galois group , @xmath15 .
the absolute value @xmath16 has a unique extension to an absolute value on @xmath17 and @xmath14 is identified with the group of automorphisms of @xmath17 which preserve @xmath16 , or equivalently the group of continuous automorphisms of @xmath17 . for each embedding @xmath18
we obtain a closed embedding @xmath19 and as the embedding @xmath18 varies we obtain a conjugacy class of closed embeddings @xmath19 .
slightly abusively , we shall consider @xmath14 a closed subgroup of @xmath3 , suppressing the fact that the embedding is only determined up to conjugacy .
this can be compared with the situation ` at infinity ' .
let @xmath20 denote the usual archimedean absolute value on @xmath1 .
the completion of @xmath1 with respect to @xmath20 is the field of real numbers @xmath21 and its algebraic closure is @xmath22 the field of complex numbers .
each embedding @xmath23 gives rise to a closed embedding @xmath24 as the embedding @xmath23 varies one obtains a conjugacy class of elements @xmath25 of order @xmath26 , which we refer to as complex conjugations .
there are however many important differences between the case of finite places ( i.e. primes ) and the infinite place @xmath27 is an infinite extension and @xmath17 is not complete .
we will denote its completion by @xmath28 .
the galois group @xmath14 acts on @xmath28 and is in fact the group of continuous automorphisms of @xmath28 . the elements of @xmath13 ( resp .
@xmath17 ) with absolute value less than or equal to @xmath29 , form a closed subring @xmath30 ( resp .
@xmath31 ) . these rings are local with maximal ideals @xmath32 ( resp .
@xmath33 ) consisting of the elements with absolute value strictly less than @xmath29 .
the field @xmath34 is an algebraic closure of the finite field with @xmath8 elements @xmath35 , and we will denote it by @xmath36 .
thus we obtain a continuous map @xmath37 which is surjective .
its kernel is called the inertia subgroup of @xmath14 and is denoted by @xmath38 .
the group @xmath39 is procyclic and has a canonical generator called the ( geometric ) frobenius element and defined by @xmath40 in many circumstances it is technically convenient to replace @xmath14 by a dense subgroup @xmath41 , which is referred to as the weil group of @xmath13 and which is defined as the subgroup of @xmath42 such that @xmath43 maps to @xmath44 we endow @xmath41 with a topology by decreeing that @xmath38 with its usual topology should be an open subgroup of @xmath41 .
we will take a moment to describe some of the finer structure of @xmath38 which we will need for technical purposes later .
first of all there is a ( not quite canonical ) continuous surjection @xmath45 such that @xmath46 for all @xmath47 .
the kernel of @xmath48 is a pro-@xmath8-group called the wild inertia group .
the fixed field @xmath49 is obtained by adjoining @xmath50{p}$ ] to @xmath51 for all @xmath52 coprime to @xmath8 and @xmath53{p } = \zeta_n^{t(\sigma ) } \sqrt[n]{p},\ ] ] for some primitive @xmath54-root of unity @xmath55 ( independent of @xmath43 , but dependent on @xmath48 ) .
i want to focus here on attempts to describe @xmath3 via its representations .
perhaps the most obvious representations to consider are those representations @xmath58 with open kernel , and these so called artin representations are already very interesting .
however one obtains a richer theory if one considers representations @xmath59 which are continuous with respect to the @xmath60-adic topology on @xmath61 .
we refer to these as _
@xmath60-adic representations_. 1 . a choice of embeddings @xmath23 and @xmath62 establishes a bijection between isomorphism classes of artin representations and isomorphism classes of @xmath60-adic representations with open kernel .
thus artin representations are a special case of @xmath60-adic representations : those with finite image .
2 . there is a unique character @xmath63 such that @xmath64 for all @xmath60-power roots of unity @xmath65 .
this is called the @xmath60-adic _ cyclotomic character_. 3 .
if @xmath66 is a smooth projective variety ( and we choose an embedding @xmath67 ) then the natural action of @xmath3 on the cohomology @xmath68 is an @xmath60-adic representation .
for instance if @xmath69 is an elliptic curve then we have the concrete description @xmath70({\overline{{{{\mathbb{q } } } } } } ) , { \overline{{{{\mathbb{q}}}}}}_l ) \cong { \overline{{{{\mathbb{q}}}}}}_l^2,\ ] ] where @xmath71 $ ] denotes the @xmath72-torsion points on @xmath73 .
we will write @xmath74 for the twist @xmath75 before discussing @xmath60-adic representations of @xmath3 further , let us take a moment to look at @xmath60-adic representations of @xmath14
. the cases @xmath76 and @xmath77 are very different . consider first the much easier case @xmath78 . here
@xmath60-adic representations of @xmath14 are not much different from representations of @xmath41 with open kernel .
more precisely define a _
wd - representation _ of @xmath41 over a field @xmath73 to be a pair @xmath79 and @xmath80 where @xmath81 is a finite dimensional @xmath73-vector space , @xmath82 is a representation with open kernel and @xmath83 is a nilpotent endomorphism which satisfies @xmath84 for every lift @xmath85 of @xmath86 .
the key point here is that there is no reference to a topology on @xmath73 , indeed no assumption that @xmath73 is a topological field .
given @xmath82 there are up to isomorphism only finitely many choices for the pair @xmath87 and these can be explicitly listed without difficulty .
a wd - representation @xmath87 is called _ unramified _ if @xmath88 and @xmath89
. it is called frobenius semi - simple if @xmath82 is semi - simple . any wd - representation
@xmath87 has a canonical frobenius semi - simplification @xmath90 ( see @xcite ) .
in the case that @xmath91 we call @xmath87 @xmath60-integral if all the eigenvalues of @xmath92 have absolute value @xmath29 .
this is independent of the choice of frobenius lift @xmath93 .
if @xmath78 , then there is an equivalence of categories between @xmath60-integral wd - representations of @xmath41 over @xmath94 and @xmath60-adic representations of @xmath14 . to describe it choose a frobenius lift @xmath85 and a surjection @xmath95 .
up to natural isomorphism the equivalence does not depend on these choices .
we associate to an @xmath60-integral wd - representation @xmath87 the unique @xmath60-adic representation sending @xmath96 for all @xmath97 and @xmath47 .
the key point is grothendieck s observation that for @xmath78 any @xmath60-adic representation of @xmath14 must be trivial on some open subgroup of the wild inertia group
. we will write @xmath98 for the wd - representation associated to an @xmath60-adic representation @xmath99 .
note that @xmath98 is unramified if and only if @xmath100 . in this case
we call @xmath99 _ unramified_. the case @xmath77 is much more complicated because there are many more @xmath8-adic representations of @xmath14 .
these have been extensively studied by fontaine and his co - workers .
they single out certain @xmath8-adic representations which they call _ de rham _ representations .
i will not recall the somewhat involved definition here ( see however @xcite and @xcite ) , but note that ` most ' @xmath8-adic representations of @xmath14 are not de rham . to any de rham representation @xmath99 of @xmath14 on a @xmath17-vector space @xmath81 they associate the following . 1 . a wd - representation @xmath98 of @xmath41 over @xmath17 ( see @xcite and @xcite ) .
2 . a multiset @xmath101 of @xmath102 integers , called the hodge - tate numbers of @xmath99 .
the multiplicity of @xmath103 in @xmath101 is @xmath104 where @xmath105 denotes @xmath28 with @xmath14-action @xmath106 times the usual ( galois ) action on @xmath28 .
we now return to the global situation ( i.e. to the study of @xmath107 ) .
the @xmath60-adic representations of @xmath3 that arise ` in nature ' , by which i mean ` from geometry ' , have a number of very special properties which i will now list .
let @xmath108 be a subquotient of @xmath74 for some smooth projective variety @xmath66 and some integers @xmath109 and @xmath110 . 1 . ( grothendieck )
the representation @xmath99 is unramified at all but finitely many primes @xmath8 .
( fontaine , messing , faltings , kato , tsuji , de jong , see e.g. @xcite , @xcite ) the representation @xmath99 is de rham in the sense that its restriction to @xmath111 is de rham .
( deligne , @xcite ) the representation @xmath99 is _ pure _ of weight @xmath112 in the following sense .
there is a finite set of primes @xmath113 , such that for @xmath114 , the representation @xmath99 is unramified at @xmath8 and for every eigenvalue @xmath115 of @xmath116 and every embedding @xmath117 @xmath118 in particular @xmath115 is algebraic ( i.e. @xmath119 ) .
an amazing conjecture of fontaine and mazur ( see @xcite and @xcite ) asserts that any irreducible @xmath60-adic representation of @xmath3 satisfying the first two of these properties arises from geometry in the above sense and so in particular also satisfies the third property .
[ cfm ] suppose that @xmath120 is an irreducible @xmath60-adic representation which is unramified at all but finitely many primes and with @xmath121 de rham .
then there is a smooth projective variety @xmath66 and integers @xmath109 and @xmath110 such that @xmath81 is a subquotient of @xmath74 . in particular
@xmath99 is pure of some weight @xmath122 .
algebraic geometers have formulated some very precise conjectures about the action of @xmath3 on the cohomology of varieties .
we do nt have the space here to discuss these in general , but we will formulate , in an as algebraic a way as possible , some of their conjectures . 1 . for each prime @xmath60 and for each embedding @xmath124 @xmath125
is an irreducible subrepresentation of @xmath126 .
2 . for all indices @xmath110 and for all primes @xmath8 there is a wd - representation @xmath127 of @xmath41 over @xmath2 such that @xmath128 for all primes @xmath60 and all embeddings @xmath129 .
3 . there is a multiset of integers @xmath130 such that 1 .
for all primes @xmath60 and all embeddings @xmath129 @xmath131 2 .
and for all @xmath132 @xmath133 is the multiplicity of @xmath11 in @xmath134 .
if one considers the whole of @xmath135 rather than its pieces @xmath136 , then part 2 .
is known to hold up to frobenius semisimplification for all but finitely many @xmath8 and part 3 .
is known to hold ( see @xcite ) .
it follows from a theorem of faltings @xcite that the whole conjecture is true for @xmath137 of an abelian variety .
the putative constituents @xmath136 are one incarnation of what people call ` pure motives ' .
if one believes conjectures [ cfm ] and [ cgeo ] then ` geometric ' @xmath60-adic representations should come in compatible families as @xmath60 varies .
there are many ways to make precise the notion of such a compatible family . here
is one . by a _
weakly compatible system of @xmath60-adic representations _ @xmath138 we shall mean a collection of semi - simple @xmath60-adic representations @xmath139 one for each pair @xmath140 , where @xmath60 is a prime and @xmath141 , which satisfy the following conditions .
* there is a multiset of integers @xmath142 such that for each prime @xmath60 and each embedding @xmath129 the restriction @xmath143 is de rham and @xmath144 .
* there is a finite set of primes @xmath113 such that if @xmath114 then @xmath145 is unramified for all @xmath60 and @xmath146 .
* for all but finitely many primes @xmath8 there is a frobenius semi - simple wd - representation @xmath147 over @xmath2 such that for all primes @xmath78 and for all @xmath146 we have @xmath148 * we call @xmath149 _ strongly compatible _ if the last condition ( the existence of @xmath147 ) holds for all primes @xmath8 .
* we call @xmath149 _ irreducible _ if each @xmath150 is irreducible .
* we call @xmath149 _ pure _ of weight @xmath151 , if for all but finitely many @xmath8 and for all eigenvalues @xmath115 of @xmath152 , where @xmath153 , we have @xmath154 for all embeddings @xmath132 .
* we call @xmath149 _ geometric _ if there is a smooth projective variety @xmath66 and integers @xmath109 and @xmath110 and a subspace @xmath155 such that for all @xmath60 and @xmath146 , @xmath156 is @xmath3 invariant and realises @xmath150 . 1 .
if @xmath157 is a continuous semi - simple de rham representation unramified at all but finitely many primes then @xmath99 is part of a weakly compatible system .
any weakly compatible system is strongly compatible .
any irreducible weakly compatible system @xmath149 is geometric and pure of weight @xmath158 .
a famous theorem of cebotarev asserts that if @xmath159 is any galois extension in which all but finitely many primes are unramified ( i.e. for all but finitely many primes @xmath8 the image of @xmath38 in @xmath160 is trivial ) then the frobenius elements at unramified primes @xmath161 are dense in @xmath160 .
it follows that an irreducible weakly compatible system @xmath149 is uniquely determined by @xmath147 for all but finitely many @xmath8 and hence by one @xmath150 .
conjectures [ cfm ] and [ ccs ] are known for one dimensional representations , in which case they have purely algebraic proofs based on class field theory ( see @xcite ) .
otherwise only fragmentary cases have been proved , where amazingly the arguments are extremely indirect involving sophisticated analysis and geometry .
we will come back to this later .
@xmath0-functions are certain dirichlet series @xmath162 which play an important role in number theory . a full discussion of the role of @xmath0-functions in number theory is beyond the scope of this talk .
the simplest example of an @xmath0-function is the riemann zeta function @xmath163 it converges to a holomorphic function in the half plane @xmath164 and in this region of convergence it can also be expressed as a convergent infinite product over the prime numbers @xmath165 this is called an _ euler product _ and the individual factors are called euler factors . lying deeper is the fact that @xmath166 has meromorphic continuation to the whole complex plane , with only one pole : a simple pole at @xmath167 .
moreover if we set @xmath168 then @xmath169 satisfies the functional equation @xmath170 encoded in the riemann zeta function is lots of deep arithmetic information .
for instance the location of the zeros of @xmath166 is intimately connected with the distribution of prime numbers .
moreover its special values at negative integers ( where it is only defined by analytic continuation ) turn out to be rational numbers encoding deep arithmetic information about the cyclotomic fields @xmath171 .
another celebrated example is the @xmath0-function of an elliptic curve @xmath73 : @xmath172 in this case the @xmath0-function is defined as an euler product ( converging in @xmath173 ) @xmath174 where @xmath175 is a rational function , and for all but finitely many @xmath8 @xmath176 with @xmath177 being the number of solutions to the congruence @xmath178 in @xmath179 .
it has recently been proved @xcite that @xmath180 can be continued to an entire function , which satisfies a functional equation @xmath181 for some explicit positive integer @xmath182 .
a remarkable conjecture of birch and swinnerton - dyer @xcite predicts that @xmath183 has infinitely many rational solutions if and only if @xmath184 .
again we point out that it is the behaviour of the @xmath0-function at a point where it is only defined by analytic continuation , which is governing the arithmetic of @xmath73 .
this conjecture has been proved ( see @xcite ) when @xmath180 has at most a simple zero at @xmath167 .
one general setting in which one can define @xmath0-functions is @xmath60-adic representations . let us look first at the local setting .
if @xmath87 is a wd - representation of @xmath41 on an @xmath73-vector space @xmath81 , where @xmath73 is an algebraically closed field of characteristic zero , we define a local l - factor @xmath185 ( @xmath186 is the subspace of @xmath81 where @xmath38 acts trivially and @xmath88 . ) one can also associate to @xmath87 a conductor @xmath187 , which measures how deeply into @xmath38 the wd - representation @xmath87 is nontrivial , and a local epsilon factor @xmath188 , which also depends on the choice of a non - trivial character @xmath189 with open kernel .
( see @xcite . )
if @xmath190 is an @xmath60-adic representation of @xmath3 which is de rham at @xmath60 and pure of some weight @xmath151 , and if @xmath191 we will define an @xmath0-function @xmath192 which will converge to a holomorphic function in @xmath193 .
for example @xmath194 and if @xmath69 is an elliptic curve then @xmath195 ( for any @xmath146 ) . note the useful formulae @xmath196 also note that @xmath197 determines @xmath198 for all @xmath8 and hence @xmath199 for all but finitely many @xmath8 .
hence by the cebotarev density theorem @xmath197 determines @xmath99 ( up to semisimplification ) .
write @xmath200 for the multiplicity of an integer @xmath103 in @xmath101 and , if @xmath201 , define @xmath202 by : @xmath203 assume that @xmath204 are integers , i.e. that @xmath205
. then we can define a @xmath206-factor , @xmath207 , which is a product of functions @xmath208 as @xmath11 runs over a set of integers depending only on the numbers @xmath200 and @xmath204 .
we can also define an epsilon factor @xmath209 which again only depends on @xmath200 , @xmath204 and a non - trivial character @xmath210 .
set @xmath211 and @xmath212 ( which makes sense as @xmath213 for all but finitely many @xmath8 ) and @xmath214 where @xmath215 .
[ cfe ] suppose that @xmath99 is an irreducible @xmath60-adic representation of @xmath3 which is de rham and pure of weight @xmath151 .
then @xmath216 for all @xmath8 , so that @xmath217 . moreover the following should hold .
it is tempting to believe that something like properties 1 .
, 2 . and 3 . should characterise those euler products which arise from @xmath60-adic representations .
we will discuss a more precise conjecture along these lines in the next section .
why galois representations should be * the * source of euler products with good functional equations seems a complete mystery .
let @xmath223 denote the profinite completion of @xmath224 , i.e. @xmath225 a topological ring . also let @xmath226 denote the topological ring of finite adeles @xmath227 where @xmath223 is an open subring with its usual topology .
as an abstract ring , @xmath226 is the subring of @xmath228 consisting of elements @xmath229 with @xmath230 for all but finitely many @xmath8
. however the topology is not the subspace topology .
we define the topological ring of adeles to be the product @xmath231 note that @xmath1 embeds diagonally as a discrete subring of @xmath232 with compact quotient @xmath233 we will be interested in @xmath234 , the locally compact topological group of @xmath235 invertible matrices with coefficients in @xmath232 .
we remark that the topology on @xmath234 is the subspace topology resulting from the closed embedding @xmath236 @xmath237 is a discrete subgroup of @xmath234 and the quotient @xmath238 has finite volume .
if @xmath239 is an open subgroup with @xmath240 , then @xmath241 note that @xmath242 is a subgroup of @xmath243 of finite index .
most of the statements we make concerning @xmath234 can be rephrased to involve only @xmath244 , but at the expense of making them much more cumbersome . to achieve brevity ( and because it seems more natural ) we have opted to use the language of adeles .
we hope that this extra abstraction will not be too confusing for the novice . before continuing our introduction of automorphic forms
let us digress to mention class field theory , which provides a concrete example of the presentational advantages of the adelic language .
it also implies essentially all the conjectures we are considering in the case of one dimensional galois representations .
indeed this article is about the search for a non - abelian analogue of class field theory .
class field theory gives a concrete description of the abelianisation ( maximal continuous abelian quotient ) @xmath245 of @xmath3 and @xmath246 of @xmath41 .
first the local theory asserts that there is an isomorphism @xmath247 with various natural properties , including the facts that @xmath248 is the image of the inertia group @xmath38 in @xmath246 , and that the induced map @xmath249 takes @xmath8 to the geometric frobenius element @xmath86 .
secondly the global theory asserts that there is an isomorphism @xmath250 such that the restriction of @xmath251 to @xmath252 coincides with the composition of @xmath253 with the natural map @xmath254 .
thus @xmath251 is defined completely from a knowledge of the @xmath253 ( and the fact that @xmath251 takes @xmath255 to complex conjugation ) and the reciprocity theorem of global class field theory can be thought of as a determination of the kernel of @xmath256 .
we now return to our ( extended ) definition of automorphic forms . for each partition @xmath257
let @xmath258 denote the subgroup of @xmath259 consisting of matrices of the form @xmath260 let @xmath261 denote the orthogonal subgroup .
let @xmath262 denote the centre of the universal enveloping of @xmath263 , the complexified lie algebra of @xmath244 ( i.e. @xmath264 with @xmath265=xy - yx$ ] ) . via the harish - chandra isomorphism
( see for example @xcite ) we may identify homomorphisms @xmath266 with multisets of @xmath52 complex numbers .
we will write @xmath267 for the homomorphism corresponding to a multiset @xmath268 .
thus @xmath262 acts on the irreducible finite dimensional @xmath263-module with highest weight @xmath269 ( @xmath270 ) by @xmath271 .
fix such a multiset @xmath268 of cardinality @xmath52 .
the space of cusp forms with infinitesimal character @xmath268 , @xmath272 is the space of smooth bounded functions @xmath273 satisfying the following conditions . 1 .
( @xmath6-finiteness ) the translates of @xmath274 under @xmath275 ( where @xmath276 denotes the orthogonal group ) span a finite dimensional vector space ; 2 .
( infinitesimal character @xmath268 ) if @xmath277 then @xmath278 ; 3 .
( cuspidality ) for each partition @xmath257 , @xmath279 one would like to study @xmath283 as a representation of @xmath234 , unfortunately it is not preserved by the action of @xmath244 ( because the @xmath6-finiteness condition depends on the choice of a maximal compact subgroup @xmath261 ) .
it does however have an action of @xmath284 and of @xmath263 , which is essentially as good .
more precisely it is an admissible @xmath285-module in the sense of @xcite .
in fact it is a direct sum of irreducible , admissible @xmath286-modules each occurring with multiplicity one .
we will ( slightly abusively ) refer to these irreducible constituents as cuspidal automorphic representations of @xmath234 with infinitesimal character @xmath268 .
@xmath287 is just the space of locally constant functions on @xmath288 and so cuspidal automorphic representations of @xmath289 with infinitesimal character @xmath290 , are just the ( finite order ) complex valued characters of @xmath291 , i.e. dirichlet characters .
@xmath292 is simply obtained from @xmath293 by twisting by @xmath294 , where @xmath295 is the product of the absolute values @xmath296 .
thus in the case @xmath297 cuspidal automorphic representations are essentially dirichlet characters .
the case @xmath298 is somewhat more representative . in this case we have @xmath299 unless @xmath300 , @xmath301 or @xmath302 .
it is conjectured that the third possibility can not arise unless @xmath303 .
let us consider the case @xmath304 a little further . if @xmath305 then it turns out that the irreducible constituents of @xmath306 are in bijection with the weight @xmath307 holomorphic cusp forms on the upper half plane , which are normalised newforms ( see for example @xcite ) .
thus in some sense cuspidal automorphic representations are are also generalisations of classical holomorphic normalised newforms .
note that if @xmath308 is a character of @xmath309 and if @xmath310 is a cuspidal automorphic representation of @xmath234 with infinitesimal character @xmath268 then @xmath311 is also a cuspidal automorphic representation with infinitesimal character @xmath268 and the contragredient ( dual ) @xmath312 of @xmath310 is a cuspidal automorphic representation with infinitesimal character @xmath313 .
one of the main questions in the theory of automorphic forms is to describe the irreducible constituents of @xmath283 .
if we are to do this we first need some description of all irreducible admissible @xmath314-modules , and then we can try to say which occur in @xmath283 . just as a character @xmath315 can be factored as @xmath316 where @xmath317 ( resp .
@xmath318 ) , so an irreducible , admissible @xmath286-module can be factorised as a restricted tensor product ( see @xcite ) @xmath319 where @xmath320 is an irreducible , admissible @xmath321-module ( see for example @xcite ) , and each @xmath322 is an irreducible smooth ( i.e. stabilisers of vectors are open in @xmath323 ) representation of @xmath323 with @xmath324 for all but finitely many @xmath8 . to the factors
@xmath325 one can associate various invariants ( see @xcite ) .
( we also remark that @xmath262 acts via a character @xmath266 on any irreducible , admissible @xmath321-module @xmath332 .
this character is called the infinitesimal character of @xmath332 . )
now we may attach to @xmath310 * a central character @xmath333 ; * an @xmath0-function @xmath334 ( which may or may not converge ) ; * an extended @xmath0-function @xmath335 ; * a conductor @xmath336 ( which makes sense because @xmath337 when @xmath324 ) ; * and an epsilon constant @xmath338 where @xmath339 is any non - trivial character .
the following theorem and conjecture describe the ( expected ) relationship between automorphic forms and @xmath0-functions with euler product and functional equation .
we suppose @xmath340 .
a similar theorem to theorem [ gjt ] is true for @xmath297 , except that @xmath341 may have one simple pole . in this case
it was due to dirichlet .
conjecture [ cct ] becomes vacuous if @xmath297 .
[ gjt ] suppose that @xmath310 is an irreducible constituent of @xmath283 with @xmath340 .
then @xmath341 converges to a holomorphic function in some right half complex plane @xmath342 and can be continued to a holomorphic function on the whole complex plane so that @xmath343 is bounded in all vertical strips @xmath344 .
moreover @xmath343 satisfies the functional equation @xmath345 [ cct ] suppose that @xmath310 is an irreducible , admissible @xmath314-module such that the central character of @xmath310 is trivial on @xmath346 and such that @xmath341 converges in some half plane .
suppose also that for all characters @xmath347 the @xmath0-function @xmath348 ( which will then converge in some right half plane ) can be continued to a holomorphic function on the entire complex plane , which is bounded in vertical strips and satisfies a functional equation @xmath349 ( @xmath350 also automatically converges in some right half plane . )
then there is a partition @xmath351 and cuspidal automorphic representations @xmath352 of @xmath353 such that @xmath354 this conjecture is known to be true for @xmath298 ( @xcite , @xcite ) and @xmath355 ( @xcite ) . for @xmath356 a weaker form of this conjecture involving twisting by higher dimensional automorphic representations
is known to hold ( see @xcite , @xcite ) .
these results are called ` converse theorems ' .
the reason for us introducing automorphic forms is because of a putative connection to galois representations , which we will now discuss . but first let us discuss the local situation .
it has recently been established ( @xcite , @xcite , @xcite ) that there is a natural bijection , @xmath357 , from irreducible smooth representations of @xmath323 to @xmath52-dimensional frobenius semi - simple wd - representations of @xmath41 over @xmath22 .
the key point here is that the bijection should be natural .
we will not describe here exactly what this means ( instead we refer the reader to the introduction to @xcite ) .
it does satisfy the following .
now suppose that @xmath365 and that @xmath99 is a de rham @xmath60-adic representation of @xmath3 which is unramified at all but finitely many primes . using the local reciprocity map @xmath357 , we can associate to @xmath99 an irreducible , admissible @xmath285-module @xmath366 where @xmath367 is a tempered irreducible , admissible @xmath321-module with infinitesimal character @xmath368 and with @xmath369 .
the definition of @xmath370 depends only on the numbers @xmath200 and @xmath204
. then we have the following conjectures . [ cgfa ]
suppose that @xmath268 is a multiset of @xmath52 _ integers _ and that @xmath310 is an irreducible constituent of @xmath283 .
identify @xmath67 .
then each @xmath371 can be defined over @xmath2 and there is an irreducible geometric strongly compatible system of @xmath60-adic representations @xmath149 such that @xmath372 and @xmath373 for all primes @xmath8 .
[ cafg ] suppose that @xmath120 is an irreducible @xmath60-adic representation which is unramified at all but finitely many primes and for which @xmath121 is de rham .
let @xmath374 .
then @xmath375 is a cuspidal automorphic representation of @xmath234 .
conjecture [ cafg ] is probably the more mysterious of the two , as only the case @xmath297 and fragmentary cases where @xmath298 are known .
this will be discussed further in the next section .
note the similarity to the main theorem of global class field theory that @xmath376 has kernel @xmath346 .
[ kt ] suppose that @xmath268 is multiset of @xmath52 _ distinct _ integers and that @xmath310 is an irreducible constituent of @xmath283 .
let @xmath191 .
suppose moreover that @xmath377 for some character @xmath378 , and that either @xmath379 or for some prime @xmath8 the representation @xmath322 is square integrable ( i.e. @xmath371 is indecomposable ) .
then there is a continuous representation @xmath380 with the following properties .
this was established by finding the desired @xmath60-adic representations in the cohomology of certain unitary group shimura varieties .
it seems not unreasonable to hope that similar techniques might allow one to improve many of the technical defects in the theorem .
however clozel has stressed that in the cases where @xmath268 does not have distinct elements or where @xmath387 , there seems to be no prospect of finding the desired @xmath60-adic representations in the cohomology of shimura varieties .
it seems we need a new technique .
let us first summarise in a slightly less precise way the various conjectures we have made , in order to bring together the discussion so far .
fix an embedding @xmath23 and let @xmath268 be a multiset of integers of cardinality @xmath340 .
then the following sets should be in natural bijection .
one way to make precise the meaning of ` natural ' is to require that two objects @xmath388 and @xmath389 should correspond if the local l - factors @xmath390 and @xmath391 are equal for all but finitely many @xmath8 . note that in each case the factors @xmath390 for all but finitely many @xmath8 , completely determine @xmath388 . 1 . [ af ] irreducible constituents @xmath310 of @xmath272 .
[ lf ] near equivalence classes of irreducible , admissible @xmath285-modules @xmath310 with the following properties .
( we call two @xmath314-modules , @xmath310 and @xmath392 nearly equivalent if @xmath393 for all but finitely many @xmath8 . ) 1 .
@xmath332 has infinitesimal character @xmath268 .
the central character @xmath359 is trivial on @xmath394 .
3 . for all characters
@xmath395 the @xmath0-function @xmath396 converges in some right half plane , has holomorphic continuation to the entire complex plane so that it is bounded in vertical strips and satisfies the functional equation @xmath397 4 .
( see @xcite for an explanation of this condition . )
there is a finite set of primes @xmath113 containing all primes @xmath8 for which @xmath371 is ramified , such that , writing @xmath398 for @xmath114 , @xmath399 is bounded as @xmath400 from the right .
+ in this case @xmath401 .
+ ( lr ) : : [ lr ] ( fix @xmath402 . ) irreducible @xmath60-adic representations @xmath403 which are unramified at all but finitely many primes and for which @xmath121 is de rham with @xmath404 . in this case
( wcs ) : : [ cs ] irreducible weakly compatible systems of @xmath60-adic representations @xmath149 with @xmath372 . in this case
( gcs ) : : [ gcs ] irreducible geometric strongly compatible systems of @xmath60-adic representations @xmath149 with @xmath372 . in this case
@xmath406 . for @xmath297 we must drop the item @xmath407 , because it would need to be modified to allow @xmath408 to have a simple pole , while , in any case condition ( lf ) ( b ) would make the implication @xmath409 trivial .
this being said , in the case @xmath297 all the other four sets are known to be in natural bijection ( see @xcite ) .
this basically follows because global class field theory provides an isomorphism @xmath410 i would again like to stress how different are these various sorts of objects and how surprising it is to me that there is any relation between them .
items ( af ) and ( lf ) both concern representations of adele groups , but arising in rather different settings : either from the theory of discrete subgroups of lie groups or from the theory of @xmath0-functions with functional equation . items ( lr ) and ( wcs ) arise from galois theory and item ( gcs ) arises from geometry . not much .
trivially one has @xmath411 .
the passage @xmath412 is ok by theorem [ gjt ] .
as discussed in section [ s3 ] we have significant partial results in the directions @xmath409 and @xmath413 , but both seem to need new ideas .
( though i should stress that i am not really competent to discuss converse theorems . )
one way to establish the equivalence of all five items would be to complete the passages @xmath409 and @xmath414 and to establish the passage @xmath415 .
it is these implications which have received most study , though it should be pointed out that in the function field case the equivalence of the analogous objects was established by looking at the implications @xmath416 ( see @xcite . it is the use of techniques from grothendieck s @xmath60-adic cohomology to prove the first of these implications which is most special to function fields . )
however it is striking that in the case of number fields all known implications from items ( lr ) , ( wcs ) or ( gcs ) to ( lf ) go via ( af ) .
for the rest of this article we will concentrate on what still seems to be the least understood problem : the passage from ( lr ) or ( wcs ) to ( af ) or ( lf ) .
although the results we have are rather limited one should not underestimate their power .
perhaps the most striking illustration of this is that the lifting theorems discussed in section [ s4.2 ] ( combined with earlier work using base change and converse theorems ) allowed wiles @xcite to finally prove fermat s last theorem .
the discussion in the rest of this paper will of necessity be somewhat more technical .
in particular we will need to discuss automorphic forms , @xmath60-adic representations and so on over general number fields ( i.e. fields finite over @xmath1 ) .
we will leave it to the reader s imagination exactly how such a generalisation is made . in this connection we should remark that if @xmath417 is a finite extension of number fields and if @xmath99 is a semi - simple de rham @xmath60-adic representation of @xmath418 which is unramified at all but finitely many primes , then ( see @xcite ) @xmath419 ( formally if the @xmath0-functions do nt converge ) .
in fact this is true euler factor by euler factor and similar results hold for conductors and @xmath420-factors ( see @xcite ) .
this observation can be extremely useful .
suppose that @xmath421 is a group , @xmath268 a normal subgroup such that @xmath422 is cyclic with generator of @xmath43 .
it is an easy exercise that an irreducible representation @xmath82 of @xmath268 extends to a representation of @xmath421 if and only if @xmath423 as representations of @xmath268 .
if one believes conjectures [ cgfa ] and [ cafg ] , one might expect that if @xmath417 is a cyclic galois extension of number fields of prime order , if @xmath43 generates @xmath424 and if @xmath310 is a cuspidal automorphic representation of @xmath425 with @xmath426 , then there should be a cuspidal automorphic representation @xmath427 of @xmath428 , such that for all places @xmath429 of @xmath0 we have @xmath430 .
this is indeed the case .
for @xmath297 we have @xmath431 , langlands @xcite proved it for @xmath298 using the trace formula and arthur and clozel @xcite generalised his method to all @xmath52 .
one drawback of this result is that if @xmath432 is a place of @xmath6 inert in @xmath0 then there is no complete recipe for @xmath433 in terms of @xmath310 .
this can be surprisingly serious .
it can however be alleviated , if we know how to associate irreducible @xmath60-adic representations to both @xmath427 and @xmath310 .
langlands used this to show that many two dimensional artin representations ( i.e. @xmath60-adic representations with finite image ) were automorphic ( i.e. associated to a cuspidal automorphic representation ) .
in fact using additional results from the theory of @xmath0-functions , particularly the converse theorem for @xmath434 ( see section [ s4.4 ] ) , he and tunnell ( @xcite ) were able to establish the automorphy of all continuous two dimensional artin representations with soluble image .
suppose that @xmath82 is a representation of a finite group @xmath421 .
then there are soluble subgroups @xmath435 , one dimensional representations @xmath436 of @xmath437 and integers @xmath438 such that as virtual representations of @xmath421 we have @xmath439 let @xmath365 .
suppose that @xmath403 is an @xmath60-adic representation with _
finite image_. then the @xmath0-function @xmath440 has meromorphic continuation to the entire complex plane and satisfies the expected functional equation .
artin s argument runs as follows .
let @xmath421 denote the image of @xmath99 and write @xmath441 as in brauer s theorem .
let @xmath442 be the galois extension with group @xmath421 cut out by @xmath99 and let @xmath443 .
then one has equalities @xmath444 by class field theory for the fields @xmath445 , the character @xmath436 is automorphic on @xmath446 and so @xmath447 has holomorphic continuation to the entire complex plane ( except possibly for one simple pole if @xmath448 ) and satisfies a functional equation .
it follows that @xmath197 has meromorphic continuation to the entire complex plane and satisfies a functional equation .
the problem with this method is that some of the integers @xmath438 will usually be negative so that one can only conclude the meromorphy of @xmath197 , not its holomorphy . as cogdell and piatetski - shapiro point out , conjecture [ cct ]
would have very important implications for galois representations .
for instance the cases @xmath298 and @xmath449 played a key role in the proof of the automorphy of two dimensional artin representations ( see [ s4.1 ] ) .
conjecture [ cct ] combined with brauer s theorem and a result of jacquet and shalika @xcite in fact implies that many ( all ? - certainly those with soluble or perfect image ) artin representations are automorphic .
a similar argument shows that in many other cases , in order to check the automorphy of an @xmath60-adic representation of @xmath3 , it suffices to do so after a finite base change .
for instance one has the following result .
_ assume conjecture [ cct ] .
let @xmath191 and let @xmath159 be a finite , totally real galois extension .
suppose that @xmath427 is a cuspidal automorphic representation of @xmath428 with infinitesimal character corresponding to a multiset @xmath268 consisting of @xmath52 distinct integers . if @xmath450 also suppose that @xmath433 is square integrable ( i.e. @xmath451 is indecomposable ) for some finite place @xmath432 of @xmath6 .
let @xmath403 be an @xmath60-adic representation such that @xmath452 for some character @xmath308 of @xmath3 , and such that @xmath453 is irreducible .
suppose finally that @xmath454 and @xmath427 are associated , in the sense that , for all but finitely many places @xmath432 of @xmath6 , we have @xmath455 then there is a regular algebraic cuspidal automorphic representation @xmath310 of @xmath234 associated to @xmath99 in the same sense . _ to describe this sort of theorem we first remark that if @xmath157 is continuous then after conjugating @xmath99 by some element of @xmath61 we may assume that the image of @xmath99 is contained in @xmath456 and so reducing we obtain a continuous representation @xmath457 the lifting theorems i have in mind are results of the general form if @xmath99 and @xmath458 are @xmath60-adic representations of @xmath3 with @xmath458 automorphic and if @xmath459 then @xmath99 is also automorphic .
very roughly speaking the technique ( pioneered by wiles @xcite and completed by the author and wiles @xcite ) is to show that @xmath460 arises from automorphic forms for all @xmath82 by induction on @xmath82 . as @xmath461 is an abelian group one
is led to questions of class field theory and galois cohomology .
i should stress that such theorems are presently available only in very limited situations .
i do not have the space to describe the exact limitations , which are rather technical , but the sort of restrictions that are common are as follows . 1 .
if @xmath190 then there should be a character @xmath462 and a non - degenerate bilinear form @xmath463 on @xmath81 such that * @xmath464 and * @xmath465 .
+ ( this seems to be essential for the method of @xcite . ) 2 .
@xmath99 should be de rham with distinct hodge - tate numbers .
( this again seems essential to the method of @xcite , but see @xcite . )
either @xmath99 and @xmath458 should be ordinary ( i.e. their restrictions to @xmath111 should be contained in a borel subgroup ) ; or @xmath99 and @xmath458 should be crytsalline ( not just de rham ) at @xmath60 with the same hodge - tate numbers and @xmath60 should be large compared with the differences of elements of @xmath101 .
( the problems here are connected with the need for an integral fontaine theory , but they are not simply technical problems .
there are some complicated results pushing back this restriction in isolated cases , see @xcite , @xcite , @xcite , but so far our understanding is very limited . )
the image of @xmath466 should not be too small ( e.g. should be irreducible when restricted to @xmath467 ) , though in the case @xmath298 there is beautiful work of skinner and wiles ( @xcite and @xcite ) dispensing with this criterion , which this author has unfortunately not fully understood .
in addition , all the published work is for the case @xmath298 .
however there is ongoing work of a number of people attempting to dispense with this assumption . using a very important insight of diamond @xcite
, the author , together with l.clozel and m.harris , has generalised to all @xmath52 the so called minimal case ( originally treated in @xcite ) where @xmath99 is no more ramified than @xmath466 .
one would hope to be able to deduce the non - minimal case from this , as wiles did in @xcite for @xmath298 . in
this regard one should note the work of skinner and wiles @xcite and the work of mann @xcite .
however there seems to be one missing ingredient , the analogue of the ubiquitous ihara lemma , see lemma 3.2 of @xcite ( and also theorem 4.1 of @xcite ) . as this seems to be an important question , but one which lies in the theory of discrete subgroups of lie groups , let us take the trouble to formulate it , in the hope that an expert may be able to prove it .
it should be remarked that there are a number of possible formulations , which are not completely equivalent and any of which would seem to suffice . we choose to present one which has the virtue of being relatively simple to state .
suppose that @xmath468 is a unitary group which becomes an inner form of @xmath259 over an imaginary quadratic field @xmath73 .
suppose that @xmath469 is compact .
let @xmath60 be a prime which one may assume is large compared to @xmath52 .
let @xmath470 and @xmath471 be distinct primes different from @xmath60 with @xmath472 and @xmath473 .
let @xmath474 be an open compact subgroup of @xmath475 and consider the representation of @xmath476 on the space @xmath477 of locally constant @xmath478-valued functions on @xmath479 ( note that @xmath480 is a discrete cocompact subgroup of @xmath481 . )
suppose that @xmath482 is an irreducible _ sub - representation _ of @xmath483 with @xmath484 generic .
then @xmath485 is also generic .
the most serious problem with applying such lifting theorems to prove an @xmath60-adic representation @xmath99 is automorphic is the need to find some way to show that @xmath466 is automorphic .
the main success of lifting theorems to date , has been to show that if @xmath73 is an elliptic curve over the rationals then @xmath486 is automorphic , so that @xmath73 is a factor of the jacobian of a modular curve and the @xmath0-function @xmath180 is an entire function satisfying the expected functional equation ( @xcite , @xcite,@xcite ) .
this was possible because @xmath487 happens to be a pro - soluble group and there is a homomorphism @xmath488 splitting the reduction map .
the artin representation @xmath489 is automorphic by the langlands - tunnell theorem alluded to in section [ s4.1 ] .
i would like to discuss one other technique which has been some help if @xmath298 and may be helpful more generally .
we will restrict our attention here to the case @xmath298 and @xmath490 .
we have said that the principal problem with lifting theorems for proving an @xmath60-adic representation @xmath491 is automorphic is that one needs to know that @xmath466 is automorphic .
this seems to be a very hard problem .
nonetheless one can often show that @xmath466 becomes automorphic over some galois totally real field @xmath159 .
( because @xmath6 is totally real , if @xmath492 and @xmath493 then @xmath494 . so this ` potential automorphy ' is far from vacuous ) . the way one does this is to look for an abelian variety @xmath495 with multiplication by a number field @xmath496 with @xmath497=\dim a$ ] , and such that @xmath466 is realised on @xmath498 $ ] for some prime @xmath499 , while for some prime @xmath500 the image of @xmath501 on @xmath502 $ ] is soluble .
one then argues that @xmath502 $ ] is automorphic , hence by a lifting theorem @xmath503 is automorphic , so that ( tautologically ) @xmath498 $ ] is also automorphic , and hence , by another lifting theorem , @xmath454 is automorphic .
one needs @xmath6 to be totally real , as over general number fields there seems to be no hope of proving lifting theorems , or even of attaching @xmath60-adic representations to automorphic forms . in practice , because of various limitations in the lifting theorems one uses , one needs to impose some conditions on the behaviour of a few primes , like @xmath60 , in @xmath6 and some other conditions on @xmath504 .
the problem of finding a suitable @xmath504 over a totally real field @xmath6 , comes down to finding a @xmath6-point on a twisted hilbert modular variety .
this is possible because we are free to choose @xmath6 , the only restriction being that @xmath6 is totally real and certain small primes ( almost ) split completely in @xmath6 . to do this ,
one has the following relatively easy result .
suppose that @xmath66 is a smooth geometrically irreducible variety .
let @xmath113 be a finite set of places of @xmath1 and suppose that @xmath505 has a point over the completion of @xmath1 at each place in @xmath113 .
let @xmath506 be the maximal extension of @xmath1 in which all places in @xmath113 split completely ( e.g. @xmath507 is the maximal totally real field ) .
then @xmath505 has a @xmath506-point . in this regard
it would have extremely important consequences if , in the previous proposition , one could replace @xmath506 by @xmath508 , the maximal soluble extension of @xmath1 in which all places in @xmath113 split completely .
i do not know if it is reasonable to expect this .
[ mer ] suppose that @xmath149 is an irreducible weakly compatible system of two dimensional @xmath60-adic representations with @xmath509 where @xmath510 .
suppose also that @xmath511 for one ( and hence for all ) pairs @xmath140 .
then there is a galois totally real field @xmath159 and a cuspidal automorphic representation @xmath310 of @xmath512 such that in particular @xmath149 is pure of weight @xmath517 .
moreover @xmath149 is strongly compatible and @xmath518 has meromorphic continuation to the entire complex plane and satisfies the expected functional equation . the last sentence of this theorem results from the first part and brauer s theorem .
we remark that conjecture [ cct ] would imply that this theorem could be improved to assert the automorphy of @xmath149 over @xmath1 .
aaaa e.artin , _ zur theorie der l - reihen mit allgemeinen gruppencharakteren _ ,
hamburg 8 ( 1930 ) , 292306 .
j.arthur and l.clozel , _ simple algebras , base change and the advanced theory of the trace formula _ , annals of math .
studies 120 , pup 1989 .
l.berger , _ reprsentations @xmath8-adiques et quations diffrentielles _ , invent .
( 2002 ) , 219284 .
p.berthelot , _ altrations de varits algbriques ( daprs a.j.de jong ) _ , astrisque 241 ( 1997 ) , 273311 .
c.breuil , b.conrad , f.diamond and r.taylor , _ on the modularity of elliptic curves over @xmath1 _ , j.a.m.s .
14 ( 2001 ) , 843939 .
r.brauer , _ on artin s l - series with general group characters _ , ann . math .
( 2 ) 48 ( 1947 ) , 502514 .
b.birch and p.swinnerton-dyer , _ notes on elliptic curves ii _ , j. reine angew . math . 218
( 1965 ) , 79108 .
k.buzzard and r.taylor , _ companion forms and weight one forms _ , annals of math 149 ( 1999 ) , 905919 .
b.conrad , f.diamond and r.taylor , _ modularity of certain potentially barsotti - tate galois representations _ , jams 12 ( 1999 ) , 521567 .
l.clozel , _ motifs et formes automorphes : applications du principe de fonctorialit _ , in `` automorphic forms , shimura varieties and @xmath0-functions i '' , academic press 1990 .
l.clozel , _ reprsentations galoisiennes associes aux representations automorphes autoduales de @xmath519 _ , pub . math .
ihes 73 ( 1991 ) , 97145 .
j.cogdell and i.piatetski-shapiro , _ converse theorems for @xmath259 _ , publ .
ihes 79 ( 1994 ) , 157214 .
j.cogdell and i.piatetski-shapiro , _ converse theorems for @xmath259 ii _
, j. reine angew . math . 507
( 1999 ) , 165188 .
p.deligne , _ la conjecture de weil i _ , publ .
ihes 43 ( 1974 ) , 273307 .
f.diamond , _ the taylor - wiles construction and multiplicity one _ , invent .
( 1997 ) , 379391 .
j.dixmier , _ algbres enveloppantes _ , gauthier villars 1974 .
g.faltings , _ endlichkeitsstze fr abelsche varietten ber zahlkrpern _ ,
73 ( 1983 ) , 349366 .
d.flath , _ decomposition of representations into tensor products _ , in `` automorphic forms , representations and @xmath0-functions i '' ams 1979 .
m.fontaine , talk at `` mathematische arbeitstagung 1988 '' , maz - planck - institut fr mathematik preprint no . 30 of 1988 .
m.fontaine , _ le corps des priodes @xmath8-adiques _ , astrisque 223 ( 1994 ) , 59111 .
m.fontaine , _ reprsentations @xmath8-adiques semi - stables _ , astrisque 223 ( 1994 ) , 113184 .
m.fontaine , _ reprsentations @xmath60-adiques potentiellement semi - stables _ , astrisque 223 ( 1994 ) , 321347 .
m.fontaine and b.mazur , _ geometric galois representations _ , in elliptic curves , modular forms and fermat s last theorem ( j.coates and s .- t.yau eds . ) , international press 1995 .
godement and h.jacquet , _ zeta functions of simple algebras _ ,
lnm 260 , springer 1972 .
m.harris , _ on the local langlands correspondence _ , these proceedings .
g.henniart , _ une preuve simple des conjectures de langlands pour @xmath519 sur un corps @xmath8-adiques _ , invent .
139 ( 2000 ) , 439455 .
m.harris and r.taylor , _ the geometry and cohomology of some simple shimura varieties _ , pup 2001 .
y.ihara , _ on modular curves over finite fields _ , in `` proceedings of an international colloquium on discrete subgroups of lie groups and applications to moduli '' oup 1975 .
l.illusie , _ cohomologie de de rham et cohomologie tale @xmath8-adique _ , astrisque 189190 ( 1990 ) , 325374 .
h.jacquet , _ principal @xmath0-functions of the linear group _ , in `` automorphic forms , representations and @xmath0-functions 2 '' ams 1979 .
h.jacquet and r.langlands , _ automorphic forms on @xmath520 _ , lnm 114 , springer 1970 .
h.jacquet , i.piatetski-shapiro and j.shalika , _ automorphic forms on @xmath521 _ , annals of math . 109
( 1979 ) , 169258 .
h.jacquet and j.shalika , _ on euler products and the classification of automorphic representations ii _ ,
103 ( 1981 ) , 777815 .
v.kolyvagin , _ on the mordell - weil group and the shafarevich - tate group of modular elliptic curves _ , in `` proceedings of the kyoto international congress of mathematicians '' math .
japan 1991 .
r.kottwitz , _ on the @xmath522-adic representations associated to some simple shimura varieties _ , invent .
( 1992 ) , 653665 .
l.lafforgue , _ drinfeld varieties and the langlands program _ , these proceedings .
r.p.langlands , _ automorphic representations , shimura varieties and motives .
ein mrchen .
_ in `` automorphic forms , representations and @xmath0-functions ii '' ams 1979 .
langlands , _ base change for @xmath523 _ , annals of math .
studies * 96 * , princeton univ . press , princeton , 1980 .
w.r.mann , _ local level - raising for @xmath259 _ , preprint .
l.moret-bailly , _ groupes de picard et problmes de skolem ii _ , ann .
ens 22 ( 1989 ) , 181194 .
t.miyake , _ modular forms _ , springer 1989 .
f.pop , _ embedding problems over large fields _ , annals of math 144 ( 1996 ) , 134 .
k.ribet , _ congruence relations between modular forms _ , in `` proceedings of the warsaw icm '' polish scientific publishers 1984 .
d.savitt , _ modularity of some potentially barsotti - tate galois representations _ ,
p.serre , _ abelian @xmath60-adic representations and elliptic curves _ , benjamin 1968 .
c.skinner and a.wiles , _ residually reducible representations and modular forms _ , inst .
hautes etudes sci .
( 1999 ) , 5126 .
c.skinner and a.wiles , _ base change and a problem of serre _ , duke math . j. 107 ( 2001 ) , 1525 .
c.skinner and a.wiles , _ nearly ordinary deformations of irreducible residual representations _ , annales de la facult de sciences de toulouse x ( 2001 ) , 185215 .
j.tate , _ number theoretic background _ , in a.borel and w.casselman `` automorphic forms , representations and @xmath0-functions '' , proc .
symposia in pure math .
33 ( 2 ) , ams 1979 .
r.taylor , _ on the meromorphic continuation of degree two @xmath0-functions _ , preprint available at http://www.math.harvard.edu/~rtaylor
. j. tunnell , _
artin s conjecture for representations of octahedral type _ , bull .
ams * 5 * ( 1981 ) , 173175 .
r.taylor and a.wiles , _ ring theoretic properties of certain hecke algebras _ , annals of math . 141 ( 1995 ) , 553572 .
n.wallach , _ representations of reductive lie groups _ , in `` automorphic forms , representations and @xmath0-functions i '' ams 1979 .
a.weil , _ ber die bestimmung dirichletscher reihen durch funktionalgleichungen _ ,
( 1967 ) , 149156 .
a.wiles , _ modular elliptic curves and fermat s last theorem _ , annals of math . 141 ( 1995 ) , 443551 .
|
in the first part of this paper we try to explain to a general mathematical audience some of the remarkable web of conjectures linking representations of galois groups with algebraic geometry , complex analysis and discrete subgroups of lie groups . in the second part we briefly review some limited recent progress on these conjectures .
4.5 mm * 2000 mathematics subject classification : * 11f80 . * keywords and phrases : * galois representations , @xmath0-function , automorphic forms .
|
the subject of hamiltonian circuits and walks plays an important role in mathematics and physics alike . given a connected undirected graph @xmath7 , a _
hamiltonian circuit _ ( or cycle ) is a cycle ( i.e. , a closed loop ) through @xmath7 that visits each of the @xmath8 vertices of @xmath7 exactly once @xcite .
in particular , a hamiltonian circuit has length @xmath8 .
similarly , a _
hamiltonian walk _
( or path ) is an open non - empty path ( i.e. , with two distinct extremities ) of length @xmath9 that visits each vertex exactly once .
note that a hamiltonian circuit can be turned into a hamiltonian walk by removing any one of its edges , whereas a hamiltonian walk can be extended into a hamiltonian circuit only if its end points are adjacent in @xmath7 . .
] we add now to this list of well - known definitions the set @xmath10 of _ hamiltonian chains _ of order @xmath0 .
each member in @xmath10 is a set of @xmath0 disjoint paths whose union visits each vertex of @xmath7 exactly once ( see fig .
[ fig : chain ] ) .
the set of hamiltonian walks is then @xmath11 , and by convention we shall let @xmath12 denote the set of hamiltonian circuits .
note that if @xmath8 is even , @xmath13 is the set of dimer coverings of @xmath7 .
the hamiltonian chain problem has been studied earlier by duplantier and david on the manhattan lattice @xcite , but never to our knowledge on an undirected lattice .
determining whether @xmath7 contains a hamiltonian circuit is a difficult ( np - complete ) problem .
an even more difficult problem is to determine _ how many _ distinct hamiltonian circuits are contained in @xmath7 . in this paper
we shall present an algorithm that efficiently enumerates hamiltonian circuits , walks , and chains for regular low - dimensional graphs .
the motivation for studying such hamiltonian structures is by no means limited to graph theory . indeed , under appropriate solvent conditions
, biopolymers such as proteins may fold to form compact conformations , the study of which is currently at the centre of an intense activity in the biophysics community .
while real biopolymers contain complicated interactions which can probably not be fully accounted for within any simple lattice model , the study of hamiltonian walks has been advocated as a first approximation for understanding qualitatively the excluded - volume mechanisms at work behind such problems as polymer melting @xcite and protein folding @xcite .
our extension to hamiltonian chains permits to study polydisperse models of several polymers .
another interest stems from the study of magnetic systems with o(@xmath14 ) symmetry in physics .
these can modelled on the lattice as self - avoiding loops ( each having the weight @xmath14 ) @xcite , which , in the limit of vanishing temperature @xmath15 , are constrained to visit all the vertices @xcite .
coupling such systems to a magnetic field @xmath16 amounts , in a perturbative expansion around @xmath17 , to inserting pairs of loop end points @xcite . in the limit @xmath18 , the partition function @xmath19 of the o(@xmath14 ) model at @xmath20 in a weak magnetic field @xmath16
can thus be expressed in terms of the number of hamiltonian chains as @xmath21 finally , the exact enumeration of configurations is useful for settling issues of ergodicity when developing algorithms that provide unbiased sampling of hamiltonian walks in two @xcite and three @xcite dimensions .
in section [ sec : alg ] we present our enumeration algorithm and discuss some aspects of its implementation . results in dimensions @xmath22 and @xmath23 are given in section [ sec : res ] . for convenience ,
we limit the discussion to the simplest lattices ( square and cubic ) , although the construction extends straightforwardly to any regular lattice .
our results are strongest in @xmath22 where we determine all @xmath10 for @xmath3 square lattices up to @xmath4 , @xmath24 up to @xmath5 , and @xmath12 up to @xmath6 . in @xmath23 the largest lattice that we were able to access has size @xmath25 .
finally , we show in section [ sec : app ] how to extract physically interesting quantities from our data .
we first present our algorithm in dimension @xmath22 and then discuss the necessary modifications for @xmath23 .
for convenience , we limit the presentation to the simplest lattices , viz . an @xmath26 square lattice and an @xmath27 cubic lattice , although the construction extends straightforwardly to any regular lattice .
the boundary conditions are free ( non - periodic ) , although it will be clear that it is easy to introduce periodic boundary conditions along _ one _ of the lattice directions .
the algorithm is based on the transfer matrix principle , according to which the lattice is cut into two parts by means of a conveniently chosen @xmath28 dimensional oriented surface @xmath29 .
the part of the lattice above ( resp . below ) @xmath29 is called the future ( resp .
the past ) .
the surface @xmath29 cuts the lattice only at mid points of edges . .
the lattice is shown in solid line style , with the dotted lines representing lattice edges beyond the bounds of the lattice .
the surface @xmath29 is shown as a dashed line . ] at the initial ( resp .
final ) step of the enumeration the whole lattice belongs to the future ( resp .
past ) , so the algorithm consists in sweeping @xmath29 over the entire lattice .
this is done by gradually pushing @xmath29 towards the future , so that in any one step of the algorithm a _ single _
vertex is transferred from the future to the past . a few subsequent steps for a @xmath30 lattice
are shown in fig .
[ fig : transfer ] .
square . ] in any step , the configuration of the system is described by some information about the edges cut by @xmath29 , and by the number @xmath0 of chains which have already been completed .
the information refered to is the connectivity of the cut edges with respect to the part of the lattice which belongs to the past , and is best illustrated by an example ( see fig .
[ fig : config ] ) . the complete description of the configuration reads in this case @xmath31 , where the first @xmath32 entries refer to the state of the cut edges which are parallel to the 2-direction ( vertical ) , and the next entry refers to the state of the one cut edge which is parallel to the 1-direction ( horizontal ) .
the last entry is the number @xmath2 of completed chains . in the connectivity part of the information , we use the following coding : 1 .
a zero entry means an empty edge .
two equal positive entries mean a pair of edges which are connected in the past by part of a chain .
each of these edges will eventually have to be linked to a chain end point in the future , so as to form a complete chain .
3 . an unpaired positive entry means an edge which forms part of a partially completed chain , one end point of which has already been fixed in the past .
the edge will eventually be linked to another end point in the future , leading to the formation of a complete chain .
it is important to avoid any redundancy in this connectivity information .
a unique coding is obtained by requiring that the positive entries ( i.e. , unpaired entries , or the leftmost member of a pair of equal entries ) be arranged in increasing order ( @xmath33 ) when reading through the coding from left to right . in step @xmath34 of the enumeration
, the configurations are transfered from `` time '' @xmath34 to time @xmath35 .
more precisely , each configuration at time @xmath34 is examined in turn , and all its descendent configurations at time @xmath35 are generated by exhausting the possible arrangements of the chain at the vertex which is transfered from the future to the past .
the information described above is necessary and sufficient for deducing the connectivity information at time @xmath35 from that at time @xmath34 .
note that since the transfered vertex is not allowed to be empty , there are four possible arrangements if it accommodates a chain end , and six arrangements if a chain passes though it .
each configuration generated at time @xmath35 is inserted in an appropriate date structure a hash table which also keeps track of its weight ( here an integer ) .
the weight of a descendent configuration is the sum of the weights of all the parent configurations that generated it .
some of the generated descendent configurations are however _ rejected _ before insertion in the hash table ( see below ) .
once step @xmath34 has been completed , the hash table storing the configurations at time @xmath34 is erased , and one can move on to step @xmath35 . in this way , only two hash tables ( at times @xmath34 and @xmath35 ) are needed in the entire process .
note that the choice of data structure is essential for the feasibility and the efficiency of the algorithm .
a hash table permits to store the configurations via a key which is obtained by reading its coding as one large integer , modulo a suitably chosen prime . storing and retrieving configurations can be done in constant time , i.e. , independently of the number of configurations being stored in the hash table .
the hash table also allows to keep track of the weight of each configuration , according to the above rule .
namely , when a descendent configuration is generated with weight @xmath36 , we first make an attempt of looking it up in the hash table at time @xmath35 .
if it is not there , it is inserted with weight @xmath36 .
if it is already there , @xmath36 is added to the weight of configuration already present . if in the transfer process the two ends of the same chain ( coded by two equal positive entries ) join up , the resulting configuration is rejected , since this would mean forming a cycle rather than a chain .
( we make an exception to this rule when the very last vertex is added , since this permits to enumerate @xmath12 . )
if an unpaired positive entry gets left behind in the past it means that a chain has been completed , and so @xmath37 .
denote now a general configuration as @xmath38 . at step @xmath39 ,
the initial state is @xmath40 and has weight @xmath41 .
when a row of the lattice is completed , any configuration with @xmath42 gets rejected . when transfering the @xmath43th vertex of the last row , any configuration with @xmath44 gets rejected .
this trick allows us to avoid having to deal with a lot of special cases when a boundary vertex is transfered and also makes it much easier in practice to implement the algorithm correctly .
after step @xmath45 the lattice belongs completely to the past , and all the configurations are of the form @xmath46 .
their respective weights are precisely the @xmath10 that we wanted to compute . the maximum lattice size that we can attain is essentially limited by the number of different intermediate configurations generated in the transfer process .
this number attains its largest value after transfering the next last vertex in the next last row . in practice
we could store at most @xmath47 configurations .
as usual in enumeration studies , the coefficients @xmath10 are much larger than the integers which are usually represented by a computer ( @xmath48 for an unsigned integer on a 32-bit machine ) .
we therefore repeat the enumeration several times , computing each time the result modulo different coprime integers ( @xmath49 , @xmath50 , @xmath51 , ) , and reconstruct the true result in the end by using the chinese remainder theorem .
note that the use of modular arithmetics is possible because the weights of configurations are constructed only by successive additions of positive integers .
the counts for systems of size @xmath26 and @xmath52 should of course coincide .
verifying that this is indeed the case is however a very strong check of the algorithm , since permuting @xmath32 and @xmath53 ( with @xmath54 ) leads to a completely different transfer process in terms of the propagation of the surface @xmath29 .
we have performed such checks both for @xmath22 and @xmath23 .
we now describe briefly how the algorithm can be adapted to a @xmath55-dimensional hypercubic lattice of size @xmath56 . the surface @xmath29
is pushed though the lattice by means of @xmath55 nested loops , of which the innermost loop ( at nesting level @xmath28 ) moves @xmath29 along the 1-direction , etc .
, and the outermost loop ( at nesting level @xmath57 ) moves @xmath29 along the @xmath55-direction .
in general the loop at nesting level @xmath58 moves @xmath29 along the @xmath59-direction .
a configuration is given by @xmath60 , where the space @xmath61 describes the edges cut by @xmath29 which are parallel to the @xmath58-direction and consists of @xmath62 entries .
when the loop at nesting level @xmath63 is executed for the last time , the entry in @xmath61 corresponding to the position of the loops at nesting levels @xmath64 must be zero ; otherwise the configuration is rejected . at each vertex
there are @xmath65 possible local arrangements if the vertex contains a chain end , and @xmath66 arrangements if it does not .
we have implemented the algorithm for @xmath22 and @xmath23 .
it is of course most efficient in low dimensions when the number of entries necessary to describe a configuration is small , and the configurations themselves are strongly constrained by topology .
we were however able to obtain useful results as well for @xmath23 ( see below ) .
we first present our results in two dimensions .
.number @xmath10 of hamiltonian chains of order @xmath0 ( i.e. , consisting of precisely @xmath0 chains ) on a @xmath67 square lattice . [ cols=">,>",options="header " , ] in two dimensions , the @xmath68 for @xmath3 square can be computed from tables [ tab : walks][tab : circuits ] .
the resulting numerical values are displayed in table [ tab : adjacent ] .
note that @xmath68 vanishes for odd @xmath69 ( as it does on any bipartite lattice having an odd number of vertices ) .
the values in table [ tab : circuits ] have been used in @xcite to test that a certain monte carlo algorithm for producing hamiltonian walks did indeed give unbiased results .
physically , on may argue that @xmath68 is proportional to the probability that the two ends of an open polymer ( walk ) join so as to form a ring polymer ( circuit ) .
it is tempting to try similarly to construct from the ratio of @xmath11 and @xmath70 the probability that the conformation of two chains is such that one end point of each are adjacent on the lattice .
unfortunately , this is not possible , since certain two - chains ( more precisely those in which an end point of one chain is adjacent to both end points of the other chain ) can be obtained from more than one one - chain by removing an internal edge in the latter . the study of phase transitions through the location of partition function zeroes in the complex magnetic field plane was initiated by yang and lee @xcite . in particular , these authors established that for the ising model these zeroes lie on the unit circle in terms of the variable @xmath71 , i.e. , they correspond to purely imaginary values of the field @xmath16 .
the zero closest to the positive real axis is denoted @xmath72 , where @xmath73 is the so - called lee - yang edge . at the critical temperature , @xmath74 in the thermodynamic limit , and
its finite - size scaling permits to access a critical exponent .
another possible approach is to study the density of zeroes @xmath75 .
creswick and kim @xcite have shown that at the critical temperature @xmath76 for @xmath77 .
we have studied the zeroes of the partition function ( [ partfunc ] ) in terms of the variable @xmath78 for @xmath3 squares with @xmath79 . for odd @xmath69
there is one trivial zero at @xmath80 ( since @xmath81 ) , and for even @xmath69 the two zeroes closest to @xmath80 form a complex conjugate pair with an imaginary part that tends to zero as @xmath82 . disregarding these `` exceptional '' zeroes , all the remaining zeroes ( for @xmath69 of any parity )
are found to lie on the positive real axis in the complex @xmath83-plane , corresponding to purely imaginary @xmath16 as in the lee - yang theorem .
we now define a finite-@xmath69 approximation to the density of zeroes in the point @xmath84 as @xmath85 , where we have arranged the zeroes of @xmath86 in increasing order @xmath87 . of partition function zeroes in the variable @xmath78 for the hamiltonian chain problem on @xmath3 squares . ]
the approximations @xmath88 are shown in fig .
[ fig : zeroes ] for @xmath89 .
one observes a clear crossover near @xmath90 , separating two regimes of power law behaviours .
for even @xmath69 the curves bifurcate for @xmath91 , which can be remedied by regrouping the zeroes two by two ( not shown ) .
the power laws extracted from the largest available sizes read @xmath92 we have no satisfying explanation for these exponents at present .
the naive application of the standard scaling laws @xmath93 ( josephson ) and @xmath94 ( rushbrooke ) , and the results @xmath95 and @xmath96 for the critical @xmath80 system @xcite , leads to @xmath97 , which is clearly off the mark .
this work was supported through the european community network enrage ( grant mrtn - ct-2004 - 005616 ) and by the agence nationale de la recherche ( grant anr-06-blan-0124 - 03 ) .
|
we present an algorithm for enumerating exactly the number of hamiltonian chains on regular lattices in low dimensions . by definition , these are sets of @xmath0 disjoint paths whose union visits each lattice vertex exactly once .
the well - known hamiltonian circuits and walks appear as the special cases @xmath1 and @xmath2 respectively . in two dimensions ,
we enumerate chains on @xmath3 square lattices up to @xmath4 , walks up to @xmath5 , and circuits up to @xmath6 . some results for three dimensions
are also given .
using our data we extract several quantities of physical interest .
|
tight correlations between the mass of supermassive black holes and the properties of their host galaxies @xcite suggest a co - evolution triggered by merging events . at the lower end of the black hole mass range , @xmath2 , where dwarf galaxies and bulges of late type spirals among other sources are located , such correlations are less clear @xcite .
for example , pseudobulges , barred galaxies and narrow line seyfert 1s ( nlsy1s ) appear to lie below the @xmath3 relation established for higher mass systems @xcite , which has been interpreted as a sign of evolution triggered by secular processes and/or not fully grown systems @xcite .
intermediate mass black holes at the center of galaxies ( imbhs ; @xmath4 ) constitute ideal cases for tracing the behavior of these correlations at the low - mass end .
the characteristics of this population are of particular interest for studying galaxy evolution and to put constraints on models of primordial black hole seed formation @xcite . during the past decade , imbhs and
their hosts have been studied in the optical mainly from _ sdss _ data , but recently also with images taken with wfpc2 on _ hst _ and in the x - rays @xcite .
we make use of the high angular and spectral resolution that can be achieved in the near - infrared ( nir ) when assisted by adaptive optics systems , to discover an imbh candidate at the center of the starburst / seyfert composite galaxy iras 01072 + 4954 ( @xmath5 ) . in the optical ,
this source displays only narrow emission lines and is classified as a transition object in the baldwin - phillips - terlevich diagnostic diagram @xcite .
this means that the main ionizing source of the emitting species could be star formation , an active nucleus ( seyfert 2 ) or both .
however , in the x - ray domain , it shows an emission typical of seyfert 1 agn : steep power - law photon - spectral index ( @xmath6 ) and very low hydrogen column density ( @xmath7 ) .
long- and short - term x - ray flux variations were also detected @xcite . in our analysis of its nuclear nir emission ( @xmath8 pc )
, we find that the main reason for the ambiguous classification is the faintness of the agn @xcite .
the low bolometric luminosity , @xmath9 , combined with a very low @xmath10 produces broad emission lines that are very faint and narrow , and therefore not detectable in the optical spectrum taken by moran et al .
@xcite from the central @xmath11 region .
the predicted flux and width of the h@xmath12 broad component are @xmath13 and _ fwhm_@xmath14 ( determined using the expressions from @xcite ; see @xcite for details ) .
nevertheless , this mini - agn seems to accrete at a high rate , @xmath15 , which makes it comparable to other nlsy1s ( e.g. , @xcite ) . in this paper
we focus on the black hole mass determination using k - band integral field observations of the bulge of this galaxy . in the following sect.[sec : obs ] , we briefly present the data . in sect.[sec :
phot ] , we show the construction and fitting of the surface brightness profile and discuss the probable presence of an inner structure disk or bar . the fitting of the line - of - sight velocity distribution of the stars in the bulge is presented in sect.[sec : kyn ] .
finally , in sect.[sec : mbh ] we use those photometric and kinematical measurements to estimate the black hole mass via scaling relations and discuss the results .
conclusions are presented in sect.[sec : fin ] .
the central @xmath16 of iras 01072 + 4954 was observed on october 2008 with the near - infrared integral field spectrometer nifs @xcite mounted on the `` frederick c. gillett '' gemini north telescope on mauna kea , hawaii .
the adaptative optics ( ao ) module altair was used in laser guide star mode .
nifs has a @xmath17 field of view and provides high spatial and spectral resolution simultaneously .
the observations cover the h- and k - bands , although in this paper we refer only to the k - band data . the spatial resolution achieved with
the ao in the k - band was @xmath18 , which corresponds to 72pc on source .
the instrumental spectral resolution is @xmath19 .
the data was reduced using the gemini iraf package ( released version 2.14 , of september 15 , 2008 ) and standard iraf tasks .
absolute flux calibration and extinction corrections were also applied to the data cubes .
the 2mass @xcite provides nir images that cover the whole galaxy . however , in the k - band
, the source is very faint at the center only @xmath20 times brighter than the background and has a small radius,@xmath21 , so that a proper 2d photometric modeling is not possible .
we also notice that the bulge size is about the same as the 2mass k - band point - spread function ( psf ) , meaning that it is not resolved in the 2mass image .
therefore , we combine the 2mass with our nifs data to construct a 1d surface brightness profile ( fig.[fig : prof ] ) .
we model the bulge as a sersic profile with an effective radius @xmath22 and index @xmath23 .
the disk is described by an exponential profile with scale radius @xmath24 .
a gaussian with the width of the nifs psf represents the emission from the agn and the hot dust at the center .
the photometric parameters resulting from the best fit of these three components are : @xmath25 , @xmath26 and @xmath27 with central surface brightness @xmath28 .
k - band brightness profile of iras 01072 + 4954 obtained by the combination of the nifs ( squares ) and 2mass ( circles ) data .
the dashed red line corresponds to a sersic profile fitted to the bulge , and the dash - dotted green line to an exponential function that fits the disk component .
the dotted magenta line represents the agn + hot dust contribution , while the solid blue line show the overall fit .
, scaledwidth=58.0% ] the presented model is only a first approximation to the description of the surface brightness .
there are two main reasons foo believing that the bulge contribution found in this way is overestimated .
first , the two data sets do not properly overlap with each other .
assuming that the nifs data represent the bulge of the galaxy reasonably well , and that the emission from @xmath29 of the 2mass image corresponds to the disk , we convolve the sersic model to the resolution of the nifs data and the exponential function to the one of the 2mass image .
however , these components are not independent of each other and the disk contribution at the center might be underestimated .
the second one is that iras 01072 + 4954 is a barred galaxy , and although the bar does not seem to dominate the emission at the center as we show in the following , its contribution has not been taken into account in the axially symmetric averaged 1d profile . at first glance ,
the nir surface brightness of iras 01072 + 4954 in the 2mass and in the nifs data look smooth and uniform . in order to recognize the presence of sub - structures in the disk or in the bulge
, we use the unsharp masking method .
it essentially consists of subtracting a smoothed image of the galaxy from the original one , and adjusting the smoothing function to strengthen the compact structures .
figure[fig : bar2mass ] shows in the second panel the result of this process applied to the @xmath30 2mass k - band image ( first panel ) .
the subtracted image was convolved with a 2d gaussian with a width of @xmath31 .
an elongated structure , that resembles a bar , emerges at the center .
to evaluate how significant this elongation is , we compare its shape with the 2mass psf ( image at the bottom right corner on the second panel ) . fitting elliptical 2d gaussians to both shapes , we obtain axis ratios ( @xmath32 ) of 1.00 for the psf , and 1.68 with orientation @xmath33 for the central structure .
this seems to indicate that the shape is intrinsic to the sub - structure .
however , the sizes are comparable , implying that the bar - like feature is not resolved in the 2mass image .
_ left panel : _ @xmath34 k - band surface brightness distribution of iras 01072 + 4954 from the 2mass image .
_ right panel : _ result of the unsharp masking procedure applied to the image .
a star in the field that represents the 2mass psf is shown at the bottom right.,scaledwidth=75.0% ] same as in fig.[fig : bar2mass ] but for the @xmath35 nifs image .
the residuals in the right panel are obtained after the removal of the strong psf contribution ( see text for details).,scaledwidth=75.0% ] when the same procedure is applied to the @xmath35 nifs image , the agn and hot dust emission at the center dominate the residuals .
their contribution corresponds to the psf , which is assumed to be well described by the image of a star close to the field of view . to remove this emission ,
the psf is subtracted from the k - band continuum image by the maximum amount possible such that the residual image is still smooth at the center .
then , the unsharp masking procedure is applied convolving the image with a gaussian of @xmath36 .
the second panel in figure[fig : barnifs ] shows the results . in the first panel
the k - band continuum image , created by taking a median of the nifs data cube over the wavelength range , is shown .
the unsharp masking method reveals an elongated structure .
the faint and more extended emission at the center probably corresponds to residuals of the psf subtraction . when fitting the elongated feature with a 2d gaussian , we find that the axis ratio @xmath37 and the orientation @xmath38 are comparable to the values found previously with the 2mass data .
this allows us to suspect that both residuals are related to the same sub - structure . in this case , the sensitivity of the observations is not high enough to obtain a better map of it , and for the same reason , can not be resolved . as shown in fig .
[ fig : barnifs ] , its surface brightness is on average @xmath39 .
this is the same brightness as the one observed in the continuum image at about 1arcsecond from the center . although it is unlikely that its contribution would change by much more than about one order of magnitude our estimation of the bulge luminosity , an exact estimation is not possible because we are missing all emission fainter than @xmath40 which includes the disk in the nifs observations .
we derive the line - of - sight velocity distribution ( losvd ) of the stars in the photometric bulge , by fitting a set of high angular resolution stellar templates to the spectrum integrated over an aperture of @xmath41 , which corresponds to the scale radius of the bulge found above .
we assume that the stellar losvd can be described by a truncated gauss - hermite series parametrized by the radial velocity @xmath42 , the velocity dispersion @xmath43 , and coefficients of the third and fourth hermite polynomials @xmath44 and @xmath45 .
the deep and narrow co absorption features at @xmath46 and @xmath47 indicate that the velocity dispersion of the stars in the bulge is low .
this implies that , despite the high spectral resolution of the nifs data , it is not possible to constrain significantly all the losvd parameters simultaneously @xcite . in this cases
, one would like to bias the solution towards the pure gaussian one and , in that way , avoid the coupling of the parameters , @xmath42 with @xmath44 and @xmath43 with @xmath45 .
the penalized pixel fitting routine ( ppxf ) of cappellari and emsellem @xcite applies such procedure , allowing to recover the losvd in moderately low signal - to noise ( @xmath48 ) and/or undersampled data .
this routine minimizes the difference between the galaxy spectrum and the combination of the stellar templates convolved with the losvd .
the algorithm allows different weights of the stellar templates .
we use a subset of the gemini spectroscopic library of nir stars observed with the nifs ifu in k - band @xcite . to reduce the effects of the template mismatch , we select the stellar types that match best the absorption features in both h and k bands by comparing the galaxy spectrum with medium resolution stellar templates of ivanov et al . @xcite .
the selected set of stellar spectra consists of 7 objects of the following types : k2iii , k5ib , k5ii , k5iii , m2iii , m3iii and m5iii .
fit of the stellar kinematics of the spectrum integrated over an aperture @xmath49 .
the observed spectrum is shown as a thin solid line , while the red solid line represents the best fit to the losvd .
the small spectral regions delimited by the vertical green lines correspond to two h@xmath50 emission lines , which were masked for the fit .
the residuals , shown at the bottom , are shifted vertically for clarity .
, scaledwidth=58.0% ] figure[fig : coppxf ] shows the best fit of the losvd with the parameters : @xmath51 , @xmath52 , @xmath53 and @xmath54 .
the fit is obtained after adopting a @xmath55 and the addition of a legendre polynomial of degree 10 . in a conservative approach ,
the errors are calculated via monte carlo simulations , where 1000 realizations of an input spectrum are fitted with the ppxf routine without any penalization ( @xmath56 ) .
the input spectrum is created by convolving a stellar template with the losvd and adding poisson noise to the level of the galaxy spectrum , in this case , to reach @xmath57 .
then , the effects of undersampling are simulated allowing the velocity scale ( @xmath58 per pixel ) of the input spectrum to vary in such a way that the second moment of the losvd , found by fitting the spectrum with the ppxf routine , varies in the range @xmath59 .
given the moderate @xmath48 of the data and that @xmath43 is low compared to the velocity scale ( less than two pixels ) , the higher moments @xmath44 and @xmath45 of the losvd suffer from large uncertainties .
fitting only the first two moments we obtain @xmath60 and @xmath61 , which are within the estimated errors .
though the bar - like structure is barely detected photometrically , the kinematics of the stars very close to the center are clearly affected by its presence . to elaborate a map of the stellar losvd , we bin the data in apertures with three different sizes that have @xmath62 per bin .
then , we fit the losvd of the spectra integrated in every aperture using the ppxf routine , and combine the results .
this procedure gives us a first estimate of the stellar velocity field ( fig.[fig : losvd ] ) .
special care has to be taken when interpreting the maps because the combination procedure smooths the velocity gradients , and the error of the values increases with the distance to the center .
given the low values of los velocity and velocity dispersion , the @xmath48 of single pixel spectra , even after binning the data only information of the central square arcsecond ( @xmath63 ) can be recovered .
_ left panel : _ los velocity .
_ middle panel : _ velocity dispersion .
_ right panel : _ los velocity and velocity dispersion curves obtained from simulated long - slit spectroscopy along the two axes shown in the left panel .
blue curves+points correspond to the slit along the axis with higher los velocity gradient , red curves+points to a slit perpendicular to the previous one . the radius to the center increases from west to east .
( these are preliminary results ) . ]
first and second panels in fig.[fig : losvd ] show the stellar los velocity and the velocity dispersion maps constructed from the nifs k - band data cube .
the first one displays a characteristic rotation pattern consistent with the expectations for a disk galaxy nearly face - on . in the dispersion map
, there is a clear evidence for an internal structure with @xmath64 . to get a better estimation of the rotation curve and the uncertainties , we simulate long slits of @xmath65arcsec width that cross the center with different orientations .
the data of every slit are binned to have the maximum number of points with @xmath48 in the range 10 - 20 .
finally , the ppxf routine is applied tuning the bias in every single bin to obtain the best fit and the uncertainties in each case .
the third panel in fig.[fig : losvd ] shows the derived rotation and dispersion curves for the slit oriented along the highest los velocity gradient ( @xmath66 ) and those for the slit perpendicular to it .
the fact that the velocity dispersion is dominated by the bar - like structure implies that the @xmath43 estimated above is an upper limit .
the mass of the black hole at the center of galaxies has been found to correlate with some properties of the host galaxy bulge . however , barred galaxies show a higher scatter around the scaling relations defined by normal disk galaxies or follow a different relation @xcite .
it seems that the vertical resonances and/or instabilities responsible for the formation of bars can affect dramatically the velocity distribution of the stars , and favor the development of pseudobulges @xcite .
here we use the derived upper limits of the stellar velocity dispersion and the k - band bulge luminosity to estimate the black hole mass @xmath10 . the validity of the scaling relations used is also discussed .
table[table : mbh ] summarizes the different black hole mass estimations .
the k - band bulge luminosity , obtained by integrating the sersic profile that best fit the surface brightness , is @xmath68 . as explained previously
, we expect this value to be overestimated by @xmath69 order of magnitude . in @xcite
we show that using the @xmath67 relation from marconi & hunt @xcite , the upper limit on the black hole mass is @xmath70 , while with the graham updated version of the same relation @xcite it is @xmath71 .
however , as graham pointed out ( private communication , also @xcite ) , such a relation is not suitable for low-@xmath10 , because it has been derived mainly based on luminous core galaxies .
core galaxies follow a luminosity - dispersion @xmath72 relation and their @xmath73 , probably as a result of a dry merging formation process @xcite ( see also @xcite ) .
in contrast , non - barred low-@xmath10 sersic galaxies follow a @xmath74 relation . given the lack of a study that investigates the @xmath67 in such systems
, we make the exercise to bend the known @xmath67 relation ( eq.14 of @xcite ) at @xmath75 assuming the stated proportionality . with the resulting relation , @xmath76
, we obtain a black hole mass of @xmath77 .
recently gltekin et al .
@xcite and graham et al .
@xcite have recalibrated the @xmath78 .
they use 49 objects and 18 upper limits , and 64 objects , respectively , with reliable parameter estimations .
also xiao et al .
@xcite studied this relation , focusing on low-@xmath10 sources .
we evaluate the black hole mass using their relations and report the results in table[table : mbh ] .
clc @xmath79 & notes on the applied scaling relation & ref .
@xmath80 & 27 bright , mainly early type objects . not suitable in the case ( see text ) . &
@xcite@xmath81 & 22 bright objects .
not suitable in the case ( see text ) . &
@xcite @xmath82 & bending the eq.14 of @xcite .
see text for details . &
@xmath83 & full sample .
intrinsic scatter 0.44 dex .
& @xcite @xmath84 & 8 + 11 upper limits barred galaxies
. might not be reliable ( see text ) .
& @xcite @xmath85 & full sample .
intrinsic scatter 0.35 dex .
& @xcite @xmath86 & 20 barred galaxies . intrinsic scatter 0.31 & @xcite@xmath87 & 155 low-@xmath10 sources .
intrinsic scatter 0.46 & @xcite @xmath88 & 25 low-@xmath10 barred galaxies .
intrinsic scatter 0.46 & @xcite from the black hole masses obtained through the @xmath78 relation , the highest value corresponds to the one calculated from the relation found by gltekin et al . for barred galaxies
however , when looking at the location of these sources in the @xmath10 vs. @xmath43 diagram ( their fig 1 . ) , the 8 galaxies with dynamically detected black holes lie well within the scatter of the correlation depicted by the full sample ( 49 sources ) . nevertheless , the @xmath10 calculated with the relation fitted to the whole sample is two orders of magnitude lower than the one obtained by using only barred galaxies ( + 18 upper limits ) .
the authors also assign a probability of 0.1809 to the case of the upper limit sources not having a black hole .
in contrast , the same probability stated for the full sample is 0.0004 .
we conclude that the difference between the two estimations is introduced by the inclusion of the upper limits to the fit which , in the case of the barred galaxies , dominate the subsample .
therefore , we do not consider this black hole mass estimation as reliable .
all other estimations of the @xmath10 are consistent with the hypothesis of the existence of an imbh at the center of iras 01072 + 4954 . in any case , it is interesting to notice the differences in the @xmath10 calculated from various versions of the scaling relations .
this of course is related to the sample selection , the method applied to fit the data , and the treatment of the errors .
the question of whether the emission detected in the nifs data corresponds to a classical bulge or to a pseudobulge is still open .
high angular resolution observations of the whole galaxy or a major portion of it are necessary to do a proper decomposition of the photometrical components and derive more accurate bulge parameters .
iras 01072 + 4954 is a starburst / seyfert galaxy that hosts an active nucleus detected in the nir and in x - rays @xcite . here
we expose the methods employed to get photometrical and kinematical information of the bulge , a more detailed description will be published in a forthcoming paper .
we present evidence for the presence of an imbh at the center of iras 01072 + 4954 .
we make use of scaling relations of the black hole mass with the k - band bulge luminosity and with the stellar velocity dispersion recently calibrated with different samples .
all the reliable @xmath10 estimations derived from those relations are consistent with @xmath1 .
we also find indications for the presence of a bar - like structure with @xmath89 .
this implies an overestimation of the bulge luminosity , the velocity dispersion and , consequently , of the black hole mass .
m. valencia - s thanks j. stern and a. laor for constructive discussions during the ahar conference , and also a. graham for his comments , in particular referring to the suitability of the scaling relations to derive black hole masses .
m. valencia - s .
, s. smajic and m. vitale are members of the international max - planck research school ( imprs ) for astronomy and astrophysics at the universities of bonn and cologne supported by the max planck society .
j. zuther , s. fischer and m. bremer gladly acknowledge the financial support by group of eight australia - germany joint research co - operation scheme via project - id 50753527 .
m. garca - marn is supported by the german federal department for education and research ( bmbf ) under the project no .
part of this work was supported by the german deutsche forschungsgemeinschaft , dfg , via grant sfb 956 and fruitful discussions with members of the european union funded cost action mp0905 : black holes in a violent universe and pecs project no .
98040 . based on observations ( programm i d : gn-2008b - q-77 ) 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 science and technology facilities council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , ministrio da cincia , tecnologia e inovao ( brazil ) and ministerio de ciencia , tecnologa e innovacin productiva ( argentina ) .
9 magorrian j , tremaine s , richstone d , et al .
1998 _ aj _ * 115 * 2285 ferrarese l and merritt d 2000 _ apj _ * 539 * l9 gebhardt k , bender r , bower g , et al . 2000 _ apj _ * 539 * l13 graham a w , erwin p , caon n and trujillo i 2001 _ apj _ * 563 * l11 marconi a and hunt l k 2003 _ apj _ * 589 * l21 hu j 2008 _ mnras _ * 386 * 2242 graham a w 2008 _ apj _ * 680 * 143 greene j e , peng c y , kim m , et al .
2010 _ apj _ * 721 * 26 kormendy j and bender r 2011 _ nature _ * 469 * 377 graham a w 2011 _ preprint _ : astro - ph/1103.0525 jiang y -f , greene j e and ho l c 2011 _ apj _ * 737 * l45 kormendy j and kennicutt jr .
r c 2004 _ ara&a _ * 42 * 603 ryan c j , de robertis m m , virani s , laor a and dawson p c 2007 _ apj _ * 654 * 799 orban de xivry g , davies r , schartmann m , et al .
2011 _ mnras _ * 417 * 2721 mathur s , fields d , peterson b m , and grupe d 2011 _ preprint _ : astro - ph/1102.0537 greene j e and ho l c 2005 _ apj _ * 627 * 721 greene j e and ho l c 2007 _ apj _ * 656 * 84 dewangan g c , mathur s , griffiths r e and rao a r 2008 _ apj _ * 689 * 762 ai y l , yuan w , zhou h y , wang t g and zhang s h 2011 _ apj _ * 727 * 31 moran e c , halpern j p and helfand d j 1996 _ apjs _ * 106 * 341 panessa f , wolter a , pellegrini s , et al .
2005 _ apj _ * 631 * 707 valencia - s .
m , zuther a , eckart a , iserlohe c and gillian g 2012 _ preprint _ : astro - ph/1203.6063 stern j and laor a 2012 _ mnras _
( accepted ; _ preprint _ : astro - ph/1203.3158 ) mcgregor p j , hart j , conroy p g , et al .
2003 _ society of photo - optical instrumentation engineers ( spie ) conf .
series _ vol .
4841 , ed .
m iye and a f m moorwood , p 15811591 skrutskie m f , cutri r m , stiening r , et al .
2006 _ aj _ * 131 * 1163 athanassoula e 2005 _ mnras _ * 358 * 1477 gltekin k , richstone d o , gebhardt k , et al .
2009 _ apj _ * 698 * 198 graham a w , onken c a , athanassoula e and combes f 2011 _ mnras _ * 412 * 2211 graham a w 2007 _ mnras _ * 379 * 711 graham a w 2012 _ apj _ * 746 * 113 kormendy j , fisher d b , cornell m e and bender r 2009 _ apjs _ * 182 * 216
|
the starburst / seyfert composite galaxy iras 01072 + 4954 ( @xmath0 ) is an enigmatic source that combines a seyfert 1-like x - ray emission with a starburst optical spectrum that lacks broad line emission .
we performed high angular resolution observations of the central kiloparsec of this galaxy in the near - infrared . combining our data with 2mass images of the whole galaxy , we obtain and model the surface brightness profile .
we find indications for the presence of an elongated bar - like structure in both data sets .
we also model the line of sight velocity distribution of the stars in the bulge .
the derived photometrical and kinematical parameters of the bulge are used to evaluate the black hole mass through scaling relations .
we find that all reliable estimations of the black hole mass are consistent with the presence of an intermediate mass black hole of @xmath1 .
|
the _ joint spectral radius _ of a set of matrices ( or linear operators ) originated in early sixties with rota and strang @xcite and found countless applications in functional analysis , dynamical systems , wavelets , combinatorics , number theory , automata , formal languages , etc .
( see bibliography in @xcite ) .
we focus on finite sets of matrices , although all the results are extended to arbitrary compact sets .
if the converse is not stated , we assume a fixed basis in @xmath3 and identify an operator with the corresponding matrix . everywhere below @xmath4
is an arbitrary family of @xmath5-matrices , @xmath6 is the set of all @xmath7 products of @xmath8 matrices from @xmath9 ( with repetitions permitted ) .
a product @xmath10 , is called _ simple _ if it is not a power of a shorter product .
[ d3 ] the joint spectral radius of a family @xmath9 is @xcite @xmath11 this limit exists and does not depend on the matrix norm . in case
@xmath12 ( i.e. , @xmath13 ) , according to gelfand s theorem , the joint spectral radius is @xmath14 , i.e. , coincides with the usual spectral radius @xmath15 , which is the maximal modulus of eigenvalues of @xmath16 ( see , for instance @xcite ) .
the joint spectral radius has the following geometrical meaning : @xmath17 is the infimum of numbers @xmath18 for which there is a norm @xmath19 in @xmath3 such that in the induced operator norm , we have @xmath20 . in particular , @xmath21 if and only if all operators from @xmath9 are contractions in some ( common ) norm .
thus , the joint spectral radius is the indicator of simultaneous contractivity of operators @xmath22 .
another interpretation is due to the discrete dynamical system : @xmath23 where each matrix @xmath24 is chosen from @xmath9 independently for every @xmath8 and @xmath25 .
then @xmath17 is the exponent of fastest possible growth of trajectories of the system : the maximal upper limit @xmath26 among all the trajectories @xmath27 is equal to @xmath28 . in particular , @xmath21 if and only if the system is stable , i.e. , @xmath29 as @xmath30 , for every trajectory .
the problem of computing or estimating the joint spectral radius is notoriously hard .
this is natural in view of the negative complexity results of blondel and tsitsiklis @xcite .
several methods of approximate computation were elaborated in the literature @xcite .
they work well in low dimensions ( mostly , not exceeding @xmath31 ) . when the dimension growth , then ether the estimation becomes rough or the the running time grows dramatically . in recent work @xcite we derived the _ invariant polytope algorithm _ that finds the exact value of @xmath17 for the vast majority of matrix families in dimensions up to @xmath32 . for sets of nonnegative matrices
it works faster and finds the exact value in dimensions up to @xmath33 and higher .
we conjectured in @xcite that the set of matrix families @xmath9 for which this algorithm finds @xmath17 within finite time is of full lebesgue measure .
several open problems from applications have been solved by using this method .
however , it can not handle one important case , which often emerges in practice : the case of several _ spectrum maximizing products_. in this paper we generalize this method making it applicable for a wider class of matrix families , including that special case . to formulate the problem
we need some more facts and notation .
the following double inequality for the joint spectral radius @xmath34 is well known : @xmath35^{1/k } \ \le \
\rho \ \le \
\max_{a \in { { \mathcal{a}}}^\ell } \|a\|^{1/\ell}\ , , \quad k,\ell \in { { \mathbb n}}\ , .\ ] ] moreover , both parts of this inequality converge to @xmath36 as @xmath30 . in fact,@xmath37
( by definition ) and @xmath38^{1/k } = \rho$ ] ( see @xcite ) .
all algorithms of approximate computation of the joint spectral radius are based on this inequality .
first , one finds the greatest value of @xmath39^{1/k}$ ] ( the left hand side of ( [ ineq ] ) ) among all reasonably small @xmath8 , then one minimizes the value @xmath40 ( the right hand side of ( [ ineq ] ) ) by choosing an appropriate matrix norm @xmath41 . thus maximizing the lower bound and minimizing the upper one we approximate the joint spectral radius .
sometimes those two bounds meet each other , which gives the exact value of @xmath36 .
this happens when one finds the _ spectrum maximizing product _
@xmath42 and the norm @xmath43 for which both inequalities in ( [ ineq ] ) become equalities .
[ d5 ] a simple product @xmath44 is called the spectrum maximizing product ( s.m.p . )
if the value @xmath45^{1/n}$ ] is maximal among all products of matrices from @xmath9 of all lengths @xmath46 .
let us remark that an s.m.p .
maximizes the value @xmath45^{1/n}$ ] among all products of our matrices , not just among products of length @xmath47 .
observe that for any s.m.p .
, @xmath44 , we have @xmath45^{1/n } = \rho({{\mathcal{a}}})$ ] .
indeed , from ( [ ineq ] ) it follows that @xmath45^{1/n } \le \rho({{\mathcal{a}}})$ ]
. if this inequality is strict , then there are @xmath48 such that @xmath39^{1/k } > [ \rho(\pi)]^{1/n}$ ] ( because of the convergence property ) , which contradicts to the maximality of @xmath49 .
thus , _ to find the joint spectral radius it suffices to prove that a given product @xmath44 is an s.m.p._. the invariant polytope algorithm @xcite proves the s.m.p .
property of a chosen product @xmath49 by recursive construction of a polytope ( or more general @xmath50 , although here we consider for simplicity real polytopes ) @xmath51 such that @xmath52^{1/n } p , \ , i = 1 , \ldots , m$ ] . in the minkowski norm @xmath53 generated in @xmath3 by this polytope , we have @xmath54^{1/n}$ ] and @xmath53 is said to be an extremal norm for @xmath9 . applying ( [ ineq ] ) for @xmath55 ( left hand side inequality ) and for @xmath56 ( right hand side )
we conclude that @xmath45^{1/n } = \rho({{\mathcal{a}}})$ ] .
note that if a product @xmath44 is s.m.p .
, then so is each of @xmath47 its cyclic permutations . if there are no other s.m.p .
, then we say that the s.m.p . is _
unique _ meaning that it is unique up to cyclic permutations .
the disadvantage of the polytope algorithm is that it is guaranteed to work only if the family @xmath9 has a unique s.m.p .
otherwise the algorithm may not be able to construct the desired polytope within finite time , even if this exists .
the uniqueness of s.m.p .
condition , although believed to be generic , is not satisfied in many practical cases . for example , it happens often that several matrices @xmath57 are s.m.p .
( of length @xmath58 ) . the extension of our algorithm presented in this paper works with an arbitrary ( finite ) number of s.m.p.s .
we prove the theoretical criterion of convergence of the algorithm and apply it to solve two long standing open problems : computing the hlder regularity of the butterfly subdivision scheme ( section [ s5 ] ) and computing the regularity of daubechies wavelets of high order ( section [ s6 ] ) .
we begin with a short description of the invariant polytope algorithm from @xcite for computing the joint spectral radius and finding an extremal polytope norm of a given family @xmath4 .
we make a usual assumption that the family is irreducible , i.e. , its matrices do not have a common nontrivial invariant subspace .
recall that for a reducible family the computation of the joint spectral radius is obtained by solving several problems of smaller dimensions @xcite . for a given set @xmath59
we denote by @xmath60 the convex hull of @xmath61 and by @xmath62 the symmetrized convex hull .
the sign @xmath63 denotes as usual the asymptopic equivalence of two values ( i.e. , equivalence up to multiplication by a constant ) . * the invariant polytope algorithm * ( see @xcite ) .
first , we fix some number @xmath64 and find a simple product @xmath65 with the maximal value @xmath45^{1/n}$ ] among all products of lengths @xmath66 .
we call this product a _ candidate s.m.p . _ and try to prove that it is actually an s.m.p . denote @xmath67^{1/n}$ ] and normalize all the matrices @xmath68 as @xmath69 .
thus we obtain the family @xmath70 and the product @xmath71 such that @xmath72 . for the sake of simplicity
we assume that the largest by modulo eigenvalue of @xmath73 is real , in which case it is @xmath74 .
we assume it is @xmath58 , the case of @xmath75 is considered in the same way .
the eigenvector @xmath76 corresponding to this eigenvalue is called _ leading eigenvector_. the vectors @xmath77 ,
are leading eigenvectors of cyclic permutations of @xmath73 . the set @xmath78 is called _
root_. then we construct a sequence of finite sets @xmath79 and their subsets @xmath80 as follows : zero iteration .
we set @xmath81 .
@xmath8-th iteration , @xmath82 . we have finite set @xmath83 and its subset @xmath84 .
we set @xmath85 and for every @xmath86 , check whether @xmath87 is an _ interior point _ of @xmath88 ( this is an lp problem ) .
if so , we omit this point and take the next pair @xmath89 , otherwise we add @xmath90 to @xmath91 and to @xmath92 . when all pairs @xmath93 are exhausted , both @xmath91 and @xmath92 are constructed .
let @xmath94 .
we have @xmath95 termination .
the algorithm halts when @xmath96 , i.e. , @xmath97 ( no new vertices are added in the @xmath8-th iteration ) . in this case @xmath98 , and hence @xmath99 is an invariant polytope , @xmath49 is an s.m.p . , and @xmath100^{1/n}$ ] .
* end of the algorithm*. actually , the algorithm works with the sets @xmath101 only , the polytopes @xmath102 are needed to illustrate the idea .
thus , in each iteration of the algorithm , we construct a polytope @xmath103 , store all its vertices in the set @xmath101 and spot the set @xmath104 of newly appeared ( after the previous iteration ) vertices . every time we check whether @xmath105 .
if so , then @xmath102 is an invariant polytope , @xmath106 for all @xmath107 , where @xmath108 is the miknowski norm associated to the polytope @xmath102 , and @xmath49 is an s.m.p .
otherwise , we update the sets @xmath101 and @xmath92 and continue .
if the algorithm terminates within finite time , then it proves that the chosen candidate is indeed an s.m.p . and gives the corresponding polytope norm .
although there are simple examples of matrix families for which the algorithm does not terminate , we believe that such cases are rare in practice .
in fact , in all numerical experiments made with randomly generated matrices and with matrices from applications , the algorithm did terminate in finite time providing an invariant polytope .
the only special case when it does not work is when there are several different s.m.p .
( up to cyclic permutations ) . in this case
the algorithm never converges as it follows from the criterion proved in @xcite .
the criterion uses the notion of dominant product which is a strengthening of the s.m.p .
property . a product @xmath44
is called _ dominant _ for the family @xmath9 if there is a constant @xmath109 such that the spectral radius of each product of matrices from the normalized family @xmath110^{-1/n}{{\mathcal{a}}}$ ] , which is neither a power of @xmath73 nor one of its cyclic permutations , is smaller than @xmath111 . a dominant product is an s.m.p . , but , in general , not vice versa . *
theorem a * @xcite .
_ for a given set of matrices and for a given initial product @xmath49 , the invariant polytope algorithm terminates within finite time if and only if @xmath49 is dominant and its leading eigenvalue is unique and simple . _ note that if there is another s.m.p . , which is neither a power of @xmath49 nor of its cyclic permutation , then @xmath49 is not dominant .
therefore , from theorem a we conclude [ c10 ] if a family @xmath9 has more than one s.m.p . , apart from taking powers or cyclic permutations , then , for every initial product , the invariant polytope algorithm does not terminate within finite time .
the problem occurs in the situation when a family has several s.m.p .
, although not generic , but possible in some relevant applications
. mostly those are s.m.p . of length @xmath58 ,
i.e. , some of matrices of the family @xmath9 have the same spectral radius and dominate the others .
this happens , for instance , for transition matrices of refinement equations and wavelets ( see sections [ s5 ] , [ s6 ] ) . in the next section
we show that the algorithm can be modified and extended to families with finitely many spectrum maximizing products .
let a family @xmath9 have @xmath112 candidate s.m.p.s @xmath113 .
these products are assumed to be simple and different ( up to cyclic permutations ) .
denote by @xmath114 the length of @xmath115 .
thus , @xmath116^{1/n_1 } = \ldots = [ \rho(\pi_r)]^{1/n_r } = \rho_c$ ] .
let @xmath117 be the normalized family , @xmath118 be a leading eigenvector of @xmath119 ( it is assumed to be real ) and @xmath120 be the corresponding roots , @xmath121 .
the first idea is to start constructing the invariant polytope with all roots simultaneously , i.e. , with the initial set of vertices @xmath122 however , this algorithm may fail to converge as the following example demonstrates .
[ ex10 ] _ let @xmath16 and @xmath123 be operators in @xmath124 : @xmath16 is a contraction with factor @xmath125 towards the @xmath126 axis along the vector @xmath127 , @xmath123 is a contraction with factor @xmath125 towards the @xmath128 axis along the vector @xmath129 .
the matrices of these operators are @xmath130 clearly , both @xmath16 and @xmath123 have a unique simple leading eigenvalue @xmath58 ; @xmath131 is the leading eigenvector of @xmath16 and @xmath132 is the leading eigenvector of @xmath123 . _ the algorithm with two candidate s.m.p.s @xmath133 and with initial vertices @xmath134 does not converge .
indeed , the set @xmath135 has an extreme point @xmath136 , which tends to the point @xmath137 as @xmath30 , but never reaches it . on the other hand , the same algorithm with initial vertices
@xmath138 terminates immediately after the first iteration with the invariant polytope ( rhombus ) @xmath139 .
indeed , one can easily check that @xmath140 .
therefore , @xmath16 and @xmath123 are both s.m.p . and @xmath141 .
this example shows that if the algorithm does not converge with the leading eigenvectors @xmath142 ( or with the roots @xmath143 ) , it , nevertheless , may converge if one multiplies these eigenvectors ( or the roots ) by some numbers @xmath144 . in example
[ ex10 ] we have @xmath145 .
we call a vector of positive multipliers @xmath146 _ balancing vector_. thus , if a family has several candidate s.m.p.s , then one can _ balance _ its leading eigenvectors ( or its roots ) i.e. , multiply them by the entries @xmath147 of some balancing vector @xmath148 and start the invariant polytope algorithm . in the next section we prove that the balancing vector @xmath149 for which the algorithm converges does exist and can be efficiently found , provided all the products @xmath150 are dominant ( meaning the natural extension of dominance from a single product to a set of products ) .
thus , the corresponding extension of the invariant polytope algorithm is given by algorithm [ algop ] .
a crucial point in algorithm [ algop ] is step * 5 * , where suitable scaling factors @xmath144 have to be given .
then algorithm [ algop ] essentially repeats the invariant polytope algorithm replacing the root @xmath151 by the union of roots @xmath152 .
finding the scaling factors that provide the convergence of the algorithm is a nontrivial problem . a proper methodology to compute them in an optimal way ( in other words , to balance the leading eigenvectors ) is derived in the next section .
[ r40 ] _ till now we have always assumed that the leading eigenvalue is real .
this is not a restriction , because the invariant polytope algorithm is generalized to the complex case as well ( algorithm c in @xcite ) . for the sake of simplicity , in this paper
we consider only the case of real leading eigenvalue . _
let @xmath150 be some products of matrices from @xmath9 of lengths @xmath153 respectively .
they are assumed to be simple and different up to cyclic permutations .
we also assume that all those products are candidates s.m.p.s , in particular , @xmath116^{1/n_1 } = \ldots = [ \rho(\pi_r)]^{1/{n_r } } = \rho_c$ ] .
we set @xmath154 and denote as @xmath61 the supremum of norms of all products of matrices from @xmath155 . since @xmath155 is irreducible and @xmath156 , this supremum is finite @xcite . by @xmath157
we denote the family of adjoint matrices , the definition of @xmath158 is analogous . to each product
@xmath159 we associate the word @xmath160 of the alphabet @xmath161 . by @xmath162 we denote concatenation of words @xmath163 and @xmath164 , in particular , @xmath165 ( @xmath47 times ) ; the length of the word @xmath163 is denoted by @xmath166 . a word is _ simple _ if it is not a power of a shorter word . in the sequel
we identify words with corresponding products of matrices from @xmath155 .
we now make the main assumption : each product @xmath119 has a real leading eigenvalue ( remark [ r40 ] ) , which is either @xmath58 or @xmath75 in this case . for the sake of simplicity , we assume that all @xmath167 ( the case @xmath168 is considered in the same way ) .
we denote by @xmath118 the corresponding leading eigenvector of @xmath115 ( one of them , if it is not unique ) .
clearly , the corresponding adjoint matrix @xmath169 also has a leading eigenvalue @xmath58 , and a real leading eigenvector @xmath170 .
if the the leading eigenvalue is unique and simple , then @xmath171 . in this case
we normalize the adjoint leading eigenvector @xmath170 by the condition @xmath172 .
take arbitrary @xmath173 and consider the product @xmath174 , where @xmath175 and @xmath176 ( for the sake of simplicity we omit the indices @xmath107 ) .
the vectors @xmath177 are the leading eigenvectors of cyclic permutations of the product @xmath115 .
the set @xmath178 is a root of the tree from which the polytope algorithm starts .
let @xmath179 be the polytope produced after the @xmath8-th iteration of the algorithm started with the product @xmath115 , or , which is the same , with the root @xmath180 .
this polytope is a symmetrized convex hull of the set @xmath181 of all alive vertices of the tree on the first @xmath8 levels .
in particular , @xmath182 and @xmath183 .
we denote by @xmath184 and @xmath185 the union of the corresponding sets for all @xmath186 .
if algorithm [ algop ] terminates within finite time , then @xmath185 is finite and @xmath184 is a polytope . if @xmath119 is an s.m.p .
, i.e , if @xmath187 , then , by the irreducibility , the set @xmath184 is bounded @xcite .
now assume that all @xmath115 have unique simple leading eigenvalues , in which case all the adjoint leading eigenvectors @xmath170 are normalized by the condition @xmath188 .
for an arbitrary pair @xmath189 , and for arbitrary @xmath190 , we denote @xmath191 thus , @xmath192 is the length of the orthogonal projection of the convex body @xmath193 onto the vector @xmath194 .
in particular , @xmath195 .
sometimes we omit the superscript @xmath8 if its value is specified . for given @xmath196 and for a point @xmath197 ,
we denote @xmath198 note that for @xmath199 , the sets @xmath200 and @xmath181 have the same symmetrized convex hull @xmath193 .
therefore , the maxima of the function @xmath201 over these two sets coincide with the maximum over @xmath193 .
hence , comparing ( [ qij ] ) and ( [ qjx ] ) gives @xmath202 for an arbitrary balancing vector @xmath146 , we write @xmath203 . our aim is to find a balancing vector such that the polytope algorithm starting simultaneously with the roots @xmath204 terminates within finite time .
[ d10 ] let @xmath190 .
a balancing vector @xmath205 is called @xmath8-admissible , if @xmath206 an @xmath207-admissible vector is called admissible .
since the value @xmath208 is non - decreasing in @xmath8 , we see that the @xmath8-admissibility for some @xmath8 implies the same holds true for all smaller @xmath8 .
in particular , an admissible vector is @xmath8-admissible for all @xmath209 .
we begin with two auxiliary facts needed in the proofs of our main results .
[ l20 ] if a @xmath210 matrix @xmath211 and a vector @xmath212 , are such that @xmath213 , then @xmath211 has an eigenvalue @xmath214 for which @xmath215 , where @xmath216 depends only on the dimension @xmath217 .
see @xcite for the proof .
the following combinatorial fact is well - known : [ l30 ] let @xmath218 be two simple nonempty words of a finite alphabet , @xmath219 be natural numbers . if the word @xmath220 contains a subword @xmath221 such that @xmath222 , then @xmath164 is a cyclic permutation of @xmath163 .
now we extend the key property of dominance to a set of candidate s.m.p.s .
[ d20 ] products @xmath150 are called dominant for the family @xmath9 if all numbers @xmath223^{1/n_i } , \ , i = 1 , \ldots , r$ ] , are equal ( denote this value by @xmath224 ) and there is a constant @xmath109 such that the spectral radius of each product of matrices from the normalized family @xmath225 which is neither a power of some @xmath119 nor that of its cyclic permutation is smaller than @xmath111 . if the products @xmath150 are dominant , then they are s.m.p . , but , in general , not vice versa . the s.m.p .
property means that the function @xmath226^{1/n}$ ] ( @xmath47 is the length of @xmath49 ) defined on the set of products of the normalized family @xmath155 attains its maximum ( equal to one ) at the products @xmath227 and at their powers and cyclic permutations .
the dominance property means , in addition , that for all other products , this function is smaller than some @xmath109 .
this property seems to be too strong , however , the following theorem shows that it is rather general .
[ th10 ] algorithm [ algop ] with the initial products @xmath150 and with a balancing vector @xmath149 terminates within finite time if and only if these products are all dominant , their leading eigenvalues are unique and simple and @xmath149 is admissible .
the proof is in appendix 1 .
[ r30 ] _ theorem [ th10 ] implies that if the algorithm terminates within finite time , then the leading eigenvalues of products @xmath115 must be unique and simple .
that is why we defined admissible balancing vectors for this case only . _
if algorithm [ algop ] produces an invariant polytope , then our candidate s.m.p.s are not only s.m.p.s but also dominant products .
a number of numerical experiments suggests that the situation when the algorithm terminates within finite time ( and hence , there are dominant products ) should be generic . by theorem [ th10 ] , if all our candidate s.m.p.s @xmath228 are dominant and have unique simple leading eigenvalues , then balancing the corresponding roots @xmath229 by weights @xmath144 we run the algorithm and construct an invariant polytope , provided the balancing vector @xmath230 is admissible .
a natural question arises if an admissible vector always exists .
the next theorem gives an affirmative answer .
before we formulate it , we need an auxiliary result .
[ l10 ] for given coefficients @xmath231 the system has a solution @xmath232 is and only if for every nontrivial cycle @xmath233 on the set @xmath234 , we have ( with @xmath235 ) @xmath236 proof . the necessity is simple : for an arbitrary cycle we multiply the @xmath47 inequalities @xmath237 , and obtain ( [ cycle ] ) . to prove sufficiency
, we slightly increase all numbers @xmath208 so that ( [ cycle ] ) still holds for all cycles .
this is possible , because the total number of cycles is finite .
we set @xmath238 and @xmath239 , where maximum is computed aver all paths @xmath240 with @xmath241 .
note that if a path contains a cycle , then removing it increases the product @xmath242 , since the corresponding product along the cycle is smaller than one .
this means that , in the definition of @xmath243 , it suffices to take the maximum over all simple ( without repeated vertices ) paths , i.e. , over a finite set .
it is easy to see that @xmath244 .
reducing now all @xmath208 back to the original values , we obtain strict inequalities .
@xmath245 [ th20 ]
if the products @xmath246 are dominant and have unique simple leading eigenvalues , then they have an admissible balancing vector . proof . in view of lemma [ l10 ]
, it suffices to show that for every cycle @xmath247 , on the set @xmath234 , we have @xmath248 .
we denote this quantity by @xmath249 and take arbitrary @xmath250 .
there is a product @xmath73 of matrices from @xmath155 such that @xmath251 . without loss of generality
we assume that this scalar product is positive .
since the product @xmath252 has a unique simple leading eigenvalue @xmath58 , it follows that for every @xmath253 we have @xmath254 as @xmath30 . applying this to the vector @xmath255
, we conclude that @xmath256 , whenever @xmath8 is large enough .
thus , for the product @xmath257 , the vector @xmath258 is close to @xmath259 .
analogously , for each @xmath260 , we find a product @xmath261 such that the vector @xmath262 is close to @xmath263 .
therefore , for the product @xmath264 , the vector @xmath265 is close to @xmath266 .
note that @xmath267 , where @xmath61 is the supremum of norms of all products of matrices from @xmath70 . if @xmath268 , then invoking lemma [ l20 ] we conclude that @xmath269 , where @xmath270 can be made arbitrarily small by taking @xmath30 . due to the dominance assumption
, it follows that @xmath271 is a power of some @xmath272 , where @xmath273 is the set of products @xmath274 and of its cyclic permutations . due to the dominance assumption
, it follows that @xmath271 is a power of some @xmath272 .
taking @xmath8 large enough we apply lemma [ l30 ] to the words @xmath275 and conclude that @xmath252 is a cyclic permutation of @xmath276 .
similarly , @xmath277 is a cyclic permutation of @xmath276 .
this is impossible , because @xmath278 , and the products @xmath279 are not cyclic permutations of each other .
the contradiction proves that @xmath280 which completes the proof of the theorem .
@xmath245 [ r100 ] _ in a just published paper @xcite , mller and reif present another approach for the computation of joint spectral radius .
developing ideas from @xcite they come up with an elegant branch - and - bound algorithm , which , in contrast to the classical branch - and - bound method @xcite , can find the exact value .
although its running time is typically bigger than for our invariant polytope algorithm @xcite ( we compare two algorithms in example below ) , it has several advantages .
in particular , it uses the same scheme for the cases of one and of many s.m.p
. it would be interesting to analyze possible application of the balancing idea for that algorithm .
_ thus , an admissible balancing vector @xmath149 does exist , provided our candidate s.m.p.s are dominant products and their leading eigenvectors are unique and simple . to find @xmath149 , we take some @xmath209 ,
compute the values @xmath208 by evaluating polytopes @xmath281 , set @xmath282 and solve the following lp problem with variables @xmath283 : @xmath284 if @xmath285 , then the @xmath8-admissible vector does not exist . in this case
, we have to either increase @xmath8 or find other candidate s.m.p.s .
if @xmath286 , then we have a @xmath8-admissible vector @xmath287 .
this vector is optimal in a sense that the minimal ratio between @xmath288 and @xmath208 over all @xmath289 is the biggest possible .
[ r10 ] _ to find an admissible vector one needs to solve lp problem ( [ lp ] ) for @xmath290 .
however , in this case the evaluation of the coefficients @xmath291 may , a priori , require an infinite time .
therefore , we solve this problem for some finite @xmath8 and then run algorithm [ algop ] with the obtained balancing vector @xmath149 . if the algorithm terminates within finite time , then @xmath149 is admissible indeed ( theorem [ th10 ] ) .
otherwise , we can not conclude that there are no admissible balancing and that our candidate s.m.p.s are not dominant .
we try to to increase @xmath8 and find a new vector @xmath149 . _ thus , _ step 5 _ of algorithm [ algop ] consists in choosing a reasonably big @xmath8 and solving lp problem ( [ lp ] ) .
if it results @xmath285 , then the balancing vector does not exist , and hence the algorithm will never converge and we have to find another candidate s.m.p . if @xmath286 , then the vector @xmath287 is @xmath8-admissible . if the algorithm does not converge with this @xmath230 , we increase @xmath8 and solve ( [ lp ] ) again .
[ r15 ] _ our approach works well also if the family has a unique s.m.p .
@xmath292 , but there are other simple products @xmath293 for which the values @xmath294^{1/n_j}$ ] , although being smaller than @xmath295^{1/n_1 } = 1 $ ] , are close to it . in this case
the ( original ) invariant polytope algorithm sometimes converges slowly performing many iterations and producing many vertices .
this is natural , because if , say @xmath296^{1/n_2 } = 1-\delta$ ] with very small @xmath297>0 , then the dominance of @xmath298 over @xmath299 plays a role only after many iterations .
our approach suggests to collect all those `` almost s.m.p .
candidates '' @xmath293 add them to @xmath292 , find the balancing multipliers @xmath147 for their roots by solving lp problem ( [ lp ] ) and run algorithm [ algop ] for the initial set @xmath300 . in most of practical cases , this modification significantly speeds up the algorithm .
_ another modification of the invariant polytope algorithm is considered in the next section .
[ ex:8p ] consider the following example introduced by deslaurier and dubuc in @xcite , associated to an eight - point subdivision scheme , @xmath301 the joint spectral radius of @xmath302 was found in @xcite , where it was shown that both @xmath16 and @xmath123 are s.m.p .
its computation required the construction of a binary tree with @xmath303 levels and considering about @xmath304 matrix products ( i.e. , vertices of the tree ) .
applying our algorithm [ algop ] with the candidates s.m.p .
@xmath16 and @xmath123 and with a balancing vector @xmath305 for the leading eigenvectors @xmath306 and @xmath307 of @xmath16 and @xmath123 , respectively , we construct the invariant polytope with @xmath308 vertices in @xmath309 steps : @xmath310 thus , in our case it suffices to construct a binary tree with @xmath309 levels and consider @xmath308 of its vertices .
the same approach developed for the case of many s.m.p . can be used to introduce extra initial vertices .
sometimes algorithm [ algop ] converges slowly because the family @xmath9 is not well - conditioned : its matrices have a common `` almost invariant subspace '' of some dimension @xmath311 . in this case the invariant polytope @xmath312 may be very flattened ( almost contained in that subspace ) . as a consequence ,
the algorithm performs many iterations because the polytopes @xmath102 , being all flattened , badly absorb new vertices . to avoid this trouble one can introduce extra initial vertices @xmath313 and run algorithm [ algop ] with the initial set @xmath314 . in next theorem we use the value @xmath315 defined in ( [ qjx ] ) .
[ th30 ] suppose algorithm [ algop ] with initial roots @xmath143 terminates within finite time ; then this algorithm with extra initial vertices @xmath313 also does if and only if @xmath316 for all @xmath317 . proof . for the sake of simplicity
, we consider the case @xmath318 , the proof in the general case is similar .
let @xmath312 denote the final polytope produced by the algorithm starting with the root @xmath319 , and let @xmath320 . necessity .
assume the algorithm terminates within finite time .
in the proof of theorem [ th10 ] we showed that the maximum of the linear functional @xmath321 on the final polytope @xmath312 is equal to one and is attained at a unique point @xmath306 .
hence , either @xmath322 , in which case the proof is completed , or @xmath323 , and hence there is a sequence of products @xmath324 such that @xmath325 as @xmath30 . therefore , @xmath326 as @xmath30 .
this implies that , for sufficiently large @xmath8 , the points @xmath327 are not absorbed in the algorithm , and hence , the algorithm can not terminate within finite time .
since the second largest eigenvalue of the matrix @xmath298 is smaller than @xmath58 in absolute value , it follows that @xmath328^t$ ] as @xmath30 .
if @xmath329 , then the matrix @xmath330^t$ ] maps the set @xmath331 to the segment @xmath332 $ ] , which is contained in @xmath333 .
hence , @xmath334 , for some @xmath8 .
therefore , every product @xmath73 of length @xmath335 containing a subword @xmath336 takes the point @xmath337 inside @xmath312 . on the other hand , for all products
@xmath73 not containing this subword , we have @xmath338 , where @xmath339 are some constants ( see ( * ? ? ? * theorem 4 ) ) .
hence , for large @xmath335 , all such products also take the point @xmath337 inside @xmath312 .
thus , all long products take @xmath337 inside @xmath312 , hence the algorithm starting with the initial set @xmath340 terminates within finite time .
@xmath245 in practice , it suffices to introduce extra initial vertices of the form @xmath341 , where @xmath342 is the canonical basis vector ( the @xmath107-th entry is one and all others are zeros ) and @xmath343 is some positive coefficient .
we fix a reasonably small @xmath270 , say , between @xmath344 and @xmath345 , a reasonably large @xmath8 , say @xmath346 , rum @xmath8 iterations of algorithm [ algop ] and compute the values @xmath347 so , @xmath348 is the length of projection of the polytope @xmath102 onto the @xmath107-th coordinate axis , or the largest @xmath107-th coordinate of its vertices .
if @xmath349 for all @xmath107 , then @xmath102 contains the cross - polytope @xmath350 , which , in turn , contains the euclidean ball of radius @xmath351 centered at the origin . in this case the polytope @xmath102 is considered to be well - conditioned , and hence we do not add any extra vertex . if , otherwise , @xmath352 for some @xmath107 , then we add an extra vertex @xmath353^{-1}e_i$ ] .
collecting all such vertices for @xmath354 ( assume there are @xmath355 ones ) we run algorithm [ algop ] with @xmath356 initial extra vertices in the set @xmath357 .
in section [ s6 ] we apply this trick to speed up algorithm [ algop ] for daubechies matrices , which turn out to be extremely ill - conditioned .
subdivision schemes are iterative algorithms of linear interpolation and approximation of multivariate functions and of generating curves and surfaces . due to their remarkable properties
they are widely implemented and studied in an extensive literature .
the butterfly scheme originated with dyn , gregory , and levin @xcite and became one of the most popular bivariate schemes for interpolation and for generating smooth surfaces ( see also the generalization given in @xcite ) .
this scheme is a generalization of the univariate four - point interpolatory scheme to bivariate functions @xcite .
first we take an arbitrary triangulation of the approximated surface and consider the corresponding piecewise - linear interpolation .
this interpolation produces a sequence of piecewise - linear surfaces with thriangular faces that converges to a continuous surface , which is considered as an interpolation of the original one . to describe this algorithm in more detail
we assume that the original surface is given by a bivariate function @xmath358 .
we consider a regular triangulation of @xmath124 and take the values of the function @xmath359 at its vertices .
so we obtain a function @xmath360 defined on a triangular mesh . in the next iteration
we define the function @xmath361 on the refined triangular mesh : the values at the vertices of the original mesh stay the same , the values at midpoints of edges are defined as a linear combination of eighth neighboring vertices as shown in the following figure ( @xmath362 is the new vertex , the coefficients of the linear combination are written in the corresponding vertices ) .
the parameter @xmath0 is the same for all vertices and for all iterations . in the next iteration
we do the same with the new mesh and produce the function @xmath363 , etc .
each function is extended from the corresponding mesh to the whole plane by linearity .
the scheme is said to converge if for every initial function @xmath359 , the functions @xmath364 converge uniformly to a continuous function @xmath365 . due to linearity and shift - invariance
, it suffices to have the convergence for the initial @xmath297-function , which vanishes at all vertices but one , where it is equal to one .
the corresponding limit function @xmath366 is called _ refinable function _ and denoted by @xmath367 .
the scheme converges for every @xmath368 and reproduces polynomial of degree one , i.e. , if @xmath369 , then @xmath370 .
the case @xmath371 is special , in this case the scheme reproduces polynomials of degree @xmath372 @xcite .
one of the most important problems in the study of any subdivision scheme is its regularity .
we use the standard modulus of continuity @xmath373 the _ hlder exponent _ of the function @xmath359 is @xmath374 where @xmath47 is the biggest integer such that @xmath375 .
the hlder regularity of a subdivision scheme is @xmath376 .
this is well - known that the exponent of hlder regularity of a bivariate subdivision scheme is equal to @xmath377 where @xmath378 is the maximal degree of the space of algebraic polynomials reproduced by the scheme , @xmath379 are restrictions of the _ transition operators _ @xmath380 of the scheme to their common invariant subspace orthogonal to the subspace of algebraic polynomials of degree @xmath378 @xcite . for the butterfly scheme , for all @xmath381
, we have @xmath56 and the operators @xmath382 are given by @xmath383-matrices . in the only `` most regular '' case @xmath371 the scheme respects the cubic polynomials , i.e. , @xmath384 , and @xmath385 are @xmath386-matrices .
for this exceptional and most important case it was conjectured in early 90th that @xmath387 , i.e. , the limit function @xmath367 of the scheme are continuously differentiable and its derivative @xmath388 has hlder exponent @xmath58 .
we are going to prove this conjecture and , moreover , we show that the derivative of the function @xmath367 is not lipschitz , but `` almost lipschitz '' with the logarithmic factor @xmath2 .
[ th40 ] the hlder regularity of the butterfly scheme with @xmath1 is equal to @xmath2 .
the derivative @xmath388 of the limit function is `` almost lipschitz '' with the logarithmic factor @xmath2 : @xmath389 [ r20 ] _ in the four - point subdivision scheme , which is a univariate parameter - dependent analogue of the butterfly scheme , the case @xmath1 is also crucial .
this is the only case when the scheme reproduces cubic polynomials .
as it was proved by s.dubuc in 1986 @xcite , the regularity of the four - point scheme in this case is equal to two and @xmath390 , i.e. , @xmath388 is almost lipschitz with the logarithmic factor @xmath58 . by theorem [ th40 ] , for the butterfly scheme , the situation is similar , but @xmath388 is almost lipschitz with the logarithmic factor @xmath2 . _ to prove theorem [ th40 ] we first show that @xmath391 . then we conclude that @xmath392 . by a more refine analysis of the matrices we establish
that @xmath393 then it will remain to refer to some known facts of the theory of subdivision schemes .
the main and most difficult part is the finding of the joint spectral radius of the matrices @xmath385 .
this is done in the next subsections .
then we conclude the proof .
the @xmath386-matrices @xmath385 can be computed exactly by the matlab program of p.oswald @xcite . to simplify the notation ,
we denote @xmath395 , and @xmath396 .
these four @xmath386-matrices are written in appendix 2 .
our goal is to prove that @xmath397 .
it is remarkable that all the matrices of the family @xmath9 and the leading eigenvector of its s.m.p .
possess rational entries , so our computations are actually done in the exact arithmetics .
factorization of the family @xmath9 to @xmath398 and @xmath399*. it appears that the matrices @xmath400 can be factored to a block lower - triangular form in a common basis . to see this
we take the leading eigenvectors @xmath306 of the matrix @xmath16 ( as it was mentioned above , it has rational entries ) : @xmath401 it is checked directly that the linear span of the following six vectors is a common invariant subspace of all matrices from @xmath9 : @xmath402 .
therefore , we can transform the matrices from @xmath9 into block lower - triangular form with diagonal blocks of dimensions @xmath403 and @xmath404 .
the transformation matrix is @xmath405 ( written by columns ) , where @xmath406 is the @xmath8-th canonical basis vector .
this gives the transformed matrices with block lower - triangular structure : @xmath407 [ bc ] with @xmath408-matrices @xmath409 and @xmath410-matrices @xmath411 .
those matrices are written down in appendix 2 .
note that they are all rational .
let @xmath412 and @xmath413 .
it is well known that the joint spectral radius of a block lower - triangular family of matrices is equal to the maximal joint spectral radius of blocks @xcite .
hence @xmath414 .
now we are going to show that @xmath415 , from which it will follow that @xmath141 .
we begin with the family @xmath398 .
analysis of the family @xmath398 * we have a family @xmath398 of @xmath410-matrices @xmath416 written in appendix 2 .
each of the matrices @xmath417 has a simple leading eigenvalue @xmath58 , the corresponding leading eigenvectors @xmath418 are all simple .
the matrix @xmath419 has spectral radius @xmath420 .
we are going to show that @xmath421 , i.e. , this family has three s.m.p .
: @xmath422 , @xmath423 and @xmath424 .
the leading eigenvectors are ( normalized in the maximum norm ) , @xmath425 \frac{1453}{3306 } \\[2 mm ] -\frac{1997}{3306 } \\[2 mm ] \frac{283}{3306 } \\[2 mm ] \frac{2023}{3306 } \\[2 mm ] \frac{787}{3306 } \\[2 mm ] \frac{2851}{11020 } \\[2 mm ] -\frac{4749}{11020 } \\[2 mm ] -\frac{613}{2204 } \\[2 mm ] -1 \\[2 mm ] 1 \end{array } \right ) , \qquad u_2 = \left ( \begin{array}{r } \frac{230}{1089 } \\[2 mm ] \frac{230}{1089 } \\[2 mm ] -\frac{370}{1089 } \\[2 mm ] -\frac{730}{1089 } \\[2 mm ] -\frac{730}{1089 } \\[2 mm ] -\frac{370}{1089 } \\[2 mm ] -\frac{221}{363 } \\[2 mm ] 1 \\[2 mm ] -\frac{221}{363 } \\[2 mm ] 0 \\ 0 \end{array } \right ) , \qquad u_3 = \left ( \begin{array}{r } \frac{66}{625 } \\[2 mm ] \frac{1158}{4375 } \\[2 mm ] \frac{2298}{4375 } \\[2 mm ] \frac{82}{4375 } \\[2 mm ] -\frac{154}{625 } \\[2 mm ] \frac{442}{4375 } \\[2 mm ] -1 \\[2 mm ] \frac{2661}{21875 } \\[2 mm ] -\frac{15959}{21875 } \\[2 mm ] \frac{2204}{4375 } \\[2 mm ] -\frac{2204}{4375 } \end{array } \right).\ ] ] solving problem ( [ lp ] ) for @xmath426 , and for the family @xmath398 , we obtain the scaling factors : @xmath427 the algorithm [ algop ] with the three candidate s.m.p.s @xmath428 , and @xmath429 and with those factors @xmath430 terminates within four iterations and produces an invariant polytope .
however , we slightly change the factors to @xmath431 , @xmath432 and @xmath433 respectively in order to preserve the rationality of the vectors ( and consequently the exactness of the computation ) .
thus , @xmath434 the algorithm still converges within four iterations producing the polytope @xmath435 with @xmath436 vertices .
here is the list of vertices : @xmath437 thus , @xmath438 and , by theorem [ th10 ] , @xmath417 are dominant products for @xmath398 .
analysis of the family @xmath399 * we have a family @xmath399 of @xmath408-matrices @xmath439 written in appendix 2 .
each of the matrices @xmath440 has a simple leading eigenvalue @xmath58 , the matrix @xmath441 has spectral radius @xmath420 .
first of all , we observe the existence of three invariant @xmath2-dimensional subspaces of all the matrices @xmath409 .
we indicate by @xmath442 and @xmath443 the unique leading eigenvectors associated to the eigenvalue @xmath58 of the matrices @xmath444 , @xmath445 and @xmath446 , ( normalized in maximum norm ) , @xmath447 \frac{1}{4 } \\[2 mm ] 0 \\[2 mm ] 0 \\[2 mm ] 1 \\[2 mm ] \end{array } \right ) , \qquad w_2 = \left ( \begin{array}{r } \frac{1}{4 } \\[2 mm ] 0 \\[2 mm ] \frac{1}{4 } \\[2 mm ] 0 \\[2 mm ] 0 \\[2 mm ] 1 \end{array } \right ) , \qquad w_3 = \left ( \begin{array}{r } -\frac{1}{7 } \\[2 mm ] -\frac{25}{28 } \\[2 mm ] -\frac{25}{28 } \\[2 mm ] -\frac{2}{7 } \\[2 mm ] 1 \\[2 mm ] 1 \\[2 mm ] \end{array } \right)\ ] ] the invariant subspaces are given by @xmath448 , @xmath449 and @xmath450 .
thus we define the matrix @xmath451 which block - diagonalizes all matrices @xmath409 , @xmath452 .
we denote the diagonal blocks of the matrices @xmath453 as @xmath454 and @xmath455 .
we obtain three families of @xmath456 matrices to analyze , @xmath457 with @xmath458 0 & -\frac{1}{4 } \end{array } \right ) , \qquad g_{12 } = g_{11 } , \qquad g_{13 } = \left ( \begin{array}{rr } -\frac{1}{4 } & 0 \\[2 mm ] -\frac{1}{4 } & 1 \end{array } \right ) , \qquad g_{14 } = \left ( \begin{array}{rr } 0 & -\frac{1}{4 } \\[2 mm ] 1 & -\frac{1}{4 } \end{array } \right).\ ] ] then @xmath459 with @xmath460 0 & -\frac{1}{4 } \end{array } \right ) , \qquad g_{22 } = \left ( \begin{array}{rr } -\frac{1}{4 } & 0 \\[2 mm ] -\frac{1}{4 } & 1 \end{array } \right ) , \qquad g_{23 } = g_{12 } , \qquad g_{24 } = \left ( \begin{array}{rr } 0 & -\frac{1}{4 } \\[2 mm ] 1 & -\frac{1}{4 } \end{array } \right).\ ] ] i.e. @xmath461 and @xmath462 with @xmath463 1 & \frac{3}{4 } \end{array } \right ) , \qquad g_{32 } = g_{31 } , \qquad g_{33 } = \left ( \begin{array}{rr } 1 & -\frac{1}{4 } \\[2 mm ] 0 & -\frac{1}{4 } \end{array } \right ) , \qquad g_{34 } = \left ( \begin{array}{rr } -1 & \frac{1}{4 } \\[2 mm ] -4 & \frac{3}{4 } \end{array } \right).\ ] ] all previous families have joint spectral radius @xmath58 .
the @xmath464-norm is extremal for @xmath465 .
this means that @xmath466 for @xmath467 .
hence @xmath468 .
since @xmath461 , it follows that @xmath469 . for the family @xmath470 , we apply algorithm [ algop ] and obtain the invariant polytope @xmath312 . in this case
@xmath312 is an octagon with vertices @xmath471 thus , @xmath472 , and hence @xmath473 .
* the proof of theorem [ th40 ] * we start with introducing some further notation .
a norm @xmath41 in @xmath3 is called _ extremal _ for a family @xmath9 if @xmath474 for all @xmath57 .
algorithm [ algop ] constructs an extremal polytope norm .
a family @xmath9 is called _ product bounded _ if norms of all products of matrices from @xmath9 are uniformly bounded ( see e.g. @xcite ) .
if a family has an extremal norm and @xmath141 , then it is product bounded
. we have shown that @xmath475 .
hence , the block lower - triangular form yields that @xmath476 , and so @xmath477 .
furthermore , all matrices @xmath382 in a special basis of the space @xmath478 have the form : @xmath479 where @xmath480 is the @xmath481-diagonal matrix with all diagonal entries equal to @xmath482 ( see @xcite ) .
therefore , the joint spectral radius of @xmath483 is equal to the maximum of the joint spectral radii of the three blocks , i.e. , the maximum of @xmath484 , of @xmath485 , and of @xmath486 .
thus , @xmath487 , and hence @xmath488 . the hlder exponent is found .
now let us analyze the regularity of the derivative @xmath388 . for any refinable function @xmath367 ,
the modulus of continuity @xmath489 is asymptotically equivalent to the logarithm of the left - hand side of the equality ( [ defect1 ] ) with @xmath490 $ ] ( see @xcite ) .
hence , to prove that @xmath491 it suffices to establish ( [ defect1 ] ) .
applying factorization ( [ factor1 ] ) and the results of steps 1 - 3 , we obtain @xmath492 in this block lower - triangular form , we have three blocks ( @xmath493 and @xmath494 ) with the joint spectral radius @xmath495 and one ( @xmath496 ) with a smaller spectral radius ( @xmath497 ) . moreover , all these former three blocks are product bounded , since they have extremal norms .
therefore @xcite , @xmath498 where @xmath499 is a constant . on the other hand ,
it is verified directly that each of the matrices @xmath500 , has two jordan blocks of size @xmath372 corresponding to the leading eigenvalue @xmath501 .
hence , the left - hand side of ( [ defect2 ] ) is bigger than or equal to @xmath502^k\| \ge c_2 \lambda^k k^2 = 4^{-k}k^2 $ ]
. therefore , it is asymptotically equivalent to @xmath503 .
this proves ( [ defect1 ] ) and hence @xmath504 .
@xmath245 the convergence analysis of the butterfly scheme can be extended to other values of @xmath505 $ ] . in this case
we have to deal with @xmath383-matrices @xmath506 the scheme converges ( to continuous limit functions ) if and only if their joint spectral radius @xmath36 is smaller than one .
the regularity of the scheme is equal to @xmath507 . for @xmath508 , each matrix @xmath509 , has two simple eigenvalues of modulus one : precisely @xmath58 and @xmath75 .
the matrix @xmath510 has a simple eigenvalue @xmath75 and the @xmath58 of multiplicity @xmath2 ( both algebraic and geometric ) .
the two leading eigenvectors corresponding to @xmath58 and @xmath75 define a common invariant subspace for the family which can be transformed into a similar block triangular form .
the @xmath456-blocks are respectively @xmath511 0 & -1 \end{array } \right ) , \qquad \left ( \begin{array}{rr } -\frac12 & \frac{\sqrt3}2 \\[2 mm ] \frac{\sqrt3}2 & \frac12 \end{array } \right ) , \qquad \left ( \begin{array}{rr } -\frac12 & -\frac{\sqrt3}2 \\[2 mm ] -\frac{\sqrt3}2 & \frac12 \end{array } \right ) , \qquad \left ( \begin{array}{rr } 1 & 0 \\[2 mm ] 0 & 1 \end{array } \right).\ ] ] they are all symmetric , hence their joint spectral radius equals to the maximal spectral radius of these matrices @xcite , i.e. , is equal to one .
the remaining @xmath512 family of @xmath513 matrices has the fourth matrix as an s.m.p .
starting from its ( unique ) leading eigenvector algorithm [ algop ] terminates within @xmath514 iterations and constructs an invariant polytope norm with @xmath515 vertices .
this proves that @xmath516 , and hence the scheme does not converge .
-5 mm @xmath2 & @xmath372 & @xmath513 & @xmath309 & @xmath403 & @xmath517 & @xmath514 + ' '' '' @xmath518 & @xmath58 & @xmath519 & @xmath520 & @xmath521 & @xmath2 & @xmath522 & @xmath521 & @xmath523 & @xmath58 + @xmath8 & @xmath524 & @xmath525 & @xmath404 & @xmath526 & @xmath527 & @xmath303 & @xmath528 & @xmath529 & @xmath530 + ' '' '' @xmath518 & @xmath531 & @xmath532 & @xmath533 & @xmath534 & @xmath535 & @xmath536 & @xmath537 & @xmath538 & @xmath530 + + we have successfully applied our procedure also for other values @xmath539 .
this leads us to conjecture that the generalized butterfly subdivision scheme is convergent in the whole interval . to support this conjecture
we report in table [ tab : omega ] the results obtained for @xmath540 , @xmath541 ( see also figure [ fig : holdergbs ] ) .
one of the most important applications of the joint spectral radius is the computation of the hlder regularity of refinable functions and wavelets . for daubechies wavelets ,
this problem was studied in many works ( see @xcite and references therein ) . let us recall that the daubechies wavelets is a system of functions @xmath542 , that constitutes an orthonormal basis in @xmath543 .
all functions of this system are generated by double dilates and integer translates of the compactly supported _ wavelet function _ @xmath544 .
i.daubechies in @xcite constructed a countable family of wavelet functions @xmath545 , each generates its own wavelet system .
the function @xmath546 is the haar function .
for all @xmath547 the functions @xmath548 are continuous , their smoothness increases in @xmath335 and @xmath549 @xcite .
so , there are arbitrarily smooth systems of wavelets . however , the price for the regularity is the length of the support , which also grows with @xmath335 : @xmath550 $ ] .
the regularity is a very important characteristics of wavelets , in particular , for their applications in functional analysis , approximation theory , image processing and in numerical pde .
there are several methods to obtain lower and upper bounds for the hlder exponents of wavelet functions ( see @xcite ) .
the matrix approach is the only one that theoretically allows to find them precisely .
it was established in @xcite that @xmath551 , where @xmath552 are special matrices of size @xmath553 .
this enabled to find the precise values of the hlder exponent for some small @xmath335 . for @xmath554 , and @xmath513 ,
the value @xmath555 were found by daubechies and lagarias in @xcite ; for @xmath556 , and @xmath514 , they were computed by g.gripenberg @xcite . every time a delicate analysis of special properties of those matrices was involved . in all the cases the s.m.p . of the family @xmath557
was one of those two matrices , and it was a general belief that this is the case for all @xmath335 .
we apply the standard routine of algorithm [ algop ] to find the precise values of @xmath555 for all @xmath558 . in particular
, we shall see that for @xmath559 , the conjecture of one matrix s.m.p .
is violated and the s.m.p .
is @xmath560 .
we need to recall key steps of construction of the matrices @xmath552 . for every @xmath561 we have a set of @xmath562 _ daubechies filter coefficients _ :
they possess some special properties , in particular , @xmath564 and the polynomial @xmath565 has zero of order @xmath335 at the point @xmath566 .
we set @xmath567 and write the transition @xmath568-matrices as follows : @xmath569 we compute @xmath570 by algorithm [ algop ] .
for some @xmath335 we have a non - unique s.m.p .
( these are the cases when @xmath571 and @xmath422 are both s.m.p . ) and find the balancing vector @xmath149 by the method in subsection [ s3.4 ] .
however , due to symmetry , the entries of that two - dimensional vector @xmath149 are equal .
another difficulty , much more significant for the daubechies matrices is that with growing @xmath335 they become very ill - conditioned .
all vertices of the constructed polytope have very small last components , which corresponds to the property of quasi - invariance of a certain subspace and determines a polytope strongly flattened along certain directions .
for @xmath559 , the last components are about @xmath572 of the values of the first components .
this creates enormous numerical difficulties in the running of algorithm [ algop ] , in particular , in the linear programming routines .
that is why we use the technique with extra initial vertices ( section [ s4 ] ) . in the next subsections we present three illustrative cases ( @xmath573 ) and
report the computed hlder regularity of daubechies wavelets for al @xmath558 .
we demonstrate the computation process for @xmath574 and @xmath526 and see the crucial changes in the behaviour of algorithm [ algop ] when the dimension grows .
the case @xmath575 was done ( by a different approach ) by daubechies and lagarias in @xcite , while the two other cases are new .
for the pair of transition matrices : @xmath576 b_1 & = & \left ( \begin{array}{rrr } -4.676287953813834 & 5.212854848820774 & 0 \\
-0.239791829285782 & 1.703224934278843 & -4.676287953813834 \\ 0 & 0 & -0.239791829285782 \end{array } \right),\end{aligned}\ ] ] the candidate s.m.p . is @xmath571 with @xmath577 . in order to apply algorithm [ algop ]
we compute the leading eigenvector of @xmath571 , @xmath578 and set @xmath579 .
observe that @xmath306 almost lies on the subspace @xmath580 spanned by the vectors @xmath581 of the canonical basis of @xmath582 . applying the normalized matrices @xmath583 and @xmath584 repeatedly to @xmath357 one observes that the resulting vectors also almost lie on the subspace @xmath585 .
this has implications on the flatness of the invariant polyhedron computed by algorithm [ algop ] . computed by the standard algorithm .
[ fig : d41],width=317 ] in its basic implementation the algorithm converges and generates a centrally symmetric polytope in @xmath582 of @xmath586 vertices , in @xmath517 iterations .
the partial polytope norms of the family @xmath587 are reported in table [ tab : k10 ] .
-5 mm @xmath403 & @xmath517 + ' '' '' @xmath588 & @xmath589 & @xmath590 & @xmath591 & @xmath58 + + the vertices ( beyond @xmath306 ) of the polytope follow ( we report only half of them ) : @xmath592 & & v_5 = \left ( \begin{array}{r } -0.7308 \\ -0.2175 \\ 0.0042 \end{array }
\right ) , \quad v_6 = \left ( \begin{array}{r } -0.7308 \\ -0.0393 \\ 0.0113 \end{array } \right)\end{aligned}\ ] ] and the corresponding unit polytope @xmath593 is shown in figure [ fig : d41 ] .
we see that @xmath312 appears to be very flat . computed with an extra initial vertex .
[ fig : d42],width=317 ] in fact , the largest singular value ( or @xmath594 ) , the _ reduced _ singular value decomposition is given by @xmath595 where @xmath596 and @xmath597 are unitary matrices and @xmath598 is a diagonal matrix with nonnegative diagonal elements @xmath599 ( the singular values , usually ordered in decreasing way ) that are the square roots of the eigenvalues of the hermitian ( semi)-positive definite matrix @xmath600 ( see e.g. @xcite ) .
] of the matrix of its vertices @xmath601 is @xmath602 , while the smallest one is @xmath603 ( note that if @xmath604 would be zero then @xmath605 would not span the whole space and the polytope would be contained in a subspace ) .
this almost @xmath606 times difference gives a numerical evidence of the flattening phenomenon . to avoid it we add an extra initial vertex @xmath607 and obtain a better behaviour of algorithm [ algop ] and a more balanced polytope ( see figure [ fig : d42 ] ) .
the polytope has now @xmath608 vertices which are computed in only two iterations .
denote @xmath609 .
the vertices beyond @xmath306 and @xmath307 of the polytope follow ( we report only half of them ) : @xmath610 the largest singular value is now @xmath611 and the smallest singular value is now @xmath612 , which demonstrate a much more balanced shape of the unit ball of the polytope extremal norm .
the computed hlder exponent of @xmath613 is @xmath614 .
we have the @xmath615-matrices @xmath552 and are going to prove that the s.m.p . is @xmath616 which interesting in itself since it contradicts to the conjectured property that @xmath617 for all @xmath335 ( see introduction ) .
let @xmath618 , @xmath619 .
applying algorithm [ algop ] we compute the starting set of vectors @xmath620 , where @xmath306 is the leading eigenvector of @xmath49 , @xmath621 .
clearly , @xmath622 .
thus , @xmath623 and @xmath624 are the leading eigenvectors of cyclic permutations of the product @xmath49 .
we have @xmath625 @xmath626 observe that the last two components of all the vectors @xmath627 are very small , i.e. , all these vectors almost lie on the subspace @xmath628 spanned by the vectors @xmath629 of the canonical basis of @xmath630 . applying @xmath583 and @xmath584 repeatedly to @xmath357
one observes that the resulting vectors also almost lie on the subspace @xmath631 .
this means that the invariant polytope computed by algorithm [ algop ] is nearly degenerate , it is close to a @xmath517-dimensional polytope .
a consequence of this is a slow convergence behaviour of the algorithm and an ill - conditioning of basic linear algebra operations .
the algorithm terminates and generates a centrally symmetric polytope in @xmath630 of @xmath632 vertices , in @xmath529 iterations .
the partial polytope norms of the family @xmath587 are reported in table [ tab : k10 ] .
-5 mm @xmath517 & @xmath514 & @xmath524 & @xmath525 & @xmath404 & @xmath526 & @xmath527 & @xmath303 & @xmath528 & @xmath529 + ' '' '' @xmath633 & @xmath634 & @xmath635 & @xmath636 & @xmath637 & @xmath638 & @xmath639 & @xmath640 & @xmath641 & @xmath642 & @xmath643 & @xmath644 & 1 + + the largest singular value of the set of vertices is @xmath645 and the smallest singular value is @xmath646 , which gives a numerical evidence of the flatness of the polytope . to correct the behavior of the algorithm we add an extra vector along the `` most narrow '' ( for the polytope @xmath312 ) direction
( see section [ s4 ] ) .
adding the vector @xmath647 , we obtain a better behaviour of the algorithm and a more balanced polytope .
the results are summarized here ; the polytope has @xmath648 vertices which are computed in @xmath525 iterations .
-5 mm @xmath513 & @xmath309 & @xmath403 & @xmath517 & @xmath514 & @xmath524 & @xmath525 + ' '' '' @xmath588 & @xmath649 & @xmath650 & @xmath651 & @xmath652 & @xmath653 & @xmath654 & @xmath655 & @xmath656 & 1 + + the largest singular value is now @xmath657 and the smallest one is @xmath658 , which demonstrate a much more balanced shape of the unit ball of the polytope extremal norm .
the computed hlder exponent of @xmath659 is @xmath660 [ r20b ] _ if we consider the adjoint family @xmath661 we have naturally @xmath662 .
it is remarkable that applying algorithm [ algop ] to @xmath661 , we do not have problems with flattening .
alas convergence remains slow ( @xmath526 iterations and @xmath663 vertices ) .
_ in this case both @xmath571 and @xmath422 are s.m.p . , i.e. @xmath665 . the balancing technique ( section [ s3 ] ) gives @xmath666 , i.e. , equal weights to the leading eigenvectors of @xmath571 and of @xmath422 .
the two vectors , say @xmath306 and @xmath307 follow : @xmath667 we observe that they almost lie on the subspace @xmath668 spanned by the vectors @xmath669 of the canonical basis of @xmath670 . applying @xmath583 and @xmath584 repeatedly to @xmath357
one observes that the resulting vectors also almost lie on the subspace @xmath671 .
note that the last components seems to vanish exponentially in the number of iterations ( from the smallest to the highest index ) .
if we add an extra initial vector @xmath672 , algorithm [ algop ] terminates after @xmath524 iterations with an invariant polytope of @xmath673 vertices .
the partial polytope norms are reported in table [ tab : k12 ] .
the largest singular value is @xmath674 and the smallest one is @xmath675 , which demonstrate a relatively balanced shape .
the computed hlder exponent is @xmath676 .
-5 mm @xmath403 & @xmath517 + ' '' '' @xmath588 & @xmath677 & @xmath678 & @xmath679 & @xmath58 + proceeding this way we have computed the exact values of hlder exponents of daubechies wavelets according to table [ tab : results ] .
we indicate the s.m.p .
, the extra initial vectors , the number of vertices of the final polytope ( @xmath680 ) , the number of iterations of algorithm [ algop ] ( @xmath681its ) and the hlder exponent @xmath230 . -5 mm & @xmath680 & @xmath230 + ' '' '' @xmath2 & @xmath571 & none & @xmath58 & @xmath682 & @xmath683 + @xmath372 & @xmath571 & none & @xmath372 & @xmath684 & @xmath685 + @xmath513 & @xmath571 & @xmath686 & @xmath2 & @xmath608 & @xmath687 + @xmath309 & @xmath688 & @xmath689 & @xmath513 & @xmath690 & @xmath691 + @xmath403 & @xmath688 & @xmath692 & @xmath309 & @xmath693 & @xmath694 + @xmath517 & @xmath688 & @xmath692 & @xmath309 & @xmath695 & @xmath696 + @xmath514 & @xmath688 & @xmath697 & @xmath309 & @xmath698 & @xmath699 + @xmath524 & @xmath688 & @xmath700 & @xmath403 & @xmath701 & @xmath702 + @xmath525 & @xmath703 & @xmath704 & @xmath525 & @xmath705 & @xmath706 + @xmath404 & @xmath688 & @xmath707 & @xmath404 & @xmath648 & @xmath708 + @xmath526 & @xmath688 & @xmath709 & @xmath517 & @xmath673 & @xmath710 + @xmath527 & @xmath688 & @xmath711 & @xmath712 & @xmath713 & @xmath714 + @xmath303 & @xmath688 & @xmath715 & @xmath528 & @xmath713 & @xmath716 + @xmath528 & @xmath717 & @xmath718 & @xmath303 & @xmath719 & @xmath720 + @xmath529 & @xmath703 & @xmath721 & @xmath527 & @xmath722 & @xmath723 + @xmath724 & @xmath688 & @xmath725 & @xmath404 & @xmath726 & @xmath727 + @xmath712 & @xmath728 & @xmath729 & @xmath527 & @xmath730 & @xmath731 + @xmath732 & @xmath688 & @xmath733 & @xmath732 & @xmath734 & @xmath735 + @xmath32 & @xmath688 & @xmath736 & @xmath308 & @xmath737 & @xmath738 +
if the algorithm terminates within finite time , then the products @xmath739 are dominant and their leading eigenvalues are unique and simple .
this is shown in the same way as in the proof of theorem 4 of @xcite for @xmath740 .
to prove that @xmath149 is admissible , we take arbitrary @xmath107 and @xmath741 and denote by @xmath742 the vertex of the final polytope @xmath312 with the largest scalar product @xmath743 . since @xmath744^kv_j^*,z ) = ( v_j^ * , z)$ ] , all the points @xmath745 also provide the largest scalar product with the vector @xmath194 .
hence , they are all on the boundary of @xmath312 , i.e. , they are not absorbed in the algorithm .
consequently , the algorithm can terminate within finite time only if @xmath742 is the leading eigenvector of @xmath227 , i.e. , @xmath746 .
thus , the maximal scalar product @xmath743 over all @xmath747 is attained at a unique vertex @xmath746 , where it is equal to @xmath748 . since @xmath749
, it follows that @xmath750 thus , @xmath751 , which proves the admissibility of @xmath149 .
denote by @xmath273 the set of products @xmath752 , and of its cyclic permutations . since this is a set of dominant products for @xmath155 , their leading eigenvectors @xmath753 are all different up to normalization , i.e. , they are all non - collinear . indeed ,
if , say , @xmath754 , then , replacing the products @xmath115 and @xmath755 by the corresponding cyclic permutations , it may be assumed that @xmath756 .
however , in this case @xmath757 for any @xmath758 .
therefore , the spectral radius of every product of the form @xmath759 is at lest one . by the dominance assumption ,
this product is a power of some product @xmath760 .
taking now @xmath758 large enough and applying lemma [ l30 ] first to the words @xmath761 and then to the words @xmath762 , we conclude that both @xmath763 must be cyclic permutations of @xmath73 , which is impossible .
thus , all the leading eigenvectors @xmath764 are non - collinear .
hence , there is @xmath270 such that for every @xmath765 the ball of radius @xmath766 centered at @xmath767 may contain leading eigenvectors of at most one matrix from @xmath273 .
if the polytope algorithm with the initial roots @xmath768 does not converge , then there is an element of some root , say , @xmath769 and an infinite sequence @xmath770 , which is not periodic with period @xmath298 , and such that every vector @xmath771 is not absorbed in the algorithm .
this implies that there is a constant @xmath772 such that @xmath773 for all @xmath8 . on the other hand , @xmath774 ,
hence the compactness argument yields the existence of a limit point @xmath775 of this sequence .
thus , for some subsequence , we have @xmath776 as @xmath30 .
let @xmath250 be a small number to specify .
passing to a subsequence , it may be assumed that @xmath777 for all @xmath778 .
denote @xmath779 .
we have @xmath780 . invoking the triangle inequality ,
we obtain @xmath781 hence , lemma [ l20 ] yields @xmath782 for all @xmath8 .
the dominance assumption implies that if @xmath250 is small enough , then all @xmath783 must be powers of matrices from @xmath273 .
each matrix from @xmath273 has a simple unique leading eigenvalue @xmath58 .
therefore , there is a function @xmath784 such that @xmath785 as @xmath786 , and for every matrix @xmath787 which is a power of a matrix from @xmath273 the inequality @xmath788 implies that there exists a leading eigenvector @xmath789 of @xmath787 such that @xmath790 .
thus , for every @xmath791 there exists a leading eigenvector @xmath792 of @xmath783 such that @xmath793 . for sufficiently small @xmath297 we have @xmath794 .
hence , for every @xmath8 , the vector @xmath792 belongs to the ball of radius @xmath795 centered at @xmath796 .
however , this ball may contain a leading eigenvector of at most one matrix from @xmath273 , say @xmath73 .
therefore , all @xmath797 , are powers of @xmath73 and @xmath796 is the leading eigenvector of @xmath73 .
thus , @xmath798 for some @xmath799 .
clearly , @xmath73 is a cyclic permutation of some @xmath227 .
if @xmath800 , then assuming that @xmath801 ( the general case is considered in the same way ) , we have @xmath802 as @xmath30 .
since the balancing vector @xmath230 is admissible and @xmath803 , it follows that @xmath804 .
therefore , the limit point @xmath805 for some @xmath806 , is interior for the initial polytope @xmath807 .
this means that for large @xmath8 , the point @xmath808 will be absorbed in the algorithm , which contradicts to the assumption .
consider the last case , when @xmath73 is a cyclic permutation of @xmath298 . since @xmath796 is the leading eigenvector of @xmath73
we see that @xmath809 for some @xmath810 and @xmath811 .
we assume @xmath812 , the case of negative @xmath813 is considered in the same way .
if @xmath814 , then we again conclude that @xmath808 are absorbed in the algorithm for large @xmath8 .
if @xmath815 , then for the product @xmath816 , we have @xmath817 , and hence @xmath818 . therefore , @xmath819 as @xmath30 . by lemma [ l20 ]
, this means that the spectral radius of the product @xmath820 tends to @xmath821 as @xmath30 .
the dominance assumption implies now that for every sufficiently large @xmath8 , this product is a power of some @xmath822 . applying lemma [ l30 ] to the words @xmath823 , we see that @xmath824 is a cyclic permutation of @xmath73 . in particular , @xmath825 .
therefore , the length @xmath826 is divisible by @xmath827 .
hence , the number @xmath828 is divisible by @xmath47 , which is impossible , because @xmath829 .
this completes the proof .
we report here the matrices of the butterfly scheme for @xmath830 . first the @xmath386 family @xmath9 , then the @xmath831 family @xmath832 and finally the @xmath833 family @xmath834 .
the four matrices @xmath835 : @xmath836 \frac{15049}{4104 } & \frac{13997}{1026 } & \frac{89}{342 } & -\frac{4225}{1368 } & -\frac{3062}{513 } & \frac{7531}{4104 } & \frac{120}{19 } & -\frac{3851}{1368 } & \frac{785}{456 } & \frac{15553}{4104 } & \frac{7957}{1026 } & -\frac{9767}{4104 } & -\frac{8171}{4104 } & -\frac{449}{216 } & -\frac{17}{54 } & \frac{47}{171 } & \frac{3911}{1368 } \\[2 mm ] -\frac{29369}{4104 } & -\frac{16895}{2052 } & -\frac{773}{228 } & \frac{4133}{1368 } & \frac{475}{108 } & \frac{5893}{4104 } & -\frac{980}{171 } & \frac{1643}{1368 } & -\frac{4919}{1368 } & -\frac{15629}{4104 } & -\frac{7639}{2052 } & \frac{5251}{4104 } & \frac{9871}{4104 } & \frac{15877}{4104 } & \frac{7631}{2052 } & -\frac{443}{228 } & -\frac{2347}{1368 } \\[2 mm ] \frac{65953}{4104 } & -\frac{17155}{4104 } & \frac{1475}{1368 } & -\frac{9643}{1368 } & \frac{1415}{4104 } & \frac{133}{216 } & -\frac{339}{152 } & -\frac{1573}{684 } & \frac{3469}{456 } & \frac{30277}{4104 } & \frac{1795}{4104 } & \frac{1897}{4104 } & -\frac{2441}{1026 } & -\frac{299}{1026 } & -\frac{7655}{4104 } & \frac{5005}{1368 } & -\frac{2347}{1368 } \\[2 mm ] -\frac{43913}{4104 } & \frac{34169}{4104 } & -\frac{1283}{456 } & \frac{4643}{1368 } & -\frac{2677}{4104 } & \frac{10537}{4104 } & \frac{209}{72 } & \frac{409}{684 } & -\frac{6971}{1368 } & -\frac{14153}{4104 } & \frac{14935}{4104 } & -\frac{7043}{4104 } & \frac{1547}{2052 } & \frac{4181}{2052 } & \frac{14425}{4104 } & -\frac{1429}{456 } & \frac{3911}{1368 } \\[2 mm ] -\frac{41957}{4104 } & -\frac{1319}{2052 } & -\frac{863}{171 } & \frac{4577}{1368 } & \frac{4501}{2052 } & \frac{14623}{4104 } & -\frac{42}{19 } & \frac{1525}{1368 } & -\frac{677}{152 } & -\frac{19439}{4104 } & -\frac{391}{513 } & -\frac{5069}{4104 } & \frac{4825}{4104 } & \frac{12691}{4104 } & \frac{4303}{1026 } & -\frac{523}{171 } & \frac{887}{1368 } \\[2 mm ] \frac{889}{513 } & -\frac{5081}{1026 } & \frac{181}{1140 } & \frac{112}{171 } & \frac{3661}{1026 } & \frac{991}{5130 } & -\frac{1571}{855 } & \frac{1309}{684 } & \frac{1657}{1140 } & -\frac{61}{540 } & -\frac{7141}{2565 } & \frac{1139}{2565 } & -\frac{1963}{10260 } & \frac{8447}{10260 } & -\frac{977}{5130 } & \frac{3871}{3420 } & -\frac{191}{855 } \\[2 mm ] -\frac{929}{270 } & \frac{2515}{1026 } & \frac{679}{380 } & \frac{365}{342 } & -\frac{3935}{1026 } & -\frac{7168}{2565 } & \frac{5017}{3420 } & -\frac{1189}{855 } & -\frac{599}{228 } & -\frac{2255}{2052 } & \frac{3863}{10260 } & \frac{965}{1026 } & \frac{13877}{10260 } & -\frac{1199}{513 } & -\frac{3193}{2565 } & -\frac{3107}{3420 } & -\frac{191}{855 } \\[2 mm ] -\frac{1003}{10260 } & -\frac{11509}{5130 } & \frac{1411}{1140 } & \frac{167}{180 } & \frac{4391}{5130 } & -\frac{13723}{10260 } & \frac{1061}{1710 } & \frac{3181}{1710 } & \frac{401}{285 } & -\frac{17719}{10260 } & -\frac{7006}{2565 } & -\frac{5029}{10260 } & -\frac{8209}{10260 } & -\frac{3826}{2565 } & -\frac{20899}{10260 } & \frac{53}{171 } & -\frac{106}{171 } \\[2 mm ] -\frac{8518}{2565 } & -\frac{42173}{10260 } & \frac{1021}{1140 } & \frac{784}{285 } & \frac{21967}{10260 } & -\frac{533}{270 } & -\frac{466}{171 } & \frac{107}{228 } & -\frac{10439}{3420 } & -\frac{25}{1026 } & -\frac{12499}{10260 } & \frac{2981}{1026 } & \frac{7787}{2565 } & \frac{20039}{10260 } & \frac{2999}{2565 } & -\frac{115}{342 } & -\frac{2429}{3420 } \\[2 mm ] -\frac{56513}{5130 } & -\frac{163009}{20520 } & -\frac{11933}{6840 } & \frac{5249}{1140 } & \frac{66761}{20520 } & -\frac{2971}{2565 } & -\frac{1481}{360 } & \frac{14827}{6840 } & -\frac{2009}{380 } & -\frac{14051}{2052 } & -\frac{118553}{20520 } & \frac{2269}{2052 } & \frac{9701}{4104 } & \frac{39097}{20520 } & \frac{32347}{20520 } & -\frac{2169}{760 } & -\frac{1658}{855 } \\[2 mm ] -\frac{51289}{10260 } & \frac{51557}{5130 } & -\frac{21}{95 } & \frac{75}{76 } & -\frac{12344}{2565 } & \frac{3443}{10260 } & \frac{18733}{3420 } & -\frac{29}{57 } & -\frac{1189}{684 } & -\frac{20527}{10260 } & \frac{6691}{2052 } & -\frac{26377}{10260 } & -\frac{9703}{10260 } & -\frac{13943}{5130 } & -\frac{793}{1026 } & -\frac{5467}{3420 } & \frac{775}{342 } \\[2 mm ] -\frac{23953}{10260 } & \frac{33437}{5130 } & -\frac{523}{855 } & \frac{509}{1140 } & -\frac{3284}{2565 } & \frac{2863}{2052 } & \frac{688}{171 } & \frac{3233}{3420 } & \frac{47}{380 } & -\frac{14779}{10260 } & \frac{9307}{5130 } & -\frac{6526}{2565 } & -\frac{4882}{2565 } & -\frac{2609}{2052 } & -\frac{2063}{5130 } & -\frac{761}{1140 } & \frac{775}{342 } \\[2 mm ] \frac{92413}{5130 } & -\frac{132199}{20520 } & \frac{3991}{2280 } & -\frac{7453}{1140 } & \frac{35951}{20520 } & \frac{205}{513 } & -\frac{11701}{6840 } & -\frac{179}{760 } & \frac{6845}{684 } & \frac{65351}{10260 } & -\frac{40487}{20520 } & -\frac{5029}{10260 } & -\frac{15133}{4104 } & -\frac{2051}{1080 } & -\frac{4313}{1080 } & \frac{6557}{1368 } & -\frac{1658}{855 } \\[2 mm ] \frac{11347}{5130 } & -\frac{3617}{2565 } & \frac{4129}{1710 } & -\frac{43}{285 } & -\frac{151}{270 } & -\frac{5068}{2565 } & -\frac{3253}{3420 } & -\frac{2231}{1710 } & -\frac{109}{380 } & \frac{22733}{10260 } & \frac{1093}{2052 } & \frac{13849}{5130 } & \frac{19073}{10260 } & \frac{415}{2052 } & \frac{160}{513 } & \frac{1193}{1140 } & -\frac{2429}{3420 } \\[2 mm ] -\frac{30997}{4104 } & -\frac{79909}{20520 } & \frac{2113}{1368 } & \frac{42259}{6840 } & \frac{78761}{20520 } & -\frac{8299}{4104 } & -\frac{2611}{2280 } & \frac{1859}{570 } & -\frac{29269}{6840 } & -\frac{77327}{20520 } & -\frac{52403}{20520 } & \frac{48463}{20520 } & \frac{32657}{10260 } & \frac{8377}{5130 } & \frac{19687}{20520 } & -\frac{69}{760 } & -\frac{637}{6840 } \\[2 mm ] \frac{44899}{20520 } & -\frac{21365}{4104 } & \frac{9313}{6840 } & -\frac{193}{1368 } & \frac{403}{216 } & -\frac{7219}{4104 } & -\frac{4751}{6840 } & \frac{6211}{3420 } & \frac{13183}{6840 } & -\frac{22907}{20520 } & -\frac{84623}{20520 } & -\frac{4067}{20520 } & -\frac{12139}{10260 } & -\frac{7481}{5130 } & -\frac{54053}{20520 } & \frac{7027}{6840 } & -\frac{7129}{6840 } \end{array } \right ) \\[2 mm ] a_2 & = & \left ( \begin{array}{rrrrrrrrrrrrrrrrr } \frac{7691}{684 } & -\frac{9157}{1368 } & \frac{11009}{4104 } & -\frac{4561}{2052 } & \frac{10501}{4104 } & -\frac{2813}{2052 } & -\frac{3673}{1368 } & \frac{397}{1368 } & \frac{286}{57 } & \frac{10133}{2052 } & -\frac{6077}{4104 } & \frac{2209}{1026 } & \frac{13}{216 } & \frac{319}{4104 } & -\frac{8237}{4104 } & \frac{5005}{1368 } & -\frac{443}{228 } \\[2 mm ] -\frac{9157}{1368 } & \frac{7691}{684 } & -\frac{2813}{2052 } & \frac{10501}{4104 } & -\frac{4561}{2052 } & \frac{11009}{4104 } & \frac{286}{57 } & \frac{397}{1368 } & -\frac{3673}{1368 } & -\frac{6077}{4104 } & \frac{10133}{2052 } & -\frac{8237}{4104 } & \frac{319}{4104 } & \frac{13}{216 } & \frac{2209}{1026 } & -\frac{443}{228 } & \frac{5005}{1368 } \\[2 mm ] -\frac{2345}{1368 } & -\frac{2099}{171 } & \frac{1132}{513 } & \frac{9409}{4104 } & \frac{2549}{513 } & -\frac{12763}{4104 } & -\frac{1939}{342 } & \frac{1603}{1368 } & -\frac{1033}{456 } & -\frac{5915}{4104 } & -\frac{5657}{1026 } & \frac{20095}{4104 } & \frac{13297}{4104 } & \frac{11569}{4104 } & \frac{1601}{2052 } & \frac{47}{171 } & -\frac{1429}{456 } \\[2 mm ] \frac{1571}{342 } & -\frac{16343}{1368 } & \frac{11663}{4104 } & \frac{1183}{1026 } & \frac{20803}{4104 } & -\frac{6641}{2052 } & -\frac{2281}{456 } & \frac{2275}{1368 } & \frac{449}{342 } & \frac{2411}{2052 } & -\frac{20279}{4104 } & \frac{101}{27 } & \frac{7321}{4104 } & \frac{5665}{4104 } & -\frac{5789}{4104 } & \frac{1099}{456 } & -\frac{523}{171 } \\[2 mm ] -\frac{16343}{1368 } & \frac{1571}{342 } & -\frac{6641}{2052 } & \frac{20803}{4104 } & \frac{1183}{1026 } & \frac{11663}{4104 } & \frac{449}{342 } & \frac{2275}{1368 } & -\frac{2281}{456 } & -\frac{20279}{4104 } & \frac{2411}{2052 } & -\frac{5789}{4104 } & \frac{5665}{4104 } & \frac{7321}{4104 } & \frac{101}{27 } & -\frac{523}{171 } & \frac{1099}{456 } \\[2 mm ] -\frac{2099}{171 } & -\frac{2345}{1368 } & -\frac{12763}{4104 } & \frac{2549}{513 } & \frac{9409}{4104 } & \frac{1132}{513 } & -\frac{1033}{456 } & \frac{1603}{1368 } & -\frac{1939}{342 } & -\frac{5657}{1026 } & -\frac{5915}{4104 } & \frac{1601}{2052 } & \frac{11569}{4104 } & \frac{13297}{4104 } & \frac{20095}{4104 } & -\frac{1429}{456 } & \frac{47}{171 } \\[2 mm ] \frac{4319}{684 } & -\frac{2377}{684 } & -\frac{20821}{5130 } & -\frac{8057}{2052 } & \frac{11303}{10260 } & \frac{17159}{5130 } & -\frac{1097}{342 } & -\frac{2771}{3420 } & \frac{14287}{3420 } & \frac{21263}{10260 } & -\frac{953}{2052 } & -\frac{6568}{2565 } & -\frac{4877}{2052 } & \frac{7367}{10260 } & \frac{1231}{2052 } & \frac{3871}{3420 } & -\frac{3107}{3420 } \\[2 mm ] \frac{8473}{3420 } & \frac{8473}{3420 } & \frac{1211}{5130 } & -\frac{16903}{10260 } & -\frac{16903}{10260 } & \frac{1211}{5130 } & \frac{4241}{1710 } & \frac{673}{855 } & \frac{4241}{1710 } & -\frac{1849}{10260 } & -\frac{1849}{10260 } & -\frac{5416}{2565 } & -\frac{24079}{10260 } & -\frac{24079}{10260 } & -\frac{5416}{2565 } & \frac{53}{171 } & \frac{53}{171 } \\[2 mm ] -\frac{2377}{684 } & \frac{4319}{684 } & \frac{17159}{5130 } & \frac{11303}{10260 } & -\frac{8057}{2052 } & -\frac{20821}{5130 } & \frac{14287}{3420 } & -\frac{2771}{3420 } & -\frac{1097}{342 } & -\frac{953}{2052 } & \frac{21263}{10260 } & \frac{1231}{2052 } & \frac{7367}{10260 } & -\frac{4877}{2052 } & -\frac{6568}{2565 } & -\frac{3107}{3420 } & \frac{3871}{3420 } \\[2 mm ] -\frac{824}{285 } & -\frac{1198}{171 } & -\frac{13063}{5130 } & \frac{2573}{2565 } & \frac{7679}{2565 } & \frac{5057}{5130 } & -\frac{6337}{1710 } & \frac{1663}{1140 } & -\frac{1333}{3420 } & -\frac{16613}{5130 } & -\frac{8501}{2052 } & -\frac{61}{135 } & -\frac{1081}{10260 } & \frac{2455}{2052 } & \frac{2041}{2052 } & -\frac{761}{1140 } & -\frac{5467}{3420 } \\[2 mm ] -\frac{1198}{171 } & -\frac{824}{285 } & \frac{5057}{5130 } & \frac{7679}{2565 } & \frac{2573}{2565 } & -\frac{13063}{5130 } & -\frac{1333}{3420 } & \frac{1663}{1140 } & -\frac{6337}{1710 } & -\frac{8501}{2052 } & -\frac{16613}{5130 } & \frac{2041}{2052 } & \frac{2455}{2052 } & -\frac{1081}{10260 } & -\frac{61}{135 } & -\frac{5467}{3420 } & -\frac{761}{1140 } \\[2 mm ] -\frac{22037}{2280 } & \frac{119207}{6840 } & -\frac{3841}{20520 } & \frac{66763}{20520 } & -\frac{122123}{20520 } & \frac{26969}{20520 } & \frac{63959}{6840 } & \frac{8}{19 } & -\frac{4843}{1368 } & -\frac{68621}{20520 } & \frac{24209}{4104 } & -\frac{78761}{20520 } & -\frac{2923}{2565 } & -\frac{8251}{2565 } & -\frac{139}{4104 } & -\frac{2169}{760 } & \frac{6557}{1368 } \\[2 mm ] \frac{1829}{1710 } & \frac{1931}{380 } & -\frac{2605}{2052 } & -\frac{8411}{5130 } & -\frac{30991}{10260 } & \frac{734}{513 } & \frac{674}{855 } & -\frac{2713}{1140 } & -\frac{139}{684 } & \frac{6062}{2565 } & \frac{40547}{10260 } & -\frac{1303}{2565 } & \frac{35}{54 } & \frac{1259}{10260 } & \frac{3188}{2565 } & -\frac{115}{342 } & \frac{1193}{1140 } \\[2 mm ] \frac{1931}{380 } & \frac{1829}{1710 } & \frac{734}{513 } & -\frac{30991}{10260 } & -\frac{8411}{5130 } & -\frac{2605}{2052 } & -\frac{139}{684 } & -\frac{2713}{1140 } & \frac{674}{855 } & \frac{40547}{10260 } & \frac{6062}{2565 } & \frac{3188}{2565 } & \frac{1259}{10260 } & \frac{35}{54 } & -\frac{1303}{2565 } & \frac{1193}{1140 } & -\frac{115}{342 } \\[2 mm ] \frac{119207}{6840 } & -\frac{22037}{2280 } & \frac{26969}{20520 } & -\frac{122123}{20520 } & \frac{66763}{20520 } & -\frac{3841}{20520 } & -\frac{4843}{1368 } & \frac{8}{19 } & \frac{63959}{6840 } & \frac{24209}{4104 } & -\frac{68621}{20520 } & -\frac{139}{4104 } & -\frac{8251}{2565 } & -\frac{2923}{2565 } & -\frac{78761}{20520 } & \frac{6557}{1368 } & -\frac{2169}{760 } \\[2 mm ] -\frac{2081}{570 } & \frac{125431}{6840 } & -\frac{31475}{4104 } & -\frac{3703}{1026 } & -\frac{121739}{20520 } & \frac{19553}{2052 } & \frac{13529}{2280 } & -\frac{23827}{6840 } & \frac{619}{1710 } & \frac{4469}{10260 } & \frac{193087}{20520 } & -\frac{17584}{2565 } & -\frac{79721}{20520 } & \frac{9343}{20520 } & \frac{106477}{20520 } & -\frac{21281}{6840 } & \frac{1487}{285 } \\[2 mm ] \frac{125431}{6840 } & -\frac{2081}{570 } & \frac{19553}{2052 } & -\frac{121739}{20520 } & -\frac{3703}{1026 } & -\frac{31475}{4104 } & \frac{619}{1710 } & -\frac{23827}{6840 } & \frac{13529}{2280 } & \frac{193087}{20520 } & \frac{4469}{10260 } & \frac{106477}{20520 } & \frac{9343}{20520 } & -\frac{79721}{20520 } & -\frac{17584}{2565 } & \frac{1487}{285 } & -\frac{21281}{6840 } \end{array } \right)\end{aligned}\ ] ] @xmath837 \frac{22307}{4104 } & \frac{54187}{4104 } & \frac{8251}{4104 } & -\frac{15943}{4104 } & -\frac{10201}{1368 } & \frac{131}{456 } & \frac{8869}{1368 } & -\frac{2399}{684 } & \frac{3299}{1368 } & \frac{18013}{4104 } & \frac{27103}{4104 } & -\frac{6887}{4104 } & -\frac{4225}{2052 } & -\frac{1870}{513 } & -\frac{7727}{4104 } & \frac{887}{1368 } & \frac{1099}{456 } \\[2 mm ] -\frac{1319}{2052 } & -\frac{41957}{4104 } & \frac{14623}{4104 } & \frac{4501}{2052 } & \frac{4577}{1368 } & -\frac{863}{171 } & -\frac{677}{152 } & \frac{1525}{1368 } & -\frac{42}{19 } & -\frac{391}{513 } & -\frac{19439}{4104 } & \frac{4303}{1026 } & \frac{12691}{4104 } & \frac{4825}{4104 } & -\frac{5069}{4104 } & \frac{887}{1368 } & -\frac{523}{171 } \\[2 mm ] \frac{34169}{4104 } & -\frac{43913}{4104 } & \frac{10537}{4104 } & -\frac{2677}{4104 } & \frac{4643}{1368 } & -\frac{1283}{456 } & -\frac{6971}{1368 } & \frac{409}{684 } & \frac{209}{72 } & \frac{14935}{4104 } & -\frac{14153}{4104 } & \frac{14425}{4104 } & \frac{4181}{2052 } & \frac{1547}{2052 } & -\frac{7043}{4104 } & \frac{3911}{1368 } & -\frac{1429}{456 } \\[2 mm ] -\frac{17155}{4104 } & \frac{65953}{4104 } & \frac{133}{216 } & \frac{1415}{4104 } & -\frac{9643}{1368 } & \frac{1475}{1368 } & \frac{3469}{456 } & -\frac{1573}{684 } & -\frac{339}{152 } & \frac{1795}{4104 } & \frac{30277}{4104 } & -\frac{7655}{4104 } & -\frac{299}{1026 } & -\frac{2441}{1026 } & \frac{1897}{4104 } & -\frac{2347}{1368 } & \frac{5005}{1368 } \\[2 mm ] -\frac{16895}{2052 } & -\frac{29369}{4104 } & \frac{5893}{4104 } & \frac{475}{108 } & \frac{4133}{1368 } & -\frac{773}{228 } & -\frac{4919}{1368 } & \frac{1643}{1368 } & -\frac{980}{171 } & -\frac{7639}{2052 } & -\frac{15629}{4104 } & \frac{7631}{2052 } & \frac{15877}{4104 } & \frac{9871}{4104 } & \frac{5251}{4104 } & -\frac{2347}{1368 } & -\frac{443}{228 } \\[2 mm ] -\frac{11509}{5130 } & -\frac{1003}{10260 } & -\frac{13723}{10260 } & \frac{4391}{5130 } & \frac{167}{180 } & \frac{1411}{1140 } & \frac{401}{285 } & \frac{3181}{1710 } & \frac{1061}{1710 } & -\frac{7006}{2565 } & -\frac{17719}{10260 } & -\frac{20899}{10260 } & -\frac{3826}{2565 } & -\frac{8209}{10260 } & -\frac{5029}{10260 } & -\frac{106}{171 } & \frac{53}{171 } \\[2 mm ] \frac{2515}{1026 } & -\frac{929}{270 } & -\frac{7168}{2565 } & -\frac{3935}{1026 } & \frac{365}{342 } & \frac{679}{380 } & -\frac{599}{228 } & -\frac{1189}{855 } & \frac{5017}{3420 } & \frac{3863}{10260 } & -\frac{2255}{2052 } & -\frac{3193}{2565 } & -\frac{1199}{513 } & \frac{13877}{10260 } & \frac{965}{1026 } & -\frac{191}{855 } & -\frac{3107}{3420 } \\[2 mm ] -\frac{5081}{1026 } & \frac{889}{513 } & \frac{991}{5130 } & \frac{3661}{1026 } & \frac{112}{171 } & \frac{181}{1140 } & \frac{1657}{1140 } & \frac{1309}{684 } & -\frac{1571}{855 } & -\frac{7141}{2565 } & -\frac{61}{540 } & -\frac{977}{5130 } & \frac{8447}{10260 } & -\frac{1963}{10260 } & \frac{1139}{2565 } & -\frac{191}{855 } & \frac{3871}{3420 } \\[2 mm ] -\frac{163009}{20520 } & -\frac{56513}{5130 } & -\frac{2971}{2565 } & \frac{66761}{20520 } & \frac{5249}{1140 } & -\frac{11933}{6840 } & -\frac{2009}{380 } & \frac{14827}{6840 } & -\frac{1481}{360 } & -\frac{118553}{20520 } & -\frac{14051}{2052 } & \frac{32347}{20520 } & \frac{39097}{20520 } & \frac{9701}{4104 } & \frac{2269}{2052 } & -\frac{1658}{855 } & -\frac{2169}{760 } \\[2 mm ] -\frac{42173}{10260 } & -\frac{8518}{2565 } & -\frac{533}{270 } & \frac{21967}{10260 } & \frac{784}{285 } & \frac{1021}{1140 } & -\frac{10439}{3420 } & \frac{107}{228 } & -\frac{466}{171 } & -\frac{12499}{10260 } & -\frac{25}{1026 } & \frac{2999}{2565 } & \frac{20039}{10260 } & \frac{7787}{2565 } & \frac{2981}{1026 } & -\frac{2429}{3420 } & -\frac{115}{342 } \\[2 mm ] -\frac{3617}{2565 } & \frac{11347}{5130 } & -\frac{5068}{2565 } & -\frac{151}{270 } & -\frac{43}{285 } & \frac{4129}{1710 } & -\frac{109}{380 } & -\frac{2231}{1710 } & -\frac{3253}{3420 } & \frac{1093}{2052 } & \frac{22733}{10260 } & \frac{160}{513 } & \frac{415}{2052 } & \frac{19073}{10260 } & \frac{13849}{5130 } & -\frac{2429}{3420 } & \frac{1193}{1140 } \\[2 mm ] -\frac{132199}{20520 } & \frac{92413}{5130 } & \frac{205}{513 } & \frac{35951}{20520 } & -\frac{7453}{1140 } & \frac{3991}{2280 } & \frac{6845}{684 } & -\frac{179}{760 } & -\frac{11701}{6840 } & -\frac{40487}{20520 } & \frac{65351}{10260 } & -\frac{4313}{1080 } & -\frac{2051}{1080 } & -\frac{15133}{4104 } & -\frac{5029}{10260 } & -\frac{1658}{855 } & \frac{6557}{1368 } \\[2 mm ] \frac{33437}{5130 } & -\frac{23953}{10260 } & \frac{2863}{2052 } & -\frac{3284}{2565 } & \frac{509}{1140 } & -\frac{523}{855 } & \frac{47}{380 } & \frac{3233}{3420 } & \frac{688}{171 } & \frac{9307}{5130 } & -\frac{14779}{10260 } & -\frac{2063}{5130 } & -\frac{2609}{2052 } & -\frac{4882}{2565 } & -\frac{6526}{2565 } & \frac{775}{342 } & -\frac{761}{1140 } \\[2 mm ] \frac{51557}{5130 } & -\frac{51289}{10260 } & \frac{3443}{10260 } & -\frac{12344}{2565 } & \frac{75}{76 } & -\frac{21}{95 } & -\frac{1189}{684 } & -\frac{29}{57 } & \frac{18733}{3420 } & \frac{6691}{2052 } & -\frac{20527}{10260 } & -\frac{793}{1026 } & -\frac{13943}{5130 } & -\frac{9703}{10260 } & -\frac{26377}{10260 } & \frac{775}{342 } & -\frac{5467}{3420 } \\[2 mm ] -\frac{21365}{4104 } & \frac{44899}{20520 } & -\frac{7219}{4104 } & \frac{403}{216 } & -\frac{193}{1368 } & \frac{9313}{6840 } & \frac{13183}{6840 } & \frac{6211}{3420 } & -\frac{4751}{6840 } & -\frac{84623}{20520 } & -\frac{22907}{20520 } & -\frac{54053}{20520 } & -\frac{7481}{5130 } & -\frac{12139}{10260 } & -\frac{4067}{20520 } & -\frac{7129}{6840 } & \frac{7027}{6840 } \\[2 mm ] -\frac{79909}{20520 } & -\frac{30997}{4104 } & -\frac{8299}{4104 } & \frac{78761}{20520 } & \frac{42259}{6840 } & \frac{2113}{1368 } & -\frac{29269}{6840 } & \frac{1859}{570 } & -\frac{2611}{2280 } & -\frac{52403}{20520 } & -\frac{77327}{20520 } & \frac{19687}{20520 } & \frac{8377}{5130 } & \frac{32657}{10260 } & \frac{48463}{20520 } & -\frac{637}{6840 } & -\frac{69}{760 } \end{array } \right ) \\[3 mm ] a_4 & = & \left ( \begin{array}{rrrrrrrrrrrrrrrrr } \frac{1453}{152 } & -\frac{11939}{1368 } & \frac{307}{152 } & -\frac{1985}{1368 } & \frac{439}{152 } & -\frac{2507}{1368 } & -\frac{6071}{1368 } & \frac{73}{342 } & \frac{5017}{1368 } & \frac{6221}{1368 } & -\frac{2933}{1368 } & \frac{3911}{1368 } & \frac{989}{684 } & \frac{517}{684 } & -\frac{2003}{1368 } & \frac{463}{152 } & -\frac{385}{152 } \\[2 mm ] -\frac{11939}{1368 } & \frac{1453}{152 } & -\frac{2507}{1368 } & \frac{439}{152 } & -\frac{1985}{1368 } & \frac{307}{152 } & \frac{5017}{1368 } & \frac{73}{342 } & -\frac{6071}{1368 } & -\frac{2933}{1368 } & \frac{6221}{1368 } & -\frac{2003}{1368 } & \frac{517}{684 } & \frac{989}{684 } & \frac{3911}{1368 } & -\frac{385}{152 } & \frac{463}{152 } \\[2 mm ] -\frac{743}{76 } & -\frac{5315}{1368 } & -\frac{21}{8 } & \frac{2693}{684 } & \frac{385}{152 } & \frac{1091}{684 } & -\frac{4721}{1368 } & \frac{859}{1368 } & -\frac{3319}{684 } & -\frac{737}{171 } & -\frac{2705}{1368 } & \frac{1027}{684 } & \frac{3997}{1368 } & \frac{4613}{1368 } & \frac{6145}{1368 } & -\frac{385}{152 } & -\frac{39}{76 } \\[2 mm ] \frac{9425}{684 } & -\frac{5}{152 } & \frac{1921}{1368 } & -\frac{431}{76 } & -\frac{95}{72 } & \frac{27}{76 } & -\frac{545}{1368 } & -\frac{3317}{1368 } & \frac{227}{36 } & \frac{1289}{171 } & \frac{3209}{1368 } & \frac{7}{18 } & -\frac{2113}{1368 } & -\frac{1301}{1368 } & -\frac{2431}{1368 } & \frac{463}{152 } & -\frac{39}{76 } \\[2 mm ] -\frac{5}{152 } & \frac{9425}{684 } & \frac{27}{76 } & -\frac{95}{72 } & -\frac{431}{76 } & \frac{1921}{1368 } & \frac{227}{36 } & -\frac{3317}{1368 } & -\frac{545}{1368 } & \frac{3209}{1368 } & \frac{1289}{171 } & -\frac{2431}{1368 } & -\frac{1301}{1368 } & -\frac{2113}{1368 } & \frac{7}{18 } & -\frac{39}{76 } & \frac{463}{152 } \\[2 mm ] -\frac{5315}{1368 } & -\frac{743}{76 } & \frac{1091}{684 } & \frac{385}{152 } & \frac{2693}{684 } & -\frac{21}{8 } & -\frac{3319}{684 } & \frac{859}{1368 } & -\frac{4721}{1368 } & -\frac{2705}{1368 } & -\frac{737}{171 } & \frac{6145}{1368 } & \frac{4613}{1368 } & \frac{3997}{1368 } & \frac{1027}{684 } & -\frac{39}{76 } & -\frac{385}{152 } \\[2 mm ] \frac{59581}{3420 } & -\frac{5923}{684 } & \frac{2881}{3420 } & -\frac{21659}{3420 } & \frac{1937}{684 } & \frac{1561}{3420 } & -\frac{5729}{1710 } & -\frac{143}{1710 } & \frac{3221}{342 } & \frac{469}{76 } & -\frac{1031}{380 } & -\frac{77}{228 } & -\frac{3797}{1140 } & -\frac{241}{228 } & -\frac{3913}{1140 } & \frac{1354}{285 } & -\frac{677}{285 } \\[2 mm ] -\frac{6187}{684 } & -\frac{6187}{684 } & -\frac{5411}{3420 } & \frac{2201}{684 } & \frac{2201}{684 } & -\frac{5411}{3420 } & -\frac{8267}{1710 } & \frac{479}{342 } & -\frac{8267}{1710 } & -\frac{2207}{380 } & -\frac{2207}{380 } & \frac{1619}{1140 } & \frac{2323}{1140 } & \frac{2323}{1140 } & \frac{1619}{1140 } & -\frac{677}{285 } & -\frac{677}{285 } \\[2 mm ] -\frac{5923}{684 } & \frac{59581}{3420 } & \frac{1561}{3420 } & \frac{1937}{684 } & -\frac{21659}{3420 } & \frac{2881}{3420 } & \frac{3221}{342 } & -\frac{143}{1710 } & -\frac{5729}{1710 } & -\frac{1031}{380 } & \frac{469}{76 } & -\frac{3913}{1140 } & -\frac{241}{228 } & -\frac{3797}{1140 } & -\frac{77}{228 } & -\frac{677}{285 } & \frac{1354}{285 } \\[2 mm ] \frac{37273}{6840 } & -\frac{1471}{380 } & -\frac{9607}{3420 } & -\frac{11699}{2280 } & \frac{56}{45 } & \frac{6017}{2280 } & -\frac{889}{342 } & -\frac{2113}{1368 } & \frac{1237}{360 } & \frac{1931}{1368 } & -\frac{4187}{3420 } & -\frac{115}{72 } & -\frac{21517}{6840 } & \frac{7879}{6840 } & \frac{1289}{1710 } & \frac{26}{57 } & -\frac{2569}{2280 } \\[2 mm ] -\frac{1471}{380 } & \frac{37273}{6840 } & \frac{6017}{2280 } & \frac{56}{45 } & -\frac{11699}{2280 } & -\frac{9607}{3420 } & \frac{1237}{360 } & -\frac{2113}{1368 } & -\frac{889}{342 } & -\frac{4187}{3420 } & \frac{1931}{1368 } & \frac{1289}{1710 } & \frac{7879}{6840 } & -\frac{21517}{6840 } & -\frac{115}{72 } & -\frac{2569}{2280 } & \frac{26}{57 } \\[2 mm ] -\frac{30023}{6840 } & \frac{12133}{2280 } & \frac{24793}{6840 } & \frac{4019}{2280 } & -\frac{23801}{6840 } & -\frac{187}{40 } & \frac{26437}{6840 } & -\frac{1183}{3420 } & -\frac{25979}{6840 } & -\frac{1507}{1368 } & \frac{9743}{6840 } & \frac{1307}{1368 } & \frac{176}{171 } & -\frac{1939}{855 } & -\frac{18397}{6840 } & -\frac{2569}{2280 } & \frac{1529}{2280 } \\[2 mm ] -\frac{6499}{2280 } & -\frac{433}{3420 } & \frac{145}{228 } & \frac{21673}{6840 } & \frac{187}{95 } & \frac{161}{1368 } & \frac{851}{855 } & \frac{17263}{6840 } & -\frac{865}{1368 } & -\frac{15277}{6840 } & -\frac{3169}{3420 } & -\frac{667}{6840 } & \frac{683}{1368 } & \frac{569}{6840 } & \frac{43}{1710 } & \frac{26}{57 } & \frac{1529}{2280 } \\[2 mm ] -\frac{433}{3420 } & -\frac{6499}{2280 } & \frac{161}{1368 } & \frac{187}{95 } & \frac{21673}{6840 } & \frac{145}{228 } & -\frac{865}{1368 } & \frac{17263}{6840 } & \frac{851}{855 } & -\frac{3169}{3420 } & -\frac{15277}{6840 } & \frac{43}{1710 } & \frac{569}{6840 } & \frac{683}{1368 } & -\frac{667}{6840 } & \frac{1529}{2280 } & \frac{26}{57 } \\[2 mm ] \frac{12133}{2280 } & -\frac{30023}{6840 } & -\frac{187}{40 } & -\frac{23801}{6840 } & \frac{4019}{2280 } & \frac{24793}{6840 } & -\frac{25979}{6840 } & -\frac{1183}{3420 } & \frac{26437}{6840 } & \frac{9743}{6840 } & -\frac{1507}{1368 } & -\frac{18397}{6840 } & -\frac{1939}{855 } & \frac{176}{171 } & \frac{1307}{1368 } & \frac{1529}{2280 } & -\frac{2569}{2280 } \\[2 mm ] -\frac{99}{38 } & \frac{32563}{6840 } & -\frac{511}{152 } & -\frac{200}{171 } & -\frac{353}{760 } & \frac{59}{18 } & \frac{16873}{6840 } & \frac{4117}{6840 } & \frac{1997}{3420 } & -\frac{7913}{3420 } & \frac{6241}{6840 } & -\frac{6919}{1710 } & -\frac{23003}{6840 } & -\frac{2851}{6840 } & \frac{4291}{6840 } & -\frac{2803}{2280 } & \frac{1091}{570 } \\[2 mm ] \frac{32563}{6840 } & -\frac{99}{38 } & \frac{59}{18 } & -\frac{353}{760 } & -\frac{200}{171 } & -\frac{511}{152 } & \frac{1997}{3420 } & \frac{4117}{6840 } & \frac{16873}{6840 } & \frac{6241}{6840 } & -\frac{7913}{3420 } & \frac{4291}{6840 } & -\frac{2851}{6840 } & -\frac{23003}{6840 } & -\frac{6919}{1710 } & \frac{1091}{570 } & -\frac{2803}{2280 } \end{array } \right),\end{aligned}\ ] ] the four matrices @xmath838 : @xmath839 -\frac{27385}{7569 } & \frac{935519}{121104 } & \frac{47249}{121104 } & \frac{40939}{60552 } & -\frac{46215}{13456 } & \frac{6745}{30276 } & \frac{150395}{40368 } & -\frac{122555}{121104 } & -\frac{56185}{30276 } & -\frac{25345}{30276 } & -\frac{5675}{7569 } \\[1.9 mm ] \frac{2425}{841 } & -\frac{200725}{13456 } & -\frac{69251}{121104 } & -\frac{49109}{60552 } & \frac{575405}{121104 } & -\frac{22085}{30276 } & -\frac{967175}{121104 } & \frac{1945}{13456 } & \frac{9595}{7569 } & -\frac{2165}{7569 } & \frac{22400}{7569 } \\[1.9 mm ] -\frac{111245}{10092 } & -\frac{36485}{40368 } & -\frac{159509}{121104 } & \frac{246115}{60552 } & \frac{183455}{121104 } & -\frac{950}{7569 } & -\frac{118385}{121104 } & \frac{51305}{40368 } & -\frac{169535}{30276 } & -\frac{162035}{30276 } & \frac{19955}{30276 } \\[1.9 mm ] -\frac{825575}{30276 } & \frac{593215}{121104 } & -\frac{53651}{13456 } & \frac{659713}{60552 } & \frac{252649}{121104 } & \frac{9925}{7569 } & \frac{144065}{121104 } & \frac{430865}{121104 } & -\frac{33280}{2523 } & -\frac{30625}{2523 } & -\frac{2305}{5046 } \\[1.9 mm ] -\frac{111785}{30276 } & -\frac{616655}{121104 } & -\frac{106391}{40368 } & \frac{43459}{60552 } & \frac{245095}{121104 } & \frac{47515}{30276 } & -\frac{421105}{121104 } & \frac{31715}{121104 } & -\frac{4355}{3364 } & -\frac{8615}{3364 } & \frac{305}{3364 } \\[1.9 mm ] -\frac{191761}{6728 } & \frac{76347}{26912 } & -\frac{718467}{134560 } & \frac{841397}{67280 } & \frac{120045}{26912 } & \frac{3143}{1682 } & \frac{13213}{26912 } & \frac{153665}{26912 } & -\frac{90001}{6728 } & -\frac{89053}{6728 } & -\frac{1307}{841 } \\[1.9 mm ] \frac{218579}{6728 } & \frac{47955}{26912 } & \frac{2017423}{403680 } & -\frac{2658161}{201840 } & -\frac{424769}{80736 } & -\frac{2486}{2523 } & \frac{217223}{80736 } & -\frac{124463}{26912 } & \frac{321065}{20184 } & \frac{307733}{20184 } & -\frac{2953}{2523 } \\[1.9 mm ] -\frac{738031}{20184 } & \frac{373229}{80736 } & -\frac{1481557}{403680 } & \frac{1203897}{67280 } & \frac{342619}{80736 } & -\frac{163}{1682 } & \frac{121883}{80736 } & \frac{556927}{80736 } & -\frac{361817}{20184 } & -\frac{326087}{20184 } & \frac{4297}{10092 } \\[1.9 mm ] \frac{1401}{29 } & -\frac{3977}{232 } & \frac{21023}{3480 } & -\frac{32659}{1740 } & \frac{1055}{696 } & -\frac{359}{174 } & -\frac{4775}{696 } & -\frac{1145}{232 } & \frac{2039}{87 } & \frac{7421}{348 } & \frac{205}{87 } \\[1.9 mm ] -\frac{4781}{174 } & \frac{13555}{696 } & -\frac{12787}{3480 } & \frac{19261}{1740 } & -\frac{893}{232 } & \frac{223}{87 } & \frac{2075}{232 } & \frac{1985}{696 } & -\frac{1111}{87 } & -\frac{949}{87 } & -\frac{1313}{348 } \end{array } \right ) \\[2 mm ] & & \hskip -7 mm b_2 \!=\ ! \left ( \begin{array}{rrrrrrrrrrr } \frac{1650899}{121104 } & -\frac{940283}{121104 } & \frac{33895}{121104 } & -\frac{499045}{121104 } & \frac{293725}{121104 } & \frac{60065}{121104 } & -\frac{465355}{121104 } & -\frac{5}{696 } & \frac{842065}{121104 } & \frac{220075}{40368 } & \frac{18395}{40368 } \\[1.9 mm ] -\frac{186367}{121104 } & \frac{896983}{121104 } & -\frac{231455}{121104 } & -\frac{35335}{121104 } & -\frac{169985}{121104 } & \frac{325415}{121104 } & \frac{384635}{121104 } & -\frac{5}{696 } & -\frac{7925}{121104 } & \frac{5185}{40368 } & -\frac{100795}{40368 } \\[1.9 mm ] \frac{383841}{13456 } & -\frac{1294319}{40368 } & \frac{462383}{40368 } & -\frac{272725}{40368 } & \frac{344065}{40368 } & -\frac{154785}{13456 } & -\frac{165345}{13456 } & \frac{235}{696 } & \frac{413965}{40368 } & \frac{420985}{40368 } & \frac{426805}{40368 } \\[1.9 mm ] -\frac{55715}{40368 } & -\frac{103615}{13456 } & -\frac{113975}{40368 } & \frac{17775}{13456 } & \frac{54145}{13456 } & \frac{57715}{40368 } & -\frac{184385}{40368 } & \frac{1075}{696 } & -\frac{695}{13456 } & -\frac{69905}{40368 } & -\frac{485}{40368 } \\[1.9 mm ] -\frac{1918919}{121104 } & \frac{819239}{121104 } & -\frac{908875}{121104 } & \frac{547705}{121104 } & \frac{99575}{121104 } & \frac{740095}{121104 } & \frac{149515}{121104 } & \frac{1075}{696 } & -\frac{708925}{121104 } & -\frac{94545}{13456 } & -\frac{171875}{40368 } \\[1.9 mm ] \frac{2091859}{121104 } & -\frac{2520247}{121104 } & \frac{686915}{121104 } & -\frac{465245}{121104 } & \frac{679265}{121104 } & -\frac{692831}{121104 } & -\frac{1055135}{121104 } & \frac{235}{696 } & \frac{808925}{121104 } & \frac{81845}{13456 } & \frac{248455}{40368 } \\[1.9 mm ] -\frac{4879959}{134560 } & \frac{3060847}{134560 } & -\frac{366567}{26912 } & \frac{263909}{26912 } & -\frac{130045}{26912 } & \frac{314715}{26912 } & \frac{185503}{26912 } & \frac{287}{464 } & -\frac{375917}{26912 } & -\frac{409385}{26912 } & -\frac{235637}{26912 } \\[1.9 mm ] \frac{1103137}{134560 } & \frac{1160719}{134560 } & \frac{486971}{80736 } & -\frac{253961}{80736 } & -\frac{454103}{80736 } & -\frac{258103}{80736 } & \frac{461357}{80736 } & -\frac{1085}{464 } & \frac{219041}{80736 } & \frac{436517}{80736 } & \frac{110665}{80736 } \\[1.9 mm ] -\frac{27188873}{403680 } & \frac{21731537}{403680 } & -\frac{1852889}{80736 } & \frac{1366043}{80736 } & -\frac{964451}{80736 } & \frac{1697333}{80736 } & \frac{1588793}{80736 } & \frac{287}{464 } & -\frac{2160035}{80736 } & -\frac{711797}{26912 } & -\frac{466593}{26912 } \\[1.9 mm ] \frac{37341}{1160 } & -\frac{37341}{1160 } & \frac{4121}{232 } & -\frac{1487}{232 } & \frac{1487}{232 } & -\frac{4121}{232 } & -\frac{2411}{232 } & 0 & \frac{2411}{232 } & \frac{2851}{232 } & \frac{3075}{232 } \\[1.9 mm ] -\frac{178027}{3480 } & \frac{178027}{3480 } & -\frac{13987}{696 } & \frac{8521}{696 } & -\frac{8521}{696 } & \frac{13987}{696 } & \frac{13207}{696 } & 0 & -\frac{13207}{696 } & -\frac{4417}{232 } & -\frac{3887}{232 } \end{array } \right ) \\[2 mm ] & & \hskip -7 mm b_3 \!=\ ! \left ( \begin{array}{rrrrrrrrrrr } \frac{661279}{121104 } & \frac{679825}{30276 } & \frac{815}{7569 } & -\frac{192245}{40368 } & -\frac{555283}{60552 } & \frac{421333}{121104 } & \frac{338935}{30276 } & -\frac{441205}{121104 } & \frac{147145}{40368 } & \frac{650225}{121104 } & -\frac{208325}{40368 } \\[1.9 mm ] \frac{749845}{121104 } & \frac{271835}{15138 } & \frac{7340}{7569 } & -\frac{192535}{40368 } & -\frac{528139}{60552 } & \frac{227323}{121104 } & \frac{138365}{15138 } & -\frac{426415}{121104 } & \frac{49725}{13456 } & \frac{558875}{121104 } & -\frac{163375}{40368 } \\[1.9 mm ] \frac{467615}{40368 } & -\frac{84875}{5046 } & \frac{29120}{7569 } & -\frac{251765}{121104 } & \frac{381797}{60552 } & -\frac{553669}{121104 } & -\frac{211075}{30276 } & \frac{66955}{40368 } & \frac{559145}{121104 } & \frac{436655}{121104 } & \frac{440815}{121104 } \\[1.9 mm ] -\frac{148535}{40368 } & \frac{29950}{2523 } & \frac{12065}{30276 } & \frac{203869}{121104 } & -\frac{248029}{60552 } & \frac{94085}{121104 } & \frac{46190}{7569 } & -\frac{16795}{40368 } & -\frac{195685}{121104 } & -\frac{15475}{121104 } & -\frac{234455}{121104 } \\[1.9 mm ] -\frac{1148275}{121104 } & \frac{212810}{7569 } & -\frac{31435}{30276 } & \frac{134675}{121104 } & -\frac{661627}{60552 } & \frac{139145}{40368 } & \frac{46065}{3364 } & -\frac{327335}{121104 } & -\frac{458135}{121104 } & -\frac{165985}{121104 } & -\frac{661045}{121104 } \\[1.9 mm ] \frac{212975}{121104 } & -\frac{310165}{30276 } & \frac{11720}{7569 } & \frac{78545}{121104 } & \frac{289229}{60552 } & -\frac{101249}{40368 } & -\frac{11125}{2523 } & \frac{186655}{121104 } & \frac{13075}{121104 } & -\frac{955}{121104 } & \frac{308285}{121104 } \\[1.9 mm ] -\frac{779423}{26912 } & \frac{24497}{3364 } & -\frac{30581}{6728 } & \frac{300223}{26912 } & \frac{8593}{67280 } & \frac{303419}{134560 } & \frac{15529}{6728 } & \frac{62849}{26912 } & -\frac{378439}{26912 } & -\frac{333781}{26912 } & -\frac{6733}{26912 } \\[1.9 mm ] \frac{285081}{26912 } & -\frac{81551}{3364 } & -\frac{3631}{20184 } & -\frac{287819}{80736 } & \frac{1697971}{201840 } & -\frac{734231}{403680 } & -\frac{247579}{20184 } & \frac{30097}{26912 } & \frac{393923}{80736 } & \frac{151169}{80736 } & \frac{312241}{80736 } \\[1.9 mm ] -\frac{2482457}{80736 } & \frac{154865}{10092 } & -\frac{17357}{6728 } & \frac{918185}{80736 } & -\frac{353907}{67280 } & \frac{236413}{403680 } & \frac{138401}{20184 } & \frac{92615}{80736 } & -\frac{1217561}{80736 } & -\frac{359881}{26912 } & -\frac{111259}{80736 } \\[1.9 mm ] \frac{3977}{232 } & -\frac{1401}{29 } & \frac{359}{174 } & -\frac{1055}{696 } & \frac{32659}{1740 } & -\frac{21023}{3480 } & -\frac{2039}{87 } & \frac{1145}{232 } & \frac{4775}{696 } & \frac{1919}{696 } & \frac{6577}{696 } \\[1.9 mm ] -\frac{13555}{696 } & \frac{4781}{174 } & -\frac{223}{87 } & \frac{893}{232 } & -\frac{19261}{1740 } & \frac{12787}{3480 } & \frac{1111}{87 } & -\frac{1985}{696 } & -\frac{2075}{232 } & -\frac{4355}{696 } & -\frac{1013}{232 } \end{array } \right ) \\[2 mm ] & & \hskip -7 mm b_4 \!=\ ! \left ( \begin{array}{rrrrrrrrrrr } \frac{79195}{15138 } & -\frac{71365}{15138 } & \frac{8855}{7569 } & \frac{10381}{30276 } & \frac{13805}{30276 } & -\frac{38465}{30276 } & -\frac{1775}{841 } & \frac{35}{58 } & \frac{12535}{5046 } & \frac{47665}{30276 } & \frac{7315}{7569 } \\[1.9 mm ] -\frac{84595}{7569 } & \frac{88510}{7569 } & -\frac{92035}{30276 } & \frac{116695}{30276 } & -\frac{92509}{30276 } & \frac{44495}{15138 } & \frac{8195}{1682 } & \frac{35}{58 } & -\frac{11350}{2523 } & -\frac{31910}{7569 } & -\frac{99065}{30276 } \\[1.9 mm ] -\frac{14795}{5046 } & -\frac{57415}{5046 } & -\frac{43595}{10092 } & \frac{25265}{10092 } & \frac{44915}{10092 } & \frac{12851}{5046 } & -\frac{24885}{3364 } & \frac{95}{58 } & -\frac{1615}{10092 } & -\frac{5035}{1682 } & \frac{305}{841 } \\[1.9 mm ] \frac{10687}{1682 } & \frac{24185}{3364 } & \frac{7235}{10092 } & -\frac{31055}{10092 } & -\frac{8285}{1682 } & \frac{1525}{1682 } & \frac{12265}{3364 } & -\frac{65}{29 } & \frac{35705}{10092 } & \frac{17335}{5046 } & -\frac{17425}{10092 } \\[1.9 mm ] -\frac{183155}{30276 } & \frac{296593}{15138 } & -\frac{9085}{15138 } & \frac{9895}{15138 } & -\frac{262085}{30276 } & \frac{67325}{30276 } & \frac{31985}{3364 } & -\frac{65}{29 } & -\frac{23455}{10092 } & -\frac{32725}{30276 } & -\frac{57965}{15138 } \\[1.9 mm ] \frac{37685}{15138 } & -\frac{254315}{15138 } & -\frac{157}{15138 } & \frac{39545}{30276 } & \frac{170995}{30276 } & -\frac{53365}{30276 } & -\frac{28655}{3364 } & \frac{95}{58 } & \frac{9695}{10092 } & -\frac{13415}{15138 } & \frac{25220}{7569 } \\[1.9 mm ] \frac{49977}{6728 } & \frac{4762}{841 } & \frac{10645}{1682 } & -\frac{15721}{3364 } & -\frac{32851}{6728 } & -\frac{29733}{6728 } & \frac{33881}{6728 } & -\frac{773}{232 } & \frac{1409}{841 } & \frac{29543}{6728 } & \frac{13711}{6728 } \\[1.9 mm ] -\frac{6891}{6728 } & -\frac{59673}{3364 } & \frac{1263}{3364 } & \frac{5295}{3364 } & \frac{62751}{6728 } & -\frac{17577}{6728 } & -\frac{57239}{6728 } & \frac{707}{232 } & -\frac{2561}{1682 } & -\frac{12589}{6728 } & \frac{27207}{6728 } \\[1.9 mm ] -\frac{515677}{20184 } & \frac{97487}{2523 } & -\frac{3967}{5046 } & \frac{36125}{10092 } & -\frac{265129}{20184 } & \frac{54409}{20184 } & \frac{130567}{6728 } & -\frac{773}{232 } & -\frac{42707}{3364 } & -\frac{151259}{20184 } & -\frac{115235}{20184 } \\[1.9 mm ] \frac{2415}{116 } & -\frac{2415}{116 } & \frac{47}{116 } & -\frac{869}{116 } & \frac{869}{116 } & -\frac{47}{116 } & -\frac{1205}{116 } & 0 & \frac{1205}{116 } & \frac{455}{58 } & \frac{291}{116 } \\[1.9 mm ] -\frac{6433}{348 } & \frac{6433}{348 } & \frac{671}{348 } & \frac{1795}{348 } & -\frac{1795}{348 } & -\frac{671}{348 } & \frac{1205}{116 } & 0 & -\frac{1205}{116 } & -\frac{2237}{348 } & -\frac{95}{87 } \end{array } \right),\end{aligned}\ ] ] and the four matrices @xmath840 : @xmath841 0 & 0 & 0 & 0 & \frac{1}{4 } & 0 \\[1.9 mm ] 0 & 0 & 0 & 0 & 0 & \frac{1}{4 } \\[1.9 mm ] 0 & 0 & 0 & -\frac{1}{4 } & 0 & 0 \\[1.9 mm ] 0 & 1 & 0 & 0 & \frac{3}{4 } & 0 \\[1.9 mm ] 0 & 0 & 1 & 0 & 0 & \frac{3}{4 } \end{array } \right ) \\[2 mm ] c_2 & = & \left ( \begin{array}{rrrrrr } 0 & \frac{1}{4 } & 0 & \frac{1}{12 } & \frac{3}{16 } & -\frac{1}{12 } \\[1.9 mm ] 1 & \frac{3}{4 } & \frac{3}{4 } & -\frac{23}{12 } & -\frac{3}{16 } & \frac{11}{48 } \\[1.9 mm ] 0 & 0 & -\frac{1}{4 } & \frac{25}{12 } & 0 & -\frac{25}{48 } \\[1.9 mm ] 0 & 0 & 0 & \frac{5}{12 } & 0 & -\frac{1}{6 } \\[1.9 mm ] 0 & 0 & -1 & \frac{5}{3 } & 1 & -\frac{5}{12 } \\[1.9 mm ] 0 & 0 & 0 & -\frac{7}{3 } & 0 & \frac{1}{3 } \end{array } \right ) \\[2 mm ] c_3 & = & \left ( \begin{array}{rrrrrr } 0 & 0 & \frac{1}{4 } & \frac{1}{12 } & -\frac{1}{12 } & \frac{3}{16 } \\[1.9 mm ] 0 & -\frac{1}{4 } & 0 & \frac{25}{12 } & -\frac{25}{48 } & 0 \\[1.9 mm ] 1 & \frac{3}{4 } & \frac{3}{4 } & -\frac{23}{12 } & \frac{11}{48 } & -\frac{3}{16 } \\[1.9 mm ] 0 & 0 & 0 & \frac{5}{12 } & -\frac{1}{6 } & 0 \\[1.9 mm ] 0 & 0 & 0 & -\frac{7}{3 } & \frac{1}{3 } & 0 \\[1.9 mm ] 0 & -1 & 0 & \frac{5}{3 } & -\frac{5}{12 } & 1 \end{array } \right ) \\[2 mm ] c_4 & = & \left ( \begin{array}{rrrrrr } 0 & -\frac{1}{12 } & -\frac{1}{12 } & -\frac{1}{4 } & \frac{5}{48 } & \frac{5}{48 } \\[1.9 mm ] 0 & \frac{5}{3 } & -\frac{25}{12 } & 0 & \frac{5}{3 } & -\frac{103}{48 } \\[1.9 mm ] 0 & -\frac{25}{12 } & \frac{5}{3 } & 0 & -\frac{103}{48 } & \frac{5}{3 } \\[1.9 mm ] 1 & \frac{1}{3 } & \frac{1}{3 } & -\frac{1}{4 } & \frac{1}{3 } & \frac{1}{3 } \\[1.9 mm ] 0 & -\frac{5}{3 } & \frac{7}{3 } & 0 & -\frac{23}{12 } & \frac{25}{12 } \\[1.9 mm ] 0 & \frac{7}{3 } & -\frac{5}{3 } & 0 & \frac{25}{12 } & -\frac{23}{12 } \end{array } \right).\end{aligned}\ ] ] the research of the first author is supported by italian indam - gncs ( gruppo nazionale di calcolo scientifico ) .
part of this work has been done during the stay of v.p . at the university of laquila and gssi ( laquila ) .
many thanks to peter oswald ( bremen ) for providing us the local subdivision matrices of the generalized butterfly subdivision scheme .
|
we generalize the recent invariant polytope algorithm for computing the joint spectral radius and extend it to a wider class of matrix sets .
this , in particular , makes the algorithm applicable to sets of matrices that have finitely many spectrum maximizing products .
a criterion of convergence of the algorithm is proved . as an application
we solve two challenging computational open problems .
first we find the regularity of the butterfly subdivision scheme for various parameters @xmath0 . in the `` most regular '' case @xmath1
, we prove that the limit function has hlder exponent @xmath2 and its derivative is `` almost lipschitz '' with logarithmic factor @xmath2 .
second we compute the hlder exponent of daubechies wavelets of high order .
* keywords : * _ matrix , joint spectral radius , invariant polytope algorithm , dominant products , balancing , subdivision schemes , butterfly scheme , daubechies wavelets _
|
recently , `` thin '' subgroups of @xmath14 have attracted some attention in number theory . by `` thin '' we mean an _ infinite index _ subgroup @xmath15
whose dimension of the limit set @xmath16 satisfies @xmath17 . while the results of bourgain - gamburd - sarnak @xcite focus on the density of almost primes found among entries of the orbits of thin subgroups , the works of bourgain and kontorovich @xcite are concerned with the density of various subsets of @xmath18 obtained through the action of @xmath0 .
one of the key steps of the proofs involves reduction ( localization ) modulo @xmath19 where @xmath19 is a square - free integer , in particular one is led to consider `` congruence subgroups '' @xmath2 defined by @xmath20 a critical ingredient is the spectral theory of the _ infinite area _
hyperbolic surfaces @xmath21 where uniform estimates on the spectrum of the laplacian are often required .
let us recall some basic known facts about the laplacian on these objects .
let @xmath22 be the hyperbolic plane endowed with its standard metric of constant gaussian curvature @xmath23 .
let @xmath0 be a convex co - compact discrete subgroup of isometries acting on @xmath22 .
this means that @xmath0 admits a finite sided polygonal fundamental domain in @xmath22 , with infinite area .
we will require that @xmath0 has no _ elliptic _ elements different from the identity and that @xmath0 has no parabolic elements ( no cusps ) . under these assumptions ,
the quotient space @xmath24 is a hyperbolic surface with infinite area whose ends are given by hyperbolic funnels .
the limit set of @xmath0 is commonly defined by @xmath25 where @xmath26 is a given point and @xmath27 is the orbit of that point under the action of @xmath0 which by discreteness accumulates on the boundary @xmath28 . the limit set @xmath29 does not depend on the choice of @xmath30 and its hausdorff dimension is commonly denoted by @xmath16 .
it is the critical exponent of poincar series @xcite ( here @xmath31 denotes hyperbolic distance ) @xmath32 let @xmath33 be the hyperbolic laplacian on @xmath34 .
its spectrum on @xmath35 has been described completely by lax and phillips in @xcite .
the half line @xmath36 is the continuous spectrum and it contains no embedded eigenvalues .
the rest of the spectrum ( point spectrum ) is empty if @xmath37 , finite and starting at @xmath38 if @xmath39 .
the fact that the bottom of the spectrum is related to the dimension @xmath40 was first discovered by patterson @xcite for convex co - compact groups . by the preceding description of the spectrum ,
the resolvent @xmath41 is therefore well defined and analytic on the half - plane @xmath42 except at a possible finite set of poles corresponding to the finite point spectrum .
resonances _ are then defined as poles of the meromorphic continuation of @xmath43 to the whole complex plane .
the set of poles is denoted by @xmath44 .
this continuation is usually performed via the analytic fredholm theorem after the construction of an adequate parametrix .
the first result of this kind in the more general setting of asymptotically hyperbolic manifolds is due to mazzeo and melrose @xcite .
a more precise parametrix for surfaces was constructed by guillop and zworski in @xcite .
it should be mentioned at this point that in the infinite area case , resonances are spread all over the half plane @xmath45 , in sharp contrast with the finite area non - compact case where resonances are known to be confined in a strip . among the known results ( and conjectures ) on the density and location of resonances we mention the following two facts which are relevant for this paper .
let @xmath46 be the counting function defined by @xmath47 from the work of guillop and zworski , we know that there exists @xmath48 such that for all @xmath49 , we have @xmath50 on the other hand , let @xmath51 be defined by @xmath52 from the work of guillop - lin - zworski @xcite , we know that one can find @xmath53 such that for all @xmath54 , we have @xmath55 it is conjectured in jakobson - naud @xcite that for all @xmath12 and all @xmath56 large enough , @xmath57 , in other words that there exists an `` essential '' spectral gap up to @xmath58 which plays the role of the critical line in infinite volume . in this paper , motivated by number theoretic works , we will restrict ourselves to the interesting case when @xmath0 is a convex co - compact subgroup of @xmath1 , and
will assume throughout the paper that @xmath0 is _ non - elementary _ i.e. is not generated by a single hyperbolic element .
i should be noticed that since @xmath0 is a _ free _
group , there is no need to distinguish @xmath0 as a subgroup of @xmath1 or viewed as a matrix subgroup of @xmath14 . is a free subgroup of @xmath14 , @xmath59 and therefore the natural projection @xmath60 is injective when restricted to @xmath0 . ] as mentioned above , a natural question is to describe the resonances of `` congruence '' surfaces @xmath2 with respect to @xmath19 . for simplicity , we will restrict ourselves in this paper to the case when _
@xmath19 is a prime number_. from the work of gamburd @xcite , it is known that for all @xmath19 large enough , the projection map @xmath61 is a surjection
. therefore we have @xmath62=\vert sl_2({\mathbb{f}}_q ) \vert = q(q^2 - 1)\asymp q^3\ ] ] since all the subgroups @xmath2 have a finite index in @xmath0 , they have all the same dimension @xmath63 .
our first result is the following . for simplicity
, we denote the counting function @xmath64 by @xmath65 .
[ lowerb ] there exist constants @xmath66 and @xmath67 such that for all @xmath68 and all @xmath69 and @xmath70 , we have @xmath71 this result shows abundance of resonances in discs with slow radius growth as @xmath72 .
this lower bound is not surprising in view of the geometric bounds obtained by borthwick in @xcite .
notice that this bound is `` almost optimal '' in term of @xmath19 , indeed , in @xmath73 , we show that @xmath74 uniformly for all @xmath75
. it may be possible ( with some more geometric work ) to derive theorem [ lowerb ] directly from the lower bound in @xcite .
however , in this paper , we use a fairly different ( and more algebraic ) route which is justified by our next statement .
we denote by @xmath5 the resonance spectrum of the surface @xmath76 . in their work on almost primes
@xcite , bourgain - gamburd - sarnak ( and also gamburd @xcite ) obtained the following `` spectral gap '' result for the family @xmath2 : * if @xmath77 then there exists @xmath78 , independent of @xmath19 such that for all @xmath19 we have @xmath79 .
* moreover , if @xmath80 then @xmath81 $ ] is independent of @xmath19 ( notice that in this case there are only eigenvalues ) . *
if @xmath82 , then there exists @xmath78 , independent of @xmath19 such that for all @xmath19 we have @xmath83 these uniform spectral gaps are obtained thanks to the work of bourgain - gamburd @xcite on cayley graphs of @xmath84 which are proved to be expanders for all finitely generated , non elementary set of generators . while in the case @xmath77 they can use the full strength of lax - phillips spectral theory , in the case @xmath82 , one has to face genuine resonances and use some transfer operators techniques from @xcite . in view of the conjecture of @xcite ,
it is natural to expect the following _ `` uniform essential spectral gap property''_. for all @xmath85 , @xmath86 is finite and independent of @xmath19 .
this conjecture seems to be out of reach ( in the finite volume case , this is selberg s eigenvalue conjecture ) , but we can prove the following .
[ upperb ] assume that @xmath85 , then there exist @xmath87 , such that we have for all @xmath75 and @xmath19 large , @xmath88 @xmath89 where for @xmath90 , @xmath91 on @xmath92 $ ] , and @xmath93 is strictly convex and decreasing on @xmath94 $ ] .
this statement is a strengthening of the main result in @xcite , extended to all congruence subgroups .
not only we have a gain over the crude bound @xmath95 but simultaneously a gain over the @xmath96 bound , as long as we count resonances in @xmath97 . notice that it can be rephrased ( for fixed @xmath56 ) as a bound of the type @xmath98 where @xmath99 is the nielsen region in @xmath100 ( the convex core of the surface ) . in the particular case
@xmath39 it gives a density theorem for the number of eigenvalues of the laplacian @xmath101 in @xmath102 $ ] which to our knowledge does not follow from previous works : @xmath103 ) = o\left ( \mathrm{vol}(n_q)^{1-\epsilon_\gamma } \right ) \ ] ] for some @xmath104 depending only on @xmath0 .
this resonance behavior in the strip @xmath105 is in sharp contrast with theorem [ lowerb ] , which shows a drastically different behavior in the half - plane @xmath106 .
we point out that the functions @xmath107 have an `` explicit '' formula in terms of topological pressure .
we do not expect this formula to produce a uniform spectral gap for @xmath108 close to @xmath40 , although we use some of the ideas of @xcite in our proof .
let us describe the organization of the paper .
the main tool in both results is to use `` congruence '' transfer operators which were already defined in @xcite . in section
@xmath109 we recall how convex co - compact subgroups of @xmath1 can be viewed as schottky groups .
we then define the so - called congruence transfer operator on a suitable space of holomorphic functions and show that its fredholm determinant is related to the selberg s zeta function of @xmath110 .
an upper bound on the growth of this determinant is then proved , using some singular values estimates . combining this result with
the trace formula leads to theorem [ lowerb ] , see @xmath111 . to prove theorem [ upperb ] , a different kind of approach is required . in @xmath112
, we use some ( modifications of ) ideas from @xcite , where we estimate the number of resonances by using a regularized hilbert - schmidt determinant related to iterates of the transfer operator . a key part of the proof comes from a variant on the lower bound on the girth of cayley graphs of @xmath84 which is proved in @xcite and a separation mechanism based on the disconnected topology of the limit set . * acknowledgments . * this work was completed while fn was visiting crm at universit ' e de montral under a cnrs funding .
both authors are supported by anr `` blanc '' gerasic .
dj is also supported by nserc , fqrnt and peter redpath fellowship .
we use the notations of @xmath113 . let @xmath22 denote the poincar upper half - plane @xmath114 endowed with its standard metric of constant curvature @xmath23 @xmath115 the group of isometries of @xmath22 is @xmath116 through the action of @xmath117 matrices viewed as mbius transforms @xmath118 below we recall the definition of fuchsian schottky groups which will be used to define transfer operators .
a fuchsian schottky group is a free subgroup of @xmath116 built as follows .
let @xmath119 be @xmath120 euclidean _ open _ discs in @xmath121 orthogonal to the line @xmath122 .
we assume that for all @xmath123 , @xmath124 .
let @xmath125 be @xmath126 isometries such that for all @xmath127 , we have @xmath128 where @xmath129 stands for the riemann sphere .
let @xmath0 be the free group generated by @xmath130 for @xmath127 , then @xmath0 is a convex co - compact group , i.e. it is finitely generated and has no non - trivial parabolic element .
converse is true _ :
up to an isometry , all convex co - compact hyperbolic surfaces can be uniformized by a group as above , see @xcite . in the particular case when @xmath0 is a convex co - compact subgroup of @xmath1 , then by using the same argument as in @xcite , one can find a set of generators as above . for all
@xmath131 we set @xmath132 and define the bowen - series map @xmath133 by @xmath134 the maximal @xmath135-invariant compact subset of @xmath136 is precisely the limit set @xmath137 , and @xmath135 is uniformly expanding on @xmath137 .
the _ topological pressure _ @xmath138 , @xmath139 , is the thermodynamical quantity given by the limit ( the sums runs over @xmath140-periodic points of the map @xmath135 ) @xmath141 the fact that this limit exists and defines a _ real - analytic decreasing strictly convex _
function @xmath142 follows from classical thermodynamical formalism , see for example @xcite for a basic reference , see also @xcite , for a justification of strict convexity based on the fact that @xmath0 is non elementary .
moreover , it has a unique zero on the real line which is exactly the dimension @xmath143 , this is a celebrated result of bowen @xcite . in particular , we have @xmath144 iff @xmath145 , which is something to keep in mind in the rest of the paper , especially in the last section . in the sequel , we will denote by @xmath146 the group @xmath84 .
let @xmath0 be a convex co - compact subgroup of @xmath1 as above and let @xmath147 be a set of schottky generators as above . for simplicity
we denote by @xmath148 for @xmath127 . for each map
@xmath149 , we fix a @xmath117 matrix representation in @xmath14 also denoted by @xmath149 .
let @xmath150 be a @xmath121-valued function , then the _ congruence transfer operator _ applied to @xmath151 is defined for all @xmath152 @xmath153 where @xmath154 is the reduction mod @xmath19 . for obvious simplicity , we will omit @xmath155 in the notations for the right factor .
we need some additional notations . considering a finite sequence @xmath156 with @xmath157 we set @xmath158
we then denote by @xmath159 the set of admissible sequences of length @xmath140 by @xmath160 the set @xmath159 is simply the set of reduced words of length @xmath140 . for all @xmath161 , we define @xmath162 by @xmath163 if @xmath164 , then @xmath165 maps @xmath166 into @xmath167 . using this set of notations
, we have the formula for @xmath168 , @xmath169 we now have to specify the function space on which the transfer operators @xmath170 will act . let @xmath171 denote the vector space of ( complex - valued ) functions @xmath151 on @xmath172 such that for all @xmath173 , @xmath174 is holomorphic on each disc @xmath175 and such that the following norm @xmath176 is finite ( @xmath177 stands for lebesgue measure on @xmath121 ) .
this function space may be viewed as a ( vector valued ) variant of the classical bergman spaces , and is a natural hilbert space . since each branch
@xmath149 acts by contraction on @xmath178 , the transfer operators are compact , _ trace class operators_. this fact is well known and dates back to ruelle @xcite , see also bandtlow - jenkinson @xcite for an in - depth analysis of spectral properties of transfer operators on holomorphic function spaces . before we carry on our analysis , it is necessary to recall a few basic facts on transfer operators acting on @xmath171 .
we start by some distortion estimates .
* _ ( uniform hyperbolicity)_. one can find @xmath179 and @xmath180 such that for all @xmath181 and @xmath164 , for all @xmath168 we have @xmath182 * _ ( bounded distortion ) .
_ there exists @xmath183 such that for all @xmath181 and all @xmath164 , for all @xmath184 @xmath185 we refer the reader to @xcite for details on proofs and references
. we will also need the following fact which is proved in @xcite . for all @xmath186 in @xmath187 with @xmath188
, one can find @xmath66 such that for all @xmath140 large enough and @xmath189 , we have @xmath190 with these preliminaries in hand one can prove the following estimate .
[ normest1 ] there exist a constant @xmath179 , independent of @xmath19 such that for all @xmath191 , we have @xmath192 a straightforward and important consequence is that the _ spectral radius _ of @xmath193 is bounded by @xmath194 .
we postpone the proof of this proposition to the appendix and move on to the central subject of this @xmath195 .
we recall that the selberg zeta function @xmath196 is defined as the analytic continuation to @xmath121 of the infinite product : @xmath197 where @xmath198 is the set of prime closed geodesics on @xmath199 , and if @xmath200 , @xmath201 is the length .
our first observation is the following .
[ detformula ] using the above notation , we have for all @xmath202 and @xmath203 , @xmath204 _ proof_. we prove this identity by analytic continuation . first some trace computations are required .
let @xmath205 denote the dirac mass at @xmath173 , i.e. @xmath206 for all @xmath207 , we write @xmath208 and we denote by @xmath209 the function defined for @xmath168 by @xmath210 it is easy to check that the family @xmath211 is a hilbert basis of @xmath171 . writing @xmath212 we obtain after several applications of fubini @xmath213 @xmath214 we recall that by the mapping property of schottky groups , there exists @xmath78 such that for all @xmath215 and all @xmath216 , @xmath217 this uniform contraction property guarantees uniform convergence of @xmath218 on @xmath219 , allowing us to write @xmath220 where @xmath221 is the bergman reproducing kernel of the disc @xmath222 , given by the explicit formula @xmath223
^ 2}.\ ] ] a standard computation involving stokes s and cauchy formula ( see for example borthwick @xcite , p. 306 ) then shows that @xmath224 where @xmath225 is the unique fixed point of @xmath226 .
moreover , @xmath227 where @xmath228 is the closed geodesic represented by the conjugacy class of @xmath229 .
there is a one - to - one correspondence between prime periodic orbits of the bowen - series map @xmath135 and prime conjugacy classes in @xmath0 ( see borthwick @xcite , p. 303 ) , therefore each prime conjugacy class in @xmath0 ( and iterates ) appears in the above sum over all periodic orbits of @xmath135 .
however , for all @xmath230 its conjugacy class in @xmath0 _ splits _ into @xmath231=\vert g \vert\ ] ] conjugacy classes in @xmath2 , with same geodesic length .
let us explain this fact .
let @xmath232 be a normal subgroup of a group @xmath233 , and let @xmath234 .
then it is a basic and general observation that the conjugacy class of @xmath235 in @xmath233 splits into possibly several conjugacy classes in @xmath232 which are in one - to one correspondence with the cosets of @xmath236 where @xmath237 is the centralizer of @xmath235 in @xmath233 .
since @xmath0 is a free group , it is obvious in our case that whenever @xmath238 , @xmath239 and therefore @xmath240 going back to our trace computations , we have _ formally _ obtained @xmath241 @xmath242 where @xmath243 is the set of primitive conjugacy classes in @xmath2 . to justify convergence ,
first observe that we have @xmath244 which is roughly bounded by @xmath245^{-1}},\ ] ] and the pressure formula ( [ pressure ] ) together with bowen s result show uniform convergence of the series @xmath246 on half - planes @xmath247 , uniformly in @xmath19 .
moreover , by using proposition [ normest1 ] , we know that the _ spectral radius _ of @xmath170 is bounded ( uniformy in @xmath19 ) by @xmath194 , therefore we do have @xmath241 for all @xmath248 . since the infinite product formula for the selberg s zeta function holds for all @xmath248 , we have the desired conclusion by analytic continuation .
@xmath249 the above formula is critical in our analysis : a result of patterson and perry @xcite says that resonances ( apart from topological zeros located at negative integers ) are the same ( with multiplicity ) as zeros of the selberg zeta function
. therefore resonances on @xmath4 are the same as non - trivial zeros of @xmath250 , with multiplicities .
such a correspondence is also pointed out in @xcite , but comes after a rather roundabout argument based on different calculations of laplace transforms of counting functions .
the goal of this section is to prove the following bound .
[ detest1 ] there exists a constant @xmath251 such that for all @xmath19 large and all @xmath202 , we have the bound @xmath252 the proof will follow from a careful estimate of singular values of the operators @xmath193 .
we need first to recall some material on singular values and weyl inequalities , our basic reference is the book of simon @xcite . let @xmath253 be two hilbert spaces . consider @xmath254 a compact operator .
the singular value sequence @xmath255 is defined as the eigenvalue sequence of @xmath256 we will need to use the following fact .
[ vector ] let @xmath257 be @xmath258 hilbert spaces with hilbert bases denoted by @xmath259 let @xmath260_{1\leq i , j\leq m}:\mathcal{h}_1\oplus \ldots \oplus \mathcal{h}_m \rightarrow \mathcal{h}_1\oplus \ldots \oplus \mathcal{h}_m\ ] ] be a compact operator where each @xmath261
. then we have for all @xmath262 , @xmath263 } \vert t_{i , j } \mathrm{e}_\ell^j \vert_{\mathcal{h}_i}\right ) .\ ] ] _ proof . _
we set @xmath264 .
we define a natural basis @xmath265 of @xmath266 by setting for all @xmath267 @xmath268 from the min - max principle for the eigenvalues of compact self - adjoint operators it follows that @xmath269 where the min is taken along all subspaces @xmath270 with codimension @xmath271 . choosing @xmath272 we have @xmath273 writing @xmath274 we obtain by the triangle inequality and cauchy - schwarz @xmath275 let @xmath276 be defined by @xmath277
so that we can write @xmath278 we have obviously @xmath279 hence @xmath280 @xmath281 the proof ends by writing @xmath282}(t)\ ] ] and applying the above formula .
@xmath249 we can now move on to the proof of proposition [ detest1 ] . viewing @xmath171 as @xmath283 where @xmath284 , the formula @xmath285 shows that in the matrix representation of @xmath170 , there are at most @xmath120 non - zero operator entries per row . using the explicit basis @xmath286 for @xmath287 , it is enough to estimate @xmath288 where @xmath289 .
using the fact that @xmath290 we obtain the bound ( for some adequate constants @xmath179 and @xmath291 ) @xmath292 applying lemma [ vector ] , we have reached @xmath293 } \rho_0^j \leq \widetilde{c } \vert g \vert e^{c \vert s \vert } \rho_0^{\frac{k}{\vert g \vert}}.\ ] ] we can now use weyl inequalities ( see @xcite theorem 1.15 ) to write @xmath294 @xmath295 setting @xmath296 we end up with @xmath297 for some suitable constant @xmath298 . the proof is done since @xmath299 .
@xmath249 [ countingest1 ] there exists a constant @xmath179 such that for all @xmath19 large enough , we have @xmath300 _ proof_. this estimate follows straightforwardly from jensen s formula ( see the end of @xmath112 , proposition [ jensen ] for details ) but a lower bound is required . indeed , using the bound ( [ bound1 ] )
, we observe that @xmath301^{-1 } } > -\infty\ ] ] which is a lower bound independent of @xmath19 . applying the classical jensen s identity on the disc @xmath302 , where @xmath303 is carefully chosen in function of @xmath304
, we end up with the above bound .
in this section , we use the global upper bound proved in the previous section to produce a lower bound , thanks to the leading singularity of the trace formula .
before we state the trace formula , we need to point out a fact .
the nielsen volume of a geometrically finite surface @xmath34 is defined as the hyperbolic area of the _ nielsen region _
@xmath215 , the geodesically convex hull of closed geodesics on the surface . in the convex co - compact case ,
the nielsen region is a compact surface with geodesic boundary . from the gauss - bonnet formula , we know
( see for example @xcite theorem 2.15 ) that @xmath305 , where @xmath306 is the euler - poincar characteristic .
going back to our surfaces @xmath76 , there is a natural covering @xmath307 with degree @xmath308=\vert g \vert$ ] .
it is a standard fact of algebraic topology that @xmath309 which translates into the formula @xmath310 the wave - trace formula stated below is due to guillop and zworski @xcite .
we denote by @xmath311 the set of _ primitive closed geodesics _ on the surface @xmath312 , and if @xmath313 , @xmath314 is the length . in the following , @xmath99 still denotes the nielsen region .
let @xmath315 i.e. a smooth function , compactly supported in @xmath316 .
we have the identity : @xmath317 @xmath318 where @xmath319 is the usual fourier transform @xmath320 to prove theorem [ lowerb ] , we choose a test function @xmath321 such that @xmath322 and @xmath323 . for all
@xmath324 we set @xmath325 where @xmath56 will be a large parameter .
since the length spectrum of @xmath110 is a subset of the length spectrum of @xmath24 ( without multiplicities ) , we can definitely find a uniform @xmath78 such that for all @xmath326 , @xmath327 . in the sequel
, we take @xmath56 large enough so that @xmath328 .
the above trace formula gives @xmath329 in view of the preceding remarks , this yields for all @xmath19 large and @xmath56 as above @xmath330 where @xmath179 is a uniform constant .
using the fact that @xmath331 and repeated integrations by parts , we have the estimate @xmath332 for all @xmath333 . we write @xmath334 @xmath335 where @xmath336 does not depend on @xmath337 .
since we have obviously @xmath338 and @xmath56 is taken large , we can write @xmath339 a stieltjes integration by parts yields @xmath340 @xmath341 where we have used the upper bound from corollary [ countingest1 ] . setting @xmath342 with @xmath68 and @xmath56 large ( but fixed ) , we have obtained @xmath343 where @xmath344 is a ( possibly large ) constant depending only on @xmath215 .
taking @xmath215 so large that @xmath345 , we get that for all @xmath19 large enough , @xmath346 which yields @xmath347 and the proof is done .
let @xmath349 and set @xmath350 then for all @xmath351 small enough , @xmath352 is a bounded subset of @xmath187 whose connected components have length at most @xmath353 where @xmath179 is independent of @xmath351 , see @xcite lemma 15.12 .
let @xmath354 denote these connected components .
the existence of a finite patterson - sullivan measure @xmath355 supported on @xmath137 shows that ( see @xcite p. 312 , first displayed equation ) the number @xmath356 of connected components is @xmath357 .
given @xmath358 , let @xmath359 be the unique euclidean open disc in @xmath121 orthogonal to @xmath187 such that @xmath360 we consider @xmath361 set @xmath362 then the norm on @xmath348 is given by @xmath363 the parameter @xmath351 will play the role of a scale parameter whose size will be adjusted according to the spectral parameter @xmath364 .
an important fact in the sequel is the following estimate taken from @xcite .
[ separation ] there exists @xmath365 such that for all @xmath366 , for all @xmath367 and all @xmath368 , there exists an index @xmath369 such that @xmath370 with @xmath371 the above lemma guarantees that the transfer operator considered previously is well defined for all @xmath140 large enough ( independently of @xmath372 ) : @xmath373 the basic norm estimate is the following .
[ normest2 ] set @xmath374 , where @xmath364 is the spectral parameter .
there exist a constant @xmath53 , independent of @xmath375 such that for all @xmath140 large and @xmath376 the proof is postponed to the appendix .
this estimate essentially shows that the spectral radius of @xmath377 can be bounded uniformly on all spaces @xmath348 in term of the topological pressure .
moreover , by following verbatim the trace computations of @xmath73 , we see that the determinants ( and hence the full spectrum of @xmath377 ) @xmath378 _ do not _ depend on the scale parameter @xmath351 ( traces depend only on periodic points of the bowen - series map ) . to count resonances on @xmath5
, we will use the following family of _ hilbert - schmidt _
determinants @xmath379 remark that if @xmath380 , then by proposition [ detformula ] , the operator @xmath381 must have @xmath382 as an eigenvalue , and so does @xmath377 , but clearly @xmath383 for all @xmath351 small , therefore @xmath384 . on the other hand , @xmath385 might have some extra zeros which are not resonances , but that s a minor issue since we are interested on upper bounds on the density . in this subsection
, we prove to lemmas which are to be used in the proof of the main estimate in a crucial way .
however , since they only play a role at the very end of the proof , the reader can skip them at first glance .
[ girth ] there exists @xmath386 such that for all @xmath207 , all @xmath387 with @xmath388 , we have @xmath389 _ proof .
_ this is a slight variation on the `` girth lower bound '' proved in @xcite for cayley graphs of @xmath84 with respect to arbitrary generators of @xmath0 .
let @xmath390 be the usual enclidean norm on @xmath391 and if @xmath392 is a @xmath117 real matrix , set @xmath393 which is an algebra norm .
given @xmath394 , assume that we have two words @xmath395 with @xmath396 consider the matrix @xmath397 , then we do have @xmath398 but @xmath399 therefore one of the two off - diagonal entries of @xmath400 is a non - zero multiple of @xmath19 , which forces @xmath401 since @xmath402 is an algebra norm , we have @xmath403 and the proof is done with @xmath404 we will also need to use the following fact . [ separate ] there exist constants @xmath405 and @xmath406 such that for all @xmath207 , for all @xmath168 and all words @xmath407 , @xmath408 where @xmath409 . _ proof_. let @xmath407 and pick @xmath168 .
since @xmath410 we have @xmath411 .
let us write @xmath412 where @xmath413 and @xmath414 , @xmath415 . since @xmath416 is a mbius transform , we can use the standard formula @xmath417 recall that for all @xmath418 , for all @xmath419 @xmath420 where @xmath421 is understood mod @xmath120 .
therefore @xmath422 implies that @xmath423 and @xmath424 belong to two different discs @xmath425 therefore we have @xmath426 using the lower bound for the derivatives from ( [ hyp1 ] ) , we end up with @xmath427 and the proof is done .
@xmath249 both of these estimates are of independent interest but we will actually combine them as follows .
[ erhenfest ] let @xmath179 be a constant
. there exists @xmath78 depending only on @xmath428 such that for all @xmath207 , for all @xmath168 and all @xmath429 with @xmath430 , we have @xmath431 _ proof_.
let @xmath432 where @xmath433 will be adjusted later on .
assume that we have two words @xmath407 such that @xmath434 by lemma [ separate ] , we get that @xmath435 which shows that @xmath436 where @xmath437 is another constant ( depending on the previous ones ) .
assuming @xmath438 we get @xmath439 it is now clear that if @xmath433 is taken small enough , we have for large @xmath19 @xmath440 hence contradicting lemma [ girth ] .
@xmath249 the main theorem will follow , after a suitable application of jensen s formula from the next statement which is the main goal of this section .
we recall that we will work with the modified zeta function @xmath441 where @xmath442 will be adjusted later on .
[ firstest ] fix @xmath443 .
then there exist constants @xmath78 , @xmath53 and @xmath444 , @xmath13 such that for all @xmath445 ( with @xmath75 ) and @xmath446 , we have for all @xmath19 large , @xmath447 where @xmath448 $ ] , @xmath449 . it is necessary to recall at this point a few facts on regularized hilbert - schmidt determinants , our reference is @xcite .
let @xmath266 be an abstract separable hilbert space , and @xmath450 a compact operator .
the operator @xmath56 is called hilbert - schmidt if @xmath451 and its hilbert - schmidt norm is @xmath452 the regularized determinant @xmath453 is defined for all hilbert - schmidt operators by @xmath454),\ ] ] where @xmath455 is a trace class operator .
if @xmath56 is itself a trace class operator , then we have actually @xmath456 a key tool for our purpose is the following inequality and use the inequality @xmath457 ] ( see @xcite , chapter 4 , theorem 7.4 ) : @xmath458 we can now give a proof of proposition [ detest1 ] .
first we will use the notation for all @xmath207 @xmath459 given @xmath460 , let @xmath461 be a hilbert basis of @xmath462 . according to inequality ( [ hsest ] )
, we need to estimate the hilbert - schmidt norm @xmath463 @xmath464 in addition , we have @xmath465 @xmath466 noticing that @xmath467 \atop 0\ \mathrm{otherwise } \right.\ ] ] we have obtained @xmath468 } \int_{\omega_j(h ) } ( \gamma_\alpha')^s \overline{(\gamma_\beta')^s } \mathbf{e}_k^\ell\circ \gamma_\alpha \overline{\mathbf{e}_k^\ell\circ \gamma_\beta } dm.\ ] ] since @xmath469 converges uniformly on compact subsets of @xmath470 to the bergman kernel @xmath471 , we can exchange summations to write @xmath472 } \int_{\omega_j(h ) } ( \gamma_\alpha'(z))^s \overline{(\gamma_\beta'(z))^s } b_{\omega(h)}(\gamma_\alpha z,\gamma_\beta z ) dm(z).\ ] ] we stress that since @xmath473 is disconnected , we have @xmath474 if @xmath30 and @xmath475 do not belong to the same connected component .
we assume from now on that @xmath449 with @xmath476 and @xmath477 .
we will choose @xmath478 of the form @xmath479.\ ] ] we recall that each disc @xmath359 has by construction diameter at most @xmath353 , and we choose @xmath480 so that the conclusion of corollary [ erhenfest ] is true .
therefore , in the above sum , there is no off - diagonal contribution .
indeed , according to corollary [ erhenfest ] there are no words with @xmath410 such that @xmath481 provided that @xmath482 . as a consequence we have actually @xmath483 using the fact as @xmath484 becomes large , see @xcite , after lemma 3.4 , p.737 . ]
that each disc @xmath359 has radius at most @xmath485 , and because of the uniform distortion estimate ( [ bdist ] ) , we have for all @xmath164 and all @xmath486 , @xmath487 on the other hand , using lemma [ separation ] and the explicit formula for the bergman kernel , we see that @xmath488 uniformly in @xmath140 .
since we have @xmath489 we obtain by the inequality ( [ pressure2 ] ) that @xmath490 now recall that because of bowen s formula , @xmath12 implies that @xmath491 and the proof is done with @xmath492 since @xmath493 . @xmath249 using proposition [ detest1 ] , we can prove theorem [ upperb ] .
we will apply the following version of jensen s formula which can be derived straightforwardly from the classical textbooks , for example @xcite , p.125 .
[ jensen ] let @xmath494 be a holomorphic function on the open disc @xmath495 , and assume that @xmath496 .
let @xmath497 denote the number of zeros of @xmath494 in the closed disc @xmath498 . for all @xmath499 , we have @xmath500 the goal is to apply the above formula to @xmath501 where @xmath140 is taken according to proposition [ detest1 ] .
we need a lower bound .
assume that @xmath502 , we get @xmath503 @xmath504 using the bound ( [ bound1 ] ) we do have ( recall that @xmath135 is the bowen - series map on the boundary ) .
@xmath505^{-1}}.\ ] ] on the other hand , formula ( [ pressure ] ) for the topological pressure gives us ( for all @xmath68 ) @xmath506 since @xmath507 , this last lower bound shows clearly that one can find @xmath508 independent of @xmath509 such that for all @xmath502 , we have @xmath510 going back to the proof of theorem [ upperb ] , fix now @xmath511 .
let @xmath512 denote the ( closed ) rectangle @xmath513 for @xmath49 , set @xmath514 , and choose @xmath304 large enough so that @xmath515 .
set @xmath516 .
clearly if @xmath517 is small enough , we do have @xmath518 .
one can also check that we have @xmath519 applying the above formula to @xmath501 with @xmath520 on the disc @xmath521 we get @xmath522 @xmath523 using proposition [ detest1 ] we get @xmath524 @xmath525 since @xmath526 are fixed , we clearly get the desired conclusion , up to a change of constants .
@xmath249 clearly the proof we have used ( based on the separation lemma [ separate ] ) not only simplifies part of the arguments in @xcite but slightly strengthens the result . moreover , we believe the technique can be carried over to higher dimensional settings , at least for schottky groups . on the other hand ,
it is clear that this trick fails for situations where the limit set is not disconnected , for example for quasi - fuchsian groups .
this is where the more sophisticated arguments ( based on off - diagonal cancellations ) used in @xcite can be usefull , this will be pursued elsewhere .
we also point out that our method should work without great modifications to deal with subgroups of arithmetic co - compact fuchsian groups , and also for schottky subgroups of @xmath527)$ ] .
the only thing required is an appropriate `` logarithmic girth '' estimate with respect to the congruence parameter .
in this section , we prove proposition [ normest2 ] , which gives a crude bound for the operator norm of @xmath528 on spaces @xmath348 . let @xmath529 .
we first write @xmath530 @xmath531 we recall that because each disc @xmath359 has size at most @xmath353 and thanks to the bounded distortion property , we do have @xmath532 therefore we have @xmath533 @xmath534 by the reproducing property of bergman kernels , we have for all @xmath535 , @xmath536 which allows us to write ( thanks to cauchy - schwarz inequality and lemma [ separation ] ) @xmath537 therefore we have @xmath538 @xmath539 since @xmath540 , we have obtained @xmath541 @xmath542 exchanging summations , we can use cauchy - schwarz again ( and translation invariance of norms with respect to the @xmath543 variable ) to get @xmath544 this concludes the proof since by lemma [ pressure2 ] , we now have @xmath545 jack button .
all fuchsian schottky groups are classical schottky groups . in _ the epstein birthday schrift _ , volume * 1 * of _ geom .
_ , pages 117125 ( electronic ) .
coventry , 1998 .
|
let @xmath0 be a convex co - compact subgroup of @xmath1 , and let @xmath2 be the sequence of `` congruence '' subgroups of @xmath0 .
let @xmath3 be the resonances of the hyperbolic laplacian on the `` congruence '' surfaces @xmath4 .
we prove two results on the density of resonances in @xmath5 as @xmath6 : the first shows at least @xmath7 resonances in slowly growing discs , the other one is a bound from above in boxes @xmath8 , with @xmath9 , where we prove a density estimate of the type @xmath10 with @xmath11 for all @xmath12 , @xmath13 .
|
the existence and uniqueness of black hole solutions of the einstein equations with various types of matter has been a rich avenue of research for many years .
the classic `` no - hair '' theorems ( see , for example , @xcite for a review ) proved the uniqueness of the kerr - newman family of metrics describing four - dimensional , asymptotically flat , black hole solutions of the einstein equations with an electromagnetic field or in a vacuum . more recently , there has been an explosion of interest in the generalization of these uniqueness results to higher - dimensional black holes , and particular , in the non - uniqueness of higher - dimensional rotating black holes ( see , for example , @xcite ) .
black hole solutions of the einstein - scalar field system have been studied for almost as long as those of the einstein - maxwell system ( see @xcite for a detailed review , and @xcite for a summary ) .
the case of a minimally coupled scalar field has been the most extensively studied , with a number of `` no - hair '' results proved , in asymptotically flat space , particularly for static , spherically symmetric black holes @xcite .
these results depend on some assumptions about the form of the self - interaction potential , typically that it is positive semi - definite . when these assumptions are not satisfied , asymptotically flat black hole and soliton solutions of the field equations can be constructed , some numerically with a particular choice of self - interaction potential @xcite , and there are also analytic solutions ( often with unusual potentials ) @xcite , which can be generated from vacuum solutions in the static case @xcite . including a cosmological constant changes the picture for minimally coupled scalar fields , providing the self - interaction potential is non - zero @xcite .
when the cosmological constant is positive , numerical @xcite and analytic @xcite black hole solutions have been found when the self - interaction potential is non - convex @xcite , although at least some of these solutions are unstable @xcite . on the other hand , with a negative cosmological constant , stable black holes with minimally coupled scalar field hair
have been found numerically @xcite , and there are also some analytic solutions @xcite .
such asymptotically anti - de sitter ( ads ) solutions have attracted much recent interest in the literature due to their interpretation via the ads / cft ( conformal field theory ) correspondence ( some works on this topic include @xcite ) , particularly for black holes with scalar hair in supergravity theories @xcite , although the latter are unstable @xcite .
when non - minimal coupling of the scalar field to the ricci scalar curvature is included in the model , the case of conformal coupling has been of particular interest .
the bbmb solution @xcite is the unique @xcite static , asymptotically flat , black hole solution of the field equations in four space - time dimensions , when the self - interaction potential is zero , but is unstable @xcite .
furthermore , the scalar field diverges on the event horizon .
an analytic solution has been found in asymptotically flat space when the scalar field self - interaction potential is non - zero @xcite .
the analogue of the bbmb solution in four dimensions when there is a positive cosmological constant has a quartic self interaction potential @xcite .
although the scalar field does not suffer from the divergences of the bbmb black hole , it is still unstable @xcite . as for minimal coupling , in the presence of a negative cosmological constant , four - dimensional stable solitons @xcite and black holes @xcite have been found when the self - interaction potential is either zero or quadratic .
the case of non - minimal coupling which is not conformal coupling has received less attention in the literature . in four - dimensional , asymptotically flat , space - time , various black hole `` no - hair '' theorems have been proved @xcite . with a zero self - interaction potential ,
four - dimensional black holes are studied in @xcite .
it is found that non - minimally- and non - conformally - coupled scalar field hair can be supported by the black hole only when there is a negative cosmological constant , in agreement with numerical work @xcite .
furthermore , the hair is stable only when the constant governing the coupling between the scalar field and the ricci scalar curvature is such that the breitenlohner - freedman bound @xcite on the `` effective mass '' is satisfied . most of the above work is concerned with static , spherically symmetric black holes in four space - time dimensions . in asymptotically ads space , as well as
spherically symmetric black holes , the well - known topological black holes exist ( see , for example , @xcite ) .
topological black holes with scalar field hair have not been widely studied , apart from an analytic solution in the minimally coupled case @xcite , and numerical solutions in the conformally coupled case @xcite .
solitons and black holes with scalar hair in more than four space - time dimensions have received more attention in the literature in recent years , particularly in asymptotically ads space @xcite . the techniques for generating minimally coupled scalar solutions from solutions of the vacuum einstein equations have been generalized @xcite , yielding new analytic solutions . for conformally coupled scalar fields , there is no higher - dimensional analogue of the asymptotically flat bbmb black hole @xcite , although the corresponding analytic solution in three - dimensional , asymptotically ads space does exist @xcite , and there are numerical soliton and black hole solutions in ads in higher dimensions @xcite , all with vanishing self - interaction potential .
a set of analytic solutions in various dimensions with a non - zero self - interaction potential and a cosmological constant has also been considered recently in the conformally coupled case @xcite . for other non - minimal coupling , saa @xcite has proved a no - hair theorem in the asymptotically flat case , although for certain values of the coupling constant assumptions on the magnitude of the scalar field are required .
our purpose in this paper is to consider non - minimal , non - conformal coupling , extending the analysis of @xcite to solitons , higher - dimensional black holes , and topological black holes in asymptotically ads space .
the structure of this paper is as follows . in sec .
[ sec : model ] we describe our scalar field model , the field equations and boundary conditions to be satisfied by soliton and black hole solutions , and the conformal map which transforms our non - minimally coupled scalar field system to one with a minimally coupled scalar field and a non - zero self - interaction potential . for most of the parameter space
, we are able to prove that non - trivial solutions can not exist , and these non - existence results are presented in sec .
[ sec : no - hair ] .
non - trivial solutions exist in the remainder of the parameter space , and in sec .
[ sec : nontrivial ] we present our numerical solutions and study their thermodynamics and stability .
finally our conclusions are presented in sec .
[ sec : conc ] .
we consider the following action , describing a scalar field @xmath0 with non - minimal coupling to gravity in @xmath1 space - time dimensions : @xmath2 , \label{eq : action}\ ] ] where @xmath3 is the ricci scalar curvature , and we have included a cosmological constant @xmath4 . here and throughout this paper , the metric has signature @xmath5 and we use units in which @xmath6 . the constant @xmath7 governs the coupling between the scalar field and the ricci scalar curvature .
two values of @xmath7 are particularly important : minimal coupling @xmath8 , and conformal coupling @xmath9 $ ] . in the action ( [ eq : action ] ) we have set any self - interaction potential to zero , for simplicity .
we are interested in static , spherically symmetric , solitons and black holes for all values of the cosmological constant @xmath4 ; and static topological black holes when @xmath13 .
we therefore use the following ansatz for the metric , in the usual schwarzschild - like co - ordinates : @xmath14 where @xmath15 denotes the line element of an @xmath16-dimensional space @xmath17 with constant curvature .
the discrete parameter @xmath18 takes the values @xmath19 , for which we have the following forms of the function @xmath20 : @xmath21 the value @xmath22 gives the usual spherically symmetric metric , with the hypersurface @xmath23 equal to an @xmath16-sphere ; when @xmath24 , the hypersurface @xmath25 has constant negative curvature , and when @xmath26 , the hypersurface @xmath27 is an @xmath16-dimensional euclidean space ( see @xcite for further details ) .
it is convenient to introduce a new metric function @xmath28 by @xmath29 with the ansatz ( [ eq : ansatz ] ) , and assuming that the scalar field @xmath30 depends only on the radial co - ordinate @xmath31 , the field equations ( [ eq : einst ] , [ eq : scalarfieldeq ] ) take the form : @xmath32 - \left(2\xi-\frac{1}{2}\right)h\phi'^2+\xi \phi \phi'(h'+2h \delta ' ) -2\xi \phi\nabla^2 \phi + \lambda ; \nonumber \\ 0 & = & \frac{n-2}{r}(1-\xi\phi^2)\delta'-(1 - 2\xi)\phi'^2 - 2\xi\phi(\delta'\phi'-\phi '' ) ; \nonumber \\ 0 & = & h \phi''+\phi'\left(h\delta'+h'+h\frac{n-2}{r}\right ) -\xi r \phi .
\label{eq : fieldeqs}\end{aligned}\ ] ] we now summarize the discussion of sec .
[ sec : intro ] , namely what is known about the existence of soliton and black hole solutions of these field equations , for various values of the cosmological constant @xmath4 and coupling constant @xmath7 .
firstly , we bring together in one table the known results for solitons and black holes in four space - time dimensions : j. d. bekenstein , phys .
d * 5 * , 1239 ( 1972 ) ; d * 5 * , 2403 ( 1972 ) ; d * 51 * , r6608 ( 1996 ) ; t. hertog , phys .
d * 74 * , 084008 ( 2006 ) ; m. heusler , class . quantum gravity * 12 * , 779 ( 1995 ) ; m. heusler and n. straumann , class . quantum gravity
* 9 * , 2177 ( 1992 ) ; d. sudarsky , class . quantum gravity * 12 * , 579 ( 1995 ) .
u. nucamendi and m. salgado , phys .
d * 68 * , 044026 ( 2003 ) ; m. alcubierre , j. a. gonzalez and m. salgado , phys .
d * 70 * , 064016 ( 2004 ) ; a. corichi , u. nucamendi and m. salgado , phys .
d * 73 * , 084002 ( 2006 ) .
o. bechmann and o. lechtenfeld , class .
* 12 * , 1473 ( 1995 ) ; k. a. bronnikov and g. n. shikin , grav .
cosmol .
* 8 * , 107 ( 2002 ) ; h. dennhardt and o. lechtenfeld , int .
j. mod .
a * 13 * , 741 ( 1998 ) .
d. birmingham , class .
quantum gravity * 16 * , 1197 ( 1999 ) ; j. p.
s. lemos , phys . lett .
b * 352 * , 46 ( 1995 ) ; class .
quantum gravity * 12 * , 1081 ( 1995 ) ; phys .
b * 353 * , 46 ( 1996 ) ; j. p. s. lemos and v. t. zanchin , phys .
d * 54 * , 3840 ( 1996 ) ; l. vanzo , phys .
d * 56 * , 6475 ( 1997 ) .
|
we study higher - dimensional soliton and hairy black hole solutions of the einstein equations non - minimally coupled to a scalar field .
the scalar field has no self - interaction potential but a cosmological constant is included .
non - trivial solutions exist only when the cosmological constant is negative and the constant governing the coupling of the scalar field to the ricci scalar curvature is positive .
at least some of these solutions are stable when this coupling constant is not too large .
|
named entity recognition ( ner ) is a very important task in natural language processing . in the ner task ,
the objective is to find and cluster named entities in text into any desired categories such as person names ( per ) , organizations ( org ) , locations ( loc ) , time expressions , etc .
ner is an important precursor to tasks like machine translation , question answering , topic modelling and information extraction among others .
various methods have been used in the past for ner including hidden markov models , conditional random fields , feature engineering approaches using support vector machines , max entropy classifiers for finally classifying outputs and more recently neural network based approaches .
development of an ner system for indian languages is a comparatively difficult task .
hindi and many other indian languages provide some inherent difficulties in many nlp related tasks .
the structure of the languages contain many complexities like free - word ordering ( which affect n - gram based approaches significantly ) , no capitalization information and its inflectional nature ( affecting hand - engineered approaches significantly ) .
also , in indian languages there are many word constructions that can be classified as named entities ( derivational / inflectional constructions ) etc and these constraints on these constructions vary from language to language hence carefully crafted rules need to be made for each language which is a very time consuming and expensive task .
another major problem in indian languages is the fact that we have scarce availability of annotated data for indian languages .
the task is hard for rule - based nlp tools , and the scarcity of labelled data renders many of the statistical approaches like deep learning unusable .
this complexity in the task is a significant challenge to solve .
can we develop tools which can generalize to other languages(unlike rule based approaches ) but still can perform well on this task ? on the other hand , rnn s and its variants
have consistently performed better than other approaches on english ner and many other sequence labelling tasks .
we believe rnn would be a very effective method compared to fixed - window approaches as the memory cell takes much larger parts of the sentence into context thus solving the problem of sentences being freely ordered to a large extent .
we propose a method to be able to model the ner task using rnn based approaches using the unsupervised data available and achieve good improvements in accuracies over many other models without any hand - engineered features or any rule - based approach
. we would learn word - vectors that capture a large number of precise semantic and syntactic word relationships from a large unlabelled corpus and use them to initialize rnns thus allowing us to leverage the capabilities of rnns on the currently available data .
we believe to the best of our knowledge , that this is the first approach capable of using rnn for ner in hindi data .
we believe learning based approaches like these could generalize to other indian languages without having to handcraft features or develop dependence on other nlp related tools .
our model uses no language specific features or gazetteers or dictionaries .
we use a small amount of supervised training data along with some unannotated corpus for training word embeddings yet we achieve accuracies on par with the state of the art results on the conll 2003 dataset for english and achieve 77.48% accuracy on icon 2013 nlp tools corpus for hindi language .
our paper is mainly divided into the following sections : * in section 1 we begin with an introduction to the task of ner and briefly describe our approach . * in section 2 , we mention the issues with hindi ner and provide an overview of the past approaches to ner . * in section 3 ,
we descibe our proposed rnn based approach to the task of ner and the creation of word embeddings for ner which are at the core of our model . * in section 4
we explain our experimental setup , describe the dataset for both hindi and english and give results and observations of testing on both the datasets . * in section 5
we give our conclusions from the experiments and also describe methods to extend our approach to other languages .
ner task has been extensively studied in the literature .
previous approaches in ner can be roughly classified into rule based approaches and learning based approaches .
rule based approaches include the system developed by ralph grishman in 1995 which used a large dictionary of named entities @xcite .
another model was built for ner using large lists of names of people , location etc . in 1996@xcite .
a huge disadvantage of these systems is that a huge list needed to be made and the output for any entity not seen before could not be determined .
they lacked in discovering new named entities , not present in the dictionary available and also cases where the word appeared in the dictionary but was not a named entity .
this is an even bigger problem for indian languages which would frequently be agglutinative in nature hence creation of dictionaries would be rendered impossible .
people either used feature learning based approaches using hand - crafted features like capitalization etc .
they gave these features to a machine learning based classifier like support vector machine ( svm)@xcite , naive bayes ( nb ) or maximum entropy ( me ) classifiers .
some posed this problem as a sequence labelling problem terming the context is very important in determining the entities .
then , the handcrafted series were used in sequences using machine learning methods such as hidden markov models ( hmm)@xcite , conditional random field ( crf ) @xcite and decision trees ( dt)@xcite .
many attempts have been made to combine the above two approaches to achieve better performance .
an example of this is @xcite who use a combination of both handcrafted rules along with hmm and me .
more recent approaches for indian language and hindi ner are based on crf s and include @xcite and @xcite .
the recent rnn based approaches for ner include ones by @xcite
. also , there are many approaches which combine ner with other tasks like @xcite ( pos tagging and ner along with chunking and srl tasks ) and @xcite ( combining entity linking and ner ) which have produced state - of - the - art results on english datasets .
owing to the recent success in deep learning frameworks , we sought to apply the techniques to indian language data like hindi .
but , the main challenge in these approaches is to learn inspite of the scarcity of labelled data , one of the core problems of adapting deep - learning approaches to this domain .
we propose to leverage the vast amount of unlabelled data available in this domain .
the recurrent neural networks rnn trained generally have to learn the recurrent layer as well as the embedding layer for every word .
the embedding layer usually takes a large amount of data to create good embeddings .
we formulate a two stage methodology to utilize the unlabelled data : in the first stage we utilize unlabelled corpora .
we learn skip - gram @xcite based embeddings and glove @xcite embeddings on those corpora .
we use the wikipedia corpus for hindi as a source to train these models . by that , we get wordvectors which will be used in the second stage . in the second stage ,
as illustrated in figure 1 , we use the deep - learning based models .
we initialize their embedding layers with the wordvectors for every word .
then , we train the network end - to - end on the labelled data . as various approaches have proved , a good initialization is crucial to learning good models and train faster @xcite .
we apply this approach to use word - vectors to counter the scarcity of labelled data .
the idea behind this is that the models would require much lesser data for convergence and would give much better results than when the embeddings are randomly initialized . to get both previous and subsequent context for making predictions we use bi - directional rnn @xcite .
we know that vanilla rnn suffers from not being able to model long term dependencies @xcite hence we use the lstm variant of the rnn @xcite which helps the rnn model long dependencies better . word2vec based approaches use the idea that words which occur in similar context are similar .
thus , they can be clustered together .
there are two models introduced by : @xcite cbow and skipgram .
the latter is shown to perform better on english corpuses for a variety of tasks , hence is more generalizable .
thus , we use the skip - gram based approach .
most recent method for generating wordvectors was glove , which is similar in nature to that of skipgram based model .
it trains embeddings with local window context using co - occurrence matrices .
the glove model is trained on the non - zero entries of a global co - occurrence matrix of all words in the corpus .
glove is shown to be a very effective method , and is used widely thus is shown to be generalizable to multiple tasks in english .
for english language , we use the pretrained word embeddings using the aforementioned approaches , since they are widely used and pretty effective .
the links for downloading the vectors are provided .
however , for hindi language we train using above mentioned methods(word2vec and glove ) and generate word vectors .
we start with one hot encoding for the words and random initializations for their wordvectors and then train them to finally arrive at the word vectors .
we use the hindi text from ltrc iiit hyderabad corpus for training .
the data is 385 mb in size and the encoding used is the utf-8 format ( the unsupervised training corpus contains 27 million tokens and 500,000 distinct tokens ) .
the training hindi word embeddings were trained using a window of context size of 5 .
the trained model is then used to generate the embeddings for the words in the vocabulary .
the data would be released along with the paper at our website along with the wordvectors and their training code . for a comparative study of performance of these methods
, we also compare between the skip - gram based wordvectors and glove vectors as embeddings to evaluate their performance on hindi language .
the architecture of the neural networks is described in figure 2 .
we trained deep neural networks consisting of either one or two recurrent layers since the labelled dataset was small . in the architecture
, we have an embedding layer followed by one or two recurrent layers as specified in the experiments followed by the softmax layer .
we experimented with three different kinds of recurrent layers : vanilla rnn , lstm and bi - directional lstm to test which one would be the most suitable for ner task . for the embedding layer ,
it is initialized with the concatenation of the wordvector and the one - hot vector indicating its pos tag .
the pos tagging task is generally considered as a very useful feature for entity recognition , so it was a reliable feature .
this hypothesis was validated when the inclusion of pos tags into the embedding improved the accuracy by 3 - 4% .
this setup was trained end - to - end using adam optimizer @xcite and batch size of 128 using dropout layer with the dropout value of 0.5 after each of the recurrent layers .
we have used dropout training @xcite to reduce overfitting in our models and help the model generalise well to the data .
the key idea in dropouts is to randomly drop units with their connections from the neural network during training .
we perform extensive experimentation to validate our methodology .
we have described the datasets we use and the experimental setup in detail in this section .
we then present our results and provide a set of observations made for those results .
we test the effectiveness of our approach on icon 2013 nlp tools contest dataset for hindi language , along with cross - validating our methodology on the well - established conll 2003 english named entity recognition dataset @xcite .
we used the icon 2013 nlp tools contest dataset to evaluate our models on hindi .
the dataset contains words annotated with part - of - speech ( pos ) tags and corresponding named entity labels in shakti standard form ( ssf ) format @xcite .
the dataset primarily contains 11 entity types : organization ( org ) , person ( per ) , location ( loc ) , entertainment , facilities , artifact , living things , locomotives , plants , materials and diseases .
rest of the corpus was tagged as non - entities ( o ) .
the dataset was randomly divided into three splits : train , development and test in the ratios 70% , 17% and 13% .
the training set consists of 3,199 sentences comprising 56,801 tokens , development set contains 707 sentences comprising 12,882 tokens and test set contains 571 sentences comprising of 10,396 tokens .
we use the f1-measure to evaluate our performance against other approaches .
we perform extensive experiments on the conll 2003 dataset for named entity recognition .
the dataset is primarily a collection of reuters newswire articles annotated for ner with four entity types : person ( per ) , location(loc ) , organization(org ) , miscellaneous ( misc ) along with non entity elements tagged as ( o ) .
the data is provided with a training set contains 15,000 sentences consisting of approximately 203,000 tokens , along with a development set containing 3466 sentences consisting of around 51,000 tokens and a test set containing 3684 sentences comprising of approximately 46,435 tokens .
we use the standard evaluation scripts provided along with the dataset for assessing the performance of our methodology .
the scripts use the f1-score to evaluate the performance of models .
we use this architecture for the network because of the constraint on the dataset size caused by scarcity of labelled data .
we used a nvidia 970 gtx gpu and a 4.00 ghz intel i7 - 4790 processor with 64 gb ram to train our models .
as the datasets in this domain expand , we would like to scale up our approach to bigger architectures .
the results obtained on icon 2013 nlp tools dataset are summarized in table 2 .
we cross - validated our approach with english language using the conll 2003 dataset .
the results are summarized in table 1 , we are able to achieve state - of - the - art accuracies without using additional information like gazetteers , chunks along with not using any hand - crafted features which are considered essential for ner task as chunking provides us data about the phrases and gazetteers provide a list of words which have high likelihood of being a named entity . the neural networks which did not have wordvector based initializations could not perform well on the ner task as predicted .
this can be attributed to the scarcity of the data available in the ner task .
we also observe that networks consisting of one recurrent layer perform equally good or even better than networks having two recurrent layers .
we believe this would be a validation to our hypothesis that increasing the number of parameters can lead to overfitting .
we could see significant improvement in performance after using lstm - rnn instead of vanilla rnn which can be attributed to the ability of lstm to model long dependencies . also , the bidirectional rnn achieved significant improvement of accuracy over the others suggesting that incorporating context of words around ( of both ahead and back ) of the word is very useful .
we provide only 1 layer in our best model to be released along with the paper .
we show that the performance of deep learning based approaches on the task for entity recognition can significantly outperform many other approaches involving rule based systems or hand - crafted features .
the bidirectional lstm incorporates features of varied distances providing a bigger context relieving the problem of free - word ordering too . also , given the scarcity of data , our proposed method effectively leverages lstm based approaches by incorporating pre - trained word embeddings instead of learning it from data since it could be learnt in an unsupervised learning setting .
we could extend this approach to many indian languages as we do not need a very large annotated corpus .
when larger labelled datasets are developed , in the new system we would like to explore more deep neural network architectures and try learning the neural networks from scratch .
|
in this paper we describe an end to end neural model for named entity recognition ( ner ) which is based on bi - directional rnn - lstm .
almost all ner systems for hindi use language specific features and handcrafted rules with gazetteers .
our model is language independent and uses no domain specific features or any handcrafted rules .
our models rely on semantic information in the form of word vectors which are learnt by an unsupervised learning algorithm on an unannotated corpus .
our model attained state of the art performance in both english and hindi without the use of any morphological analysis or without using gazetteers of any sort .
|
granular media can flow like a fluid under a certain situation . in the case of _ the rapid granular flow _
, where the density is relatively low and interactions are dominated by the instantaneous collisions , the kinetic theory of dense gases @xcite is extended to the inelastic hard spheres to derive the constitutive relations @xcite . in the theory , the density correlations is taken into account to some extent but not the velocity correlations in most of the cases . as the flow gets denser
, however , the molecular chaos assumption becomes questionable .
in addition , the interactions may no longer be approximated by the instantaneous collisions but enduring contacts take place around the random closed packing fraction .
the comprehensive granular rheology including the rather complicated dense regime has not been established yet . during the last several years
, careful experiments and large - scale molecular dynamics simulations have been done on the dense granular flows @xcite .
one of the important model systems that has been intensively studied is the steady flow down a slope under the gravity , where we can control the ratio of the shear stress @xmath0 to the normal stress @xmath1 by changing the inclination angle @xmath2 . in this system
, it has been found that the packing fraction @xmath3 in the bulk of the flow is constant and is determined solely by the inclination angle @xmath2 ; in other words , @xmath3 is independent of the total flow hight @xmath4 and/or the roughness of the slope @xcite .
this interesting feature has been qualitatively understood by using _ the bagnold scaling _
@xcite , which states the shear stress @xmath0 is proportional to the square of the shear rate @xmath5 : @xmath6 here , @xmath7 is the particle mass , and @xmath8 is the particle diameter .
this scaling can be understood by dimensional analysis of the rigid granular flow , where the inverse of the shear rate @xmath9 is the only time scale in the system .
this scaling applies to the normal stress @xmath1 also , which gives @xmath10 in the slope flow under gravity , the force balance gives @xmath11 .
thus we finally have @xmath12 i.e. , the packing fraction @xmath3 is determined by the inclination angle @xmath2 .
this dimensional analysis does not hold when the time scales other than @xmath9 come into the problem , e.g. the time scales of the particle deformation @xcite , but not only the constant density profile but also the bagnold scaling itself has been found in the numerical simulations of dense steady flow down a slope for hard enough particles @xcite . in the slope flow simulations ,
the value of the packing fraction @xmath3 has been shown to increase upon decreasing the inclination angle @xmath2 , and eventually the flow stops at a finite angle @xmath13 ; namely , @xmath14 is an decreasing function of @xmath3 in the dense region @xcite .
one can interpret the transition at @xmath13 as the jamming transition @xcite .
the theoretical analysis of the functional form of @xmath14 has been done by louge @xcite using the kinetic theory , but he found the opposite dependence in the dense region , namely , the theory gives increasing packing fraction @xmath3 upon increasing inclination angle @xmath2 as shown in fig .
[ friction_packing1 ] , where curves from a kinetic theory @xcite is shown by symbols connected by dashed lines for the various restitution coefficients @xmath15 .
several explanations for this discrepancy have been proposed , such as the enduring contact @xcite , the burnett order ( the second order of the spatial gradients ) effect @xcite , and the particle roughness @xcite etc . , but the subject is still under debate . recently , the present authors @xcite have made a detailed comparison between the simulation results of the dense slope flow and the kinetic theory by jenkins and richman @xcite .
in contrast with the rather good agreement for the stresses , it has been found that the kinetic theory overestimate the energy dissipation rate @xmath16 , and this discrepancy is responsible for the contradicting behavior in the kinetic theory , i.e. @xmath14 increases with the packing fraction @xmath3 .
the authors conjectured that the discrepancy in the energy dissipation rate @xmath16 should be caused by the velocity correlations enhanced by the inelastic collisions ; the decrease of the relative normal velocity through the inelastic collisions results in reduction of the energy loss per collision .
such an effect has been noticed in granular gas simulations without shear @xcite , and the velocity correlations has been investigated analytically @xcite .
however , the situation is rather complicated under shear , because the shear tends to break the correlations .
the spatial velocity correlations in granular flow under shear has not been carefully studied so far @xcite . versus the packing fraction @xmath3 from the simulation data ( @xmath17 ( @xmath18 ) , @xmath19 ( @xmath20 ) , and @xmath21 ( @xmath22 ) ) and the plot of eq .
( [ soverp ] ) from the kinetic theory ( @xmath17 ( @xmath23 connected by dashed line ) , @xmath19 ( @xmath24 connected by dashed line ) , and @xmath21 ( @xmath25 connected by dashed line ) . for the simulation data ,
the average normal stress @xmath26 is used .
, scaledwidth=23.0% ] in this paper , we study the velocity correlation in the sheared granular flow , focusing its effects on the energy dissipation rate and the stress .
we adopt the simple shear flow of the inelastic hard spheres as a model system , in accordance with most of the kinetic theory analysis .
note that the enduring contacts is not allowed in the hard sphere model , whose effects are often under debate in the soft - sphere model simulations of the slope flow @xcite .
this paper is organized as follows . in section [ model ] , we briefly summarize the inelastic hard sphere model and the constitutive relations based on the kinetic theory .
we summarize our simulation method and present the results in section [ simulation ] .
the discussion and the summary are given in section [ discussion ] .
the inelastic hard sphere model is one of the simplest and widely - used models of granular materials @xcite .
the particles are infinitely rigid , and they interact through instantaneous two - body collisions .
we adopt the simplest collision rule for the monodisperse smooth hard spheres with diameter @xmath8 , mass @xmath7 , and a constant normal restitution coefficient @xmath15 in three dimensions as follows : the particle @xmath27 at the position @xmath28 with the velocity @xmath29 collides with the particle @xmath30 if @xmath31 and @xmath32 , and their post - collisional velocities @xmath33 and @xmath34 are given by @xmath35\bm n_{ij } , \\
\bm c_j^*&=&\bm c_j + \frac{1+e_p}{2 } \left[\bm n_{ij}\cdot \left(\bm c_i-\bm c_j\right)\right]\bm n_{ij } , \label{colrule}\end{aligned}\ ] ] respectively . here , @xmath36 is a unit vector defined as @xmath37 .
the collision is elastic when @xmath38 , and inelastic when @xmath39 . in the inelastic case ,
the particles lose the kinetic energy every time they collide , thus external drive is necessary to keep particles flowing .
we compare the simulation results of the inelastic hard spheres with the constitutive relations obtained from the chapman - enskog method @xcite , which has been developed in the kinetic theory of gases . in this paper
, we employ those by garz and dufty @xcite , who have improved the previous studies @xcite , that is limited to the weakly inelastic case ( @xmath40 ) , to include the case with any value of the restitution constant @xmath15 under the assumption that the state is near the local homogeneous cooling state @xcite .
in the following , we briefly summarize the kinetic theory to derive the constitutive relations .
the hydrodynamic variables are the number density field @xmath41 , the velocity field @xmath42 , and the granular temperature field @xmath43 , defined in terms of the single - particle distribution function @xmath44 as @xmath45 the hydrodynamic equations for these variables are given by @xmath46 where @xmath47 is the stress tensor , @xmath48 is the heat flux , and @xmath49 is the symmetrized velocity gradient tensor : @xmath50 .
note that the energy dissipation rate @xmath16 in eq .
( [ ebalance0 ] ) appears due to the energy loss through the inelastic collisions , which gives peculiar features to the granular hydrodynamics .
the constitutive relations for @xmath47 , @xmath48 , and @xmath16 are determined by the single - particle distribution @xmath44 .
its time evolution depends on the two - particle distribution function @xmath51 through the two - particle collision ; the @xmath52-particle distribution function depends on the ( @xmath53)-particle distribution function .
this is known as the bbgky hierarchy @xcite . in the enskog approximation ,
the two - particle distribution at collision is approximated as @xmath54 to close the bbgky hierarchy at the single - particle distribution @xcite .
here , @xmath55 is the radial distribution function at distance @xmath8 , and depends on the packing fraction @xmath56 .
the term @xmath55 represents the positional correlations , and the actual procedure to determine the functional form of @xmath55 is presented in subsection [ radial ] .
the correlations in the particle velocities are neglected under the molecular chaos assumption .
.the dimensionless functions in the constitutive relations from ref .
@xcite . [ cols="^,^ " , ] the constitutive relations for the hydrodynamic equations have been obtained in ref .
@xcite by the chapman - esnkog method with the approximation ( [ eq : enskog ] ) up to the navier - stokes order ( i.e. the first order of the spatial gradients ) . in the simple steady shear flow with constant @xmath52 , constant @xmath57 , and @xmath58 , the nonzero terms are the pressure , or the normal stress @xmath59 the shear stress @xmath60 and the energy dissipation rate @xmath61 the dimensionless functions @xmath62 are listed in table [ constitutiveeq ] . in the simple shear flow , eqs .
( [ contituity ] ) and ( [ motion ] ) are automatically satisfied with the constant normal stress @xmath1 and the constant shear stress @xmath0 .
the energy balance equation ( [ ebalance0 ] ) gives @xmath63 because there is no heat flux @xmath64 .
equation ( [ ebalance ] ) means that the granular temperature is locally determined by the balance between the viscous heating and the energy dissipation .
equation ( [ ebalance ] ) with eqs .
( [ shear ] ) and ( [ dissipation ] ) gives @xmath65 substituting eq .
( [ temperaturegamma ] ) into eqs .
( [ pressure ] ) and ( [ shear ] ) , we get @xmath66^{3/2}}{[f_3(\nu)]^{1/2}}\dot \gamma^2,\end{aligned}\ ] ] which are exactly what we have anticipated from the bagnold scaling eqs .
( [ bagnolds ] ) and ( [ bagnoldp ] ) .
the above derivation of the bagnold scaling by the kinetic theory gives the definite expression for eq .
( [ slope ] ) , @xmath67 as a function of the packing fraction @xmath3 .
this is plotted in fig .
[ friction_packing1 ] by symbols connected by lines , along with the simulation data .
one can see clear discrepancy between the theory and the simulation especially in the higher density region .
the kinetic theory gives increasing functions @xmath68 of @xmath3 , which means that the flow down steeper slope is denser .
in this section , we compare the expressions eqs .
( [ pressure])-([dissipation ] ) with the simulation results of simple shear flow of inelastic hard spheres .
the simulation is done under the constant volume condition with a uniform shear in a rectangular box of the size @xmath69 .
the shear is applied by the lees - edwards shearing periodic boundary conditions in the @xmath70 direction @xcite ; the periodic boundary condition is employed in the @xmath71 and @xmath72 directions .
we employ the event driven method , using the fast algorithm developed by isobe @xcite .
a steady shear flow with the mean velocity @xmath73 is prepared as follows .
first , a random configuration is prepared by the compressing procedure proposed by lubachevsky and stillinger @xcite in the elastic system without shear under the periodic boundary condition .
secondly , the initial shear flow is constructed from the above random configuration by giving the initial mean velocity @xmath73 and setting the initial temperature @xmath74 .
lastly , the steady shear flow of the inelastic system is obtained by relaxing the initial flow under the lees - edwards shearing periodic boundary condition @xcite . with the present parameter and system size ,
the final steady state is the simple shear flow with uniform packing fraction @xmath75 and mean velocity @xmath76 @xcite .
all the following data are taken in the steady state , and averaged over the space and time ( typically over 10,000 collisions per particle ) unless otherwise noted . in the following ,
all the quantities are given in the dimensionless form with the unit mass @xmath7 , the unit length @xmath8 , and the unit time @xmath9 .
most of the data are from the simulations with system size @xmath77 , @xmath78 , and @xmath79 .
several simulations has been done with @xmath80 to check the system size effect .
we measure the temperature @xmath57 , the normal stress @xmath1 , the shear stress @xmath0 , and the energy dissipation rate @xmath16 for various values of the packing fraction @xmath3 .
these are compared with eqs.([pressure])-([dissipation ] ) from the kinetic theory . with @xmath81 ( dashed line ) , @xmath19 ( dotted line ) , and @xmath21 ( solid line ) . the peak values of contact are 136 ( out of range ) , 30.6 ( @xmath20 ) , and 17.4 ( @xmath22 ) for @xmath15=0.70 , 0.92 , and 0.98 , respectively . the theoretical value at contact , @xmath82 , is shown by an arrow .
( b ) plot of @xmath83 versus the packing fraction @xmath3 for @xmath81 ( @xmath18 ) , @xmath19 ( @xmath20 ) , and @xmath21 ( @xmath22 ) .
@xmath84 is shown by a solid line .
, scaledwidth=46.0% ] for the constitutive relations with table [ constitutiveeq ] , we need to know the radial distribution function at the particle diameter , @xmath55 , as a function of the packing fraction . for elastic hard spheres ( @xmath38 ) in equilibrium , the well known expression of @xmath55 is the carnahan - starling formula @xcite @xmath85 for @xmath86 , where @xmath87 is the freezing packing fraction and @xmath88 @xcite .
torquato @xcite proposed the formula that include the higher packing fraction up to the random closed packing fraction @xmath89 as @xmath90 as for the inelastic hard spheres under shear , a generally accepted form of @xmath55 does not exist , but it has been found in several simulations that @xmath55 is larger for stronger inelasticity @xcite .
figure [ gr - plot](a ) shows the radial distribution @xmath91 averaged over the all directions obtained from our shear flow simulation with the packing fraction @xmath92 for various values of @xmath15 .
the spatial mesh to measure @xmath91 was taken as @xmath93 , and the peak values of @xmath91 around @xmath94 ( at the distance of the particle diameter ) are marked by symbols for @xmath95 and @xmath19 .
we can see that the peak value strongly increases for smaller @xmath15 , and can be much larger than the value from eq .
( [ torq ] ) ( @xmath82 shown by an arrow ) .
it is quite difficult to evaluate the precise value of @xmath55 from this direct measurement of @xmath91 because of the strong increase of @xmath91 in the limit of @xmath96 .
the way we determine @xmath55 from the simulation is through the expression of the collision frequency @xmath97 @xcite from the kinetic theory @xcite , @xmath98 where @xmath99 is given in table [ constitutiveeq ] . by measuring @xmath97 and @xmath57 for each @xmath3 from the simulation
, we can evaluate @xmath100 @xmath101 is plotted versus @xmath3 for various values of @xmath15 in fig .
[ gr - plot](b ) , where @xmath84 in eq .
( [ torq ] ) is shown by a solid line for reference .
@xmath101 shows stronger increase upon increasing the packing fraction @xmath3 as @xmath15 gets smaller ; by comparing it with fig .
[ gr - plot](a ) , we see that this indirect estimate gives an reasonable @xmath15 dependence of @xmath55 . in the following , we use @xmath101 as @xmath55 in table [ constitutiveeq ] unless otherwise noted .
( a ) and the normal stress @xmath1 ( b ) .
the simulation data for @xmath95 ( @xmath22 ) , @xmath19 ( @xmath20 ) , and @xmath102 ( @xmath18 ) are compared with the values from the kinetic theory ( @xmath103 and @xmath104 ) shown by symbols connected by dashed lines for @xmath95 ( @xmath25 ) , @xmath19 ( + ) , and @xmath102 ( @xmath23 ) .
@xmath105 and @xmath106 with @xmath107 are denoted by the solid lines , which agree with the simulation data ( see text ) .
, title="fig:",scaledwidth=23.0% ] ( a ) and the normal stress @xmath1 ( b ) .
the simulation data for @xmath95 ( @xmath22 ) , @xmath19 ( @xmath20 ) , and @xmath102 ( @xmath18 ) are compared with the values from the kinetic theory ( @xmath103 and @xmath104 ) shown by symbols connected by dashed lines for @xmath95 ( @xmath25 ) , @xmath19 ( + ) , and @xmath102 ( @xmath23 ) . @xmath105 and @xmath106 with @xmath107
are denoted by the solid lines , which agree with the simulation data ( see text ) .
, title="fig:",scaledwidth=23.0% ] in fig .
[ constitutive ] , the energy dissipation rate @xmath16 ( a ) and the normal stress @xmath1 ( b ) are shown for various values of the packing fraction @xmath3 and the restitution coefficient @xmath15 .
for the normal stress , we find in the simulation that @xmath108 depends on the direction @xmath109 , but the differences among them are at most @xmath110 in the plotted region and are not significant compared to the difference from the kinetic theory that we will study in the following .
thus , here we plot the average @xmath111 .
the value from the kinetic theory are shown in fig . [ constitutive](a ) and ( b ) by symbols connected by dashed lines .
we see in fig .
[ constitutive](a ) that the energy dissipation rate is overestimated by the theory in the dense region , and the disagreement is larger for smaller @xmath15 .
the normal stress in fig . [ constitutive](b ) also shows a similar tendency , although the relative disagreements are smaller than those in the energy dissipation rate @xmath16 . [ [ the - energy - dissipation . ] ] the energy dissipation .
+ + + + + + + + + + + + + + + + + + + + + + + we first focus on the discrepancy in the energy dissipation rate @xmath16 . from the collision rule eq .
( [ colrule ] ) , the energy dissipated per collision is given by @xmath112 where @xmath113 $ ] is the relative normal velocity of colliding particles just before the collision .
thus , @xmath16 is given by @xmath114 here , @xmath115 denotes the average of a quantity @xmath116 over all collisions ; if the value of @xmath116 is @xmath117 at the @xmath118-th collision , @xmath119 , where @xmath120 is the total number of collisions .
note that eq .
( [ eq : gammaex ] ) is the exact expression for @xmath16 . on the other hand , the expression ( [ dissipation ] ) from the kinetic theory with eq .
( [ g0 m ] ) gives @xmath121\cdot \frac{1}{2 } n\omega_0(\nu , t ) .
\label{gammakinetic}\ ] ] to interpret this expression , let us consider the random collision of particles whose velocity fluctuation is given by the maxwellian . in this case , @xmath122 , then eq . (
[ eq : gammaex ] ) gives @xmath123 the difference between this and eq .
( [ gammakinetic ] ) comes from the deviation of the velocity distribution from the maxwellian , but the difference is found to be small in the parameter region studied in the present paper .
therefore , from the comparison of the exact expression ( [ eq : gammaex ] ) with the kinetic theory expression eq .
( [ gammakinetic ] ) , we conclude that the deviation found in fig .
[ constitutive](a ) comes from the fact that @xmath124 .
[ [ the - collisional - temperature . ] ] the collisional temperature .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + and `` the collisional temperature '' @xmath125 vs. the packing fraction @xmath3 . @xmath57 and @xmath125 are denoted by @xmath18 and @xmath23 with the dashed lines for @xmath81 , respectively , and by @xmath20 and @xmath24 with the dashed lines for @xmath126 .
the inset shows @xmath57 and @xmath125 for @xmath95 represented by @xmath22 and @xmath25 with the dashed lines.,scaledwidth=23.0% ] to confirm this idea , we define `` the collisional temperature '' @xmath127 .
figure [ temperature ] shows @xmath125 and @xmath57 as functions of @xmath3 .
one can see that @xmath125 is substantially smaller than @xmath57 for @xmath128 as is concluded above . to demonstrate that the discrepancy is actually resolved by @xmath125 , we plot @xmath129 of eq .
( [ dissipation ] ) with @xmath130 of ( [ g0 m ] ) in @xmath131 replacing @xmath57 by @xmath125 ( the solid lines in fig.[constitutive](a ) ) ; this is equivalent to replace @xmath57 in eq .
( [ gammakinetic ] ) with @xmath125 and use the measured value of the collision frequency for @xmath97 .
the agreement is quite good .
[ [ normal - stress . ] ] normal stress .
+ + + + + + + + + + + + + + now , we consider the effect of @xmath132 on the normal stress @xmath1 .
the value of @xmath133 should also play an important role in the collisional component of the normal stress @xmath120 , because @xmath133 directly determines the momentum transfer from the particle @xmath27 to the particle @xmath30 through a collision : @xmath134\ c_{n , ij}\bm n_{ij}=-\delta \bm p_i$ ] .
thus , we expect that @xmath120 is approximately proportional to @xmath135 .
in addition , the collisional part @xmath120 is dominant in the dense region . in fig .
[ constitutive](b ) , @xmath106 of eq .
( [ pressure ] ) is plotted by the solid lines , where @xmath107 is used as @xmath55 .
we see that the solid lines show reasonably good agreement with the data for the whole density region . .
@xmath136 and @xmath137 are compared along with @xmath138 and @xmath139 for @xmath126 with @xmath140 ( a ) and @xmath92 ( b ) .
we see that the difference between @xmath136 ( @xmath22 ) and @xmath137 ( @xmath141 ) is small for @xmath140 , but for @xmath92 , the @xmath136 has sharper distribution .
the solid lines show @xmath138 and the dashed lines show @xmath139 .
see text for details .
, title="fig:",scaledwidth=23.0% ] .
@xmath136 and @xmath137 are compared along with @xmath138 and @xmath139 for @xmath126 with @xmath140 ( a ) and @xmath92 ( b ) .
we see that the difference between @xmath136 ( @xmath22 ) and @xmath137 ( @xmath141 ) is small for @xmath140 , but for @xmath92 , the @xmath136 has sharper distribution .
the solid lines show @xmath138 and the dashed lines show @xmath139 .
see text for details .
, title="fig:",scaledwidth=23.0% ] ( @xmath22 ) and @xmath137 ( @xmath141 ) compared along with @xmath138 ( solid line ) and @xmath139 ( dashed line ) , respectively , for @xmath17 with @xmath92 .
, scaledwidth=23.0% ] one of the possible origins of the pre - collisional velocity correlation that makes @xmath132 is the inelasticity , which makes the relative normal velocity smaller upon collision . in this subsection
, we examine how the pre - collisional velocity correlation develops in the shear flow .
it is expected that the correlation grows when particles collide with same colliding partners inelastically many times within a short period of time . under the shear , however , this correlation will be lost when they are forced to pass each other and collide with new partners . the typical time scale that a pair of particles pass
each other is the unit time , i.e. , @xmath9 .
this argument explains the smaller @xmath125 in the denser region , because particles collide more frequently with same partners before they move far apart @xcite .
this argument tells that the collision does not have memories of the previous collisions earlier than the unit time @xmath9 . to confirm this
, we compare the following two distributions of the pre - collisional velocity : ( i)@xmath142 , which is the distribution of @xmath133 for all collisions between all pairs of particles , and ( ii ) @xmath143 , which is the distribution of @xmath133 of the collisions whose colliding pairs of particles did not collide with each other during the last unit time @xmath9 .
if the velocity correlation mainly comes from the multiple collision with same partners within the unit time scale , then @xmath137 should have the width determined not by @xmath125 but by the average temperature @xmath57 .
the results are shown in fig .
[ colfreq ] for @xmath126 with @xmath140 ( a ) and @xmath144 ( b ) , where @xmath142 is denoted by @xmath22 and @xmath145 is denoted by @xmath141 .
we see that @xmath143 is wider than @xmath142 for the denser case ( b ) .
if the particles with the maxwellian velocity distribution with temperature @xmath146 collide among themselves randomly , the distribution of @xmath147 is given by @xmath148 .
\label{prand}\ ] ] in fig .
[ colfreq ] , @xmath139 and @xmath138 are shown by the dashed and solid lines , respectively .
they are indistinguishable for @xmath140 in fig .
[ colfreq](a ) , but show clear difference for @xmath92 in fig .
[ colfreq](b ) .
we find that @xmath139 fits @xmath145 , and @xmath138 fits @xmath149 , which further confirms that colliding partners are correlated in the way characterized by @xmath125 .
for smaller @xmath15 , the shape of the distributions deviates from eq .
( [ prand ] ) based on the random collision .
figure [ colfreq7 ] shows @xmath136 and @xmath137 compared along with @xmath138 and @xmath139 , respectively , for @xmath17 with @xmath92 .
@xmath136 has sharper distribution than @xmath137 , but neither of them fit well with @xmath138 nor @xmath139 .
this suggests stronger correlation than the case of @xmath126 . and @xmath150 .
( a ) the correlations in the @xmath71 direction @xmath151 .
one sees that the longitudinal component @xmath152 ( solid line ) has the larger amplitude than others ( shown by dashed lines ) .
( b ) the longitudinal velocity correlations @xmath153 for @xmath154 ( solid line ) , @xmath72 ( dashed line ) , and @xmath70 ( dotted line ) . ,
title="fig:",scaledwidth=23.0% ] and @xmath150 .
( a ) the correlations in the @xmath71 direction @xmath151 .
one sees that the longitudinal component @xmath152 ( solid line ) has the larger amplitude than others ( shown by dashed lines ) .
( b ) the longitudinal velocity correlations @xmath153 for @xmath154 ( solid line ) , @xmath72 ( dashed line ) , and @xmath70 ( dotted line ) . ,
title="fig:",scaledwidth=23.0% ] to understand the the velocity correlations in more detail , we study the spatial velocity correlation function defined as @xmath155 > } { < \sum_{i , j}\left [ \delta(r-|(\bm r_i-\bm r_j)\cdot \bm e_\gamma| ) \right]>},\ ] ] where @xmath109 , @xmath156 , and @xmath157 take @xmath158 or @xmath70 , @xmath159 , @xmath160 denotes the time average , and @xmath161 represents the unit vector in the @xmath157 direction .
@xmath162 is one when @xmath163 and @xmath164 for @xmath165 , and it is zero otherwise .
we calculated @xmath166 for the system with @xmath126 , both for the small system with @xmath167 and the large system with @xmath168 .
we find that the correlation extends over the whole system in the case of the small system , but it goes to zero for the large system . in the following , we present the spatial correlation measured in the large system , but we confirmed that the hydrodynamic quantities presented in the previous subsections did not show any differences . in fig . [ correlation](a )
, the various components of the correlation in the @xmath71-direction @xmath151 are shown .
we find that the longitudinal correlation in the @xmath71-direction , @xmath152 , has larger amplitude than other components ; this tendency is also found in the @xmath72- and @xmath70-direction ( data not shown ) .
the longitudinal correlation at the particle diameter distance ( @xmath169 ) is positive , which is consistent with the fact that @xmath132 .
it is evident that the correlation shows an oscillation , whose wavelength is order of the particle diameter , which will be discussed in section [ discussion ] .
the longitudinal components in @xmath71 , @xmath72 and @xmath70 directions are shown in fig .
[ correlation](b ) .
all of them show oscillations in the particle diameter scale .
we also found that the longitudinal correlation shows larger amplitude for smaller restitution coefficient @xmath15 and/or larger packing fraction @xmath3 ( data not shown ) .
we find that the shear stress @xmath0 shows more complicated packing fraction @xmath3 dependence than those of the energy dissipation rate @xmath16 and the normal stress @xmath1 . in fig.[shearfig ] , the simulation data of the shear stress @xmath0 are denoted by symbols , and @xmath170 from the kinetic theory ( eq .
( [ shear ] ) with table [ constitutiveeq ] ) are denoted by symbols with the dashed lines .
we find that , for @xmath95 , the shear stress is _ underestimated _ by the theory , while for @xmath126 and @xmath102 , the shear stress is _
overestimated_. actually , in the case of the elastic ( @xmath38 ) hard sphere system , the enskog theory is known to underestimate the shear viscosity in dense region @xcite , and this tendency is seen in the result for @xmath95 .
the results for @xmath17 shows that the inelasticity reduces the shear stress to the value smaller than the one expected from the kinetic theory , but we do not understand the reason of this reduction yet . rather good agreement in between for the case of @xmath171 seems to be accidental . vs. the packing fraction @xmath3 .
the simulation data are plotted for @xmath95 ( @xmath22 ) , @xmath19 ( @xmath20 ) , and @xmath102 ( @xmath18 ) .
the kinetic theory constitutive relations @xmath170 is shown by the symbols connected by dashed lines ( @xmath25 for @xmath95 , @xmath24 for @xmath19 , and @xmath23 for @xmath17 ) .
, scaledwidth=23.0% ]
in contrast to the energy dissipation rate @xmath16 and the normal stress @xmath1 , the discrepancy in the shear stress @xmath0 can not be understood just by the pre - collisional velocity distribution averaged over all directions , but the anisotropy of the pre - collisional correlations in both the velocity and the position should by important in the shear stress .
these anisotropies are not taken into account in the kinetic theory employed in the preset analysis .
in fact , for the soft - sphere system in two - dimensional , sheared flow , it has been found that the contact force distribution strongly depends on direction @xcite .
our preliminary results also show a similar direction dependence in the collisional momentum transfer per unit time .
the detailed analysis is left for future studies .
[ [ the - packing - fraction - dependence - of - the - ratio - of - the - shear - stress - to - the - normal - stress ] ] the packing fraction dependence of the ratio of the shear stress to the normal stress ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ as we have seen in fig .
[ friction_packing1 ] , @xmath68 in the simulation is a decreasing function of the packing fraction @xmath3 for larger packing fraction @xmath3 and/or smaller restitution coefficient @xmath15 , while eq .
( [ soverp ] ) , from the kinetic theory , @xmath68 always increases with @xmath3 .
kumaran argued that the particle roughness is necessary for @xmath68 to have a decreasing part upon increasing @xmath3 in the dense region @xcite .
however , even for the smooth particles , the present simulations show that @xmath68 has a decreasing part in the dense region for the inelastic hard sphere system , although the particle roughness may well amplify the decreasing part of @xmath68 .
the present authors have suggested @xcite that the origin that leads the kinetic theory to the increasing @xmath68 on @xmath3 even for the denser region is that @xmath131 in the energy dissipation @xmath16 of eq .
( [ dissipation ] ) increases too sharply for larger @xmath3 . in this paper in section [ precol ] , we showed that the sharp increase in @xmath16 can be weaken by using @xmath125 instead of @xmath57 . in the present treatment , however , it is not possible to extract the @xmath3-dependence out of @xmath172 and to compare it directly with @xmath131 because @xmath57 and @xmath125 are determined by @xmath3 and @xmath5 in the steady state simulations , therefore , the quantity that corresponds to @xmath131 in eq .
( [ dissipation ] ) can not be defined from the simulation data .
finally , let us comment on the fact that @xmath68 does increase with @xmath3 in a certain parameter range in our simple shear flow simulation , in contrast to the fact that the increasing @xmath3 upon increasing @xmath11 has never been observed in the granular flow down a slope .
this suggests that the steady flow in this parameter region is unstable in the slope flow configuration .
it is interesting to study the relation between the stability of the flow and the @xmath3 dependence of @xmath68 . as shown in fig .
[ correlation ] , the spatial velocity correlation is found to oscillate in the scale of the particle diameter .
although we have not yet understood the origin of this oscillation , it is plausible that the oscillation comes from the coupling between the density correlation and the velocity correlation .
analysis on the sheared langevin system suggests that the spatial velocity correlation is related to the radial distribution function @xcite , which oscillates in the particle diameter scale .
it is likely that similar coupling also exists in the granular shear flow .
we have simulated the simple shear flow of the smooth inelastic hard sphere system by molecular dynamics simulations .
we have found that the energy dissipation rate @xmath16 and the normal stress @xmath1 are smaller than those expected from the kinetic theory .
we have showed that the relative pre - collisional normal velocity of colliding pairs of particles , @xmath133 , is smaller than the one expected from random collisions , and this reduces @xmath16 and @xmath1 . by examining the distributions of @xmath133 for all collisions ( @xmath136 ) and for only the first collisions of the new pairs during the last period of time @xmath9 ( @xmath137 ) ,
we have concluded that the reduction of the relative velocity is caused by the multiple inelastic collisions during the time period @xmath9 . to understand the velocity correlation in more detail , we have studied the spatial velocity correlation .
it has been found that the longitudinal components of the correlations have larger amplitude with the oscillation in the scale of the particle diameter .
nm thanks a. yoshimori for his insightful discussion on the spatial velocity correlation in the langevin system .
nm is supported in part by the inamori foundation .
part of this work has been done when nm was supported by grant - in - aid for young scientists(b ) 17740262 from the ministry of education , culture , sports and technology ( mext ) , and hn and nm were supported by grant - in - aid for scientific research ( c ) 16540344 from japan society for the promotion of science ( jsps ) .
van noije , m.h .
ernst , r. brito and j.a.g .
orza , phys .
lett . * 79 * , 411 ( 1997 ) ; t.p.c .
van noije , m.h .
ernst , and r. brito , pysica a , * 251 * , 266 ( 1998 ) ; t.p.c .
van noije and m.h .
ernst , pys .
e * 61 * , 1765 ( 2000 ) .
experimentally , the spatial `` velocity correlation '' was studied by pouliquen @xcite in the granular flow down a slope . in the experiments
, however , the velocity is measured from displacement between certain time intervals ( limited by camera frames ) , and it is different from the simultaneous velocity correlations studied in the present paper . in the case of the small restitution coefficient ( e.g. @xmath81 ) , the system often encounters the inelastic collapse @xcite before reaching the steady state unless we start the simulation with the initial mean flow . with the initial mean flow ,
the system may reach a steady shear flow , but the inelastic collapse may eventually occur even with shear @xcite . in this paper , we focus on the results of the simulations where the inelastic collapse did not occur within the run . in the denser region of the shear flow ,
i.e. with smaller free volume , the particles tends to move collectively and this may also create the velocity correlation in the dense regime even without inelasticity . in any case , the collective motion is broken by the shear , and the time scale that the shear breaks the correlation is about the unit time ( @xmath9 ) .
|
we study the effect of pre - collisional velocity correlations on granular shear flow by molecular dynamics simulations of the inelastic hard sphere system .
comparison of the simulations with the kinetic theory reveals that the theory overestimates both the energy dissipation rate and the normal stress in the dense flow region .
we find that the relative normal velocity of colliding particles is smaller than that expected from random collisions , and the discrepancies in the dissipation and the normal stress can be adjusted by introducing the idea of the collisional temperature , from which we conclude that the velocity correlation neglected in the kinetic theory is responsible for the discrepancies .
our analysis of the distributions of the pre - collisional velocity suggests that the correlation grows through multiple inelastic collisions during the time scale of the inverse of the shear rate . as for the shear stress ,
the discrepancy is also found in the dense region , but it depends strongly on the particle inelasticity .
|
the theoretical study of the formation and evolution of regions in non uniform media is motivated by the observational result that the molecular clouds where massive star form generally possess density gradients ( e.g. , * ? ? ? * ) . the assumed density law is taken to be a power law @xmath2 with exponents ranging from 1 to 3 @xcite obtained from molecular line studies .
submillimeter continuum dust emission observations of the molecular material around embedded ultracompact regions gives an exponent in the range 1.25 to 2.25 .
the most commonly found value for the power - law exponent is @xmath3 , which correponds to an isothermal , self - gravitating sphere .
however , steeper density laws , @xmath4 have been inferred from radio continuum spectra of a sample of ultracompact regions @xcite .
numerical studies of region expansion near the edge of a molecular cloud ( the so - called `` champagne flows '' ) , were carried out by tenorio - tagle and collaborators . in this model , an region formed near
the edge of the cloud `` breaks out '' into the low density , intercloud medium during the expansion stage and the ionized gas can achieve supersonic velocities of around 30 km s@xmath5 in the accelerating flow .
this problem was studied in more detail by @xcite , and applied to the orion nebula region .
they found that a long - lived quasi - stationary phase exists in which the ionized flow is approximately steady in the frame of reference of the ionization front . during this phase ,
the flow structure is determined entirely by the distance of the ionizing star from the front and the curvature of the front .
the curvature determines whether the flows are more `` champagne like '' or `` globule like '' .
the curvature of the ionization front depends on the lateral density distribution in the neutral gas
. developed analytical solutions for the formation and expansion of regions in spherically symmetric , power - law density distributions , where the ionizing star is assumed to be at the center of the density distribution .
the main result of their paper is that for spherical clouds with power - law density distributions @xmath6 , there is a critical value of the exponent , @xmath7 above which the cloud becomes completely ionized during the expansion phase of the region . for power - law exponents higher than this critical value
, two regimes were identified : a slow regime , in which the expansion of the dense cloud core is mildly supersonic and has almost constant velocity , corresponding to @xmath8 , and the fast regime , where the core expansion drives a strongly accelerating shock , which corresponds to @xmath9 .
this problem was revisited by @xcite , who developed a self - similar solution for the internal motions in the ionized gas in the @xmath10 cases .
also considered self - gravitating isothermal disk density distributions where the density fall - off is steeper than @xmath11 . in this case , the region becomes unbounded in a conical section of the disk , where the opening angle of the cone depends on the ratio of the strmgren radius in the midplane density to the density distribution scale height . during the expansion stage of such regions ,
the flattening of the internal ionized gas density distribution can lead to the trapping of an initially unbounded ionization front .
more recently , photoionized regions in a variety of axisymmetric density distributions were modeled numerically by @xcite , who also included the central star s stellar wind . in that work
, it was found that the hot bubble created by the shocked stellar wind results in a champagne flow being set up outside the swept - up stellar wind shell , leading to mixed kinematics from an observational point of view .
@xcite also considered the possibility of stellar motion up the density gradient , which introduces the added complication of bowshock kinematics to the problem .
@xcite performed low - resolution three - dimensional numerical simulations of region dynamical evolution in a collapsing molecular cloud . here
, the density distribution had a core - envelope structure , where the envelope had a power - law ( @xmath0 ) density distribution .
both turbulent and non - turbulent cases were modeled . in the non - turbulent case ,
even for ionizing sources off - center from the core , the region expansion was nearly spherical for the parameters adopted and the timescale of the simulation .
the photoionized regions produced by moving stars were studied numerically by @xcite . in that paper , the regions produced by a star off - center in a spherically symmetric density distribution were modeled , where a range of stellar velocities ( between 0 and 12 km s@xmath5 ) were considered . it was found that the region produced by a star off - center in an underlying @xmath0 density distribution did not become unbounded , while that in a @xmath1 density distribution did , for the case of a stationary star .
this appears to contradict the main result of @xcite .
this result is the motivation for the present work , where we aim to provide simple tools for analyzing the formation and expansion of regions off - center in spherically symmetric density distributions .
the structure of this paper is as follows : in [ sec : prob ] we outline the problem and discuss the initial formation stage of an region off - center in a spherically symmetric density distribution and find a criterion for the region to remain bounded . in [ sec : expand ] we investigate the expansion stage and derive a simple differential equation that describes the expansion of the ionization front along the symmetry axis .
we assess the validity of our assumptions about the internal structure of the ionized gas by comparing our analytical model with the results of numerical simulations in
[ sec : numerical ] .
finally , in [ sec : summary ] we summarize our results .
from the density center . for a parcel of gas at x , distance @xmath12 from the density center ,
the star - centered coordinates are @xmath13 , where @xmath14 is the distance from the star and @xmath15 is measured clockwise from the extension of the line joining the star to the density center .
rotational symmetry is assumed around the axis joining the star to the center of the density distribution .
the position of the ionization front in this coordinate system is @xmath16 .
an approximate shape for the ionization front is also shown.,scaledwidth=45.0% ] the underlying , spherically symmetric density distribution is described by @xmath17 where @xmath18 is the number density at the position of the star , @xmath19 .
we now change our reference system and rewrite the density distribution in axisymmetric coordinates centered on the star position , with the distance from the star , @xmath14 , found from @xmath20 where @xmath15 is measured from the symmetry axis and the center of the density distribution is in the direction @xmath21 , see figure [ fig : geom ] .
the density distribution of equation [ eq : denorig ] , in terms of this new reference system , @xmath13 , becomes @xmath22 along the symmetry axis , @xmath23 , this can be written @xmath24 where we have defined @xmath25
. for @xmath26 , expansion to linear terms in a taylor series shows that @xmath27 which is no longer a power law . for @xmath28
we find @xmath29 , that is , the offset density distribution resembles the original power - law density distribution .
consequently , if the initial strmgren radius , @xmath30 , is smaller than the stellar offset distance , then we expect the initial region to remain confined , even for power laws steeper than @xmath10 .
we can develop a quantitative criterion for deciding whether the initial photoionized region will remain bounded in an offset power - law density distribution by balancing recombinations and ionizations , assuming pure hydrogen , along the symmetry axis @xmath23 : @xmath31 where @xmath32 is the case b recombination coefficient .
this equation can be written @xmath33 where @xmath34 is the non - dimensionalized strmgren radius for a uniform medium of number density @xmath18 , i.e. , @xmath35 in the `` uphill '' direction , @xmath21 , the equivalent non - dimensionalized ionization balance equation is @xmath36 equations [ eq : nondimstrom ] and [ eq : uphill ] define integral equations for the extent of the ionized region , @xmath37 , along @xmath23 and @xmath21 , respectively , as functions of @xmath34 and @xmath38 . to uniform density strmgren radius , @xmath39 , as a function of uniform density strmgren radius , @xmath34 , for power - law density distributions with exponent @xmath3 ( _ solid line _ ) and @xmath40 ( _ dashed line _ ) .
the asymptotes are @xmath41 for @xmath42 and @xmath43 for @xmath44 .
, scaledwidth=45.0% ] integrating equation [ eq : iontidy ] for @xmath10 we obtain @xmath45^{3 - 2\alpha } ) \nonumber \\ & - \frac{2}{(1 - 2\alpha)(2 - 2\alpha)}(y[1+y]^{2 - 2\alpha } ) \nonumber \\ & + \frac{1}{(1 - 2\alpha)}(y^2[1+y]^{1 - 2\alpha } ) \ .
\label{eq : intresult}\end{aligned}\ ] ] in general , this yields a positive , real root for @xmath37 ( the non - dimensionalized ionization front radius ) as long as @xmath46 for example , when @xmath3 , the requirement is @xmath47 , while for @xmath44 , the condition is @xmath48 , i.e. , @xmath49 , in order that there be a real , positive root for @xmath37 . in figure
[ fig : newt ] we plot the ratio @xmath39 against @xmath34 for @xmath34 values between 0.01 and 1.0 for the two examples above , @xmath3 and @xmath44 .
we see that for @xmath34 less than the critical value , the value of the root @xmath37 is not too different from @xmath34 , but close to the critical value @xmath50 .
in order to study the expansion of the ionized gas after the initial formation of the photoionized region we follow the same analysis as @xcite and @xcite . as the region expands , a neutral shock is sent out ahead of the ionization front into the ambient medium . across the isothermal shock ,
the relevant equations are mass conservation @xmath51 and momentum conservation @xmath52 where the subscript @xmath53 refers to preshock conditions and @xmath54 refers to postshock conditions and all velocities are in the frame of the shock .
we also have the equation for the density jump across a weak - d ionization front @xmath55^{1/2 } \right\ } \label{eq : ifront}\ ] ] where the subscript @xmath56 refers to conditions conditions in front of the ionization front and @xmath57 refers to conditions in the photoionized gas just behind the ionization front . here
, @xmath58 is the sound speed in the neutral gas and @xmath59 is the sound speed in the ionized gas , and @xmath60 refers to the velocity of the gas ahead of the ionization front in the frame of the ionization front .
furthermore , we assume that the region between the shock front and the ionization front is thin , i.e. @xmath61 , where @xmath62 and @xmath63 are the velocities of the gas ahead of the ionization front and behind the shock front in the frame of the star , and that the density of the ionized gas inside the photoionized region is spatially uniform but decreases with time , i.e. @xmath64 .
this is valid if the sound - crossing time of the ionized region is short compared to the expansion timescale . as a first approximation to the non - spherical shape of the ionization front in a non - uniform density distribution
, we assume that the radius of the ionization front can be described by @xmath65 to first order in @xmath66 , where @xmath67 and @xmath68 with @xmath69 corresponding to the shell radius when @xmath70 and @xmath71 to the case @xmath72 . the volume contained within the ionization front @xmath16 is @xmath73 which in this approximation becomes @xmath74 assuming uniform expansion , then the mass contained within a volume , @xmath75 , interior to an internal shell of radius @xmath14 must be conserved during the expansion .
that is , @xmath76 and the total volume of ionized gas satisifies the ionization balance equation , where the ionizing photon output of the star is assumed to be constant , that is @xmath77 taking the time derivatives of equations [ eq : shellmass ] and [ eq : ibalance ] we find that @xmath78 where we have evaluated the derivative for a shell located just inside the ionization front radius . writing @xmath79
then we find that the gas velocity behind the ionization front is half the ionization front velocity , @xmath80 , where @xmath81 and @xmath82 .
this assumes that the expansion is uniform and that there are no lateral gas motions within the photoionized region . with the additional assumption that @xmath83 , then equations [ eq : cont ]
, [ eq : mom ] , [ eq : ifront ] , together with @xmath80 , can be combined to give @xmath84 where we have used the thin shell approximation @xmath85 .
these assumptions are the same as those of @xcite .
the density ratio can be found from ionization balance @xmath86 where @xmath87 is the initial strmgren radius formed in a medium of uniform mass density @xmath88 .
we also have the offset power - law initial density distribution , written here for @xmath89 : @xmath90 where @xmath19 is the offset distance of the star from the center of the density distribution and @xmath88 is the density at this point .
thus @xmath91 in the directions @xmath92 along the symmetry axis .
equation [ eq : denrat ] can be written in the form @xmath93 where @xmath94 .
this constitutes a differential equation for the radius of the ionization front , @xmath16 , as a function of time and can be solved numerically with standard techniques ( e.g. , * ? ? ?
* ) , by considering directions @xmath95 and @xmath70 together and using equations [ eq : radius ] , [ eq : a0 ] , [ eq : a1 ] , [ eq : volume ] and [ eq : denrat2 ] , with initial conditions @xmath96 when @xmath97 , @xmath98 and @xmath99 . here
, it is assumed that @xmath87 is much less than the critical value found from solving equation [ eq : root ] above , and so the initial ionized region radius ( @xmath16 at @xmath97 ) is taken to be that for a uniform medium . ) against time ( in units of the sound crossing time , @xmath100 ) .
_ top _ : @xmath72 direction . _
bottom _ : @xmath70 direction .
density power laws are @xmath101 ( _ solid line _ ) , @xmath102 ( _ dashed line _ ) , @xmath3 ( _ dot - dashed line _ ) , and @xmath44 ( _ dotted line_).,title="fig:",scaledwidth=45.0% ] + ) against time ( in units of the sound crossing time , @xmath100 ) . _ top _ : @xmath72 direction .
_ bottom _ : @xmath70 direction .
density power laws are @xmath101 ( _ solid line _ ) , @xmath102 ( _ dashed line _ ) , @xmath3 ( _ dot - dashed line _ ) , and @xmath44 ( _ dotted line_).,title="fig:",scaledwidth=45.0% ] direction . _
bottom _ : @xmath70 direction .
linetypes same as in fig .
[ fig : radupdown].,title="fig:",scaledwidth=45.0% ] + direction . _
bottom _ : @xmath70 direction .
linetypes same as in fig .
[ fig : radupdown].,title="fig:",scaledwidth=45.0% ] in figures [ fig : radupdown ] and [ fig : velupdown ] we can see the result of solving equation [ eq : emsq ] numerically for the cases @xmath101 , 2 and 3 . in these figures , all distances are scaled to @xmath19 and times are scaled with the sound crossing time of this distance in the ionized gas , @xmath100 .
the results presented are particular to the case @xmath103 , which corresponds to an ionizing source with @xmath104 photons s@xmath5 and a density @xmath105 @xmath106 , where we have also assumed that the case b recombination coefficient takes the value @xmath107 @xmath108 s@xmath5 .
these values of the parameters are motivated by @xcite . in the `` upward '' direction ( @xmath70 , density increasing ) ,
as expected the radius grows more slowly in the @xmath3 and @xmath109 cases than for the uniform medium , classical case @xmath101 , as shown in figures [ fig : radupdown ] and [ fig : velupdown ] . in the `` downward '' direction , @xmath72
, figures [ fig : radupdown ] and [ fig : velupdown ] clearly show that in the @xmath44 case , the velocity starts to increase after about 5 sound crossing times and by 100 sound crossing times the velocity of the ionization front has become supersonic with respect to the ionized gas .
this means that the ionization front has overtaken the shock wave and will ionize out to infinity . in the @xmath3 case , although the velocity does start to increase again after about 13 sound crossing times , the increase is only very gradual , and within the timescale studied the ionization front does not become supersonic with respect to the ionized gas . direction .
_ bottom _ : @xmath70 direction .
linetypes same as in fig .
[ fig : radupdown ] .
the horizontal line marks equal pressures.,title="fig:",scaledwidth=45.0% ] + direction . _
bottom _ : @xmath70 direction .
linetypes same as in fig .
[ fig : radupdown ] .
the horizontal line marks equal pressures.,title="fig:",scaledwidth=45.0% ] in figure [ fig : presupdown ] we show the ratio of the pressure in the ionized gas to that in the ambient medium , assuming that the latter is isothermal with a temperature of 100 k. from these figures we see that in the @xmath70 direction the pressures equalize in about 1 sound crossing time for both @xmath3 and @xmath44 . in the @xmath72 direction
, the @xmath44 case shows that the pressures do not equalize before the ionization front begins to accelerate . in the @xmath3 case
, the pressures do equalize at around 7 sound crossing times for the adopted parameters ( i.e. , initial strmgren radius , @xmath87 ) .
we remark that once the pressures become similar , the analysis of this section does not really hold , since the external pressure should be taken into account .
however , we also comment that the fact that the pressures do become equal before the shock begins to accelerate in the @xmath3 case means that , in practice , the region will remain confined for this choice of initial strmgren radius . , 0.02 , 0.05 , 0.1 , 0.2 , 0.5 . _ bottom _ : ratio of internal to external pressures for the same initial strmgren radius values .
the horizontal line denotes equal pressures.,title="fig:",scaledwidth=45.0% ] + , 0.02 , 0.05 , 0.1 , 0.2 , 0.5 .
_ bottom _ : ratio of internal to external pressures for the same initial strmgren radius values .
the horizontal line denotes equal pressures.,title="fig:",scaledwidth=45.0% ] in figure [ fig : multidown ] we show how the breakout time of the ionization front in the @xmath72 direction varies with initial strmgren radius , for values of @xmath110 between 0.01 and 1.0 for the case @xmath3 . as @xmath111 the position of the velocity minimum moves to earlier times and higher velocities
also in this figure , we show how the internal to external pressure ratio varies with initial strmgren radius . only for @xmath112
is there a regime for which pressure balance could possibly halt the expansion and subsequent breakout of the region .
numerical solution after 60,000 yrs . _
bottom _ : @xmath44 numerical solution after 50,000 yrs .
the black triangle marks the position of the star and the position @xmath113 denotes the center of the density distribution in each case .
the heavy black curve represents the shape of the ionization front obtained from the approximation of equation [ eq : radius ] . , title="fig:",scaledwidth=45.0% ] + numerical solution after 60,000 yrs . _ bottom _ : @xmath44 numerical solution after 50,000 yrs .
the black triangle marks the position of the star and the position @xmath113 denotes the center of the density distribution in each case .
the heavy black curve represents the shape of the ionization front obtained from the approximation of equation [ eq : radius ] .
, title="fig:",scaledwidth=45.0% ] in order to test our assumption that the density within the ionized gas is approximately constant during the expansion stage , we perform numerical simulations of the expansion of an off - center region in a power - law density distribution .
the density law for this simulation is taken from @xcite , namely @xmath114 & \mbox{for $ r \leq r_c$ } , \\
n_0 e^{-\alpha/2 } \left ( r / r_c \right)^{-\alpha } & \mbox{for $ r \geq r_c$ } , \end{array } \right .
\label{eq : numsimden}\ ] ] where @xmath115 is the central density and @xmath19 is the stellar offset distance from the center of the density distribution . at @xmath19
the density has the value @xmath116 .
this density distribution is flatter for radii @xmath117 than the simple power - law density distribution discussed in
[ sec : prob ] and [ sec : expand ] above .
it is obtained by assuming that the acceleration due to gravity of the hydrostatic equilbrium solution in the halo , @xmath118 , matches that of the core ( defined as @xmath117 ) at @xmath119 , with the added requirement that the acceleration due to gravity must be zero at the center of symmetry .
this particular density law assumes that the acceleration due to gravity is linear with radius for @xmath117 .
we adopt this density distribution rather than a simple power law because the density does not go to infinity at the center of symmetry and the `` core - halo '' structure is more representative of real molecular clouds . although we do not expect the ionization front expansion of our numerical simulations to agree with the analytic expansion found in
[ sec : expand ] in the region between the stellar position and the center of the density distribution ( i.e. , @xmath117 ) , because the density laws are very different here , we do expect to be able to make a direct comparison in the halo region , @xmath118 , where the density laws are the same .
since we are primarily interested in whether the region remains bounded in the @xmath72 direction during the expansion stage , then it is sufficient that the density distributions are the same in this direction for radii @xmath118 .
the radiation - hydrodynamics code employed in these simulations has been described previously in @xcite .
the model parameters specific to this work are a central density @xmath120 @xmath106 , and distance from density center to the star of @xmath121 pc .
we assume isothermal , hydrostatic equilibrium in the initially neutral 100 k gas , and include an external gravity field ( see @xcite for details ) .
the gravitational acceleration does not have any noticeable dynamical effect on the expansion of the region .
the ionizing star produces @xmath122 photons s@xmath5 .
the grid size for these calculations is @xmath123 cells , representing a spatial size of @xmath124 pc . for these parameters , @xmath125 for the @xmath3 case and @xmath126 for the @xmath44 case .
axis of the @xmath3 numerical simulation shown in fig .
[ fig : numden23 ] . _
top _ : density .
_ middle _ : velocity .
_ bottom _ : pressure .
the thick solid line represents the @xmath127 direction and the thin solid line represents the @xmath70 direction .
units are as indicated on the axes.,scaledwidth=45.0% ] axis of the @xmath44 numerical simulation shown in fig .
[ fig : numden23 ] . _
top _ : density . _
middle _ : velocity .
_ bottom _ : pressure .
the thick solid line represents the @xmath127 direction and the thin solid line represents the @xmath70 direction .
units are as indicated on the axes.,scaledwidth=45.0% ] in figures [ fig : numden23 ] , [ fig : numcut2 ] and [ fig : numcut3 ] we show the results of calculations of off - center regions in @xmath3 and @xmath44 power - law density distributions .
figure [ fig : numden23 ] shows the ionized gas density , i.e. the shape of the regions , for both power laws after 60,000 and 50,000 yrs respectively . the density within the ionized regions is approximately constant .
we note that for both models , the ionized volumes are approximately spherical , although the star is off - center .
this was also noted by @xcite in their three - dimensional numerical simulations .
overplotted on figure [ fig : numden23 ] is the shape the ionization front would have according to the approximation of equation [ eq : radius ] .
we can see that the approximation slightly underestimates the volume of the ionized gas with respect to the numerical simulations . in figures [ fig : numcut2 ] and [ fig :
numcut3 ] we show cuts along the symmetry axis , i.e. the @xmath92 directions , for both cases . from these figures
we see that the assumption of constant density is reasonable for the @xmath72 ( `` downhill '' ) direction but not so good in the @xmath70 direction , since the density increases with distance from the star .
the pressure follows the same pattern as the density , as is to be expected in for an isothermal , @xmath128 k region . outside the ionized region , in both directions ,
the pressure is high because the neutral shock sent out ahead of the ionization front pressurizes the gas .
the velocities in the ionized gas are subsonic in the @xmath70 direction and become supersonic in the @xmath72 direction .
the ripples in the @xmath70 subsonic gas are due to the ionization front not being well resolved in this case , since the high densities lead to a very thin transition zone . the spike in the pressure is due to the heating and cooling being out of equilibrium at the unresolved ionization front .
these numerical problems are not dynamically important for these simulations , since we are interested in the expansion in the @xmath72 direction , where the gas is supersonic and hence not affected by fluctuations in the thermal pressure .
this problem was noted by @xcite .
their solution to the problem was to reduce the effective ionization cross section , @xmath129 , by a factor of 60 . in this paper , we reduce @xmath129 by a factor of 10 .
reducing the ionization cross section makes the ionization front broader and so allows it to be resolved by the numerical scheme . since the width of the ionization front is of the order of a few photon mean free paths , where @xmath130 , then the ionization front is narrower in high density gas than in low density gas , and hence more of a problem to resolve numerically in the former . in the @xmath72 direction , the much lower densities mean that the ionization front is easily resolved by the numerical scheme .
direction , showing ionization front radius ( _ thick solid line _ ) , shock front radius ( _ thin solid line _ ) , analytical solution for the same parameters ( _ dot - dashed line _ ) and uniform density analytical solution ( _ dashed line _ ) .
_ top _ : @xmath3 numerical simulation and corresponding analytical solution . _
bottom _ : @xmath44 numerical simulation and corresponding analytical solution.,title="fig:",scaledwidth=45.0% ] + direction , showing ionization front radius ( _ thick solid line _ ) , shock front radius ( _ thin solid line _ ) , analytical solution for the same parameters ( _ dot - dashed line _ ) and uniform density analytical solution ( _ dashed line _ ) . _
top _ : @xmath3 numerical simulation and corresponding analytical solution . _ bottom _ : @xmath44 numerical simulation and corresponding analytical solution.,title="fig:",scaledwidth=45.0% ] in figure [ fig : numanal23 ] a comparison of the numerical simulation and the analytic solution for the same parameters in the @xmath72 direction is plotted . here
, the distance scale is in units of the strmgren radius in a medium of uniform density @xmath131 , and time is given in units of the sound crossing time of the strmgren sphere .
since the density distribution in the @xmath70 direction of the numerical simulation is different to the analytical model discussed in section [ sec : expand ] , we do not show a comparison graph for this case , as it is meaningless . instead , we focus on the @xmath72 direction , for which the numerical simulation and analytical model density distributions are the same .
we plot both the shock wave and the ionization front trajectories , together with the analytical result . for times less than about half a sound crossing time
, there is not a good agreement between analytical and numerical results .
this can be partly attributed to resolution effects , since the initial strmgren radius in this high density medium is very small , and the details of the hydrodynamics are not particularly well resolved at this scale . for times greater than half a sound crossing time , the agreement between numerical and analytical solutions is very close for the remainder of the calculation .
the analytical solution falls between the shock front and ionization front paths for the whole of this time .
we do not follow the numerical simulation beyond around 4 sound crossing times because part of the ionized gas volume leaves the computational grid at this point .
in this paper we have developed an analytical treatment for the formation and expansion of an region off center in a steep power - law density distribution .
we find that during the initial formation stage , the region will remain bounded as long as the ratio of the strmgren radius in the equivalent uniform medium at the stellar position to the stellar offset position is less than a value that depends only on the power - law exponent ( see eq .
[ eq : root ] ) . in the expansion phase
, the ionization front will break out ( turn supersonic with respect to the ionized gas ) unless pressure balance between the internal photoionized gas and the external medium is achieved before this point . for strmgren radius to stellar offset radius ratios of more than 0.02 ,
pressure balance is not possible for an isothermal 100 k medium in hydrostatic equilibrium and so the region will become unbounded , eventually , in these cases .
we performed numerical simulations to test the assumption of constant density within the ionized gas during the expansion phase adopted in [ sec : expand ] .
we found that this assumption is valid in the `` downhill '' ( @xmath23 ) direction from the star , but is not so good in the opposite ( `` uphill '' , @xmath21 ) direction . in the @xmath23 direction ,
the numerical solution and analytical solution follow the same expansion law once the elapsed time is greater than about half a sound crossing time .
we also assessed the validity of the simple analytical approximation to the ionization front shape .
although the volume of ionized gas is slightly underestimated in the analytical approximation , the effect is not significant , as evinced by the agreement between the numerical and analytic expansion laws .
this gives us confidence that our simple description of the region expansion can provide insight into this complicated problem .
we thank will henney for a critical reading of the manuscript and pepe franco and stan kurtz for valuable comments .
sja would like to acknowledge support from dgapa - unam through papiit grant in112006 .
this work has made use of nasa s astrophysics abstract data service and the astro - ph archive .
arquilla , r. , & goldsmith , p. f. 1985 , , 297 , 436 arthur , s. j. , & hoare , m. g. 2006 , , 165 , 283 bedijn , p. j. , & tenorio - tagle , g. 1981 , , 98 , 85 bodenheimer , p. , tenorio - tagle , g. , & yorke , h. w. 1979 , , 233 , 85 dyson , j. e. 1977 , , 59 , 161 franco , j. , tenorio - tagle , g. , & bodenheimer , p. 1989 , revista mexicana de astronoma y astrofsica , vol . 18 , 18 , 65 franco , j. , tenorio - tagle , g. , & bodenheimer , p. 1990
, , 349 , 126 franco , j. , garca - barreto , j. a. , & de la fuente , e. 2000 , , 544 , 277 franco , j. , garca - segura , g. , & kurtz , s. e. 2007 , , in press ( astro - ph/0508467 ) hatchell , j. , & van der tak , f. f. s. 2003 , , 409 , 589 henney , w. j. , arthur , s. j. , & garca - daz , m. t. 2005 , , 627 , 813 lada , c. j. , alves , j. f. , & lombardi , m. 2007 , protostars and planets v , eds . b. reipurth , d. jewitt , & k. keil , ( tucson : university of arizona press ) , 3 mac low , m .-
, toraskar , j. , oischi , j. s. & abel , t. 2006 , , submitted ( astro - ph/06055001 ) press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. 1992
, numerical recipes in fortran : the art of scientific computing , 2nd ed . , cambridge : university press shu , f. h. 1992 , physics of astrophysics , vol .
ii , university science books shu , f. h. , lizano , s. , galli , d. , cant , j. , & laughlin , g. 2002 , , 580 , 969 spitzer , l. 1968 , diffuse matter in space , new york : interscience publication tenorio - tagle , g. 1979 , , 71 , 59
|
the expansion of ionization fronts in uniform and spherically symmetric power - law density distributions is a well - studied topic .
however , in many situations , such as a star formed at the edge of a molecular cloud core , an offset power - law density distribution would be more appropriate . in this paper a few of the main issues of the formation and expansion of hii regions in such media
are outlined and results are presented for the particular cases where the underlying power laws are @xmath0 and @xmath1 . a simple criterion is developed for determining whether the initial photoionized region will be unbounded , which depends on the power - law exponent and the ratio of the equivalent strmgren radius produced by the star in a uniform medium to the stellar offset distance . in the expansion stage , the ionized volumes will eventually become unbounded unless pressure balance with the external medium is reached before the ionization front velocity becomes supersonic with respect to the ionized gas .
|
kaonic atoms and kaonic nuclei carry important information concerning the @xmath5nucleon interaction in nuclear medium .
this information is very important to know the kaon properties at finite density and , for example , to determine the constraints on kaon condensation in high density matter . in recent years
, there have been important developments in the studies of kaonic nuclear states , which are kaon nucleus bound systems by the strong interaction inside the nucleus .
experimental studies of the kaonic nuclear states using in flight ( @xmath6 ) reactions were proposed and performed by kishimoto and his collaborators @xcite . and
the first theoretical results of the energy spectra of the in flight ( @xmath6 ) reaction were obtained in ref .
experiments employing stopped ( @xmath6 ) reactions were carried out by iwasaki , t. suzuki and their collaborators and reported in refs . @xcite . in these experiments , they found some possible indications of the existence of kaonic nuclear states with significantly narrow widths .
another indication of @xmath7 bound state was reported by the finuda experiment @xcite .
there are also theoretical studies of the structure and formation of kaonic nuclear states related to these experimental activities @xcite .
it should be noted that these theoretical studies predict the possible existence of ultra
high density states in kaonic nuclear systems @xcite .
a critical analysis of the latest stopped @xmath8 experimental data was also reported by oset and toki @xcite .
we think that existence of the narrow kaonic nuclear states is still controversial and we need further studies .
another very interesting feature of the kaon nucleus bound systems is based on the fact that the properties of kaons in nuclei are strongly influenced by the change undergone by @xmath9 in nuclear medium , because @xmath9 is a resonance state just below the kaon - nucleon threshold .
in fact , there are studies of kaonic atoms carried out by modifying the properties of @xmath9 in nuclear medium @xcite .
these works reproduce the properties of specific kaonic atoms reasonably well . in ref .
@xcite , the phenomenological study of kaonic atoms are performed comprehensively , where the density
dependent potentials are considered for @xmath10 fitting to take into account possible non linear effects which could be due to @xmath9 resonance .
recently , there have been significant developments in the description of hadron properties in terms of the @xmath11 chiral lagrangian .
the interpretation of the @xmath9 resonance state as a baryon meson coupled system proposed in ref .
@xcite is also supported by the studies with the chiral lagrangian @xcite .
subsequently , the properties of @xmath9 in nuclear medium using the @xmath11 chiral unitary model were also investigated by waas et al .
@xcite , lutz @xcite , ramos and oset @xcite , and ciepl@xmath12 et al .
all of these works considered the pauli effect on the intermediate nucleons .
in addition , in ref .
@xcite , the self - energy of the kaon in the intermediate states is considered , and in ref .
@xcite , the self - energies of the pions and baryons are also taken into account .
these approaches lead to a kaon self - energy in nuclear medium that can be tested with kaonic atoms and kaonic nuclei .
medium @xmath13 properties have been studied also in refs .
@xcite based on meson exchange j@xmath14lich @xmath15 interaction , and in ref .
@xcite for the environment in heavy ion collisions . in a previous work @xcite
, we adopted the scattering amplitude in nuclear medium calculated by ramos and oset @xcite for studies of kaonic atoms , and demonstrated the ability to reproduce the kaonic atom data reasonably well .
baca et al .
@xcite modify the potential by adding the small phenomenological term and improve the fit to the atomic data significantly , which is comparable to the phenomenological fit @xcite .
we then calculated the deeply bound kaonic atoms for @xmath2o and @xmath16ca , which have narrow widths and are believed to be observable with well - suited experimental methods .
we also obtained very deep kaonic nuclear states , which have large decay widths , of the order of several tens of mev .
similar results were also obtained in refs .
@xcite . in this paper
, we study the in flight ( @xmath0 ) reactions systematically using the green function method @xcite for the formation of kaon nucleus bound systems like kaonic atoms and kaonic nuclei .
the green function method is known to be suited for evaluations of formation rates both of stable and unstable bound systems .
the ( @xmath0 ) reaction for the @xmath13nucleus system formation was proposed in refs .
@xcite and @xcite , and first theoretical results of the reaction energy spectra were obtained in ref .
@xcite using the same theoretical approach of ref .
@xcite for the deeply bound pionic atom formation reaction @xcite .
we use the green function method in this paper to get more realistic reaction spectra theoretically than our previous results @xcite where we used the effective number approach which is best suited for discrete states formation . by the green function method we can consider the unstable states , which have large widths and have some overlap with neighbor states , as well as stable states , which are isolated well from other states , simultaneously in realistic theoretical formalism @xcite . we can also include the contribution from quasi free kaon production in the final states in this formalism .
we think this evaluation is very interesting and important to know the experimental feasibilities of the ( @xmath0 ) reaction and to understand the deeper meanings of the observed spectra @xcite .
we believe that the realistic calculations of the formation spectra are necessary for all observed results to study the kaon properties in nuclear medium , and to get the decisive conclusions . in sec .
ii , we describe the theoretical models of the kaon nucleus optical potential , and of the ( @xmath0 ) reactions .
numerical results are presented and discussed in sec .
iii for @xmath1c and @xmath2o target cases .
we give conclusions of this paper in sec . iv .
we have improved our theoretical formalism in our previous work in ref .
@xcite in the following three points ; 1 . the energy dependence of the kaon nucleus optical potential is correctly taken into account , 2 .
the green function method is adopted to calculate the ( @xmath0 ) reaction spectra , 3 . the quasi
free kaon production is evaluated and included in the calculated spectra , as described in detail below . in order to investigate the in flight ( @xmath0 ) spectra theoretically
, we consider the klein
gordon equation @xmath17\phi({\bm r})=[\omega - v_{\rm coul}(r)]^2 \phi({\bm r}).\ ] ] here , @xmath18 is the kaon nucleus reduced mass and @xmath19 is the coulomb potential with a finite nuclear size : @xmath20 where @xmath21 is the charge distribution of the nucleus . we employ the empirical woods saxon form for the density as @xmath22},\ ] ] where we use @xmath23 fm and @xmath24 fm with @xmath25 , the nuclear mass number . to evaluate the kaon nucleus optical potential , we use the point nucleon density distributions deduced from the @xmath26 in eq .
( [ rho_ch ] ) by the same prescription described in sect . 4 in ref .
the shapes of the density distributions of the proton and neutron are assumed to be same in this paper .
we use the green function method @xcite to calculate the formation cross sections of the @xmath13nucleus system in the ( @xmath0 ) reactions .
the details of the application of the green function method are found in refs .
@xcite .
the present method starts with a separation of the reaction cross section into the nuclear response function @xmath27 and the elementary cross section of the @xmath28 with the impulse approximation @xmath29 the forward differential cross section of the elementary process @xmath28 in the laboratory frame @xmath30 is evaluated to be 8.8 mb / sr at @xmath31 using the @xmath32 elastic cross section data in ref .
we should mention here that the corrections to this evaluation were reported in ref .
@xcite , which reduce the elementary cross section to be 3.6 mb / sr effectively . in this paper
, we show the all calculated results with assuming the elementary cross section to be 8.8 mb / sr .
the calculation of the nuclear response function with the complex potential is formulated by morimatsu and yazaki @xcite in a generic form as @xmath33 where the summation is taken over all possible final states .
the amplitude @xmath34 denotes the transition of the incident particle ( @xmath13 ) to the proton hole and the outgoing ejectile ( @xmath35 ) , involving the proton hole wavefunction @xmath36 and the distorted waves @xmath37 and @xmath38 , of the projectile and ejectile , taking the appropriate spin sum @xmath39_{jm}\chi_i({\bm r})~,~\ ] ] with the meson angular wavefunction @xmath40 and the spin wavefunction @xmath41 of the ejectile .
the distorted waves are written with the distortion factor @xmath42 as @xmath43 with the momentum transfer @xmath44 .
the distortion factor @xmath42 is defined as @xmath45 where @xmath46 is the averaged distortion cross section defined as , @xmath47 with the total cross sections of the incident kaon and emitted proton with the nucleons in the nucleus .
the averaged nuclear density @xmath48 ) in eq .
( [ distortionfactor ] ) is defined as , @xmath49}~,\ ] ] in the polar coordinates with the averaged radial parameter @xmath50 and diffuseness parameter @xmath51 defined as , @xmath52 and @xmath53 with the density parameters of the nuclei in the initial and final states .
+ the green function @xmath54 contains the kaon nucleus optical potential in the hamiltonian @xmath55 as , @xmath56 where @xmath57 is the meson creation operator , @xmath58 the proton hole state , and @xmath59 the kaon energy defined as @xmath60 using the kinematical variables defined in the formation reaction as , @xmath61 is the incident kaon kinetic energy , @xmath62 the emitted proton kinetic energy , and @xmath63 the proton separation energy from the each proton single particle level which is compiled in table @xmath64 in ref
. @xcite . obtaining the green function with the optical potential is essentially the same as solving the associated klein
gordon equation .
we can calculate the nuclear response function @xmath27 from @xmath65 by performing appropriate numerical integrations for the variables @xmath66 and @xmath67 . in the green function formalism
, we can calculate the response function @xmath27 for both bound and quasi
free kaon production energy regions , and we can also perform the summation of the kaon final states without assuming the existence of the discrete kaon bound states , which could disappear in the cases with the strongly absorptive optical potential . as for the kaon nucleus interaction , we consider two different energy dependent optical potentials , that obtained with the chiral unitary approach @xcite , and that obtained with a phenomenological fit @xcite with an appropriate phase space factor for the imaginary part @xcite , which will be explained below .
we take into account the energy dependence of both potentials in the present calculation .
the optical potentials of the chiral unitary approach , which is obtained by the kaon self energy in nuclear matter with the local density approximation , is described in detail in ref .
@xcite , and we use the latest results of the model @xcite . the optical potential obtained in the phenomenological fit @xcite is written as , @xmath68 where @xmath69 is a density and energy dependent effective scattering length and @xmath70 .
the @xmath69 is parameterized as , @xmath71 and @xmath72 f^{\rm mfg}(e)~{\rm fm}.\ ] ] the @xmath73 is a phase space factor introduced to evaluate the phase space volume of decay channels and defined in ref .
@xcite by mare@xmath74 , friedman , and gal as , @xmath75 where @xmath76 and @xmath77 are the phase space factors for @xmath78 and @xmath79 decay , respectively .
these factors are defined as @xmath80[m^2_1-(m_{\sigma}-m_{\pi})^2 ] } { [ m^2_{01}-(m_{\pi}+m_{\sigma})^2][m^2_{01}-(m_{\sigma}-m_{\pi})^2 ] } } \times \theta(m_1-m_{\pi}-m_{\sigma } ) , \label{eq : mfg1}\ ] ] and @xmath81[m^2_2-(m_{\sigma}-m _ n)^2 ] } { [ m^2_{02}-(m _ n+m_{\sigma})^2][m^2_{02 } -(m_{\sigma}-m _ n)^2]}}\theta(m_2-m_{\sigma}-m _ n ) .
\label{eq : mfg2}\ ] ] here , the branching ratios of mesic decay and non - mesic decay are assumed to be 80@xmath82 and 20@xmath82 in eq .
( [ eq : mfg ] ) .
the masses are defined as @xmath83 , @xmath84 , @xmath85 , @xmath86 , and @xmath59 is the kaon energy appeared in eq .
( [ green1 ] ) . in this paper
, we introduce this factor to reduce the strength of the absorptive potential of the phenomenological fit , due to the phase space suppression for deeply bound kaonic states . to perform more realistic estimations
, we should introduce the energy dependence for each absorptive process separately and we should also consider the medium effects for @xmath87 and @xmath88 produced by the kaon absorption . in the present calculations , we have not included these effects for simplicity .
in the present paper , we take into account the energy dependence of the optical potentials correctly .
we show the kaon nucleus optical potentials for the @xmath89b case in fig .
[ fig : pot_chi ] for the chiral unitary model and in fig .
[ fig : pot_ph ] for the phenomenological fit . because the real part of the optical potential changes its sign at a certain nuclear density at @xmath90 mev in both the chiral unitary and phenomenological models ,
the kaon nucleus optical potential is attractive , while keeping the repulsive sign for the kaon
nucleon scattering length in free space at @xmath90 mev .
the real part of the optical potential of the chiral unitary model has the energy dependence as shown in the upper panel in fig .
[ fig : pot_chi ] and the potential depth varied from around @xmath91 mev to @xmath92 mev for the kaon energies @xmath93 mev
@xmath94 mev . as for the imaginary part
, the strength of the nuclear absorption reduced from around @xmath95 mev to @xmath96 mev for the kaon energies @xmath93 mev
@xmath94 mev mainly due to the threshold effects , which is naturally included in the coupled channel calculation in the chiral unitary model @xcite . as shown in fig .
[ fig : pot_ph ] , the real part of the phenomenological optical potential does not have the energy dependence since the parameters are determined by the experimental data of kaonic atoms which have the information only around @xmath97 mev in the present energy scale .
the effects of the possible energy dependence of the real part will be discussed later in fig .
[ fig : omega ] .
the depth of the real potential of the phenomenological fit is much deeper than that of the chiral unitary model .
there are also phenomenological potentials which provide the similar depth with the chiral unitary potential .
however , we consider deeper phenomenological potential here because we can expect to have larger possibilities to form deep kaonic nuclear states , and expect to have deeper insight for the kaon bound states by comparing the results obtained from much different potentials . the imaginary part of the optical potential of the phenomenological fit has the energy dependence due to the threshold effects of the decay channels , which are evaluated in eqs .
( [ eq : mfg])([eq : mfg2 ] ) . the imaginary part of the both potentials have the qualitatively similar depths and similar energy dependence . the kaon nucleus optical potential of the chiral unitary approach @xcite for @xmath89b as a function of the radial coordinate @xmath98 .
the upper and lower panels show the real and imaginary parts , respectively .
the solid , dotted and dashed lines indicate the potential strength for the kaon energies @xmath99 , respectively.,width=245,height=302 ] the kaon nucleus optical potential of the phenomenological fit @xcite with the phase space factor @xcite for @xmath89b as a function of the radial coordinate @xmath98 .
the upper and lower panels show the real and imaginary parts , respectively .
the solid , dotted and dashed lines indicate the potential strength for the kaon energies @xmath99 , respectively .
the real part does not have the energy dependence.,width=245,height=302 ] in order to calculate the formation spectra of the kaon nucleus bound systems in the ( @xmath0 ) reactions , we use the green function method @xcite in this paper as described in sec .
[ formalism ] .
the green function method is suited to evaluate the formation rates both of the stable and unstable systems , and is suited to include simultaneously the contribution of the quasi
free kaon production in the final states to the spectra .
for these two reasons , the green function method is superior to the effective number approach which we have used in the previous paper @xcite . on the other hand , the effective number approach is better suited to calculate the formation spectra of the stable states with narrow widths than the green function method because of much better efficiency for getting the numerical results . in fig .
[ fig : gvse ] , we show the calculated spectra by the green function method and by the effective number approach for the exactly same cases to show clearly the differences of the both methods .
the incident kaon energy is fixed to @xmath31 , which corresponds to the in flight experiment @xcite . in the upper panel ,
we show the calculated results with the phenomenological optical potential eq .
( [ v_opt_ph ] ) with the reduced strength of the imaginary part @xmath100 .
we find clearly that each peak corresponding to a bound state has only small width and is well isolated . in this case , both methods provide quite similar formation spectra for the bound states formation and are considered to be equivalent to each other . for the quasi
free kaon production energy region , we can see that the quasi
free contributions to the ( @xmath0 ) spectra are automatically included in the green function method as described in sec .
[ formalism ] . the quasi
free contributions are not included in the results of the effective number approach . in the lower panel , we show the calculated results by both methods for the phenomenological optical potential eq .
( [ v_opt_ph ] ) .
we find that there is no peak structure in the spectra corresponding to the bound states formation because of the large widths of the states . in this case , the spectra calculated by both methods provide significantly different shapes even in the energy region of the kaon bound states .
hence , the green function method and the effective number approach are not equivalent to each other for the formation of the states with large widths .
this discrepancy between the theoretical models is a natural consequence of the fact that the existence of well isolated eigenstates is assumed in the effective number approach formalism .
we use the green function method in this paper to get the ( @xmath0 ) spectra for the formation of kaon bound states which are expected to have large widths .
we also show the comparison of the numerical results with the green function method and the effective number approach for atomic @xmath4 and @xmath101 states formation in fig .
[ fig : atomicstate ] . in the upper panels ,
we show the results with reduced strength of the imaginary potential @xmath102 , where we can see that the spectra obtained with the both methods are qualitatively same for both atomic states .
however , for larger imaginary potentials , the results with both methods show the significant differences as shown in middle and lower panels in fig .
[ fig : atomicstate ] .
actually we can find very interesting , even peculiar structures in the results with the full chiral unitary optical potential .
especially , the atomic @xmath4 state formation makes the @xmath3resonance dip@xmath3 instead of the @xmath3resonance peak@xmath3 in the spectrum .
similar results are expected for the phenomenological potential . by all these results
show in fig .
[ fig : atomicstate ] , we think the origin of the interesting structures for atomic states formation is the interference with the kaon absorptive processes induced by the large imaginary potential .
subcomponents of the kaonic nucleus formation cross sections in the @xmath1c(@xmath0 ) reactions at @xmath103 are plotted as functions of the emitted proton energy @xmath62 at @xmath104 ( deg . ) for the phenomenological optical potential eq.([v_opt_ph ] ) .
the upper panel shows the results with the reduced imaginary potential @xmath100 , while the lower panel shows those with eq .
( [ v_opt_ph ] ) .
the solid lines show the results obtained by the green function method for the sum of the subcomponents @xmath105 $ ] and @xmath106 $ ] .
the dashed lines show those by the effective number approach for the sum of the subcomponents @xmath107 $ ] , @xmath108 $ ] and @xmath109 $ ] .
the vertical dashed line indicates the kaon production threshold.,width=245,height=302 ] subcomponents of the kaonic atom formation cross sections in the @xmath1c(@xmath0 ) reactions at @xmath110 mev are plotted as functions of the emitted proton energy @xmath62 at @xmath104 ( deg . ) for the chiral unitary optical potential @xcite . left and
right panels show the calculated results at energies of kaonic atom @xmath4 and @xmath111 states formation , respectively .
the upper and middle panels show the results with the reduced imaginary potential @xmath112im@xmath113 and @xmath114im@xmath113 , while the lower panels show those with the full chiral unitary potential .
the solid lines show the results obtained by the green function method for the sum of the subcomponents @xmath105 $ ] and @xmath115 $ ] .
the dashed lines show those by the effective number approach for the sum of subcomponents @xmath116 $ ] and @xmath117 $ ] .
the vertical dashed lines indicate the binding energies of kaonic atom @xmath4 state ( left panels ) and @xmath111 state ( right panels ) . ,
width=302,height=302 ] in fig .
[ fig : atom_eoff ] , we show the fine structure of the total @xmath1c(@xmath0 ) spectra around the kaon production threshold , where we can observe the contributions of atomic states formation . here
, we have used the optical potential of the chiral unitary model , and we confirm numerically that the spectrum shape is almost the same as that calculated with the phenomenological optical potential in this energy region .
we find again that the spectrum around atomic states formation region is very interesting , where the resonance contributions do not make simple peak structures but make complicated structures as shown by the solid line in fig .
[ fig : atom_eoff ] because of the large imaginary potential . for comparison
, we also show the calculated results by the effective number approach by dashed line , which show the simple peak structures .
we think that it is very interesting to observe experimentally the spectrum shape for kaonic atom formation region .
we mention that these deeply bound kaonic atom states have not been observed yet .
kaonic atom formation cross sections in @xmath1c(@xmath0 ) reactions at @xmath110 mev plotted as functions of the emitted proton energy @xmath62 at @xmath118 ( deg . ) for the chiral unitary optical potential @xcite .
solid line indicates the result of the green function method and dashed line that of the effective number approach.,width=245,height=264 ] we show firstly the calculated @xmath1c(@xmath0 ) spectrum in fig . [
fig : green_ph_eoff ] for the energy independent phenomenological optical potential evaluated at @xmath90 in eq .
( [ v_opt_ph ] ) .
we also show the subcomponents of the spectrum in the same figure . though there exist a few kaonic nuclear states in this case , we find that we can not see any peak structures in the ( @xmath0 ) spectra because of the large widths .
we also find that the subcomponents of the substitutional states do not dominate the total spectrum because of the finite momentum transfer of this reaction .
calculated spectra of the @xmath1c(@xmath0 ) reaction at @xmath103 plotted as functions of the emitted proton energy @xmath62 at @xmath118 ( deg . ) .
the energy independent phenomenological potential evaluated at @xmath90 in eq .
( [ v_opt_ph ] ) is used .
each subcomponent is indicated in the figure .
the vertical dashed line indicates the kaon production threshold.,width=245,height=188 ] in order to see the effects of the energy dependence of the potential on the ( @xmath0 ) spectra , we show the calculated spectra with energy dependent optical potential in fig .
[ fig : eon_eoff ] ( upper panels ) and compare those with energy independent optical potential fixed at @xmath90 for @xmath1c target case .
the energy dependence of the optical potential is expected to be important since one can think that the strength of the imaginary part of the optical potential could be reduced significantly because of the phase space suppressions of the decay channels , and that the widths of the kaonic nuclear states could be so narrow that one can see the peak structures in the ( @xmath0 ) spectra . as can be seen in the figure , the effect of the energy dependence of the optical potential is tiny for the chiral unitary model and the shape of the ( @xmath0 ) spectrum is almost unchanged .
as for the phenomenological optical potential , we can find there appear some small structures around @xmath119 in the spectrum .
the calculated binding energies corresponding to the kaonic nuclear states are also indicated by solid arrows in the figure for the energy dependent potentials .
this structure could be observed in experiments , though it seems rather difficult .
we should mention here that the small structures in the spectrum show the indications of the existence of the kaonic nuclear states .
however , we should be careful to deduce the properties of the kaonic nuclear states from the observed spectra because the shapes and the positions of the structure could be changed from those corresponding to the binding energies and widths of the kaonic states by the energy dependence of the optical potential and by the contributions from other subcomponents .
thus , we think that we need the help of the realistic theoretical calculations of the spectrum to deduce the properties of the kaonic nuclear states from the observed spectrum correctly . calculated spectra of the @xmath1c(@xmath0 ) ( upper panels ) and the @xmath2o(@xmath0 ) ( lower panels ) reactions at @xmath110 mev plotted as functions of the emitted proton energy @xmath62 at @xmath118 ( deg . ) for the chiral unitary optical potential ( left panels ) and the phenomenological optical potential ( right panels ) .
solid lines show the results with the energy dependent potentials and dashed lines those with the energy independent potentials fixed at @xmath90 mev .
the solid arrows indicate the calculated binding energies of the kaonic nuclear states coupled to the ground states of the final state nucleus , namely @xmath120 state for c target case and @xmath121 state for o target case , respectively .
the binding energies are calculated with the energy dependent potential .
the dashed arrows in lower panels indicate the binding energies of the kaonic nuclear states suggested by the experimental group @xcite . the vertical dashed line indicates the kaon production threshold.,width=302,height=302 ] in fig .
[ fig : eon_eoff ] ( lower panels ) , we show the results for @xmath2o target case .
the experimental data for this case were reported in ref .
we show calculated results both with the chiral unitary potential and the phenomenological potential .
the dashed arrows in the figure indicate the energies of the possible peak structures suggested by the experimental group @xcite .
we also showed the calculated binding energies by the solid arrows .
we find that our calculated results do not have peak structures at the energies suggested by experiment @xcite . at present exploratory level , it is rather difficult to deduce conclusions from this comparison because @xmath122 the accuracy of the experimental data seems not enough , and @xmath123 the effects of nuclear structure changes are not included in the present calculations , which could be important even for c and o as discussed in ref .
@xcite . hence , in order to perform meaningful comparison between the data and the calculated results , we need accurate data such that we could distinguish the left ( chiral unitary ) and right ( phenomenology ) figures in fig .
[ fig : eon_eoff ] and that we could observe the existence ( or non existence ) of the small structures appearing in the spectra obtained with the phenomenological potential .
and we also need better theoretical results which incorporate the nuclear structure change effects in order to deduce decisive information on kaonic nuclear states . finally , to see the dependence of the spectrum shape on the assumptions in the theoretical calculations with the phenomenological optical potential , we show in fig .
[ fig : gal0901 ] the calculated results using the phase space factor with different branching ratios of the decay channels as , @xmath124 here , we assume the branching ratios of mesic decay and non mesic decay to be 90@xmath82 and 10@xmath82 . we find that the structures in the spectrum due to the kaonic nuclear states can be seen clearer around @xmath125 than those in the right panels in fig .
[ fig : eon_eoff ] .
furthermore , we show the calculated results with the phenomenological optical potential by replacing the potential term in the klein gordon equation eq .
( [ kg_eq ] ) as @xmath126 to consider the another type of the energy dependence . due to the additional energy dependence ,
the effects of the optical potential are smaller for both real and imaginary parts in the deeper bound region because of the smaller value of the factor @xmath127 in eq .
( [ eq : omega ] ) .
namely , the effective potential strength is small for deeply bound region , which means that the potential is less attractive and less absorptive in smaller @xmath128 region effectively .
hence , as we can see in fig .
[ fig : omega ] , the strength of the calculated ( @xmath0 ) spectra is smaller for deeper bound region than the results in fig .
[ fig : eon_eoff ] ( right panels ) .
we can find small structures again in the spectra as the indications of the kaonic nuclear states .
calculated spectra of the @xmath1c(@xmath0 ) reaction at @xmath110 mev plotted as functions of the emitted proton energy @xmath62 at @xmath118 ( deg . ) for the energy dependent phenomenological optical potential with the different branching ratio of the decay channels defined in eq .
( [ eq : f_0901 ] ) ( see text in detail ) .
each subcomponent is indicated in the figure .
the vertical dashed line indicates the kaon production threshold.,width=245,height=188 ] calculated spectra of the @xmath1c(@xmath0 ) and @xmath2o(@xmath0 ) reactions at @xmath110 mev plotted as functions of the emitted proton energy @xmath62 at @xmath104 ( deg . ) for the phenomenological optical potential with different energy dependence defined in eq .
( [ eq : omega ] ) ( see text in detail ) .
each subcomponent is indicated in the figure .
the vertical dashed lines indicate the kaon production threshold.,width=302,height=188 ]
we have calculated the in flight ( @xmath0 ) reaction spectra systematically for the formation of kaon nucleus bound systems by the green function method .
we have used two different kaon
nucleus optical potentials and compared the calculated results .
one is the optical potential of the chiral unitary model which is based on a microscopic theory and is proved to reproduce the existing atomic data reasonably well . the energy dependence of the chiral unitary potential is automatically included in the theoretical framework .
the other one is the phenomenological optical potential which is obtained by the @xmath129 fitting to the kaonic atom data and has a much deeper real part than that of the chiral unitary potential .
the energy dependence of the phenomenological potential is also considerd .
we should mention here that both potentials provide the qualitatively same energy spectra for the kaonic atoms though they have much different potential depths in the real part .
thus , it is extremely interesting to compare the calculated results for kaonic nuclear states with these potentials .
the calculated results newly reported in this paper are superior to our previous results reported in ref .
@xcite in 3 points as described in sec .
namely , ( 1 ) energy dependence of the potential , ( 2 ) green function method , and ( 3 ) quasi
free kaon production contributions are considered .
we have shown numerically that the effective number approach used in ref .
@xcite provides the equivalent results for bound states formation spectra as those provided by the green function method @xcite in cases with the less absorptive optical potentials . if the potential is so strongly absorptive that resonance peaks overlap with each other , we need to use the green function method @xcite . 1 .
the ( @xmath0 ) spectra calculated with the energy independent kaon nucleus optical potentials do not show any peak structures indicating the kaonic nuclear states formation , even if such states exist , because of the large widths of the states ( fig .
[ fig : green_ph_eoff ] ) , 2 .
the ( @xmath0 ) spectra calculated with the chiral unitary optical potential do not show any structures indicating the kaonic nuclear states formation in all cases calculated in this paper ( fig .
[ fig : eon_eoff ] left panels ) , 3 .
the ( @xmath0 ) spectra calculated with the energy dependent optical potentials of the phenomenological model eq . ( [ v_opt_ph ] )
show the small structures around the energy region of the kaonic nuclear states ( fig .
[ fig : eon_eoff ] right panels ) , 4 .
the small structures mentioned in ( 3 ) are not necessarily corresponding to the correct binding energies of the kaonic nuclear states .
the energy dependence of the imaginary part of the optical potential and the contributions of many subcomponents can shift the observed positions of the small structures from those corresponding to the binding energies of the kaonic states , 5 . the calculated ( @xmath0 ) spectra do not show peak structures at energies suggested by the experimental group @xcite for either kaon nucleus optical potential cases ( fig .
[ fig : eon_eoff ] lower panels ) .
we think that we need more accurate data and better theoretical results incorporating the effects of nuclear structure changes due to the existence of a kaon in order to make decisive conclusions from this observation , 6 .
the calculated ( @xmath0 ) spectra have certain dependence on input parameters like branching ratios of decay channels ( fig .
[ fig : gal0901 ] ) , and on the assumed functional form of the energy dependence of the potential ( fig .
[ fig : omega ] ) in the phenomenological potential case , 7 .
the spectra for the kaonic atom states formation show very interesting behavior like the @xmath3resonance dip@xmath3 for the @xmath4 state formation because of the interference with the absorption processes ( fig .
[ fig : atomicstate ] and [ fig : atom_eoff ] ) , 8 . the elementary cross section in eq
( [ crosss ] ) is assumed to be 8.8 mb / sr in all calculated results shown in this paper .
the corrections to this value are reported in ref .
@xcite , which reduce the elementary cross section to be 3.6 mb / sr .
when we take this value , all the calculated cross sections in this paper have to be reduced by roughly a factor of 2 .
by considering all results described above , we conclude that we need first to have new experimental data which are so accurate that we can identify the existence ( or non existence ) of the small structures appearing in the theoretical calculations in the kaonic nuclear states energy region .
then , if it exists , we must investigate the origins of the structures very carefully in order to conclude the existence ( or non existence ) of the kaonic nuclear states decisively in the studies of the in flight ( @xmath0 ) reactions on the targets like c and o. finally , we would like to add a few words for the perspectives in formation reactions of kaonic nuclear states .
the stopped ( @xmath0 ) reactions on the targets like c and o will provide the similar information as the in flight reaction cases on the same targets which are reported in this paper . as shown in refs .
@xcite , it may be interesting to consider lighter nuclei as targets . in any cases ,
we need to evaluate the effects of the nuclear structure changes to the reaction cross sections , which have not been included in our calculations so far .
the possibilities of the significant changes of the nuclear structure due to kaon were reported not only for lighter nuclei @xcite but also for heavier nuclei like c and o @xcite .
the possible changes of the nuclear structures may be expected to reduce the formation cross sections of kaonic nuclear states significantly because of the suppression due to the nuclear transition form factor . on the other hand ,
a kaon wavefunction confined in smaller spatial dimension is expected to be able to bear the larger momentum transfer in the ( @xmath0 ) reactions , and helps to make the cross sections larger .
thus , the quantitative evaluation is quite necessary in order to conclude the size of the observed formation cross sections in refs .
@xcite and the predicted changes of the nuclear structure in refs .
@xcite are really consistent with each other or not .
we acknowledge e. oset and a. ramos for providing us their calculated results of kaon self
energy at finite densities in the chiral unitary model and for stimulating discussions on kaon bound systems .
we also thank e. oset and a. ramos for their careful reading of our manuscript and useful comments .
we would like to thank h. toki , a. hosaka , y. akaishi , y. abe , a. ohnishi and t. kishimoto for valuable discussions .
this work is partly supported by grants in
aid for scientific research of monbukagakusho and japan society for the promotion of science ( no .
16540254 ) , and by nara women s university
intramural grant for project research .
99 t. kishimoto , phys .
83 * , 4701 ( 1999 ) .
t. kishimoto et al . , prog .
no.*149 * , 264 ( 2003 ) , t. kishimoto et al .
, nucl . phys . *
a754 * , 383c ( 2005 ) .
j. yamagata , h. nagahiro , y. okumura and s. hirenzaki , prog .
114 , 301 ( 2005 ) .
( errata 114 , 905 ( 2005 ) . )
m. iwasaki et al .
, nucl - ex/0310018 .
+ t. suzuki et al . , phys .
b * 597 * , 263 ( 2004 ) ; nucl .
* a754 * 375c ( 2005 ) . m. agnello et al . , phys .
* 94 * , 212303 ( 2005 ) .
y. akaishi and t. yamazaki , phys .
c * 65 * , 044005 ( 2002 ) .
a. dot@xmath131 , h. horiuchi , y. akaishi and t. yamazaki , phys .
b * 590 * , 51 ( 2004 ) .
e. oset and h. toki , nucl - th/0509048 .
m. alberg , e. m. henley and l. wilets , ann .
( n.y . ) * 96 * , 43 ( 1976 ) .
r. brockmann , w. weise and l. tauscher , nucl . phys . *
a308 * , 365 ( 1978 ) .
m. mizoguchi , s. hirenzaki and h. toki , nucl .
* a567 * , 893 ( 1994 ) . c. j. batty , e. friedman and a. gal , phys
. rep . * 287 * , 385 ( 1997 ) .
r. h. dalitz , t. c. wong and g. rajasekaran , phys .
rev . * 153 * 1617 ( 1967 ) .
n. kaiser , p.b .
siegel and w. weise , nucl .
a594 * , 325 ( 1995 ) .
e. oset and a. ramos , nucl . phys .
a635 * , 99 ( 1998 ) . t. waas , n. kaiser and w. weise , phys .
b * 365 * , 12 ( 1996 ) ; * 379 * , 34 ( 1996 ) ; t. waas and w. weise , nucl . phys .
* a625 * , 287 ( 1997 ) . m. lutz , phys . lett .
b * 426 * , 12 ( 1998 ) .
a. ramos and e. oset , nucl .
a671 * , 481 ( 2000 ) .
a. ciepl@xmath12 , e. friedman , a. gal and j. mare@xmath132 , nucl . phys . *
a696 * , 173 ( 2001 ) .
l. tol@xmath133s , a. ramos , a. polls , t. t. s. kuo , nucl . phys . *
a690 * , 547 ( 2001 ) .
l. tol@xmath133s , a. ramos , a. polls , phys .
c * 65 * , 054907 ( 2002 ) .
j. schaffner
bielich , v. koch , m. effenberger , nucl . phys . *
a669 * , 153 ( 2000 ) .
s. hirenzaki , y. okumura , h. toki , e. oset and a. ramos , phys .
c * 61 * , 055205 ( 2000 ) .
a. baca , c. garc@xmath134a recio , j. nieves , nucl .
a673 * , 335 ( 2000 ) .
e. friedman and a. gal , phys .
b * 459 * , 43 ( 1999 ) .
e. friedman and a. gal , nucl .
a658 * , 345 ( 1999 ) .
o. morimatsu and k. yazaki , nucl . phys . *
a435 * , 727 ( 1985 ) ; * a483 * , 493 ( 1988 ) .
s. hirenzaki , h. toki and t. yamazaki , phys .
c * 44 * , 2472 ( 1991 ) , + h. toki , s. hirenzaki and t. yamazaki , nucl
a530 * , 679 ( 1991 ) .
h. toki and t. yamazaki , phys .
b * 213 * , 129 ( 1988 ) , + h. toki , s. hirenzaki , r. s. hayano and t. yamazaki , nucl .
* a501 * , 653 ( 1989 ) . h. gilg et al . , phys .
c * 62 * , 025201 ( 2000 ) , + k. itahashi et al .
c * 62 * , 025202 ( 2000 ) .
j. nieves , e. oset , c. garcia - recio , nucl .
a554 * , 509 ( 1993 ) .
r. s. hayano , s. hirenzaki , a. gillitzer , eur .
j. a * 6 * , 99 ( 1999 ) .
f. klingl , t. waas , w. weise , nucl . phys .
* a650 * , 299 ( 1999 ) , + e. marco and w. weise , phys .
b * 502 * , 59 ( 2001 ) .
d. jido , h. nagahiro and s. hirenzaki , phys .
c * 66 * , 045202 ( 2002 ) , + h. nagahiro , d. jido and s. hirenzaki , phys .
c * 68 * , 035205 ( 2003 ) ; nucl .
a761 * , 92 ( 2005 ) .
b. conforto et al . , nucl . phys . *
b105 * , 189 ( 1976 ) .
j. mare@xmath74 , e. friedman and a. gal , phys .
b * 606 * , 295 ( 2005 ) ; nucl .
a770 * , 84 ( 2006 ) .
e. oset and a. ramos , private communication .
|
we study theoretically the kaonic atom and kaonic nucleus formations in the in flight ( @xmath0 ) reactions using the green function method , which is suited to evaluate formation rates both of stable and unstable bound systems .
we consider @xmath1c and @xmath2o as the targets and calculate the spectra of the ( @xmath0 ) reactions .
we conclude that no peak structure due to kaonic nucleus formation is expected in the reaction spectra calculated with the chiral unitary kaon nucleus optical potential . in the spectra with the phenomenological deep kaon
nucleus potential , we may have possibilities to observe some structures due to the formation of the kaonic nucleus states .
for all cases , we find clear signals due to the kaonic atom formations in the reaction spectra , which show the very interesting structures like the @xmath3resonance dip@xmath3 instead of the @xmath3resonance peak@xmath3 for the atomic @xmath4 state formation .
|
the field of sonoluminescence has undergone an explosion in recent years since the discovery of single bubble sonoluminescence ( sbsl ) [ 1 - 4 ] . in sbsl , a gas bubble in liquid is trapped at a velocity node of an acoustic field . under certain conditions ,
the bubble emits intense flashes of light ( @xmath0 photons / flash ) .
the shape of the spectrum is the subject of much investigation ; superficially it may be described as roughly that of a black - body at temperatures in excess of 20000 k. although the conditions for sbsl are now well - characterized for certain systems such as air bubbles in water , investigators have been denied a detailed look at the physical origins of this puzzling phenomenon due to the transient nature of the light flash . for example , attempts to directly measure the duration of the flash have resulted only in an upper limit of 50 - 150 ps [ 5 ] . likewise , while it is possible to make nanosecond scale measurements of the bubble size as a function of time by mie scattering , this technique can resolve only the liquid - gas interface [ 6 ] ; giving a value around 1 @xmath1 . in any case , the suggested mechanism of shock wave separation and implosion would result in a much smaller radius for the physical process which converts the phonons to photons .
we propose the use of bose - einstein correlations to measure the spatial and temporal characteristics of the light - emitting region .
bose - einstein correlations have been used for many years in astronomy and nuclear physics to extend the accessible range of time and length scales . in these fields
, the usefulness of this tool has been limited by the low brightness and non - reproducible nature of the sources , as well as final state interactions and unwanted correlations inherent in the particle production process [ 7 ] .
these limitations are absent in the present case , and sbsl is ideally suited for application of this method .
to illustrate these techniques , we now outline a specific two - photon experiment to measure the size of the light - emitting region in sbsl by the hanbury brown - twiss ( hbt ) effect [ 7 - 10 ] .
the two - photon correlation function @xmath2 can be defined as @xmath3 where @xmath4 is the probability density of detecting photons of 4-momenta @xmath5 and @xmath6 in the same pulse and @xmath7 is the single photon probability density .
because the total wavefunction for identical bosons is symmetric , the maximum value of this function will occur at @xmath8 .
when @xmath5 is not equal to @xmath6 , the strength of the correlation will depend on the degree of overlap between the single particle wavefunctions of the two particles .
this `` exchange density '' is determined by the size of the emitting source .
hence knowledge of the variation of @xmath9 with the momentum difference carries information about the source distribution . for completely uncorrelated photons , @xmath10
thus , the hbt effect is maximal when the particles are emitted randomly .
correlation functions have been derived with varying degrees of rigor and generality [ 7,10 ] .
many of these concerns , however , have to do with mutual interactions of charged particles in the final state or residual coherence in the production process .
the absence of these effects in sbsl greatly simplifies the evaluation of equation 1 . assuming the photons are emitted with random relative phases and polarization states from a source with a normalized space - time density @xmath11 , the two - particle probability density can be written as an incoherent sum over the source amplitudes @xmath12 for a pair of photons : @xmath13 while for the single particle densities @xmath14 neuhauser [ 11 ] has derived an expression for the correlation function using symmetrized plane waves with unit directional vectors @xmath15 and @xmath16 as @xmath17 [ | \rho ( k_{1 } + k_{2 } ) |^{2 } + | \rho ( k_{2 } - k_{1 } ) |^{2 } ] .\ ] ] where @xmath18 is the fourier transform of the source density .
the first factor in square brackets arises from the fact that there can be no interference between photons in different polarization states . for the photon energies considered here @xmath19 and
can safely be neglected .
thus , the correlation of colinear photons of equal energy is @xmath20 . to compare with experimental data , it is convenient to model the source .
here we assume the photons to be emitted from a fireball with a stationary gaussian distribution of radius @xmath21 and lifetime @xmath22 [ 12 ] @xmath23 .\ ] ] giving @xmath24 exp [ - q^{2}r^{2 } / 2 - \delta e^{2}\tau ^{2 } / 2 ] .\ ] ] where @xmath25 is the relative momentum , @xmath26 is the energy difference of the photons and @xmath27 is the angle between them . from the form of equation 6 , we can see that the width of the correlation function is inversely related to the source size @xmath21 and the lifetime @xmath28 . to make the experimental situation more explicit ,
we begin with the data on sbsl reported in ref . [ 5 ] by hiller , putterman and barber .
here , a single air bubble trapped in water at room temperature emits @xmath29 500,000 photons / flash at for a driving frequency of 27 khz .
the measured spectral density has a high brightness in the visible region , which makes it possible to explore the most plausible `` minimum bubble sizes '' ( 10 nm - 10,000 nm ) using commercially available photomultipliers , optical filters and other components .
the `` minimum size '' of the bubble , of course , refers strictly to the diameter of the region where the light produced by as yet unknown mechanisms ceases to interact with the surrounding material and escapes .
presumably , this occurs when the expanding material cools sufficiently to become transparent to the radiation .
a further step of interpretation must be made when one considers the refraction and polarization of the outgoing radiation through the mantle of material around it , as well as at the gas - liquid interface .
however , these details had best be studied in accompaniment with direct measurements so as to be appropriate to the specific experimental conditions .
estimates for the size and duration of the flash were obtained in a model investigated by wu and roberts [ 13 ] in which both the surface of the bubble and the enclosed gas were simulated .
they integrated the rayleigh - plesset equation using a hard - core van der waals equation of state for the gas . from several reasonable scenarios
presented one can expect typical `` minimum bubble sizes '' to range from 0.3 - 0.6 @xmath1 and the duration of the event to be about 1 ps .
if the light emitting region were about 0.1 - 1 @xmath1 in radius , experiments to determine this number would be quite easy to perform .
figure 1 shows a schematic of an experimental setup .
the momentum difference @xmath30 is solely a function of the energy and the separation angle of the two photons . for a fixed energy one can thus measure the correlation function by examining how equation 1 varies with the separation angle . using equation 6 ,
the expected result of this measurement is seen in figure 2 and is seen to extend over a large range of separation angles .
we emphasize once more the transparent interpretation of hbt results using photons compared to charged particles .
final state strong interactions as well as coulomb repulsions do not distort the signal when one uses photons .
in addition , there is no ambiguity as to whether the photons one measures come from the source or arise from secondary processes such as @xmath31 decays .
realistic counting rates can be estimated for the geometry of figure 1 .
two phototubes with high photoefficiency ( 5 - 10@xmath32 ) at wavelengths approaching 200 nm could be masked with commercially available filters ( @xmath33 nm ) .
if the tubes are placed 20 cm from the source and further defined by 1 mm circular apertures , the real singles rate in each detector would be typically @xmath34 0.1 photons / flash and the coincidence rate @xmath29 0.01 photons / flash or about 200 - 300 counts / sec . for a source radius of 500 nm ,
the width of the correlation function corresponds to a detector separation of 20 mm .
noise and random coincidences may be greatly reduced by triggering the phototube readout on the flash itself .
the choice of aperture size is somewhat arbitrary , and is governed by considerations of count rate and the scale of details one wishes to examine in the correlation function .
the high instantaneous brightness of the source ( @xmath35 photons / sec ) overcomes the objections to the coincidence method voiced in ref . [ 8 ] .
however , different experimental modes may be contemplated if it becomes necessary to improve the signal to noise ratio in order to examine details of the correlation function . in this case , the number of photons per flash striking each detector may be increased by using a larger aperture .
if these values are stored in an adc for each flash , then the system can be run as an intensity spectrometer , where correlations between the intensities for each detector are computed for each flash and summed over many events .
this would be a digital version of the intensity correlator ( linear multiplier ) constructed by hanbury brown and twiss [ 8 ] .
once the size of the bubble is known , it would be possible to directly measure the time duration of the flash in a conjugate experiment .
if the phototube positions are fixed within a region of large spatial correlation ( @xmath36 20 mm in the above example ) then the intensity correlation as a function of energy difference would give the lifetime .
again , with commercially available filters , the photon energy difference could be stepped through a range of values which could probe flash lifetimes from 1 fs to 1 ps .
several extensions of the hbt experiment are immediately obvious .
first , the partition of energy in the spectrum may be checked by determining the source size for different wavelengths of light .
the observed spectrum may be the overlap of several spectra from different epochs of bubble dynamics or shock waves . in this case , it is likely that higher photon energies result from the earlier , hotter phases of the light - production mechanism , and will appear to come from a smaller radius
. of even greater interest would be to perform the hbt experiment in those systems exhibiting discrete spectral lines from the host liquid or substances dissolved in it .
secondary processes in the vicinity of the bubble such as the catalysis of chemical reactions , induced fluorescence or excitation by the passage of shock waves may be explored by a suitable choice of aperture size and filters .
it is likely that the production mechanisms will result in interesting structure in the correlation function .
high precision measurements of the correlation function will make it possible to differentiate among various radiating structures .
the number of photons in each flash is so enormous that it is possible to perform most of the above measurements on an event - by - event basis .
this would , of course , involve the use of position - sensitive detectors with fine resolution .
the advantage of this technique is that the flash characteristics are not averaged over an ensemble , hence , event by event fluctuations can be examined directly and correlated with other event by event signatures of the source such as total intensity .
this would be of particular relevance in the understanding of period - doubling timing bifurcations observed for off - resonance excitation of sbsl [ 14 ] . despite the great confidence in the spherical symmetry of the collapsing bubble
, taylor instabilities may lead to deformations of the shock front .
one possible mechanism for this could be the uneven distribution of photon energies among the normal modes of the cavity . since the size of the cavity is only several times the wavelength of the radiation , the excitation of non spherically - symmetric modes of the cavity would result in the nodes of the electric field dividing the surface of the sphere according to the dominant tessoral harmonic .
the resulting roughly - faceted surface could be crudely imaged using 3 or 4 photon correlations .
we emphasize the ideal nature of single bubble sonoluminescence for study by means of bose - einstein correlations .
the high brightness , reproducibility and weakness of final state interactions combine to allow us to infer for the first time , in detail , the structure of a radiating body by means of an intensity interferometer . in turn
, sbsl may be a strong impetus to the development of the theory of these correlations ; especially those of higher order .
|
sonoluminescence may be studied in detail by intensity correlations among the emitted photons . as an example , we discuss an experiment to measure the size of the light - emitting region by the hanbury brown - twiss effect .
we show that single bubble sonoluminescence is almost ideally suited for study by this method and that plausible values for the physical parameters are within easy experimental reach .
a sequence of two and higher order photon correlation experiments is outlined .
|
a long - standing puzzle in the particle physics is the nature of the pomeron ( @xmath0 ) .
this object , with the vacuum quantum numbers , was introduced phenomenologically in the regge theory as a simple moving pole in the complex angular momentum plane , to describe the high - energy behaviour of the total and elastic cross - sections of the hadronic reactions @xcite . due to its zero color charge
the pomeron is associated with diffractive events , characterized by the presence of large rapidity gaps in the hadronic final state .
the diffractive processes have attracted much attention as a way of amplifying the physics programme at hadronic colliders , including searching for new physics ( for a recent review see , e.g. ref .
the investigation of these reactions at high energies gives important information about the structure of hadrons and their interaction mechanisms .
the diffractive processes can be classified as inclusive or exclusive events ( see e.g. @xcite ) . in exclusive events , empty regions in pseudo - rapidity , called rapidity gaps ,
separate the intact very forward hadron from the central massive object .
exclusivity means that nothing else is produced except the leading hadrons and the central object .
the inclusive processes also exhibit rapidity gaps .
however , they contain soft particles accompanying the production of a hard diffractive object , with the rapidity gaps becoming , in general , smaller than in the exclusive case .
moreover , the inclusive processes can also be classified as single or double diffractive , which is directly associated to the presence of one or two rapidity gaps in the final state , respectively
. the diffractive physics has been tested in hadron - hadron collisions considering distinct final states like dijets , electroweak vector bosons , dileptons , heavy quarks , quarkonium + photon , and different theoretical approaches ( see , e.g. , refs .
one of these approaches is the resolved pomeron model , proposed by ingelman and schlein in ref .
@xcite , which assumes the validity of the diffractive factorization formalism and that the pomeron has a partonic structure .
the basic idea is that the hard scattering resolves the quark and gluon content in the pomeron @xcite , which can be obtained by analysing the experimental data from diffractive deep inelastic scattering ( ddis ) at hera , providing us with the diffractive distributions of singlet quarks and gluons in the pomeron @xcite .
however , other approaches based on very distinct assumptions , for example the bfkl pomeron , are also able to describe the current scarce experimental data .
consequently , the present scenario for diffractive processes is unclear , motivating the study of alternative processes which could allow to constrain the correct description of the pomeron . in ref .
@xcite we have proposed , for the first time , the study of the photon production as a complementary test of diffractive processes and the pomeron structure . in this paper
we present a detailed analysis of the rapidity and transverse momentum dependence of the cross section for the diffractive double photon and photon+jet production .
moreover , a comparison with the inclusive production is presented .
our motivation is associated to the fact that the production of prompt photons is dominated at leading order by the qcd compton subprocess @xmath1 ( see , e.g. , @xcite ) , which makes this process a probe of the diffractive gluon _ and _ quark distributions , in contrast to diffractive heavy quark production which probes mainly the diffractive gluon distribution .
the content of this paper is as follows . in the next we present a brief review of the formalism for the prompt photon and double photon production in inclusive and diffractive processes . in section [ results ]
we present our predictions for the rapidity and transverse momentum distributions as well as for the total cross sections and event rates . in section [ conc ]
we summarize our main conclusions .
in the following we apply the resolved pomeron model @xcite for the diffractive photon production .
this model assumes that the pomeron has a well defined partonic structure and that the hard process takes place in a pomeron - proton or proton - pomeron ( single diffractive ) , or pomeron - pomeron ( double diffractive ) processes . at leading order the prompt photon production the hard process
is determined by the compton process @xmath2 , the annihilation process @xmath3 and the processes @xmath4 ( pure em ) , @xmath5 and @xmath6 @xcite .
higher order contributions are not considered here and can be taken into account effectively with a @xmath7 factor . in order to estimate the hadronic cross sections we have to convolute the cross sections for these partonic subprocesses with the inclusive and/or diffractive parton distribution functions , depending on the process under analysis .
the photon + jet production in single and double diffractive processes is described by the diagrams like those presented in fig .
[ single ] .
moreover , we include the contributions associated to the production of a photon and an unobserved photon ( @xmath8 ) . in single
prompt photon diffractive production the pomeron might be emitted from one of the two protons and one should include both @xmath9 and @xmath10 interactions .
the final state will be characterized by the presence of one rapidity gap .
the corresponding cross section may be written as ( see e.g. ref .
@xcite ) @xmath11 \frac{x_ax_b}{2x_a - x_te^y}\frac{d\hat{\sigma}}{d\hat{t}}(ab\rightarrow cd)\end{aligned}\ ] ] where @xmath12 and @xmath13 are the inclusive and diffractive parton distribution functions , respectively .
the momentum fraction variables are @xmath14 and @xmath15 , with @xmath16 and @xmath17 .
@xmath18 are the lo partonic cross sections for all considered subprocesses . in the case of
double diffractive prompt photon production ( also called central diffractive ) one has @xmath19 interactions and one looks for events with two rapidity gaps and a @xmath20jet at the central region .
the corresponding cross section is given by @xmath21 which is strongly sensitive to the pomeron structure due to the quadratic dependence on the diffractive parton distributions .
the double photon production in single and double diffractive processes are described by the diagrams like those shown in fig .
[ figsdddouble ] , taking into account the @xmath4 and @xmath5 subprocesses . besides the rapidity and transverse momentum distributions given above it is useful in the double photon production to analyze the distribution on the invariant mass of the @xmath22 system , for which the single diffractive processes can be written as @xmath23 \frac{x_ax_b m_{\gamma \gamma}}{1+\cosh(y_1-y_2)}\frac{d\hat{\sigma}}{d\hat{t}}(ab\rightarrow \gamma \gamma)\ , , \label{mass}\end{aligned}\ ] ] where @xmath24 $ ] , @xmath25 $ ] , and @xmath26 , @xmath27 are the rapidities of the outgoing photons . for double diffractive processes , the factor in the brackets in eq .
( [ mass ] ) is replaced by @xmath28 . in the present work ,
the diffractive parton distributions in the proton are taken from the resolved pomeron model @xcite , where they are defined as a convolution of the pomeron flux emitted by the proton , @xmath29 , and the parton distributions in the pomeron , @xmath30 , @xmath31 , where @xmath32 is the momentum fraction carried by the partons inside the pomeron .
the pomeron flux is given by @xmath33 , where @xmath34 and @xmath35 , @xmath36 are kinematic boundaries .
the pomeron flux factor is motivated by regge theory , where the pomeron trajectory assumed to be linear , @xmath37 , and the parameters @xmath38 , @xmath39 and their uncertainties are obtained from fits to h1 data @xcite .
the diffractive quark and gluon distributions are then given by @xmath40 in our analysis we use the diffractive parton distributions obtained by the h1 collaboration at desy - hera @xcite .
moreover , we use the inclusive parton distributions as given by the cteq6l parametrization @xcite . in order to obtain reliable predictions for the diffractive cross sections one should take into account that the qcd hard scattering factorization theorem for diffraction is violated in @xmath41 collisions by soft interactions which lead to an extra production of particles that destroy the rapidity gaps related to pomeron exchange .
the inclusion of these additional absorption effects can be parametrized in terms of a rapidity gap survival probability , @xmath42 , which corresponds to the probability of the scattered proton not to dissociate due to the secondary interactions .
these effects have been calculated considering different approaches giving distinct predictions ( see , e.g. ref .
an usual approach in the literature is the calculation of an average probability @xmath43 and after to multiply the cross section by this value . as previous studies for single and double diffractive production @xcite we also follow this simplified approach assuming @xmath44 for single diffractive processes and @xmath45 for double diffractive processes .
these values are taken from ref .
in what follows we present our results for the diffractive photon production at lhc energies .
initially we analyse the dependence of our predictions on the diffractive parton distribution used in the calculations . the h1 collaboration proposed two different parametrizations for the diffractive parton distributions ( denoted a and b in ref .
@xcite ) , which differ in the form of the parton densities at the starting scale for the qcd evolution and in the value of the effective pomeron intercept , being @xmath46 in fit a and @xmath47 in fit b. in fig .
[ h1fits ] ( left panel ) we present our results for the rapidity distribution for the photon+jet production in double diffractive processes ( denoted double diffractive prompt @xmath48 hereafter ) .
we consider this process for this analysis due to the quadratic dependence on @xmath49 .
the difference between the predictions is small ( @xmath50% ) , mainly in @xmath51 region .
similar results are obtained for the transverse momentum dependence , with the difference being at large-@xmath52 [ see fig .
[ h1fits ] ( right panel ) ] . in the following calculations
we will use the diffractive parton distributions as given by the fit b in ref .
@xcite .
lets analyse now the contribution of the different partonic subprocesses for the double diffractive prompt @xmath48 production . in fig .
[ canais ] the distinct contributions are explicitly presented .
as expected from the inclusive case , the cross section for this process is dominated by the qcd compton subprocess @xmath2 , with the annihilation channel @xmath3 being smaller by one order of magnitude . moreover ,
the double photon production is dominated by the @xmath5 subprocess . in fig .
[ fig : prompt ] we present our predictions for the prompt photon production in single and double diffractive processes . for the sake of comparison , we also show the inclusive ( non - diffractive ) lo predictions .
one observes a reduction of two orders of magnitude when going from the inclusive to single diffractive , and from single to double diffractive .
however , as verified in table [ tab1 ] , the diffractive cross sections and event rates / second are still sizeable and could be measured at the lhc . in our calculations of the event rates
we have assumed the design luminosity @xmath53 mb@xmath54s@xmath54 .
the results for double photon production are shown in figs .
[ fig : double ] and [ fig : doublem ] . compared with prompt photon production , one has much cleaner processes because of the absence of a produced jet . on the other hand ,
the cross sections are much smaller ( by three orders of magnitude ) , since only the subdominant subprocesses contribute . on the single diffractive case , one could in principle measure photon pairs with invariant mass up to higher values . for the double diffractive case ,
the distribution drops fast and then only lower values for the diphoton invariant mass would be accessible .
the total cross sections and event rates / second are presented in table [ tab1 ] .
.total cross sections and event rates / second for photon+jet and double photon production . [ cols="^,^,^,^",options="header " , ] some comments are in order .
firstly , the double photon production in central exclusive diffractive processes , @xmath55 , where nothing else is produced except the leading hadrons and the two photons , have been studied in @xcite , observed in @xmath56 collisions at tevatron @xcite and searched for at the lhc @xcite . at @xmath57 tev
, the cross section is predicted to be equal to 20 pb for diphoton masses
@xmath58 gev @xcite .
moreover , the exclusive double photon production in @xmath41 collisions was recently studied in ref .
@xcite considering the incident protons as photon emitters and the elastic light - by - light scattering ( @xmath59 ) , obtaining @xmath60 = 12 $ ] fb at @xmath57 tev .
in contrast , for the double diffractive @xmath22 production considered in this paper , which is characterized by the presence of remnants of the pomeron beyond the leading hadrons and the two photons , we predict @xmath61 pb .
if we consider the tevatron energy and the rapidity range probed in @xcite , given by @xmath62 , we predict @xmath63 = 6.6 pb , in contrast with the central exclusive prediction of 0.2 - 2 pb , depending on the unintegrated gluon distribution used in the calculations , which reasonably describes the experimental data .
it is important to emphasize that it is not obvious if the double diffractive and the central exclusive mechanisms could be differentiated experimentally at the lhc .
unfortunately , due to the high luminosity and large pile - up environment the separation of the diffractive processes considering the rapidity gaps and the detection or not of the remnants of the pomeron will be a hard task .
another comment is that distinctly from the heavy quark and dijet production , which can be produced in exclusive and inclusive double diffractive processes , the production of photon + jet can only occur in the latter case , which makes the experimental analysis of this final state ideal to study the validity of the resolved pomeron model .
moreover , as already emphasized in the introduction , in the photon + jet production we also are probing the diffractive quark distributions .
therefore , the combined analysis of this process and heavy quark or double photon production can be useful to constrain these distributions .
as a summary , in this paper we have presented a detailed analysis for the diffractive photon production in @xmath41 collisions at the lhc . we have discussed this process in the framework of the resolved pomeron model corrected for absorption effects , as used in the estimation of several other diffractive processes .
our results indicate that the experimental analysis is feasible and it would help in constrain the underlying model for the pomeron and the diffractive parton distributions .
machado , phys .
d * 76 * , 054006 ( 2007 ) ; m. b. gay ducati , m. m. machado , m. v. t. machado , phys .
rev . * d81 * , 054034 ( 2010 ) ; m. b. gay ducati , m. m. machado and m. v. t. machado , phys .
c * 83 * , 014903 ( 2011 ) c. brenner mariotto and v. p. goncalves , _ single and double diffractive prompt photon production at the lhc_. talk presented to the xxi international workshop on deep - inelastic scattering and related subjects , 22 - 26 april 2013 , marseille , france . to be published in the proceedings of science ( pos ) .
|
in this paper we study the photon production in single and double diffractive processes considering the resolved pomeron model .
we estimate the rapidity and transverse momentum dependence of the cross section for the diffractive double photon and photon+jet production . a comparison with the inclusive production is presented .
we predict large values for the total cross sections , which makes the experimental analysis of these observables feasible at lhc energies .
|
the phase structure of quantum many - body systems has in recent years been a subject of great experimental and theoretical interest .
models based upon algebraic hamiltonians are well - suited to the study of phase transitions . they possess a well - defined classical limit @xcite , allowing classical order parameters to be determined . and for certain specific forms of their hamiltonians , algebraic models exhibit dynamical symmetries , which correspond to qualitatively distinct ground - state equilibrium configurations . these constitute the phases of the system @xcite .
algebraic models have found extensive application to the spectroscopy of many - body systems , including nuclei @xcite and molecules @xcite . in the present work ,
the phase structure of a system comprised of two interacting fluids is investigated .
the phase structure of one - fluid algebraic models , especially the interacting boson model ( ibm ) @xcite for nuclei , has been studied in detail @xcite .
however , algebraic models may also be used to describe multi - fluid systems , with multiple interacting constituent species . while one - fluid systems are described by a single elementary lie algebra , usually @xmath1 ,
multi - fluid systems are described by a coupling of such lie algebras , @xmath2 @xcite .
a richer phase structure arises for multi - fluid models , not just from the greater number of control and order parameters afforded by the more complicated model , but more fundamentally from the coupling of multiple subsystems , each of which can exist in a different phase or can drive phase transitions of the composite system . here
we consider the proton - neutron interacting boson model ( ibm-2 ) @xcite , in which proton pairs and neutron pairs are treated as distinct constituents .
the one - fluid ibm , with @xmath3 algebraic structure , has three dynamical symmetries , separated by first and second order phase transitions @xcite .
the ibm-2 , with @xmath4 algebraic structure , supports four dynamical symmetries @xcite .
its phase diagram is therefore more involved and is found to possess qualitatively new features . due to the complexity of the problem ,
a combination of analytic and numerical methods have been applied in the present work .
preliminary results were reported in refs .
numerical studies of the ibm-2 phase structure have also been carried out by arias , dukelsky , and garca - ramos @xcite .
the ibm-2 and its classical limit are summarized in sec .
[ secdefs ] .
the phase diagram of the ibm-2 is first investigated for an essential hamiltonian with few parameters , for which the most complete analytic results are obtained ( sec .
[ secschematic ] ) .
the treatment is then extended to a more general hamiltonian , incorporating realistic quadrupole and majorana interactions ( sec .
[ secgeneral ] ) .
connection of the ibm-2 phase diagram with experimental data requires knowledge of the spectroscopic predictions across the phase transitions .
these are addressed in sec .
[ secspectro ] .
the results of sec .
[ secspectro ] are of relevance in the search for triaxial shapes in nuclei .
specific signatures of two - fluid triaxial deformation , and of the phase transition between axially symmetric and triaxial structure , are presented .
those signatures involving low - lying states are applicable both to current experiments and to experiments planned for next generation radioactive beam facilities .
those involving the high - lying magnetic dipole mode might be most directly investigated through resonance fluorescence experiments at high intensity gamma - ray facilities .
let us first summarize the ibm-2 hamiltonian and the dynamical symmetries it supports .
operators in the ibm-2 are constructed from the generators of the group @xmath5 , realized in terms of the boson creation operators @xmath6 and @xmath7 ( where @xmath8 represents @xmath9 or @xmath10 , and @xmath11@xmath12@xmath13 ) and their associated annihilation operators , acting on a basis of good boson numbers @xmath14 and @xmath15 .
the physically dominant interactions are contained in a hamiltonian @xmath16 where @xmath17 , @xmath18 , @xmath19 , and conventional spherical tensor coupling notation ( _ e.g. _ , ref .
@xcite ) has been used .
this hamiltonian contains one - fluid contributions , arising from like - nucleon pairing ( @xmath20 ) and quadrupole ( @xmath21 ) interactions , as well as two - fluid coupling terms , arising from proton - neutron quadrupole ( @xmath22 ) and majorana ( @xmath23 ) interactions .
the physically relevant ranges of the hamiltonian parameters are @xmath24 , @xmath25 , @xmath26 , and @xmath27 @xcite .
a dynamical symmetry occurs when , for certain values of the parameters , the hamiltonian is constructed from the casimir operators of a chain of subalgebras of @xmath5 .
three of the ibm-2 dynamical symmetries occur for @xmath28 and have direct analogues in the one - fluid ibm @xcite . when @xmath29 , the @xmath30 dynamical symmetry is realized , with subalgebra chain @xmath31 the geometric interpretation is that the proton and neutron fluids undergo oscillations about a spherical equilibrium configuration .
parameter values @xmath32 with @xmath33 produce the @xmath34 dynamical symmetry @xmath35 yielding deformed , @xmath36-unstable structure . and @xmath32 with @xmath37 gives the @xmath38 dynamical symmetry @xmath39 for which prolate axially symmetric structure is obtained . the complementary case @xmath40 , giving oblate axially symmetric structure , is distinguished by the notation @xmath41 .
however , a symmetry special to the ibm-2 , denoted @xmath0 , is obtained for @xmath32 with @xmath42 and @xmath43 @xcite . in this case
the hamiltonian is constructed from casimir operators of the subalgebra chain @xmath44 the equilibrium configuration consists of a proton fluid with axially symmetric prolate deformation coupled to a neutron fluid with axially symmetric oblate deformation , with their symmetry axes orthogonal to each other @xcite .
this yields an overall composite nuclear shape with triaxial deformation , as shown in fig .
[ figortho ] . to avoid ambiguity
, we shall adopt here the notation @xmath45 for the complementary case @xmath46 and @xmath47 , in which the proton and neutron deformations are interchanged . dynamical symmetry .
a prolate deformed proton fluid ( light gray ) and oblate deformed neutron fluid ( dark gray ) are coupled with orthogonal symmetry axes .
figure adapted from ref .
[ figortho ] ] the phase structure of an algebraic model is determined from the classical limit of the model , which is obtained through the coherent state formalism @xcite .
a coherent state of good boson number is constructed , and the parameters of this state are directly related to the classical coordinates of the system .
this coherent state is used as a variational trial state : the global minimum of the energy surface @xmath48 is used to determine the ground state energy and equilibrium classical coordinates of the system . in the limit of infinite boson number , the coherent state variational
energy converges to the true ground state energy @xcite .
detailed coherent state analyses of the one - fluid ibm may be found in refs .
@xcite .
once the equilibrium properties of the model have been established , ground state phase transitions can be identified and categorized according to the ehrenfest classification @xcite : a phase transition is first order if the first derivative of the system s energy is discontinuous with respect to the control parameter being varied , second order if the second derivative is discontinuous , _ etc . _ where the system s energy is obtained , as in the present classical analysis , as the global minimum of an energy function @xmath49 , a first order transition is usually associated with a discontinuous jump in the equilibrium coordinates ( `` order parameters '' ) between distinct competing minima .
second or higher order transitions are associated instead with a continuous evolution of the equilibrium coordinates , as when an initially solitary global minimum becomes unstable ( possessing a vanishing second derivative with respect to some coordinate ) and evolves into two or more minima .
it should be noted that , whenever the order of a phase transition is obtained by numerical analysis , application of the ehrenfest criterion is limited by the ability to numerically resolve sufficiently small discontinuities .
this is especially a consideration for points of first - order transition very close to a point of second - order transition .
moreover , problems with the classification scheme , not addressed here , may arise at the boundaries of the parameter space or when the hamiltonian possesses additional symmetries .
for a two - fluid model , the boson number for each constituent fluid is conserved separately .
the coherent state is constructed not only with good total boson number but with good boson number for each fluid .
for the ibm-2 , the number operators are @xmath50 .
the ibm-2 coherent state @xcite is defined in terms of proton and neutron condensate bosons @xmath51 where @xmath52 , as @xmath53 the @xmath54 are interpreted geometrically as quadrupole shape variables @xcite for the proton and neutron fluids .
they are related to the four deformation parameters ( @xmath55 , @xmath56 , @xmath57 , and @xmath58 ) and to the six euler angles ( @xmath59 , @xmath60 , @xmath61 , @xmath62 , @xmath63 , and @xmath64 ) specifying the orientations of the proton and neutron intrinsic frames by @xmath65,\end{gathered}\ ] ] where @xmath66 is the wigner @xmath67 function @xcite .
calculation of the expectation value of @xmath68 with respect to the coherent state ( [ eqncoherentpn ] ) yields an energy surface latexmath:[$\scre\equiv \langle n_\pi,\alpha^{(2)}_\pi;n_\nu,\alpha^{(2)}_\nu| h invariance , @xmath49 can only depend upon the _ relative _ euler angles @xmath70 between the proton and neutron fluid intrinsic frames , not the @xmath71 and @xmath72 separately .
for evaluation of @xmath49 , the euler angles may thus simply be chosen to be @xmath73 and @xmath74 .
investigations of the ibm-2 coherent state energy surface have been carried out in refs .
@xcite . the expectation value with respect to the ibm-2 coherent state of an operator constructed from the boson operators of one fluid only ( _ e.g. _ , @xmath75 or @xmath76 ) can be calculated as in the one - fluid ibm , by the methods of refs .
the expectation values of the one - fluid operators appearing in ( [ eqnhgeneralraw ] ) are @xcite @xmath77\\ & \qquad \qquad \qquad + \frac{n_\rho(n_\rho-1)}{(1+\beta_\rho^2)^2 } \bigl [ 4\beta_\rho^2 - 4\sqrt\frac{2}{7}\chi_\rho\beta_\rho^3\cos 3\gamma_\rho + \frac{2}{7}\chi_\rho^2\beta_\rho^4\bigr ] .
\notag\end{aligned}\ ] ] the expectation value of a two - fluid operator constructed as the scalar product of one - fluid operators ( _ e.g. _ , @xmath78 ) can be obtained using a factorization result ( * ? ? ?
. however , for more complicated operators , the general method presented in appendix [ appcoherentme ] provides a convenient means of calculation .
the expectation value of @xmath79 is a function of all seven possible coordinates ( @xmath55 , @xmath56 , @xmath57 , @xmath58 , @xmath80 , @xmath81 , and @xmath82 ) , @xmath83 \\
\cdot [ \alpha_\nu^{(2)\,*}+\tilde\alpha_\nu^{(2)}+\chi_\nu(\alpha_\nu^{(2)\,*}\times\tilde\alpha_\nu^{(2)})^{(2 ) } ] , \end{gathered}\ ] ] with the @xmath84 as defined in eqn .
( [ eqnalpharho ] ) , which for vanishing euler angles simplifies to @xmath85 \bigr].\end{gathered}\ ] ] the expectation value of the majorana operator is a complicated function of the coordinates . for vanishing euler angles , it reduces to @xmath86 .\ ] ] a detailed study of the possible equilibrium euler angle and @xmath87 values for the most general two - body ibm-2 hamiltonian has been presented in ref .
so long as the multipole decomposition of the proton - neutron interaction contains no hexadecapole component [ @xmath88 , it is found that the equilibrium configuration must have euler angles which are vanishing or multiples of @xmath89 .
thus , the proton and neutron intrinsic frames are `` aligned '' , to within a possible relabeling of axes . [ for clarity , we note that in the @xmath0 equilibrium configuration ( fig .
[ figortho ] ) , even though the proton and neutron _ symmetry _ axes are orthogonal , the intrinsic frame _ coordinate _ axes are actually parallel , _
i.e. _ , the euler angles vanish .
the proton ( @xmath90 ) symmetry axis is the @xmath91 axis , while the neutron ( @xmath92 ) symmetry axis is the @xmath93 axis . ]
if , instead , a hexadecapole contribution is present , `` oblique '' equilibrium configurations are possible .
the operator @xmath22 contains no hexadecapole component , but the majorana operator does ( sec . [ subsecmajorana ] ) .
the ibm-2 is thus characterized by either four order parameters ( @xmath55 , @xmath56 , @xmath57 , and @xmath58 ) or seven order parameters ( @xmath55 , @xmath56 , @xmath57 , @xmath58 , @xmath80 , @xmath81 , and @xmath82 ) , depending upon the form of the proton - neutron interaction .
a simple , schematic hamiltonian which retains the essential dynamical symmetry features of the ibm-2 , obtained as a special case of eqn .
( [ eqnhgeneralraw ] ) , is @xmath95 this hamiltonian is often preferred for theoretical studies due to its invariance under @xmath94-spin rotations ( _ e.g. _ , ref .
@xcite ) for @xmath96 . to make the analysis of the ibm-2 phase structure as transparent as possible ,
let us first consider in detail this @xmath94-spin invariant hamiltonian before proceeding in section [ secgeneral ] to the more general form .
it is also convenient for the moment to restrict attention to the case @xmath97 . to obtain the classical limit
, it is necessary to take @xmath98 , where @xmath99 . in this limit , the lower - order terms with respect to @xmath100 in eqn .
( [ eqnqrhoqrho ] ) become negligible , and it is seen that the quantities @xmath101 are linear in @xmath100 , while the @xmath102 are all quadratic in @xmath100 . to prevent the @xmath101 contributions from vanishing identically when the limit is taken , it is necessary to rescale the hamiltonian coefficients by appropriate powers of @xmath100 .
it is thus convenient to reparametrize the hamiltonian ( [ eqnhdieperinkraw ] ) as @xmath103 so that the energy function @xmath49 is independent of @xmath100 .
this definition also condenses the full range of possible ratios @xmath104 onto the finite interval @xmath105 .
an overall normalization parameter for @xmath68 has been discarded as irrelevant to the extremum structure of the energy surface .
there are three control parameters @xmath106 , @xmath107 , and @xmath108 for the hamiltonian ( [ eqnhdieperink ] ) .
it is convenient to alternatively introduce `` scalar '' and `` vector '' parameters @xmath109@xmath110@xmath111 and @xmath112@xmath110@xmath113 .
the parameter space is outlined in fig .
[ figtetraspace ] .
-spin invariant hamiltonian of eqn .
( [ eqnhdieperink ] ) . for @xmath114 ( shaded plane )
the analysis reduces to that for the one - fluid ibm ( see text ) , yielding a point of second - order phase transition embedded in a curve ( [ eqnxipccurve ] ) of first - order phase transition points .
the axes cover ranges @xmath115 , @xmath116 , and @xmath117 . the @xmath109 and @xmath112 axes are scaled by @xmath106 , so as to converge to a point at the @xmath30 limit , where the hamiltonian is independent of @xmath109 and @xmath112 . figure from ref .
@xcite . ] for @xmath114 , indicated by the shaded triangle in fig . [ figtetraspace ] , the analysis essentially reduces to that for the one - fluid ibm .
the equilibrium configuration occurs for @xmath118 and @xmath119 and is identical to that obtained for the ibm hamiltonian @xmath120 \nhat_d -(\xi'/n^2)\qhat^{\chi}\cdot \qhat^{\chi}$ ] , the phase structure of which is well known @xcite , with @xmath121 . for @xmath122 ,
the energy surface has a minimum at @xmath123 . at @xmath124 ,
this minimum becomes unstable with respect to @xmath125 , _
i.e. _ , @xmath126 . if @xmath127 , this instability leads to a second - order transition between undeformed ( @xmath123 ) and deformed ( @xmath128 ) structure at @xmath124
. however , for any other value of @xmath129 , the minimum at @xmath123 is preempted as global minimum by a distinct minimum with nonzero @xmath125 before @xmath124 .
this leads to a first - order phase transition , at a parameter value @xmath130 given by @xmath131 easily derived from ref .
* ( 2.9b ) ) .
thus , the point of second - order phase transition at @xmath124 and @xmath127 lies on a trajectory ( [ eqnxipccurve ] ) of first - order transition points ( fig .
[ figtetraspace ] ) .
it simplifies the analysis of the phase diagram for the remainder of the parameter space to note the presence of reflection symmetries with respect to the @xmath132 parameters .
each term contributing to @xmath49 is seen by inspection of eqns .
( [ eqnndrho ] ) , ( [ eqnqrhoqrho ] ) , and ( [ eqnqpqnaligned ] ) to be invariant under a simultaneous transformation of coordinates and hamiltonian parameters @xmath133 thus , the equilibrium deformation at a point @xmath134 in parameter space is simply related to that at the point @xmath135 by eqn .
( [ eqnchiinversion ] ) .
the phase diagram is symmetric under simultaneous negation of @xmath107 and @xmath108 or , equivalently , simultaneous negation of @xmath109 and @xmath112 .
moreover , for the @xmath94-spin invariant hamiltonian ( [ eqnhdieperink ] ) with @xmath136 , the numerical coefficients on corresponding proton and neutron terms in @xmath49 are equal .
thus , @xmath49 is symmetric under interchange of all proton and neutron variables , and the phase diagram is symmetric under interchange of @xmath107 and @xmath108 or , equivalently , negation of @xmath112 .
let us begin the investigation of the phase structure of the hamiltonian ( [ eqnhdieperink ] ) with an analytic study for @xmath138 , corresponding to the rightmost plane of the parameter space diagram in fig . [ figtetraspace ] .
this plane encompasses the @xmath34 , @xmath38 , and @xmath0 dynamical symmetries .
surrounding the @xmath38 dynamical symmetry is a region of parameter space in which the equilibrium deformations are axially symmetric ( @xmath119 ) , and a similar region surrounds the @xmath41 dynamical symmetry ( @xmath139 ) . in either of these cases ( @xmath119 or @xmath140 )
, the energy surface can be written in an especially simple form in terms of the function @xmath94 defined in eqn .
( [ eqnf ] ) , @xmath141 ^ 2,\ ] ] where @xmath142 . the energy function in ( [ eqnscreaxial ] ) is minimized when the magnitude of the quantity within brackets is maximized . in the vicinity of @xmath38 , including in the entire quadrant for which @xmath107 and @xmath108 are both negative , it follows from the results of appendix [ appf ] that the minimum in @xmath143 occurs at coordinate values @xmath144 ^{1/2}\ ] ] with @xmath145 ( _ i.e. _ , @xmath146 ) .
similary , in the vicinity of @xmath41 , including in the entire quadrant for which @xmath107 and @xmath108 are both positive , the minimum in @xmath143 occurs at coordinate values @xmath147 ^{1/2}\ ] ] with @xmath148 ( _ i.e. _ , @xmath149 ) . in either case , the value of the energy function at the global minimum is @xmath150 with @xmath55 and @xmath57 given by eqn .
( [ eqnbetaprolate ] ) or ( [ eqnbetaoblate ] ) as appropriate . as @xmath107 and @xmath108
are varied away from their @xmath38 values , eventually axial equilibrium deformation gives way to triaxial deformation , with @xmath56 and/or @xmath58 nonzero .
this transition occurs on the locus of points at which the minimum given by ( [ eqnbetaprolate ] ) first becomes unstable with respect to @xmath36 deformation .
since @xmath49 depends upon both @xmath56 and @xmath58 , instablility occurs when the _ directional _ second derivative of @xmath49 first vanishes along some `` direction '' in @xmath151 coordinate space , which may generally happen before either @xmath152 or @xmath153 vanishes individually .
the equation describing the boundary curve in @xmath107 and @xmath108 is most compactly expressed in terms of the corresponding equilibrium values @xmath55 and @xmath57 from eqn .
( [ eqnbetaprolate ] ) or ( [ eqnbetaoblate ] ) , as @xmath154 this curve is shown in fig .
[ figboundary ] . the direction in @xmath151 space in which the instability occurs , at any given point on this boundary ,
is given by @xmath155 where @xmath156 along the @xmath38-@xmath0 line in particular , for which @xmath42 , the transition occurs at @xmath157 , obtained from eqn .
( [ eqnboundary ] ) as the root of a quartic equation . in this special case ,
the global minimum becomes soft purely with respect to @xmath58 at fixed @xmath90 .
-@xmath158-@xmath159 plane ( @xmath138 ) in the parameter space of the hamiltonian of ( [ eqnhdieperink ] ) , for @xmath136 , showing the curve ( [ eqnboundary ] ) of second - order phase transition .
the diagram is rotated to allow more direct comparison with figs .
[ figtetraspace ] and [ figquadrant ] .
figure adapted from ref .
@xcite . ]
the analytic results of this section indicate a continuous transition between axially symmetric and triaxial structure on the curve ( [ eqnboundary ] ) . in sec .
[ subsecnumerical ] it is verified numerically that this transition is second , rather than higher , order , and that no prior phase transition to a distinct minimum occurs before the curve ( [ eqnboundary ] ) is reached .
the transition between spherical and deformed equilibrium for the ibm-2 for arbitrary @xmath112 is closely related to the transition which occurs in the special plane @xmath114 ( shaded in fig . [ figtetraspace ] ) .
the energy function @xmath160 obeys the identity @xmath161 _ i.e. _ , for a deformation which is axial and has @xmath162 , the value of @xmath49 is independent of @xmath112 . recall that the equilibrium deformation for @xmath114 is always axial with @xmath162 ( sec .
[ subsecdieperink ] ) .
it is straight - forward to demonstrate from these observations ( see below ) that a nonzero equilibrium deformation for the parameter point @xmath163 in the @xmath114 plane implies a nonzero equilibrium deformation for all the out - of - plane points @xmath164 .
thus , the @xmath106 value @xmath165 for which the spherical - deformed transition occurs at a given @xmath109 and @xmath112 obeys , by eqn .
( [ eqnxipccurve ] ) , @xmath166 the detailed argument proceeds as follows .
let the equilibium energy at a general point in parameter space be denoted by @xmath167 , and let @xmath168 be the equilibrium value of @xmath55 and @xmath57 for @xmath114 as determined , _ e.g. _ , from ref .
* ( 2.12 ) ) . then @xmath169 $ ] . the global minimum energy for arbritrary
@xmath112 then is subject , by eqn .
( [ eqnscreslice ] ) , to the upper bound @xmath170\\ & = \scre[\xi',\chi_s,0;\beta_0(\xi',\chi_s),0,\beta_0(\xi',\chi_s),0]=\scre_0(\xi',\chi_s,0 ) .
\end{aligned}\ ] ] note that @xmath49 evaluated at zero deformation is zero for the hamiltonian considered here ( [ eqnhgeneralraw ] ) , so @xmath171 if and only if the configuration is deformed . if @xmath172 , then @xmath173 for all @xmath112 by eqn .
( [ eqnscrebound ] ) , and the equilibrium deformation for parameter point @xmath164 is also nonzero . to further investigate the nature of the spherical - deformed transition ,
let us consider the stability of the minimum of @xmath49 at @xmath174 .
instability occurs when the second derivative of @xmath49 vanishes along some `` direction '' , which we denote @xmath175 , in @xmath176 coordinate space , for some fixed values of @xmath56 and @xmath58 .
the second derivative of @xmath49 along the ray parametrized as @xmath177 and @xmath178 is @xmath179 this is independent of @xmath109 and @xmath112 and depends upon the @xmath87 coordinates only through their difference @xmath180 .
the smallest @xmath106 value at which the second derivative ( [ eqnbetasoft ] ) vanishes is @xmath124 , with @xmath181 , indicating instability of the minimum against deformations with @xmath162 and @xmath182 .
as the second derivative is independent of @xmath36 , the system is unstable against deformations of all possible @xmath36 values , representing prolate , oblate , and intermediate triaxial deformations , simultaneously . for @xmath183
, it is apparent from eqn .
( [ eqnxipcbound ] ) that the spherical - deformed phase transition occurs _ before _
the spherical minimum becomes unstable ( @xmath124 ) , and so the transition is a first - order transition to a distinct minimum .
thus , the curve of first - order phase transition occuring in the plane @xmath114 `` propagates '' out of this plane to form a surface of first - order transition . in the approximation that @xmath184 for the deformed minimum , the surface would be a vertical extension of the one - fluid ibm transition trajectory in fig .
[ figtetraspace ] .
the deviation of the surface from this limiting location is studied numerically in sec .
[ subsecnumerical ] . for @xmath185
, however , it is possible for the minimum at zero deformation to remain the global minimum until @xmath124 , leading to a second - order phase transition .
let us consider the likely properties of any alternative lower , deformed global minimum ( these are confirmed numerically in sec .
[ subsecnumerical ] ) . by symmetry in the proton and neutron parameter magnitudes ( @xmath186 ) , it is reasonable to expect this minimum to have @xmath162 .
a minimum with axial deformation has been excluded by eqn .
( [ eqnxipccurve ] ) , but a minimum with triaxial deformation must be considered .
since @xmath185 represents the `` reflection '' plane in parameter space between prolate and oblate structure ( [ eqnchiinversion ] ) , it is natural to expect the minimum to have @xmath187 .
inspection of the restricted energy function @xmath188 reveals that a triaxial deformed minimum does preempt the minimum at zero deformation as global minimum [ at @xmath189 but that this only occurs for @xmath190 , which lies outside the parameter range @xmath191 of interest in the present study . in conclusion
, a line of second - order phase transition between spherical and deformed equilibrium occurs at @xmath124 and @xmath185 , subsuming the one - fluid ibm second - order transition point .
this line is embedded in a surface of first - order transition points .
the remainder of the phase diagram is obtained by numerical minimization of the energy surface with respect to @xmath55 , @xmath56 , @xmath57 , and @xmath58 .
this minimization provides the equilibrium coordinate values at any point in the parameter space of fig .
[ figtetraspace ] . for robust identification of the global minimum
, @xmath49 is first evaluated at each point on a fine mesh in the deformation coordinates ( @xmath192 and @xmath193 provides sufficient resolution for most of the calculations shown here ) .
all points which are discrete local minima of @xmath49 relative to the neighboring mesh points are identified .
the coordinate values for these minima are then refined iteratively by a conjugate gradient method .
the global minimum is identified from among the refined values .
the phase diagram for the @xmath94-spin invariant hamiltonian with @xmath194 , obtained numerically in this fashion , is shown in fig . [ figquadrant ] .
[ only one quadrant of parameter space is included in this plot , since the others may be obtained by reflection ( sec .
[ subsecdieperink ] ) . ]
the line of second - order transition between spherical and deformed equilibrium at @xmath124 and @xmath195 is the locus of simultaneous contact of four regions of the phase diagram , those of spherical , prolate axially symmetric , oblate axially symmetric ( not shown , in adjacent quadrant ) , and triaxial equilibria
. numerical investigation of the behavior of @xmath49 at points on the axial - triaxial transition surface allows the ehrenfest criterion to be applied , and it appears that the transition is everywhere second order . -spin invariant hamiltonian ( [ eqnhdieperink ] ) with @xmath136 , as obtained by numerical minimization of @xmath49 .
the surfaces of first - order and second - order transition between regions of undeformed , axially symmetric deformed , and triaxially deformed equilibria are shown .
only one `` quadrant '' of parameter space ( @xmath115 , @xmath196 , and @xmath197 ) is included in this plot , since the others may be obtained by reflection .
the @xmath109 and @xmath112 axes are scaled by @xmath106 , so as to converge to a point at the @xmath30 limit .
figure from ref .
@xcite . ]
we now consider in detail the evolution of the equilibrium energy and coordinates at selected locations throughout the phase diagram . within the plane @xmath138 ,
the boundary curve separating axially symmetric and triaxial equilibria was established in sec .
[ subsecbackplane ] .
the equilibrium properties along the @xmath38-@xmath0 transition line are shown in fig .
[ figrayssu3su3star ] .
several characteristics may be noted .
the second derivative of @xmath49 with respect to the control parameter @xmath198 is discontinuous , as in an ehrenfest second - order phase transition .
the first derivative of one order parameter , @xmath199 , is infinite at the critical point , with an approximately square - root dependence upon the control parameter after this point , as in a landau second - order phase transition @xcite .
the first derivative of @xmath200 at this point is , in contrast , discontinous but finite . in the special case of the @xmath38-@xmath0 line , as already discussed , the two other order parameters @xmath201 and @xmath202 completely decouple from the transition process , remaining constant throughout . , ( b ) @xmath57 , and ( c ) @xmath58 along the line in parameter space between the @xmath38 and @xmath0 dynamical symmetries ( @xmath203 , @xmath204 ) , obtained numerically .
the dashed line indicates the second - order transition point , @xmath157 . ] in fig .
[ figrayshorizontal ] , the equilibrium properties on lines of constant @xmath205 in the plane @xmath138 , which would appear as horizontal lines across the diagram of fig .
[ figboundary ] , are shown .
these lines cross both the prolate - triaxial and triaxial - oblate second - order transition curves .
the equilibrium coordinate values for positive @xmath206 are related to those for negative @xmath206 by the transformations of sec .
[ subsecdieperink ] . observe that the order parameters ( @xmath55 , @xmath56 , @xmath57 , and @xmath58 ) are coupled in their behavior at the transition points , all simultaneously exhibiting discontinuities in their derivatives at each transition point . in the limiting case @xmath207 [ fig .
[ figrayshorizontal](right ) ] , the two points of second order transition at positive and negative @xmath206 converge to a single point at @xmath195 .
the discontinuities in the second derivative of @xmath49 with respect to @xmath206 at the two transition points [ fig .
[ figrayshorizontal](left , middle ) ] combine to form a `` cusp '' in @xmath49 . the discontinuity of @xmath208 at @xmath127 was noted in the context of the one - fluid ibm by jolie _ et al . _
such a discontinuity in the slope of @xmath49 would , according to the ehrenfest criterion , indicate a first - order phase transition .
however , the present case provides an example of the great variety of phenomena possible in multi - parameter problems , as is does not naturally fit the one - dimensional laundau model for a first - order phase transition .
the conventional landau first - order phase transition @xcite involves `` competition '' between two distinct local minima simultantously present in the energy surface , one overtaking the other as global minimum ; whereas , in the present example , a single minimum with @xmath209 is present for @xmath210 , a single minimum with @xmath211 is present for @xmath212 , and a continuous trajectory of minima of arbitary @xmath213 , arising from the @xmath214 symmetry of the hamiltonian , are present for @xmath195 .
this example suggests the need for a comprehensive extension of the classification schemes for single - parameter phase transitions to cover phase transitions with multiple order parameters ( see also ref .
* chapter 17 ) ) . in the @xmath138 plane of parameter space , passing from the region of axially symmetric prolate equilibrium to that of axially symmetric oblate equilibrium .
results are shown for the @xmath94-spin invariant hamiltonian ( [ eqnhdieperink ] ) with @xmath194 , along the rays of ( left ) @xmath215 , ( center ) @xmath216 , and ( right ) @xmath207 .
proton fluid variables ( @xmath55 and @xmath56 ) are represented by solid curves , while neutron fluid variables ( @xmath57 and @xmath58 ) are represented by dashed curves .
the two points of second order phase transition ( left , middle ) , converging towards the single point @xmath195 in the limit @xmath207 ( right ) , are marked with arrows . ] finally , we consider the transition between spherical and deformed equilibrium configurations . in sec .
[ subsecspherical ] , an upper bound , @xmath217 was placed upon the location of the spherical - deformed transition surface [ see eqn .
( [ eqnxipcbound ] ) ] , equality holding in the case @xmath218 .
the actual shape of this surface , visible on a coarse scale in fig .
[ figquadrant ] , is shown in detail in fig .
[ figu5cuts ] .
several `` vertical '' slices through the surface , at constant @xmath206 , are plotted .
the line of second - order phase transition at @xmath124 appears rightmost in the figure .
the curves vary little in @xmath106 ( @xmath219 ) below their upper bound . .
the @xmath106 value at @xmath207 is the value @xmath220 from the one - fluid ibm ( [ eqnxipccurve ] ) , and the recession of each curve towards smaller @xmath106 above @xmath207 indicates the extent to which @xmath221 recedes below the upper limit ( [ eqnxipcbound ] ) obtained using @xmath218 .
the curves shown are for @xmath195 to @xmath222 in steps of @xmath223 ( right to left ) . ] the evolution of equilibrium properties across the spherical - deformed transition is shown in fig .
[ figraysu5 ] .
the second - order transition along the @xmath30-@xmath34 line and the first - order transition along the @xmath30-@xmath38 line , quantitatively identical to the corresponding transitions in the one - fluid ibm , are shown for reference [ fig .
[ figraysu5](left , middle ) ] . the transition along the @xmath30-@xmath0 line [ fig .
[ figraysu5](right ) ] is strongly constrained by the proton - neutron interchange symmetry about @xmath195 , with @xmath218 and with @xmath202 and @xmath199 symmetric to each other about @xmath224 .
-spin invariant hamiltonian ( [ eqnhdieperink ] ) with @xmath194 , along ( left ) the @xmath225 transition line ( @xmath195 , @xmath207 ) , ( middle ) the @xmath226 transition line ( @xmath227 , @xmath207 ) , and ( right ) the @xmath228 transition line ( @xmath195 , @xmath229 ) .
proton fluid variables are represented by solid curves , while neutron fluid variables are represented by dashed curves .
equilibrium @xmath230 values are undefined , and therefore not shown , when @xmath231 ( h , i ) and along the entire @xmath225 transition line ( g ) .
points of phase transition are marked with arrows . ]
in the preceding section , the general features of the ibm-2 phase diagram were established using a schematic hamiltonian , and only the case @xmath136 was considered .
we now address the more general hamiltonian of eqn .
( [ eqnhgeneralraw ] ) and also consider arbitrary values of the ratio @xmath232 .
these two generalizations are closely related , as the coefficients in the energy surface @xmath49 depend upon the hamiltonian coefficients and the boson numbers in combination . to make explicit
the dependence of each term contributing to @xmath49 upon parameters , boson numbers , and coordinates , we express @xmath49 as @xmath233 the functions @xmath234 ( _ e.g. _ , @xmath235 ) may be read off directly from eqns .
( [ eqnndrho])([eqnmaligned ] ) .
the limit of both large @xmath14 and large @xmath15 has been taken , so that in @xmath236 the term linear in @xmath237 is suppressed .
the energy surface ( [ eqnegeneralraw ] ) is entirely determined by the boson - number - weighted coefficients @xmath238 , @xmath239 , and @xmath240 , together with the @xmath132 , in terms of which @xmath241 the dominant role of the proton - neutron quadrupole interaction in producing collective nuclear deformation has been well established @xcite .
consequently , the like - nucleon quadrupole interactions are often neglected in the ibm-2 hamiltonian ( see refs .
microsopic estimates suggest that the shell model proton - proton and neutron - neutron quadrupole interactions each have @xmath2421/10 to 1/5 the strength of the proton - neutron quadrupole interaction @xcite . however , within the ibm-2 , significant further strength is added to the effective like - nucleon interactions by renormalization effects arising from elimination of @xmath243-wave nucleon pairs from the model space
. this may yield like - nucleon coupling strengths comparable to the proton - neutron coupling strength @xcite .
the actual coupling strengths are a subject for further phenomenological study .
examples spanning a considerable range of values for @xmath244 are included in the following analysis . for investigation of the phase diagram for the hamiltonian ( [ eqnhgeneralraw ] )
, it is convenient to again define a transition parameter @xmath106 controling the relative weights of the @xmath245 operator and quadrupole operator in the hamiltonian .
this provides a coordinate system for the parameter space like that in fig .
[ figtetraspace ] .
the contributions of the different terms within each of these operators can then be specified by parameters @xmath246 and @xmath247 defined such that @xmath248 yielding a @xmath237-independent expression @xmath249 for the energy surface . to unambiguously define @xmath106 and avoid redundancy in the parameters , we adopt the normalization convention @xmath250 and @xmath251 .
this choice gives @xmath252 and @xmath253 and is consistent with the definition of @xmath106 in eqn .
( [ eqnhdieperink ] ) . for the @xmath94-spin invariant hamiltonian with arbitrary @xmath254 ,
the parameters are @xmath255 , @xmath256 , @xmath257 , @xmath258 , and @xmath259 .
analytic results for the phase structure of the hamiltonian ( [ eqnhgeneralraw ] ) with arbitrary quadrupole coupling coefficients can be obtained in the case @xmath138 by a straight - forward extension of the analysis described in sec .
[ subsecbackplane ] . surrounding the @xmath38 and @xmath41 dynamical symmetries are regions of parameter space in which the equilibrium deformation is axially symmetric ( @xmath119 or @xmath140 ) .
the energy surface in this case can be expressed in terms of the function @xmath94 ( of appendix [ appf ] ) as @xmath260,\end{gathered}\ ] ] with @xmath261 .
this expression is simply a quadratic form in the quantities @xmath94 , and the extremization problem reduces to a constrained minimization of the quadratic form with respect to these quantities .
since the @xmath262 are limited to positive values , the @xmath94 are restricted to the rectangular region @xmath263 and @xmath264 for the case @xmath265 [ @xmath38-like ] or to @xmath266 and @xmath267 for @xmath268 [ @xmath41-like ] .
provided @xmath269 , @xmath270 , and @xmath271 are all negative , the global minimum is again given by eqn .
( [ eqnbetaprolate ] ) for @xmath265 or eqn .
( [ eqnbetaoblate ] ) for @xmath268 .
the energy at the minimum is , in terms of the equilibrium @xmath262 values , @xmath272 the boundary curve separating axial and triaxial deformations is found as in sec .
[ subsecbackplane ] from analysis of the second derivatives of @xmath49 .
the axial minimum becomes unstable with respect to @xmath36 deformation along a curve in parameter space , again most conveniently expressed in terms of the equilibrium @xmath262 values , given by @xmath273 the instability occurs against deformations with a ratio @xmath274 given by eqn .
( [ eqngammadir ] ) with the values @xmath275 along the @xmath38-@xmath0 line in parameter space , the location of the phase transition is obtained by solution of the quartic equation @xmath276 and a similar relation with interchanged proton and neutron labels holds for the @xmath38-@xmath45 line .
observe that the @xmath108 value at which the transition occurs depends only upon the ratio @xmath277 .
the dependence of the second - order transition curve upon @xmath232 is illustrated in fig .
[ figboundaryquadnratio ] .
since @xmath278 , the phase diagram is not symmetric under reflection with respect to @xmath206 or @xmath205 but remains symmetric under inversion ( [ eqnchiinversion ] ) .
it is seen that increasing @xmath232 from unity moves the @xmath38-@xmath0 transition farther away in parameter space from @xmath38 while moving the @xmath38-@xmath45 transition closer .
the @xmath232 depencence of the boundary curve raises the possibility of considering phase transitions at a fixed point in the hamiltonian parameter space with @xmath232 as the control parameter , _
i.e. _ , phase transitions as a function of particle number .
, shown for the @xmath94-spin invariant hamiltonian with different boson number ratios @xmath232 : ( a ) @xmath279 and ( b ) @xmath280 . the boundary curve for @xmath281 ( fig . [ figboundary ] ) is shown in each panel ( dashed curve ) for comparison . ] for the @xmath94-spin invariant hamiltonian considered in sec .
[ secschematic ] , the transition between prolate axially symmetric equilibrium [ @xmath38-like ] and oblate axially symmetric equilibrium [ @xmath41-like ] always proceeds through an intermediate stage of triaxial equilibrium , except at the point in parameter space corresponding to @xmath34 .
however , if @xmath282 and @xmath283 are sufficiently small relative to @xmath284 , it is possible for a direct transition to occur between the prolate and oblate equilibria of eqns .
( [ eqnbetaprolate ] ) and ( [ eqnbetaoblate ] ) without either minimum becoming unstable with respect to triaxiality and thus for the regions of prolate and oblate equilibrium in the phase diagram to share a common boundary .
the boundary curve is given by @xmath285 where @xmath286 . over the range of @xmath287 and parameter values considered ,
this curve differs little from its small-@xmath287 approximation , the line @xmath288 . for any proton - neutron symmetric energy surface ( @xmath289 ) ,
the boundary curve ( [ eqnflipquad ] ) reduces to the line @xmath195 .
the general problem of determining whether or not the prolate and oblate regions of the phase diagram share a common boundary involves solving for an intersection of the curves ( [ eqnboundaryquad ] ) and ( [ eqnflipquad ] ) .
for the proton - neutron symmetric case ( @xmath289 ) , a prolate - oblate boundary arises for @xmath290 . the dependence of the boundary curve upon the relative strengths of like - nucleon and proton - neutron quadrupole coupling coefficients is shown in fig . [ figboundaryquadkapparatio ] . as the @xmath291
are reduced in strength relative to @xmath292 , the region of triaxial equilibrium contracts .
its separation from the origin and the onset of a direct prolate - oblate transition for @xmath293 is illustrated in fig .
[ figboundaryquadkapparatio](b ) .
the parameter values chosen for fig . [ figboundaryquadnratio](a ) and for fig .
[ figboundaryquadkapparatio](a ) are such that @xmath277 is the same ( @xmath294 ) in both cases . thus , though the curves in these two figures are quite different overall , by eqn .
( [ eqnquarticquad ] ) they share the same endpoint ( @xmath295 ) on the @xmath38-@xmath0 line .
numerical results for the equilibrium values of the energy and coordinates along the @xmath38-@xmath0 line for different relative strengths of like - nucleon and proton - neutron quadrupole interactions are shown in fig .
[ figraysquad ] .
although the transition to triaxial equilibrium is progressively delayed as @xmath277 decreases , the triaxial configuration of fig . [ figortho ] , with orthogonal symmetry axes ( @xmath296 , @xmath297 ) , is still ultimately achieved in the @xmath0 limit . , shown for different relative strengths of like - nucleon and proton - neutron quadrupole interactions : ( a ) @xmath298 and ( b ) @xmath299 .
both curves are for @xmath136 . the boundary curve for @xmath300 ( fig . [ figboundary ] )
is shown in each panel ( dashed curve ) for comparison . ] and @xmath0 dynamical symmetries , for different relative strengths of like - nucleon and proton - neutron quadrupole interactions : ( left ) @xmath300 ( for reference ) , ( middle ) @xmath298 and ( right ) @xmath299 .
the curves are for @xmath136 .
proton fluid variables are represented by solid curves , while neutron fluid variables are represented by dashed curves .
points of phase transition are marked with arrows . ] a qualitative understanding of the mechanism underlying the parameter dependences observed in figs .
[ figboundaryquadnratio ] and [ figboundaryquadkapparatio ] is easily obtained . among the quadrupole terms in the energy surface
, the @xmath301 term has a dominant influence on the proton fluid equilibrium deformation , the @xmath302 terms drives the neutron fluid deformation , and the @xmath303 term couples the two deformations . the @xmath0 configuration with orthogonal symmetry axes ( fig .
[ figortho ] ) arises since the @xmath301 term stabilizes the proton fluid about a prolate deformation and the @xmath302 term stabilizes the neutron fluid about an oblate deformation , while the @xmath303 term is responsible for the relative orientation of the symmetry axes .
all three terms are essential to producing the @xmath0 triaxial equilibrium configuration . if the strengths of the like - nucleon quadrupole interactions are both reduced relative to that of the proton - neutron interaction ( figs .
[ figboundaryquadkapparatio ] and [ figraysquad ] ) , the ability of the like - nucleon terms to stabilize the proton and neutron fluids about distinct deformations is reduced , against the tendency of the proton - neutron term to favor equal deformations . the onset of triaxiality is thus delayed . for @xmath232
greater than unity ( fig .
[ figboundaryquadnratio ] ) , the relative weight of the @xmath301 contribution to @xmath49 increases , while that of the @xmath302 contribution decreases .
thus , a larger larger positive @xmath108 ( oblate tendency for the neutron fluid ) is needed for the neutrons to undergo a transition to an oblate configuration against the restraining influence of the protons , while only a small positive @xmath107 ( oblate tendency for the proton fluid ) is necessary for the protons to undergo a transition to an oblate configuration .
limited analytic results can also be obtained for the transition between spherical and deformed equilibrium .
consider the stability of the minimum at @xmath304 .
the second derivative of @xmath49 along the ray @xmath177 and @xmath178 is @xmath305 \\
+ 2u_\nu^2[\en(1-\xi')-4\knn\xi ' ] -8u_\pi u_\nu\kpn\xi'\cos(\gamma_\pi-\gamma_\nu).\end{gathered}\ ] ] as in the special case considered in sec .
[ subsecspherical ] , this quantity is independent of @xmath109 and @xmath112 and depends upon the @xmath87 coordinates only through their difference @xmath180 . as @xmath106
is increased from 0 , vanishing of this second derivative for some value of @xmath306 and @xmath180 indicates the onset of instability . since the coefficient of @xmath307 is negative , instability must first occur for @xmath308 . with @xmath180 set to zero
, @xmath309 vanishes at @xmath310 instability against deformations with _ equal _ @xmath201 and @xmath200 therefore occurs at @xmath124 for _ any _ values of the energy surface parameters .
however , the minimum at @xmath123 can in general become unstable against deformations with _ unequal _ @xmath201 and @xmath200 at a smaller value of @xmath106 . the direction in @xmath311 space in which instability first sets in
is given by a quadratic equation for @xmath306 , @xmath312 the @xmath106 value at which instability occurs follows from this @xmath306 value by eqn .
( [ eqnxipinstabilityquad ] ) .
there are two important special cases in which instability first occurs at @xmath124 : ( 1 ) for the @xmath94-spin invariant hamiltonian ( [ eqnhdieperink ] ) with arbitrary @xmath254 , and ( 2 ) for energy surface parameters which are proton - neutron symmetric ( _ i.e. _ , @xmath313 and @xmath289 ) , as are obtained whenever a proton - neutron symmetric hamiltonian is considered with @xmath194 . the evolution of equilibrium properties across the @xmath30-@xmath34 second - order transition point for a case asymmetric in @xmath201 and @xmath200 is shown in fig .
[ figraysunequal ] . from eqns .
( [ eqnxipinstabilityquad ] ) and ( [ eqnuratioquad ] ) , the phase transition for the parameter values used in the figure ( @xmath314 , @xmath315 , and @xmath316 ) occurs at @xmath317 , with @xmath318 . the evolution of the ratio @xmath319 past this point is shown in fig .
[ figraysunequal](c ) . observe that the proton and neutron equilibrium properties in this example are unequal even within the plane @xmath207 , unlike the case of sec .
[ subsecdieperink ] , illustrating that the the ibm-2 equilibrium problem in the general proton - neutron asymmetric case does not reduce to the one - fluid ibm problem .
-@xmath34 line in parameter space , illustrating instability against deformation with unequal @xmath201 and @xmath200 .
results are shown for the hamiltonian ( [ eqnhquad ] ) with @xmath314 and @xmath320 for ( left ) the symmetric case @xmath321 and ( right ) the asymmetric case @xmath316 .
the variable @xmath55 is represented by the solid curve , while @xmath57 is represented by the dashed curve .
the point of phase transition is marked with an arrow . ]
the minimum at zero deformation always becomes unstable at the @xmath106 value determined by eqns .
( [ eqnxipinstabilityquad ] ) and ( [ eqnuratioquad ] ) , independent of @xmath206 and @xmath205 .
however , this gives rise to a second order phase transition only if a first - order transition to a distinct minimum has not already occured at smaller @xmath106 for those values of @xmath206 and @xmath205 , as discussed in sec .
[ subsecspherical ] . for the @xmath94-spin invariant hamiltonian ( [ eqnhdieperink ] ) with arbitrary @xmath254 , an extension of relation ( [ eqnscreslice ] )
is readily found , namely that @xmath322 is invariant along any line of constant @xmath323 . by arguments analogous to those of sec .
[ subsecspherical ] , a line of second order phase transition occurs for @xmath124 and @xmath324 , embedded in a surface of first order phase transition . for a hamiltonian with arbitrary coupling constants ,
however , no such simple result is obtained .
the full phase diagram of the ibm-2 hamiltonian ( [ eqnhquad ] ) for a set of parameters involving proton - neutron symmetric couplings ( @xmath298 ) but unequal boson numbers ( @xmath280 ) , obtained by numerical mimimization , is shown in fig .
[ figtetraquad ] . ) , with @xmath298 and @xmath280 , as obtained by numerical minimization of @xmath49 .
the surfaces of first - order and second - order transition between regions of undeformed , axially symmetric deformed , and triaxially deformed equilibria are shown .
the axes are defined as in fig .
[ figtetraspace ] . ]
the majorana operator @xmath23 is the analogue in the ibm-2 hamiltonian to the proton - neutron symmetry energy of the liquid drop model .
this operator arises through a combination of direct shell model effects and renormalization effects @xcite . as is evident from eqn .
( [ eqnmaligned ] ) , this operator energetically discourages configurations with @xmath325 or @xmath326 .
the approximate strength of the majorana contribution can be deduced from the energies of mixed symmetry excitations @xcite , including the scissors mode excitation in axial rotor nuclei ( _ e.g. _ , ref .
comparison of the @xmath125-vibrational or @xmath36-vibrational energy scale ( @xmath2421mev ) with the scissors mode energy scale ( @xmath2422.5mev ) indicates @xmath327 to be a generally reasonable estimate . for axially symmetric configurations ,
the contribution of the majorana operator to the energy surface is , from eqn .
( [ eqnmaligned ] ) , @xmath328 . if the equilibrium configuration without the majorana operator already has @xmath218 , as in the @xmath195 plane for the proton - neutron symmetric case of fig .
[ figtetraspace ] , then introduction of the majorana term has no further effect .
otherwise , the effect of the majorana contribution is , naturally , to bring the equilibrium values @xmath201 and @xmath200 closer to each other . in the plane @xmath138
, this invalidates the simple results ( [ eqnbetaprolate ] ) and ( [ eqnbetaoblate ] ) giving the equilibrium @xmath201 purely as a function of @xmath329 and the equilibrium @xmath200 as a function of @xmath198 .
the derivation of the simple equations ( [ eqnboundary ] ) and ( [ eqnboundaryquad ] ) for the curve on which the axial minimum becomes unstable with respect to the @xmath230 depended upon the use of these results to eliminate @xmath287 in favor of @xmath330 at the minimum , a simplification which is no longer possible with a majorana contribution .
investigation of the equilibrium properties in the presence of a majorana operator must therefore rely upon numerical minimization .
the evolution of the equilibrium properties between the @xmath38 and @xmath0 points in parameter space , for different strengths of the majorana term in the hamiltonian , is shown in fig .
[ figraysmajorana ] .
the majorana operator is seen to have two main effects on the transition to triaxial equilibrium .
the triaxial configuration of fig . [ figortho ] is highly proton - neutron asymmetric , and thus penalized energetically by the majorana operator , as seen from eqn .
( [ eqnmaligned ] ) .
thus , first , the transition to towards such triaxial structure is delayed by the majorana operator .
second , the majorana operator has the effect of bringing the proton and neutron coordinate values closer together throughout the transition .
the @xmath38-@xmath0 line is an extreme case . without the majorana operator [ fig .
[ figrayssu3su3star ] or [ figraysmajorana](a ) ] , the evolution of the proton coordinates is completely decoupled from that of the neutron coordinates , and their equilibrium values are constant along the line .
a majorana strength of @xmath331 is sufficent , however , to cause the two fluids @xmath330 and @xmath230 coordinates to remain nearly equal throughout the evolution .
the triaxial configurations produced are thus essentially one - fluid , like those described by the davydov model @xcite or by the one - fluid ibm with three - body or four - body operators in the hamiltonian @xcite .
as the majorana strength increases , the global minimum at triaxial deformation becomes very shallow , and the difference in energy between the @xmath332 triaxial minimum and axial deformations ( @xmath209 ) approaches zero , as shown in fig .
[ figmajoranasoftness ] .
this indicates that the equilibrium structure for large majorana strengths is essentially one - fluid _
@xmath36-unstable _ , or @xmath333-like , rather than rigidly triaxial . and
@xmath0 dynamical symmetries , for different majorana operator strengths : ( left ) @xmath334 ( for reference ) , ( middle ) @xmath335 , and ( right ) @xmath336 .
the curves are for the quadrupole coefficient values @xmath300 with @xmath136 .
proton fluid variables are represented by solid curves , while neutron fluid variables are represented by dashed curves .
points of phase transition are marked with arrows . ] ) deformations ( dashed curve ) , shown as a function of @xmath337 , at the @xmath0 point in parameter space .
the difference is a measure of the @xmath36-stiffness of the triaxial minimum .
the curves are for quadrupole coefficient values @xmath300 with @xmath136 . ] the decomposition of the majorana operator into multipole components , easily obtained from ref .
* ( a2 ) ) , includes a nonzero hexadecapole contribution .
the analysis of ref .
@xcite therefore allows the existence of `` oblique '' equilibrium configurations , in which the proton and neutron intrinsic frames are not aligned . for axially symmetric aligned configurations ( @xmath70 and @xmath230 all vanishing ) , instability with respect to the euler angles is simple to investigate
the hessian matrix of @xmath49 with respect to the coordinates @xmath55 , @xmath57 , @xmath56 , @xmath58 , and @xmath81 decomposes as a direct sum @xmath338 , with @xmath339 involving only derivatives with respect to the @xmath330 , _ etc._. the dependence upon the azimuthal euler angles @xmath80 and @xmath82 vanishes by axial symmetry .
thus , instability would be indicated simply by a negative second derivative of @xmath49 with respect to the opening angle @xmath81 between the proton and neutron frames . for
the parameter ranges considered ( @xmath340 , @xmath341 , @xmath342 ) and coordinate values encountered ( @xmath343 ) in the present study , this quantity is always positive . the possibility of a first order transition to a distinct minimum with nonzero euler angles must also be considered .
numerical searches provide no evidence for a such a transition , but such searches are necessarily not exhaustive due to the large number of parameters involved .
to provide a connection between the ibm-2 phase structure considered so far and observable quantities , let us now consider the evolution of the predicted spectroscopic properties eigenvalues and electromagnetic transition strengths across the phase transitions .
the present discussion emphasises the gross spectroscopic features which emerge in the transitions between the different regions of the phase diagram of fig .
[ figquadrant ] .
numerical studies of several observables in the vicinity of the @xmath0 dynamical symmetry may also be found in refs .
@xcite .
electric quadrupole and magnetic dipole transition matrix elements are calculated using transition operators @xcite @xmath344,\end{aligned}\ ] ] where @xmath345 .
the transition strengths are @xmath346 .
schematic values @xmath347 , @xmath348 , and @xmath349 are used for the effective charges ( see ref .
@xcite for further discussion ) .
the @xmath350 parameters are taken equal to their counterparts @xmath287 in the hamiltonian , following the consistent quadrupole formalism @xcite .
the hamiltonian used for the present discussion is the simple form ( [ eqnhdieperink ] ) with the addition of the majorana operator .
diagonalization is carried out using the computer code npbos @xcite , for boson numbers @xmath351 .
the energy spectrum for the @xmath0 symmetry is known analytically @xcite . the operator @xmath352 with @xmath353 and @xmath354 can be reexpressed in terms of the quadratic casimir operators @xcite of the subalgebra chain ( [ eqnchainsu3star ] ) , as @xmath355-(3/16)c_2[\grpsothreepn]$ ] .
for @xmath356 , the eigenvalues are thus given by @xmath357 where @xmath358 are the @xmath0 elliott quantum numbers @xcite and @xmath359 is the angular momentum quantum number .
this is the @xmath360 energy formula , and level energies within an @xmath0 representation follow @xmath361 rotational spacings .
however , the allowed @xmath360 representations ( see refs .
@xcite ) are different from those for @xmath38 .
the ground state representation is the @xmath362 representation , rather than the usual @xmath38 ground state representation @xmath363 . according to the @xmath364 branching rules @xcite
, the ground state representation thus contains multiple degenerate rotational bands , with @xmath365 quantum numbers @xmath366 , @xmath367 , @xmath368 , @xmath369 , as shown in fig .
[ figrepnscheme ] .
[ it has become customary to label the bands of an @xmath360 representation by the elliott @xmath365 quantum number @xcite , in part by analogy with the rigid rotor spin - projection quantum number @xmath365 @xcite , which yields a band of the same band head spin . however , the elliott @xmath360 basis is not orthogonal , and the orthonormal states obtained from diagonalization of a hamiltonian are thus not in general elliott states . throughout this article , the label @xmath365
is used simply to indicate the spin content of a band . ]
other @xmath0 representations appearing at low energy include the @xmath370 , @xmath371 , and @xmath372 representations ( fig .
[ figrepnscheme ] ) . in the classical limit ,
the first two of these correspond to coupled @xmath125 and @xmath36 vibrations of the fluids @xcite .
the last corresponds to an `` orthogonal '' scissors mode , in which the proton and neutron symmetry axes oscillate about their equilibrium perpendicular relative orientation @xcite . dynamical symmetry , following the energy relation ( [ eqnesu3 ] ) .
the lowest - energy @xmath0 representations , labeled by the @xmath358 quantum numbers , are shown for @xmath351 . ]
strengths for ( a ) the @xmath0 dynamical symmetry ( @xmath353 and @xmath354 ) , ( b ) @xmath335 , and ( c ) @xmath336 .
the level scheme for the davydov model with @xmath373 is shown for comparison ( d ) .
levels of positive and negative @xmath374 parity are indicated by solid and dashed lines , respectively .
all @xmath375 strengths are given for the spin - descending transition direction , to allow direct comparison between panels , regardless of the ordering of level energies , and are normalized to @xmath376 .
energies are normalized separately in each panel . a minimal majorana term ( @xmath377 )
is used in the calculation of part ( a ) to lift the degeneracy of the @xmath0 multiplets .
calculations are for @xmath351 . ]
the @xmath375 strengths for transitions between levels of the @xmath0 ground state representation , calculated numerically , are shown in fig .
[ figtriaxschemes](a ) .
these strengths follow selection rules dictated by the presence of a discrete , parity - like symmetry .
consider the operation consisting of negation of the @xmath378 bosons followed by interchange of all proton and neutron bosons , together yielding @xmath379 and @xmath380 .
this was denoted the @xmath374 parity operation by otsuka @xcite .
for @xmath381 , the ibm-2 hamiltonian ( [ eqnhgeneralraw ] ) with @xmath314 and @xmath382 is invariant under the @xmath374 parity operation .
thus @xmath374 parity is a good symmetry throughout the central vertical plane of fig .
[ figtetraspace ] , including the @xmath0 dynamical symmetry .
the operator @xmath383 carries negative @xmath374 parity .
therefore , @xmath384 transitions occur only between states of opposite @xmath374 parity , and all electric quadrupole moments vanish .
the operator @xmath385 decomposes into a part of positive @xmath374 parity @xmath386 , which generates magnetic dipole moments , and a part of negative @xmath374 parity @xmath387 , which induces transitions between different states @xcite .
thus @xmath388 transitions follow the same @xmath374 parity selection rule as @xmath384 transitions , but magnetic moments are allowed . in applying the selection rules , it should be noted that the @xmath384 selection rule only holds in its exact form for @xmath389 . also , the @xmath374 parity operation involves interchange of all proton and neutron bosons , so it is only well defined for @xmath194 , but the selection rules persist approximately even for @xmath390 .
the classical interpretation of the @xmath374 parity operation is obtained by observing that , in the definition of the coherent state ( [ eqncoherentpn ] ) , negation of the @xmath378 boson is equivalent to negation of the @xmath391 deformation coordinates . in the geometric model ,
this is the @xmath36 parity operation of bs @xcite , which exchanges prolate and oblate liquid drop deformations .
thus , the @xmath374 parity operation exchanges prolate and oblate deformations for each fluid and then interchanges proton and neutron fluids . the @xmath0 triaxial configuration of fig .
[ figortho ] is seen to be invariant under this combined transformation .
we now note the general characteristics of the @xmath375 strengths for the @xmath0 symmetry shown in fig .
[ figtriaxschemes](a ) .
while the energy levels for the @xmath0 symmetry fall naturally into rotational quasi - bands , the transition strengths are far from those for an @xmath38 rotor .
many of the interband transitions have strenths of the same order as in - band transitions , while some of the in - band transitions vanish due to the @xmath374 parity selection rule .
( levels of positive and negative @xmath374 parity are indicated by solid and dashed lines , respectively , in fig .
[ figtriaxschemes ] . )
transitions between non - adjacent bands , not shown in the figure , are weaker by an order of magnitude or more .
magnetic dipole transitions arise in collective models from separation between the proton and neutron fluid distributions ( _ e.g _ , ref .
the static asymmetry between these distributions for the @xmath0 dynamical symmetry gives rise to extremely large @xmath388 admixtures whenever spin - allowed , with @xmath392 for most of the transitions in the ground state representation .
these @xmath388 strengths are comparable to the @xmath388 decay strength of the scissors excitation .
the @xmath0 dynamical symmetry has previously been loosely associated with the @xmath373 triaxial rotor of the davydov model @xcite , on the basis of two similarities : the angular momentum content of the ground state representation and the quadrupole moment components found in the classical limit @xcite .
the level energies in the two cases differ considerably , as the davydov model bands exhibit a characteristic @xmath393 clustering or `` staggering '' of energies [ fig .
[ figtriaxschemes](d ) ] . however , a detailed comparison of the @xmath384 transition strengths of the two models [ fig .
[ figtriaxschemes](a , d ) ] reveals an extraordinary similarity , with many @xmath384 strengths differing by only a few per cent .
the full relationship between the models has not been established ( an approximate @xmath360 algebra underlying the dynamics of the triaxial rotor has been discussed in ref .
@xcite ) .
if the majorana operator is introduced into the @xmath0 hamiltonian , the energy spectrum is radically altered , as depicted in fig .
[ figtriaxschemes](b , c ) .
the @xmath94-spin @xcite , formally analogous to isospin , of a state provides a measure of its proton - neutron asymmetry . for sufficiently pure @xmath94-spin @xmath394 ,
the ibm-2 effectively reduces to the one - fluid ibm @xcite .
the @xmath0 eigenstates are highly mixed in their @xmath94-spin content ( @xmath395 ) , reflecting the proton - neutron asymmetry of the classical equilibrium configuration .
the majorana operator , by energetically penalizing states of @xmath396 , purifies the @xmath94-spin contents of the low - energy states .
_ e.g. _ , the @xmath0 ground state is left with an @xmath396 content of only about @xmath397 for @xmath398 [ fig .
[ figtriaxschemes](c ) ] . a small majorana contribution in the hamiltonian , @xmath399 , results in a @xmath393 staggering of level energies [ fig .
[ figtriaxschemes](b ) ] like that of the davydov model [ fig .
[ figtriaxschemes](d ) ] , as noted in ref .
@xcite . a larger majorana contribution [ fig .
[ figtriaxschemes](c ) ] leads to the reverse @xmath400 staggering , characteristic of @xmath36-soft structure .
the @xmath384 strengths evolve toward @xmath333-like values @xcite with increasing majorana strength , and the @xmath388 admixtures decrease rapidly from their large @xmath0 values , with a typical scale @xmath401 for fig . [
figtriaxschemes](b ) and @xmath402 for fig .
[ figtriaxschemes](c ) .
the evolution observed of the apparent structure with increasing majorana contribution from @xmath0 triaxialiaty , through one - fluid rigid triaxiality , to one - fluid @xmath36-softness is as expected from the classical limit analysis of sec .
[ subsecmajorana ] .
the evolution of the ibm-2 predictions along the @xmath38@xmath0 transition is shown in figs .
[ figevolnsu3energies][figevolnsu3majorana ] . since the extremely low lying @xmath403 band is characteristic of the @xmath0 level scheme , the evolution of the relevant energies is plotted in fig . [ figevolnsu3energies ] . for simplicity ,
we first consider the transition without a majorana interaction . in this case ,
rotational @xmath361 energy spacings within bands are almost exactly preserved throughout the transition .
( solid ) , @xmath404 ( dashed ) , and @xmath405 ( dotted ) levels along the @xmath38@xmath0 transition , ( a ) with no majorana operator and ( b ) for @xmath406 .
the change in slope for the excited band level energies at @xmath407 in part ( a ) arises from the crossing of the scissors and @xmath36 bands .
calculations are for @xmath353 and @xmath351 , with minimal perturbations @xmath408 and for part ( a ) @xmath409 to remove numerical degeneracies . ]
@xmath0 transition with no majorana operator : ( a ) the @xmath38 limit at @xmath410 , ( b ) the second order transition point at @xmath411 , ( c ) the approximate ground state crossing point at @xmath412 , and ( d ) the @xmath0 limit at @xmath354 .
the @xmath384 branching patterns are shown for selected band head states , and the ground state band @xmath413 and @xmath414 transitions are included for scale [ shaded arrows , widths proportional to @xmath375 strength ] .
the @xmath388 transitions deexciting the scissors band head states are shown as well [ dashed arrows , widths proportional to @xmath415 strength ] . the @xmath243 ( ground ) , @xmath125 , @xmath36 , and @xmath416 ( scissors ) band designations are schematic ( see text ) .
all panels share a common energy scale .
calculations are for @xmath353 and @xmath351 , with minimal perturbations @xmath408 for part ( a ) and @xmath377 for part ( d ) to remove numerical degeneracies . ]
@xmath0 transition with majorana contribution @xmath417 : ( a ) the @xmath38 limit at @xmath410 , ( b ) the second order transition point at @xmath418 , and ( c ) the @xmath0 limit at @xmath354 . for the transition point ,
@xmath375 and @xmath415 strengths among selected low - lying levels are given if part ( d ) .
parts ( a)(c ) share a common energy scale , while energies in part ( d ) are normalized to the @xmath419 level energy .
dashed arrows in parts ( a)(c ) indicate @xmath388 transitions from the scissors excitations . @xmath375 and @xmath415 strengths in part ( d )
are shown spin - descending , with @xmath375 strengths normalized to @xmath376 [ @xmath420 . the @xmath243 ( ground ) , @xmath125 , @xmath36 , and @xmath416 ( scissors ) band designations are schematic
calculations are for @xmath353 and @xmath351 , with a minimal @xmath245 perturbation @xmath408 for part ( a ) to remove numerical degeneracies . ] in the @xmath38 limit [ fig .
[ figevolnsu3](a ) ] , the lowest excitation above the ground state band ( @xmath421 ) is the scissors mode ( @xmath422 ) , followed at higher energy by a cluster of degenerate @xmath423 and @xmath367 excitations . these arise from proton - neutron symmetric beta and gamma oscillations ( @xmath421 ) , asymmetric beta and gamma oscillations ( @xmath422 ) , and two - phonon scissors oscillations ( @xmath424 ) .
all @xmath384 transitions from the excited bands to the ground state band vanish exactly for the @xmath38 dynamical symmetry , but for small breakings of the symmetry they approximately follow the rotational alaga rules .
@xmath388 transitions vanish as well , except those involving excitations of the scissors mode . the level scheme for the @xmath38@xmath0 second order transition point ( @xmath411 ) is shown in fig . [ figevolnsu3](b ) .
as @xmath198 is increased from @xmath425 , one of the @xmath36 bands ( no longer of pure @xmath94 spin ) rapidly descends to lower excitation energy . at the transition point
, this band is connected to the ground state band by @xmath384 transitions @xmath426 as strong as in - band transitions .
the @xmath388 admixtures are extremely large , with @xmath392 , both for the @xmath36 to ground and the in - band transitions .
the @xmath36 band is followed in its descent by a series of @xmath427 , @xmath428 , @xmath369 bands , originating as two - phonon , three - phonon , _ etc .
_ , @xmath36 excitations .
however , the @xmath423 two - phonon @xmath36 excitation does not descend as rapidly and thus has a positive anharmonicity , @xmath429 at the transition point .
positive anharmonicity of this band has also recently been discussed as a signature of the axial - triaxial transition both in the one - fluid ibm @xcite and in the @xmath430 geometric model @xcite . [ the irregular level spacing seen in this band in fig .
[ figevolnsu3](b ) is not a fundamental signature but rather arises from mixing with the nearly degenerate scissors band for these parameter values .
] as the @xmath36 vibrational energy scale decreases , the @xmath431 and @xmath432 @xmath36 exitations of the scissors band similarly approach the scissors band in energy .
although the coherent state energy surface first becomes soft to triaxial deformations at the second order transition point , it is only for @xmath433 that highly triaxial configurations like that of fig .
[ figortho ] become lower in energy than axial configurations .
a level scheme obtained in this regime , at @xmath412 , is shown in fig .
[ figevolnsu3](c ) .
the ground state band and first @xmath434 and @xmath368 bands are close in energy , resembling the @xmath0 ground state representation .
the relative @xmath384 transition strengths are intermediate between those for the @xmath38 and @xmath0 symmetries .
the first excited @xmath423 band and following @xmath434 band , in contrast , have relative energies more appropriate to a ground state and its associated @xmath36 excitation , and @xmath375 values closely matching the @xmath38 alaga rules .
the energy spacing scales within these bands is less ( by @xmath435 ) than that of the ground state family of bands , suggesting a different moment of inertia or different triaxial nature to the rotation .
the excitation energy of the @xmath436 state reaches its lowest value along the @xmath38@xmath0 transition at approximately this @xmath198 value .
the observables are thus highly suggestive of an avoided crossing between an @xmath0-like configuration and an @xmath38-like configuration , although the actual eigenstate structure is presumably more complicated .
an apparent avoided crossing also occurs between @xmath437 scissors excitations based upon these configurations .
we now briefly reconsider the @xmath38@xmath0 transition for a realistic majorana operator strength ( @xmath417 ) [ figs .
[ figevolnsu3energies](b ) and [ figevolnsu3majorana ] ] . at the @xmath38 limit ,
the eigenstates are unchanged , but the levels with @xmath396 are raised in energy [ fig .
[ figevolnsu3majorana](a ) ] .
the lowest - lying excitations are thus proton - neutron symmetric @xmath125 and @xmath36 excitations , followed at higher energies by the symmetric @xmath125 and @xmath36 multiphonon excitations .
the second - order transition between axial and triaxial equilibrium occurs at much larger @xmath198 than without the majorana operator , @xmath418 ( fig .
[ figraysmajorana ] ) . the level scheme for the transition point [ fig .
[ figevolnsu3majorana](b ) ] shows large differences between ground state and excited families of levels .
the lowest @xmath423 , @xmath367 , and @xmath368 quasi - bands have level spacings intermediate between those for the @xmath38 and @xmath34 symmetries .
the next @xmath423 quasi - band is far less rotational in spacing and has a @xmath438-@xmath439 energy spacing about twice as large as for the ground state .
the levels above it are assembled into an approximate @xmath34 multiplet structure .
all @xmath388 strengths are attenuated relative to the cases with no majorana contribution , with a typical admixture scale @xmath440 .
qualitatively , little further changes as @xmath198 increases to @xmath441 [ fig .
[ figevolnsu3majorana](c ) ] , except that the ground state family of levels takes on more distinctly @xmath34-like energy spacings and @xmath375 strengths .
@xmath0 transition and ( b ) the @xmath30@xmath34 transition , calculated for a majorana operator strength of @xmath417 .
the inset to part ( a ) shows the corresponding level scheme with no majorana operator .
energies are indicated normalized to the @xmath419 level energy .
@xmath375 strengths are shown spin - descending , normalized to @xmath376 separately in each panel [ @xmath442 for part ( a ) , @xmath443 for part ( b ) ] , and selected @xmath415 strengths are indicated as well . dashed double arrows indicate selected strong @xmath388 transitions .
calculations are for @xmath351 , with @xmath353 , @xmath354 , and @xmath444 for part ( a ) and with @xmath445 , @xmath446 , and @xmath444 for part ( b ) . ]
the basic spectroscopic features along the @xmath30@xmath0 transition differ fundamentally depending upon whether or not a realistic majorana contribution is included in the hamiltonian .
the level scheme for the @xmath30 symmetry of the ibm-2 @xcite exhibits extreme degeneracies if no majorana operator is present : there are two degenerate one - phonon @xmath438 levels ( one fully symmetric and one of mixed symmetry ) and eleven degenerate levels of various symmetry in the two - phonon multiplet . by the second order transition point ( @xmath124 ) , these highly degenerate levels evolve into a cluster of low - lying excitations of widely varying @xmath94-spin content , shown in the inset to fig . [ fige5schemes](a ) . among these levels ,
incipient forms of the @xmath0 ground - state and scissors quasi - band structures are apparent .
the electromagnetic transition properties are dominated by the ubiquity of strong @xmath388 transitions , arising from the mixing of @xmath30 states of different @xmath94 spin .
for instance , both the @xmath447 and @xmath448 transitions have @xmath415 strengths @xmath449 , typical of mixed symmetry state decays .
if instead a realistic majorana operator strength @xmath450 is considered , the low - energy excitation spectrum simplifies considerably , as shown in fig .
[ fige5schemes](a ) .
the low - lying levels have nearly pure @xmath451 , with a few per cent admixtures of other @xmath94 values , and are grouped into approximate @xmath452 multiplets .
mixed symmetry levels of @xmath453 are present at higher energy ( @xmath437 , @xmath438 , and @xmath404 levels at top of figure ) .
the @xmath374 parity is a good quantum number throughout the @xmath30@xmath0 transition , so @xmath374 parity selection rules apply to the electromagnetic transitions . the level scheme and @xmath384 transition strengths are nearly identical to those obtained for the second order transition point between @xmath30 and @xmath34 , shown for comparison in fig .
[ fige5schemes](b ) .
there is thus also considerable resemblance to the predictions of the geometric @xmath454 model @xcite .
two features of the @xmath30@xmath0 transition point spectrum allow it to be distinguished from the @xmath30@xmath34 case .
first , a slight breaking of the degeneracy of the @xmath452 multiplets occurs , in the sense necessary to create the @xmath0 quasi - bands .
thus , the @xmath455 level moves below the @xmath456 level , the @xmath457 and @xmath458 levels move below the @xmath459 level , _ etc . _
this effect becomes much more marked just past the transition point .
second , sizeable @xmath388 transition strengths , @xmath460 to @xmath461 , are present among the low - lying levels [ fig .
[ fige5schemes](a ) ] , while these vanish when @xmath195 .
such @xmath388 admixtures in general play an essential role in identifying deviations from @xmath207 @xcite .
the ibm-2 phase diagram investigated here provides a framework for studying the transition between axial and triaxial structure in nuclei .
triaxial deformation might arise from several possible sources : distinct deformations of the proton and neutron fluids as considered in the present work , higher - order interactions in an essentially one - fluid nucleus @xcite , or the presence of configurations involving hexadecapole nucleon pairs @xcite .
the main spectroscopic features of proton - neutron triaxial structure are exemplified by the @xmath0 dynamical symmetry , shown in fig .
[ figtriaxschemes](a ) .
proton - neutron triaxiality is characterized by a low - lying @xmath434 band , as in the other forms of nuclear triaxiality , but with level energies following a rotational @xmath361 sequence .
the @xmath375 strength pattern is remarkably similar to that of the classic rigid triaxial rotor of the davydov model , including the unusual feature that @xmath462 ( this however only strictly holds for @xmath194 and @xmath389 ) .
the microscopic conditions leading to two - fluid triaxial structure , which requires @xmath329 and @xmath198 of opposite sign , are found when the proton bosons are particle like ( below mid - shell ) and the neutron bosons are hole - like ( above mid - shell ) or _
vice versa_. nuclei in this category include the heavy @xmath463 and @xmath464 isotopes @xcite , the light rare earth nuclei below the @xmath465 shell closure , and the extremely light @xmath466 , @xmath467 , and @xmath468 nuclei .
the @xmath0 dynamical symmetry probably does not occur in its pure form in any actual nuclei . in this article , we have seen how the features of @xmath0 proton - neutron triaxiality are modified by the majorana interaction [ fig .
[ figtriaxschemes](b , c ) ] , which tends to alter the spectrum towards that for the @xmath34 dynamical symmetry . of most importance from a phenomenological viewpoint
, we have investigated the transition from axially symmetric deformed [ @xmath38-like ] to triaxially deformed [ @xmath0-like ] structure , either without [ fig .
[ figevolnsu3 ] ] or with [ fig .
[ figevolnsu3majorana ] ] a majorana interaction .
in particular , fig .
[ figevolnsu3majorana ] represents a realistic study of this transition , including the behavior of the lowest @xmath431 , or scissors , band .
it is found that , despite the attenuation of proton - neutron triaxial features by the majorana interaction , two - fluid triaxiality can serve as the underlying mechanism driving one - fluid triaxiality , and recognizable features of its two - fluid nature can remain even in the presence of realistic majorana interactions .
the two - fluid phenomena are likely to take on more importance as very neutron rich nuclei become accessible to experimental study .
when the valence protons and neutrons occupy well - separated orbitals , the proton - neutron interaction strengths are reduced , yielding a situation seen in sec .
[ secgeneral ] to be much more conducive to true two - fluid triaxiality .
the present analysis may also serve as a model for the study of other multi - fluid bosonic systems . within nuclear physics ,
the @xmath469 description of core - skin collective modes in neutron rich nuclei @xcite is most directly analogous . in molecular physics , the @xmath470 vibron model with two vibronic species
@xcite may be studied similarly .
the appropriate coherent state formalism is discussed in ref .
the phase transition in this model is relevant to coupled vibronic bending modes in `` floppy '' molecules such as acetylene @xcite .
an extreme case of a multi - fluid algebraic model is found in applications to polymers , in which an arbitarily large number of vibronic fluids are coupled @xcite .
yet another case is that of atomic condensates , for which scissors modes of a single - constituent bose - einstein condensate relative to an anisotropic potential have been observed @xcite .
experiments have been planned to produce condensates of two atomic species .
the exotic features of the ibm-2 phase diagram are likely to be encountered for other multi - fluid systems as well .
they highlight the need for a classification scheme beyond the simple ehrenfest or one - parameter landau models to address the phase structure of systems which simultaneously possess multiple control parameters and multiple order parameters .
discussions with j. m. arias , r. bijker , a. leviatan , and n. pietralla are gratefully acknowledged .
this work was supported by the us doe under grant de - fg02 - 91er-40608 and was carried out in part at the european centre for theoretical studies in nuclear physics and related areas .
for calculations in the coherent state formalism involving multi - fluid operators , it is useful to deduce a general formula for the matrix element of an arbitrary @xmath471-body operator between arbitrary multi - fluid coherent states .
define @xmath472 different coherent boson species @xmath473 ( @xmath474 ) in terms of the basic boson operators @xmath475 ( @xmath476 ) as orthonormal linear combinations @xmath477 where @xmath478 .
then the multi - species coherent states @xmath479 are normalized , are orthogonal to each other for different @xmath480 values , and have good total boson number @xmath481 , where the total boson number operator is @xmath482 .
coherent states of the type ( [ eqnmultispecies ] ) also arise in the study of intrinsic excitations in algebraic models . in this context , the different coherent boson operators @xmath473 represent a ground state condensate and one or more orthogonal excitation modes ( see , _ e.g. _ , refs .
@xcite ) .
the matrix element of an arbitrary @xmath471-body operator ( @xmath483 ) between two arbitrary multi - species coherent states is @xcite @xmath484 \bigl ( \prod_{i=1}^m \alpha_{t_i',r_i'}^ * \alpha_{t_i , r_i } \bigr),\end{gathered}\ ] ] where @xmath485 , @xmath486 , and an underlined superscript indicates the falling factorial [ @xmath487 @xcite .
the multiple sum in ( [ eqncoherentme ] ) nominally contains @xmath488 terms , but only those with @xmath489 and @xmath490 can yield nonzero contributions .
three stages are involved in evaluating the matrix element of a general operator : reexpression of the operator in terms of normal - ordered @xmath471-body terms , evaluation of the matrix elements of these by ( [ eqncoherentme ] ) , and simplification of the result .
these steps can all readily be carried out though computer - based symbolic manipulation .
the ibm-2 energy surface considered in secs . [ secschematic ] and [ secgeneral ] involves terms of the form @xmath492 we summarize here the extremum properties of this expression . for any value of the parameter @xmath493 , @xmath491 has two extrema with respect to @xmath494 , at @xmath495 the global minimum of @xmath491 is located at @xmath496 , and the global maximum is located at @xmath497 .
note that @xmath498 , @xmath499 , and the two extremum positions are related by @xmath500 .
the extremal values of @xmath94 are simply @xmath501= \frac{1}{2 } x_\pm(a).\ ] ] if @xmath502 then @xmath503}{\hspace{-0.11111em}<\hspace{-0.11111em}}\abs{f[a;x_+(a)]}$ ] , while for @xmath504 the inequality holds in the opposite sense .
|
the phase diagram of a two - fluid bosonic system is investigated .
the proton - neutron interacting boson model ( ibm-2 ) possesses a rich phase structure involving three control parameters and multiple order parameters .
the surfaces of quantum phase transition between spherical , axially - symmetric deformed , and @xmath0 triaxial phases are determined , and the evolution of classical equilibrium properties across these transitions is investigated .
spectroscopic observables are considered in relation to the phase diagram . and
two - fluid systems , algebraic models , phase transitions , proton - neutron interacting boson model ( ibm-2 ) , triaxial nuclear deformation 21.60.fw , 21.60.ev , 21.10.re
|
consider _ private - message _ broadcasting over the three - receiver additive white gaussian noise ( awgn ) broadcast channel where each receiver may know some of the messages requested by other receivers as side information .
we investigate the capacity region of the channel for _ all possible side information configurations_. broadcast channels model communication networks where one transmitter wishes to transmit a number of messages to multiple receivers @xcite .
the capacity region of broadcast channels is not known in general , except for a few special classes , e.g. , degraded broadcast channels , which include awgn broadcast channels @xcite . in broadcast channels
, the receivers may know some of the source messages a priori , referred to as receiver message side information .
this is motivated by applications such as multimedia broadcasting with packet loss , and the downlink phase of multi - way relay channels @xcite .
the capacity region of broadcast channels with receiver message side information is known where each receiver needs to decode _ all _ the source messages ( or equivalently , all the source messages not known a priori ) @xcite .
however , the case where the receivers need not decode all the source messages remains unsolved to date .
kramer et al .
established the capacity region of the two - receiver _ memoryless _ broadcast channel where one of the receivers need not decode all the source messages @xcite .
the capacity region of the two - receiver _ awgn _ broadcast channel is known for all message requests and side information configurations @xcite . extending the results to three or more receivers is `` highly nontrivial '' @xcite .
oechtering et al .
characterized the capacity region of the _ three - receiver _ less noisy broadcast channel where ( i ) only two receivers possess side information and ( ii ) the request of the third receiver is only restricted to a common message demanded by all the receivers @xcite .
are the source messages , each demanded by one receiver , and @xmath0 is the set of messages known to receiver @xmath1 a priori.,scaledwidth=42.0% ] this paper considers the three - receiver awgn broadcast channel where the receivers have private - message requests . in the channel model under consideration , as depicted in fig .
[ awgnbcmodelfig ] , the channel - output sequence at receiver @xmath1 , @xmath2 , is the sum of the transmitted codeword , @xmath3 , and an independent and identically distributed ( i.i.d . ) noise sequence , @xmath4 , with a normal distribution , @xmath5 .
we represent random variables using upper - case letters , and their realizations using the corresponding lower - case letters . the input and the output alphabets of the channel
are denoted by @xmath6 and @xmath7 , respectively .
this channel is stochastically degraded , and without loss of generality , we can assume that receiver @xmath8 is the strongest and receiver @xmath9 is the weakest in the sense that @xmath10 .
the transmitted codeword has a power constraint of @xmath11 and is a function of source messages , @xmath12 , which are independent .
@xmath13 is intended for receiver @xmath1 , and is an @xmath14-bit message , i.e. , its transmission rate is @xmath15 bits per channel use .
@xmath13 is uniformly distributed over the set @xmath16 which is the set of all binary vectors of length @xmath14 , i.e. , the cardinality of @xmath16 is @xmath17 .
we define the set @xmath18 as the set of messages known to receiver @xmath1 a priori .
a @xmath19 code for the channel consists of an encoding function @xmath20 where @xmath21 denotes the cartesian product , and @xmath22 denotes the @xmath23-fold cartesian product of @xmath6 , i.e. , a codeword , @xmath24 , is generated for each @xmath25 .
this code also consists of decoding functions @xmath26 where , e.g. , if @xmath27 , we have @xmath28 .
the decoded @xmath13 at receiver @xmath1 is @xmath29 , and the average probability of error for this code is defined as @xmath30 , and receiver 3 knows @xmath31.,scaledwidth=15.0% ] a rate triple @xmath32 is said to be _ achievable _ for the channel if there exists a sequence of @xmath19 codes with @xmath33 as @xmath34 .
the _ capacity region _ of the channel is the closure of the set of all achievable rate triples @xmath32 . the side information configuration of the channel is represented by a _
side information graph _ , @xmath35 , where @xmath36 is the set of _ vertices _ and @xmath37 is the set of _ arcs_. an arc from vertex @xmath1 to vertex @xmath38 , denoted by @xmath39 , exists if and only if receiver @xmath1 knows @xmath40 .
the set of out - neighbors of vertex @xmath1 is then @xmath41 . a sample side information graph is shown in fig .
[ fig : samplesigraph ] .
we classify all 64 possible side information configurations into eight groups , each consisting of eight members .
any side information graph is the union of @xmath42 , depicted in fig .
[ fig : groupleaders ] , and @xmath43 , depicted in fig .
[ fig : subgraphs ] , for some unique @xmath1 and @xmath38 where @xmath44 . according to our proposed classification , the side information graphs @xmath45 are considered as the _ group leaders _ , and group @xmath1 consists of the side information graphs formed from the union of the group leader , @xmath42 , with each of the @xmath46 .
for instance , group 6 is the set @xmath47 . , capturing if each receiver knows the message(s ) requested by stronger receiver(s).,scaledwidth=35.0% ] , capturing if each receiver knows the message(s ) requested by weaker receiver(s).,scaledwidth=35.0% ] the capacity region of the channel is known where there is no side information at the receivers @xcite , i.e. , in our notation @xmath48 .
oechtering et al .
@xcite established the capacity region for two side information configurations , which in our notation are @xmath49 and @xmath50 ; they @xcite also derived the best known inner bound for @xmath51 .
these side information configurations correspond to the cases where two of the receivers know each other s requested messages and the third receiver has no side information .
the best known inner and outer bounds for the remaining 60 side information configurations , and the best known outer bound for @xmath51 are within a constant gap of the capacity region @xcite .
the inner bound uses a separate index and channel coding scheme , developed based on the deterministic approach @xcite .
the outer bound is a polyhedron developed using fano s inequality .
we investigate the capacity region of the channel for all 64 possible side information configurations .
one of the difficulties in deriving the capacity region for all side information configurations is to find a unified scheme .
considering the three - receiver awgn broadcast channel without side information , i.e. , @xmath48 , it is well - known that the simple superposition of individual codewords , each carrying one message , can achieve the capacity region . for @xmath48
, it can be easily verified that a stronger receiver is always able to decode the message of a weaker receiver if the weaker receiver is able to decode it ; this means that knowing the message of a weaker receiver as side information does not help , and hence the same transmission scheme is optimal for all the members of group 1 .
our proposed classification , developed based on this idea , facilitates the problem of having a unified scheme by grouping the side information configurations that lead to the same capacity - achieving transmission scheme for each group .
we construct transmission schemes , and derive inner and outer bounds for different groups . for six groups , i.e. ,
all the groups except groups 4 and 7 , we establish the capacity region for all the group members .
this result also shows the looseness of both the best known inner and outer bounds for these groups . for group 4
, we improve the best known inner bound and/or outer bound for all the group members ; our bounds coincide at certain regions which can be characterized by two thresholds . for group 7 , we establish the capacity region for four members . for the remaining four members , we reduce the gap between the best known inner and outer bounds by improving both . fig .
[ fig : resultsummary ] shows a summary of our results . in a concurrent work with the preliminary published versions @xcite of this work , sima et al .
@xcite established the capacity region for 46 out of 64 possible side information configurations using different transmission schemes .
these 46 configurations are a subset of the 52 side information configurations whose capacity region is established in our work . and @xmath43 in its corresponding row and column ( each row represents one group ) .
the dotted cells show configurations whose capacity region was known prior to this work , the hatched cells show configurations whose capacity region is established in this work , and the blank cells show configurations with improved inner bound and/or outer bound but with capacity region remaining unknown.,scaledwidth=45.0% ] [ cols="<,<,<",options="header " , ] [ memberswithindexcoding ]
in this section , we provide some remarks on index coding , dirty paper coding , and simultaneous decoding , which highlight the roles that these coding techniques play in our transmission schemes .
we then extend our methodology to awgn broadcast channels with more than three receivers and conjecture the groups for which we can establish the capacity region for all the group members .
index coding has been used in the transmission schemes of eight side information configurations , shown in table [ memberswithindexcoding ] .
if we replace index coding with multiplexing coding for these members , we can not achieve the same region .
this observation is consistent with our previous study on the non - interchangeability of index and multiplexing coding for some side information configurations in awgn broadcast channels with three or more receivers @xcite . in all of these transmission schemes
, there exists a receiver who decodes the xor of two messages , both not known a priori . from the standpoint of this receiver , replacing index coding with multiplexing coding increases the amount of uncertainty to be resolved in the resultant message from @xmath52 to @xmath53 bits .
the transmission schemes for group 4 ( transmission schemes and ) employ dirty paper coding . in this group , @xmath54 ( i.e.
, the message intended for the strongest receiver ) and @xmath55 ( i.e. , the message intended for the weakest receiver ) are multiplexed , and receiver 2 ( the in - between receiver ) does not know @xmath54 .
at each receiver , dirty paper coding allows us to avoid decoding a message intended for a stronger receiver ( which otherwise imposes additional rate constraints ) , and to cancel some parts of the signal carrying a message intended for a weaker receiver ( which increases the signal - to - noise ratio ) .
for example , at receiver 2 , we avoid decoding @xmath54 , and at the same time we cancel a part of the signal which carries @xmath55 , i.e. , @xmath56 .
dirty paper coding can also be used to provide a unified scheme for group 2 , avoiding the exception @xmath49 in that group .
the following transmission scheme can achieve the capacity region for all the group members , @xmath57\right).\end{aligned}\ ] ] in this scheme , we consider @xmath58 as interference for receiver 1 known non - causally at the transmitter .
the first subcodebook is then constructed using dirty paper coding with the auxiliary random variable @xmath59 where @xmath60 is independent of @xmath61 , and @xmath62 . to achieve the capacity region , receivers 2 and 3 treat @xmath63 as noise , and decode @xmath58 using their side information .
then receiver 2 can reliably decode @xmath31 if @xmath64 , and receiver 3 can reliably decode @xmath55 if @xmath65 . receiver 1 decodes @xmath66 without being affected by @xmath58 due to dirty paper coding .
then this receiver , irrespective of its side information , can reliably decode @xmath54 if @xmath67 .
using , since we need not decode @xmath58 at receiver 1 , there is no longer any need for the index coding of @xmath68 and @xmath69 wherever possible .
simultaneous decoding is utilized for groups 5 and 6 , and four members of group 7 .
for each of these side information configurations , there is at least one message that is an operand in both subcodebooks of the transmission scheme . as an example , for @xmath70 , the two subcodebooks of the transmission scheme are both functions of @xmath68 . for group 6 , simultaneous decoding can be replaced with rate splitting and successive decoding . for this group
, we can also achieve the capacity region by using the alternative transmission scheme @xmath71\right)+x_2^{(n)}\left([m_{12},m_3]\right),\end{aligned}\ ] ] where rate splitting is utilized , and using successive decoding alone ( where , at receivers 1 and 2 , @xmath58 is first decoded while @xmath63 is treated as noise , and then @xmath63 is decoded . at receiver 3 , @xmath58
is only decoded while @xmath63 is treated as noise ) . for groups 5 and 7 , as opposed to group 6 ,
there exist some cases for which simultaneous decoding can not similarly be replaced with rate splitting and successive decoding . as an example , consider group 5 s leader , @xmath70 . by utilizing the same approach as for group 6 ,
suppose that we use the transmission scheme @xmath72\right)+x_2^{(n)}\left([m_{22},m_3]\right),\end{aligned}\ ] ] and employ successive decoding ( where , at receivers 1 and 2 , @xmath58 is first decoded while @xmath63 is treated as noise , and then @xmath63 is decoded . at receiver 3
, @xmath58 is only decoded while @xmath63 is treated as noise ) .
this leads to a strictly smaller achievable rate region when @xmath73 .
we can verify that other decoding orders ( of successive decoding ) also lead to suboptimal results .
our use of simultaneous decoding has helped us establish the capacity region for six more side information configurations ( in groups 5 and 7 ) compared to the parallel work by sima et al .
@xcite ( in which rate splitting and successive decoding are utilized , but not simultaneous decoding ) . in this subsection , we consider private - message broadcasting over awgn broadcast channels with @xmath74 receivers where without loss of generality @xmath75 .
we perform the same grouping method as for the three - receiver case . in our proposed classification
, all the side information graphs with the same subgraph obtained by removing all the arcs from a stronger receiver to a weaker receiver , are classified into one group ; the resultant common subgraph for each group is considered as the group leader .
we then , using the insights gained from the three - receiver case , conjecture a condition that we can use to characterize some groups for which we can establish the capacity region for all the members .
[ conjecture1 ] if for every receiver @xmath1 in a group leader , each side information message @xmath40 ( i.e. , @xmath76 ) is also known to all the receivers that are ( i ) stronger than receiver @xmath1 , and ( ii ) weaker than receiver @xmath38 ( i.e. , all receivers @xmath77 where @xmath78 ) , we can establish the capacity region for all the group members .
the capacity region is achieved using _ codebook construction a _ and _ decoding scheme a _ which are defined as follows .
_ codebook construction a _ : the transmission scheme for a group leader is constructed using the following method .
a subcodebook is constructed associated with a receiver @xmath1 if there is at least one message in @xmath79 that is not known a priori to any weaker receiver .
this condition is true even when ( i ) @xmath80 and @xmath13 is not known a priori to any weaker receiver , or ( ii ) receiver @xmath1 is the weakest receiver , i.e. , @xmath81 .
this subcodebook is formed by multiplexing @xmath13 with @xmath82 .
the transmission scheme is finally formed from the linear superposition of the subcodebooks .
for some members of the group which are not the group leader , index coding may also be required following the form outlined in section [ subsection : discussindex ] .
_ decoding scheme a _ : at each receiver @xmath1 , simultaneous decoding is performed over all the subcodebooks that contain only @xmath13 , messages intended for weaker receivers ( i.e. , @xmath40 , @xmath83 ) , or messages known to the receiver . as an example , we here establish the capacity region for a group leader of the four - receiver awgn broadcast channel with receiver message side information shown in fig .
[ fourreceiverconfiguration ] .
[ capacityfoureceiver ] the capacity region of the four - receiver awgn broadcast channel with the side information configuration shown in fig .
[ fourreceiverconfiguration ] is the closure of the set all rate tuples @xmath84 , each satisfying @xmath85 for some @xmath86 .
( _ codebook construction _ ) using _ codebook construction a _ , the capacity - achieving transmission scheme takes the form @xmath87\right)+x_2^{(n)}\left([m_2,m_3,m_4]\right),\end{aligned}\ ] ] where the first subcodebook consists of i.i.d .
codewords generated according to @xmath60 for each @xmath88 $ ] , the second subcodebook consists of i.i.d .
codewords generated according to @xmath89 for each @xmath90 $ ] , and @xmath91 is independent of @xmath92 .
( _ achievability proof _ ) at receivers 1 and 2 , simultaneous decoding is performed over @xmath93 , and at receivers 3 and 4 , @xmath58 is decoded while @xmath63 is treated as noise
. receiver 1 can reliably decode @xmath54 if conditions and hold .
receiver 2 can reliably decode @xmath31 if condition holds .
receiver 3 can reliably decode @xmath55 if condition holds , and receiver 4 can reliably decode @xmath94 if condition holds ( the error events are similarly written as for groups 5 and 6 in section [ section : capacity ] ) .
( _ converse proof _ )
we just define @xmath95 for this side information configuration , and the remainder of the converse proof uses the same techniques introduced for group 2 in the appendix . for this side information configuration , @xmath96 is upper bounded as @xmath97 for some @xmath86 where @xmath98 and @xmath99 follow from similar reasons as @xmath98 and @xmath99 in .
we considered the problem of private - message broadcasting over the three - receiver awgn broadcast channel with receiver message side information .
we first classified all 64 possible side information configurations into eight groups , each consisting of eight members .
we then derived inner and outer bounds for different groups , and established the capacity region for 52 out of 64 possible side information configurations . for six groups , i.e. ,
all the groups except groups 4 and 7 , we established the capacity region for all the group members , and proved the looseness of both the best known inner and outer bounds . for group 4 , we improved the best known inner bound and/or outer bound for all the group members . for this group ,
our bounds coincide at certain regions , which can be characterized by two thresholds . for group 7 , we established the capacity region for four members .
for the remaining four members , we proved that our inner and outer bounds are both tighter than the best known inner and outer bounds .
in this section , we prove the converse part of theorem [ maintheorem ] . the proof is based on those for awgn broadcast channels without side information @xcite . in the converse ,
we use fano s inequality and the entropy power inequality ( epi ) . based on fano s inequality
, we have @xmath100 where @xmath101 as @xmath102 . for the sake of simplicity , we use @xmath103 instead of @xmath104 for the remainder .
we also use the fact that the capacity region of a stochastically degraded broadcast channel without feedback is the same as its equivalent physically degraded broadcast channel @xcite for which we have the markov chain @xmath105 , i.e. , @xmath106 and @xmath107 , where @xmath108 .
_ group 1 _ : the converse proof for all the members of this group is the same as the group leader , i.e. , the three - receiver awgn broadcast channel without receiver message side information for which the converse proof exists @xcite . _ group 2 _ : for the members where receiver 2 does not know @xmath55 as side information , @xmath109 is upper bounded as @xmath110 for some @xmath86 where @xmath98 follows from adding the following inequalities resulted from the physical degradedness of the channel and .
@xmath111 in , @xmath99 follows from @xmath112 then since @xmath113 there must exist an @xmath114 such that @xmath115 for these members , we also have @xmath116 and @xmath117 where @xmath98 follows from , @xmath99 from using the conditional epi @xcite for @xmath118 , and substituting @xmath119 and , @xmath120 from using the conditional epi for @xmath121 , and substituting @xmath122 and , and @xmath123 from @xmath124 from , , and since @xmath103 goes to zero as @xmath34 , the converse proof for the members where receiver 2 does not know @xmath55 a priori is complete . for the members where receiver 2 knows @xmath55 a priori
, we just need to modify as follows , and reuse and .
@xmath125 where @xmath98 follows from , and @xmath99 from the same reason as @xmath99 in . _ group 3 _ : for this group , we just define @xmath95 , and the remainder of the converse proof uses the same techniques employed for group 2 . in this group , @xmath126 is upper bounded as @xmath127 for some @xmath86 where @xmath98 follows from , and @xmath99 from similar reason as @xmath99 in , i.e. , since @xmath128 there must exist an @xmath114 such that @xmath129 _ groups 5 and 6 : _ the outer bound for these two groups is formed from the intersection of the following outer bound and the best known outer bound given in .
if a rate triple @xmath32 is achievable for a member of groups 5 and 6 , then it must satisfy @xmath130 for some @xmath131 , where @xmath132 , @xmath133 , and @xmath134 for group 5 , and @xmath135 , @xmath133 , and @xmath136 for group 6 .
conditions and follow from the capacity region of the two - receiver awgn broadcast channel ( from the transmitter to receivers @xmath1 and @xmath38 ) where only the stronger receiver ( receiver @xmath1 ) may know the requested message of the weaker receiver ( receiver @xmath38 ) as side information .
the side information of receivers @xmath1 and @xmath38 about each other s requested messages has this property .
condition is due to the point - to - point channel capacity between the transmitter and receiver @xmath77 .
_ group 8 _ : best known outer bound given in is tight .
t. j. oechtering , m. wigger , and r. timo , `` broadcast capacity regions with three receivers and message cognition , '' in _ proc .
inf . theory ( isit ) _ , cambridge , ma , july 2012 , pp .
388392 .
t. j. oechtering , h. t. do , and m. skoglund , `` capacity - achieving coding for cellular downlink with bidirectional communication , '' in _ proc . int .
itg conf . source channel coding ( scc ) _ , siegen , germany , jan
. 2010 .
b. asadi , l. ong , and s. j. johnson , `` the capacity of three - receiver awgn broadcast channels with receiver message side information , '' in _ proc .
inf . theory ( isit ) _ , honolulu , hi , june / july 2014 , pp .
28992903 .
j. sima and w. chen , `` joint network and gelfand - pinsker coding for 3-receiver gaussian broadcast channels with receiver message side information , '' in _ proc .
inf . theory ( isit ) _ , honolulu , hi , june / july 2014 , pp .
8185 [ revised version ] available : http://arxiv.org/abs/1407.8409v2 .
h. weingarten , y. steinberg , and s. shamai , `` the capacity region of the gaussian multiple - input multiple - output broadcast channel , '' _ ieee trans .
inf . theory _ ,
52 , no . 9 , pp . 39363964 , sept . 2006 .
behzad asadi(s14 ) received the m.sc .
degree in electrical engineering from the university of tehran , iran , in 2008 . from 2008 to 2010
, he was with wmc laboratory at the university of tehran as a research associate , and from 2011 to 2013 , he was with zte corporation as a solution manager .
currently , he is pursuing his ph.d .
degree in electrical engineering at the university of newcastle , australia .
his research interests include information theory , and signal processing in wireless networks .
lawrence ong ( s05m10 ) received the beng degree ( 1st hons ) in electrical engineering from the national university of singapore ( nus ) , singapore , in 2001 .
he subsequently received the mphil degree from the university of cambridge , uk , in 2004 and the phd degree from nus in 2008 .
he was with mobileone , singapore , as a system engineer from 2001 to 2002 .
he was a research fellow at nus , from 2007 to 2008 . from 2008 to 2012 , he was a postdoctoral researcher at the university of newcastle , australia .
he was awarded a discovery early career researcher award ( decra ) in 2012 and a future fellowship in 2014 , both by the australian research council .
he is currently a future fellow at the university of newcastle .
sarah j. johnson ( s01m04 ) received the b.e .
( hons ) degree in electrical engineering in 2000 , and phd in 2004 , both from the university of newcastle , australia .
she then held a postdoctoral position with the wireless signal processing program , national ict australia before returning to the university of newcastle where she is now an australian research council future fellow .
sarah s research interests are in the fields of error correction coding and network information theory .
she is the author of a book on iterative error correction published by cambridge university press .
|
this paper investigates the capacity region of the three - receiver awgn broadcast channel where the receivers ( i ) have private - message requests , and ( ii ) may know some of the messages requested by other receivers as side information .
we first classify all 64 possible side information configurations into eight groups , each consisting of eight members .
we next construct transmission schemes , and derive new inner and outer bounds for the groups .
this establishes the capacity region for 52 out of 64 possible side information configurations . for six groups ( i.e. , groups 1 , 2 , 3 , 5 , 6 , and 8 in our terminology )
, we establish the capacity region for all their members , and show that it tightens both the best known inner and outer bounds . for group 4 ,
our inner and outer bounds tighten the best known inner bound and/or outer bound for all the group members .
moreover , our bounds coincide at certain regions , which can be characterized by two thresholds . for group 7 ,
our inner and outer bounds coincide for four members , thereby establishing the capacity region . for the remaining four members ,
our bounds tighten both the best known inner and outer bounds .
broadcast channel , awgn , capacity , side information .
|
atoms in reduced dimensional lattice , for example , in two - dimensional ( 2d ) honeycomb lattice , has low coordinate number and small spin number . together with magnetic frustration
, these electrons may experience strong spin fluctuations and probably form extremely low - dimensional ( low - d ) antiferromagnet ( afm ) , or even exotic spin liquid , quantum disorder , or quantum spin hall state , so it has attracted great interest in recent theories @xcite .
a recent synthesized honeycomb compound in@xmath0cu@xmath1vo@xmath2 @xcite is such a system with reduced dimensionality and frustration . according to the chemical valence analysis ,
the copper ions with 3@xmath13 configuration contribute a spin-1/2 magnetic moment , while the 3@xmath6 orbitals in vanadium ions are empty and contribute no magnetic moment . due to the large separation between cu
/ v - o layers , the copper spins form a quasi-2d honeycomb lattice @xcite .
this compound exhibits unusual magnetic behavior : it shows neither an afm long - range order in magnetic susceptibility measurement , nor any magnetic phase transition peak in specific heat data @xcite .
however , zn or co doping can induce a weak or strong three - dimensional ( 3d ) afm long - range order observed experimentally @xcite .
it inspires great interest in such a topic : whether the ground state of this compound is low - dimensional afm , or spin liquid predicted by meng _ et al .
_ @xcite .
additionally , it is well known that noninteracting or weakly interacting electrons in hexagonal graphene is a typical example of _ dirac _ fermions .
it is not very clear that how the properties of dirac fermions evolve with the increase of coulomb interactions .
therefore , in@xmath0cu@xmath1vo@xmath2 with hexagonal cu ions is probably a realistic material to testify how the electronic correlations affect the _ dirac _ fermions . in the meantime , through the magnetic susceptibility and specific heat measurements , it s hard to distinguish possible ground state between a low - dimensional afm and a spin liquid @xcite .
all of these drive us to explore the electronic properties and magnetism in in@xmath0cu@xmath1vo@xmath2 by combining the first - principles electronic structure calculations and analytic methods , since it could provide us first insight to this compound . utilizing the local density functional ( lda ) and the correlation correction approaches , together with the analytical perturbation method , we show that in@xmath0cu@xmath1vo@xmath2 is an afm charge transfer insulator with copper d@xmath3 orbitals half filled and contributing @xmath4 = 1/2 spins
. however , vanadium ions are nonmagnetic ( nm ) .
the oxygen @xmath14 orbitals dominate the density of states ( dos ) near @xmath7 .
we extract not only tight - binding parameters between copper ions , but also various spin coupling strengths between cu spins .
furthermore , we show that the absence of 3d afm long - range order observed experimentally in in@xmath0cu@xmath1vo@xmath2 arises from the magnetic frustration of cu spins along the @xmath12-axis , leading to a quantum disorder phase . meanwhile , in terms of this new scenario , we address the evolution of the magnetic properties on the co and zn dopings in in@xmath0cu@xmath1vo@xmath2 , giving a plausible explanation of the experimental behavior .
we also suggest the possibility that doping in oxygen sites may drive the system to a spin liquid or superconducting phase . the rest of this paper is organized as follows : in sec .
ii , we briefly describe the unit cell of in@xmath0cu@xmath1vo@xmath2 and numerical calculation methods ; in sec .
iii , we present the major numerical results and our analysis ; sec .
iv is devoted to the remarks and conclusion .
we first study the electronic state properties of the honeycomb compound in@xmath0cu@xmath1vo@xmath2 by employing the first - principles electronic structure calculation approach so as to elucidate its groundstate properties .
the electronic structure calculations were performed using the self - consistent full potential linearized augmented - plane - wave ( fp - lapw ) scheme in the wien2k programme package .
we used 36 @xmath15 points in the irreducible part of the first brillouin zone .
the muffin - tin sphere radii were selected to be 2.36 , 1.93 , and 1.71 a.u .
for in , cu ( v ) , and o in in@xmath0cu@xmath1vo@xmath2 .
the plane - wave cutoff parameter @xmath16 was 7.0 , and the cutoff between the core and valence states was @xmath177.0 ry in all calculations .
exchange and correlation effects were taken into account in the generalized gradient approximation ( gga ) by perdew , burk , and ernzerhof ( pbe ) @xcite . in order to explicitly take into account the correlated effect of the 3@xmath6 electrons of cu ( v ) ions
, we also performed gga+@xmath18 calculations for in@xmath0cu@xmath1vo@xmath2 .
the effective coulomb interaction @xmath19 = @xmath20 ( @xmath18 and @xmath21 are the on - site coulomb interaction and the hund s rule coupling , respectively ) was used instead of @xmath18 .
we take @xmath19 = 8.0 ev for copper and 4.0 ev for vanadium according to the similar compounds @xcite . to model the low - energy process in in@xmath0cu@xmath1vo@xmath2 , the tight - binding parameters in the nm ( or paramagnetic ( pm ) ) phase are necessary .
we also perform the local density functional ( lda ) calculations for nm in@xmath0cu@xmath1vo@xmath2 . the nm band structures of in@xmath0cu@xmath1vo@xmath2 are obtained by using the full - potential linearized augmented plane - wave plus local orbitals ( fp - lapw+lo ) scheme implemented in the wien2k package @xcite . in order to compare our numerical results with the experimental data ,
we adopt the experimental structural data of in@xmath0cu@xmath1vo@xmath2 measured by neutron powder diffraction @xcite .
this compound has an orthorhombic structure with space group @xmath22 and lattice constants @xmath23 = 10.0491 @xmath24 , @xmath25 = 5.8019 @xmath24 , and @xmath12 = 11.8972 @xmath24 , as shown in fig .
it was found that in@xmath0cu@xmath1vo@xmath2 has a honeycomb layered structure with cu atoms forming hexagonal net and v atoms in the center of an hexagon .
a cu atom is surrounded by five o atoms in a trigonal bipyramidal environment .
detailed results of band structure are addressed in what follows .
we first present the nm electronic structures of in@xmath0cu@xmath1vo@xmath2 in fig .
it shows that the major orbital character of the band structures near @xmath26 is a @xmath27 symmetry of copper 3@xmath6 orbitals .
this cu 3@xmath28 orbital in the honeycomb lattice , similar to the @xmath29 orbital of carbon in graphene , contributes an outstanding property of the band structures , _
i.e. _ , a _ dirac _
cone with the approximately linear spectrum is observed around the @xmath30 point , as seen in fig .
the linear energy spectrum ranges from @xmath170.2 to 0.2 ev .
we notice that the original _ dirac _ point around the @xmath31 points opens a small energy gap , about 0.2 ev , also as seen in fig .
2 , which is attributed to the weak interlayer coupling between cu / v - o hexagonal planes . though parts of empty v 3@xmath6 orbital are near the top of cu 3@xmath28 orbitals , they do not directly hybridize with each other considerably .
a part of filled oxygen 2@xmath5 orbitals are close to the bottom of cu 3@xmath28 orbitals , and these 2@xmath5 orbitals also do not considerably mix with 3@xmath28 orbitals , as we see the partial density of states ( pdos ) of in@xmath0cu@xmath1vo@xmath2 in fig .
3 . thus the low - energy physics in in@xmath0cu@xmath1vo@xmath2 could be approximately described by the cu 3@xmath28 orbitals across@xmath7 .
based on the fact that cu 3@xmath28 orbitals mainly distribute from @xmath171.0 ev to 0.4 ev , we extract these four cu 3@xmath28 bands ( notice that we have four cu atoms in primitive cell ) from the full band structures and fit these bands with a single - orbital tight - binding model with the nearest - neighbor ( nn ) , next nn ( nnn ) and @xmath32 nn hopping integrals in the cu / v - o plane and between the planes .
these tight - binding parameters are listed in table i. notice that the lda methods deal with the nm state rather than the pm one .
but as we know , the overlap and hopping integrals of electrons depend on the spatial wavefunctions , and are independent of the spin degree of freedom .
thus the hopping parameters obtained in the nm state are the same to those in the pm state .
.tight - binding parameters @xmath33 and superexchange spin coupling strengths @xmath34 in in@xmath0cu@xmath1vo@xmath2 .
subscripts @xmath35 and @xmath36 represent the nn , nnn , @xmath32 nn hopping integrals or spin coupling strengths in the planes , and the nn ones between the planes , respectively .
all the energies are measured in units of mev . [ cols="<,>,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] one can easily find that the intraplane nn hopping integral @xmath37 is about 7.5 , 31 and 3.7 times of the intraplane nnn hopping @xmath38 , 3@xmath39 hopping @xmath40 and the interplane nn hopping @xmath41 , respectively , suggesting the nn hopping and nn spin coupling dominate the transfer process and magnetic correlations in the ground state of in@xmath0cu@xmath1vo@xmath2 . to understand the magnetism and insulating nature of undoped compound , we further adopt the spin - polarized gga and correlation - corrected gga+@xmath18 schemes to uncover the groundstate properties of strongly correlated in@xmath0cu@xmath1vo@xmath2 .
we consider five kinds of different magnetic configurations , nm , ferromagnet ( fm ) , a - type afm ( _ i.e. _ , interlayer afm and intralayer fm ) , and g1- and g2-type afm as the possible candidate ground state . here
g1- and g2-type afm configurations shown in fig .
6 are two different magnetic structures , both of which are derived from a conventional g - type afm . in comparison with the g1-type afm structure ,
the lower cu / v - o hexagon ( or cu / v - o layer ) in the g2-type structure rotates @xmath42/3 degree .
thus the g1-type spin configuration differs from the g2-type one .
our gga+@xmath18 numerical results show that the total energy of the nm state is the highest , and that of the g1- and g2-type afm states are the lowest .
interestingly , we find that the g1- and g2-type afm states are degenerate .
this arises from the interlayer magnetic frustration of cu spins , as will be discussed in detail later .
the total energy difference between the fm and a - type afm configurations is rather small .
the energy difference between the lowest g1(g2)-type afm and the second lowest a - type afm or fm states is about 58.5 mev , suggesting that the afm magnetic couplings between cu spins in in@xmath0cu@xmath1vo@xmath2 are dominant . to examine the effect of the on - site coulomb interaction parameter @xmath18 on the calculation results
, we also perform the spin - polarized gga calculations for the purpose of comparison .
the pdos of in@xmath0cu@xmath1vo@xmath2 for g1(g2)-type afm state is shown in fig .
it is found that the system is an insulator with an energy gap of 0.4 ev .
however the calculated magnetic moment is only 0.4 @xmath43 , smaller than that observed by the experiment , about 0.5 @xmath43 @xcite .
considering the fact that the lsda results usually overestimate the magnitude of a magnetic moment , while the experimental results underestimate it due to the significant disorder and fluctuations in realistic compounds , one can expect that there exists a larger magnetic moment than 0.5 @xmath43 . in a similar system cu@xmath1v@xmath1o@xmath44 with 0.73 @xmath43 for @xmath19= 6.52 ev @xcite ,
the gga+@xmath18 calculations are also needed for in@xmath0cu@xmath1vo@xmath2 .
we choose the coulomb parameters lying in a physical parameter range for in@xmath0cu@xmath1vo@xmath2 , _ i.e. _ @xmath19= 8 ev for cu ions , and 4 ev for v ions according to the similar compounds , such as cu@xmath1v@xmath1o@xmath44 @xcite and in@xmath1vo@xmath45 @xcite , as well as the general compounds , such as cuprates and vanadium oxides ( v@xmath1o@xmath0 @xcite ) , _
etc._. notice that we perform the numerical calculations for several different coulomb parameters for cu and v atoms to systematically investigate their influence on our results , _
i.e. _ @xmath19= 6 @xmath46 8 ev for cu @xcite and @xmath19= 3 @xmath46 5 ev for v @xcite . from our numerical calculations
, we find that the band gap and the magnetic moment vary from 1.37 to 1.5 ev and from 0.72 to 0.79 @xmath43 , respectively , when u@xmath47 changes from 6 to 8 ev for cu and from 3 to 5 ev for v. this clearly shows that the electronic and magnetic properties do not crucially depend on the values of the parameter u in the physical range . to further uncover the insulating character of in@xmath0cu@xmath1vo@xmath2 , the pdos for g1(or g2)-type afm state by the gga+@xmath18 approach is plotted , as seen in fig .
the system is obviously an insulator with an energy gap of 1.5 ev .
we found that oxygen 2@xmath5 orbitals consist of the major bands near @xmath7 , ranging from 0 to @xmath174.0 ev . while the gravity center of cu 3@xmath6 orbitals lie far from @xmath7 .
the center of the spin - up 3@xmath6 orbital lies at about @xmath175.9 ev , and that of the spin - down 3@xmath6 orbitals lies at about @xmath174.8 ev .
this gives rise to the exchange splitting of cu 3@xmath6 orbitals about @xmath48 @xmath9 1.1 ev , implying that the hund s rule coupling @xmath21 = @xmath49 @xmath9 1.1 ev . such a large exchange splitting also shows a strong coulomb correlation on cu 3@xmath6 orbitals .
note that the charge transfer energy between o 2@xmath5 orbital and cu 3@xmath6 orbital in in@xmath0cu@xmath1vo@xmath2 , @xmath50 = @xmath51 , which has the same order as the insulating energy gap , is considerably smaller than the on - site coulomb interaction of cu 3@xmath6 electrons , demonstrating that in@xmath0cu@xmath1vo@xmath2 is a charge transfer insulator .
the charge distribution and occupation numbers in the 3@xmath6 orbitals of cu and v are essential to understand the electronic properties of in@xmath0cu@xmath1vo@xmath2 .
our results show that in the cu 3@xmath6 orbitals , the hole mainly occupies the active @xmath52 orbital .
it has a total number of 1.17 electrons , or 0.83 holes , in the @xmath52 orbital . in contrary to the chemical valence analysis
, v has totally 1.87 electrons in its 3@xmath6 orbitals , though it shows nm or no spin polarization in the present gga+@xmath18 ansatz .
this arises from a strong @xmath5-@xmath6 hybridization between v and o atoms , and a fraction of electrons transfer from oxygen 2@xmath5 orbitals to vanadium 3@xmath6 orbitals . fig .
5 displays that the main dos of copper lies from @xmath177 ev to @xmath174 ev , though a tiny portion of 3@xmath6 electrons lie between @xmath174 ev to @xmath7 which arises mainly from the hybridization of cu 3@xmath6 orbital with the 2@xmath5 orbitals near @xmath7 of surrounding o anions .
it s observed from fig .
5 that the 3@xmath52 orbital in cu is almost fully spin polarized .
the magnetic moment of cu spin is almost contributed from the well localized spin - up 3@xmath52 orbital at the position of about 7 ev below @xmath7 .
we find that the orbital ground state is the cu 3@xmath52 character , in agreement with the electron spin resonance experiment by kataev _
_ @xcite .
meanwhile , a small but finite dos distributed from @xmath176 ev to @xmath7 is seen in v 3@xmath6 orbitals , showing that the two 3@xmath6 electrons form a wide energy band below the fermi energy .
obviously , other v 3@xmath6 electrons contribute the highest empty conduction bands , as we see in fig .
the pdos of v shows that two 3@xmath6 electrons in v are not spin polarized and allocate in five 3@xmath6 orbitals with almost equal weight , in sharp contrast to those of cu atoms .
5 shows that in 5@xmath5 orbitals give rise to a small dos in a wide energy range .
in contrast , the dos of oxygen mainly distributes near @xmath7 and is greatly larger than that of in .
such a strongly correlated character in in@xmath0cu@xmath1vo@xmath2 is similar to that in the parent phases of high - t@xmath53 cuprates .
this implies that the doping in o sites can readily affect the transport properties in in@xmath0cu@xmath1vo@xmath2 , and easily drive the system transit to metallic or superconducting phase once hole carriers are doped in oxygen 2@xmath5 bands .
the groundstate magnetic properties of in@xmath0cu@xmath1vo@xmath2 is the central issue of this paper . as described above
, our numerical results demonstrate that the stablest phase is the degenerate g1- and g2-type afm states .
the magnetic moment mainly contributes from copper spins with a total net magnetic moment of 0.79 @xmath43 per cu atom in which the average electron number is 0.98 in the spin - up band , and 0.19 in the spin - down one .
the magnetic moment mainly arises from the @xmath52 orbital of copper . apparently , a strong electron correlation and a large hund s rule coupling result in such a large spin polarization , leading to a well defined local moment in coppers . in contrast , vanadium is nm , though it is 3@xmath54 configuration with two electrons distributing from @xmath7 to @xmath177.0 ev .
its nm character may arise from the strong _ p - d _ hybridization between v and o , resulting in a weak correlation between v 3@xmath6 electrons .
we do not rule out the possibility of spin frustrations in the present cu - v geometry , which may deserve further discussion in the future .
due to the full fillings of the oxygen 2@xmath5 orbitals and in 5@xmath5 orbitals , both o and in are not spin polarized .
the insulating nature in in@xmath0cu@xmath1vo@xmath2 suggests that the local spins of coppers interact through a superexchange couplings mediated via oxygens and vanadium .
one expects that the contributions to the nn and nnn cu - cu superexchange couplings arise from the direct hopping between @xmath28 electrons , the indirect hopping between cu spins through the @xmath5-@xmath6 hybridization of intermediate oxygen anions and through the o - v - o bridge .
to quantitatively obtain the superxchange coupling strengths , we utilize the tight - binding parameters obtained in table i to estimate the nn and nnn superexchange couplings between cu spins in the cu / v - o plane and between the planes , taking the effective coulomb interaction @xmath18 as 8 ev .
a @xmath55/@xmath18 expansion of the hubbard model is used to obtain an effective @xmath8-@xmath10-@xmath56 spin @xmath4=1/2 model on a honeycomb lattice : @xmath57 where @xmath58 , @xmath59 and @xmath60 are the nn , nnn intralayer and the nn interlayer magnetic couplings , respectively .
we find that the intraplane and interplane nn superexchange couplings are about @xmath61 16.2 mev and @xmath62 1.2 mev , respectively ; and the intraplane nnn superexchange couplings are about @xmath63 0.3 mev .
more long - range spin couplings are so small to be negligible .
this gives rise to @xmath10/@xmath8 @xmath9 0.02 , in addition to a rather weak interlayer coupling of @xmath56/@xmath8 @xmath9 0.07 .
all of these spin coupling strengths have been listed in table i. from the literature available , we find that the spin coupling strengths of in@xmath0cu@xmath1vo@xmath2 fall in the parameter range of a 2d heisenberg afm @xcite , showing that in@xmath0cu@xmath1vo@xmath2 is an insulator with 2d afm . as we will illustrate later , such a weak interlayer magnetic coupling and the spin frustration prevent from the long - range afm order in in@xmath0cu@xmath1vo@xmath2 .
it is observed experimentally that neither the specific heat nor magnetic susceptibility exhibits afm - pm transition @xcite : the former is almost t@xmath64 dependent , and the latter linearly increases with the lift of temperature .
consequently , one naturally suspects that the ground state of in@xmath0cu@xmath1vo@xmath2 might be a gapless spin liquid phase , a long - time searched exotic quantum phase both theoretically and experimentally , since such a phase in _
kagome _ lattice also demonstrates similar unusual temperature - dependent behavior in the low t regime @xcite .
meanwhile , a strongly low - d afm may exhibit similar behaviors @xcite . though both a low - dimensional afm and a spin liquid have local permanent magnetic moments , and do not display any sign of ordering in specific heat and magnetic susceptibility down to the lowest temperatures despite of comparable strong afm interactions .
a low - dimensional afm could be different from a spin liquid in many aspects .
for example , a 2d afm has a long - range correlation and long - range order at t= 0 k as well as short - range one at finite t , while a spin liquid only has short - range correlation , even at t= 0 k ; this leads to distinct low - temperature behavior in magnetic susceptibility for a 2d afm and a spin liquid , _ etc._. therefore our theoretical analysis above suggests 2d afm ground state rather than a spin liquid , though it is still mysterious why in@xmath0cu@xmath1vo@xmath2 exhibits neither an afm long - range order in the low temperature regime nor a curie - weiss law in the high temperature regime @xcite , which is exotic and distinctly different from undoped cuprates .
an earlier experimental study by m@xmath65ller _
_ @xcite attributed the absence of a 3d afm long - range order to the disorder stacking of 2d cu@xmath66v order domains along the @xmath12-axis .
however , from our preceding data and the particular geometry of in@xmath0cu@xmath1vo@xmath2 , as an alternative explanation , we propose that it is the strong afm frustration along the @xmath12-axis leads to the absence of the 3d afm long - range order in in@xmath0cu@xmath1vo@xmath2 . as seen the cu - v geometric configuration among two cu - v layers in fig .
6 , a cu spin has two nn layers along the @xmath12-axis , it has the same distance to other two cu spins in the upper or lower layer , forming an isosceles triangle which is a notorious spin configuration with magnetic frustration .
our analysis above has shown that both the spin couplings of the interlayer and intralayer spins are afm , indicating the spin fluctuations along the @xmath12-axis will be very large .
thus in in@xmath0cu@xmath1vo@xmath2 the intralayer cu spins are afm correlated , while the interlayer cu spins are quantum disordered .
such a frustration scenario could unifiedly account for both the presence of low - d afm correlation @xcite and the absence of 3d afm order @xcite in recent experiments .
furthermore , we apply our frustration scenario to uncover the evolution of the magnetic properties on the co and zn dopings in in@xmath0cu@xmath1vo@xmath2 , giving a possible explanation of the experimental behavior , as shown in fig .
the zn and co dopings break the inter - layer frustration , while the nm zn@xmath67 ions destroy the intra - layer afm . on the contrary ,
high spin-3/2 co@xmath67 ions enhance it .
thus the zn doping induces a weak afm at low doping , but suppresses the afm state completely at high doping . in comparison , the co doping results in a strong afm with large effective magnetic moment .
all the results are in accordance with the experimental observations . moreover , as we point out above
, the hole doping in o sites of in@xmath0cu@xmath1vo@xmath2 may introduce carriers near fermi energy . with the increase of carriers ,
one expects that the screening effect can greatly reduce the coulomb interaction , and weaken the electronic correlation .
this may drive the in@xmath0cu@xmath1vo@xmath2 system from an afm ground state to a spin liquid phase @xcite , since in the intermediate correlation regime , a half - filled hubbard system with the honeycomb structure exists a spin - liquid ground state @xcite .
nevertheless , we do nt expect that doped in@xmath0cu@xmath1vo@xmath2 will form the well - known zhang - rice singlet in the cu - o plane of cuprates , since the latter arises from the hybridization between cu 3@xmath68 and o 2@xmath5 orbitals , forming delocalized states in the xy plane ; while in the former , the 3@xmath52 orbital mixes with o 2@xmath5 orbital , forming a delocalized state in the @xmath69-direction .
this results in completely different transport properties in doped compounds , as yan _
et al . _ observed in in@xmath0cu@xmath70co@xmath71vo@xmath2 and in@xmath0cu@xmath70zn@xmath71vo@xmath2 @xcite . in summary
, the investigation of the electronic structure and magnetic properties in quasi-2d honeycomb compound in@xmath0cu@xmath1vo@xmath2 demonstrates that the undoped phase is a charge transfer insulator with a gap of 1.5 ev , and oxygen 2@xmath5 orbitals dominate the bands near @xmath7 .
spin-1/2 local moment antiferromagnetically interacts and occupies the 3@xmath72-@xmath73 orbital in copper .
in contrast , the nm v ions have two electrons in the 3@xmath6 orbitals . the tight - binding parameters and estimated spin coupling strengths suggest that in@xmath0cu@xmath1vo@xmath2 is a strongly 2d _ nel _ afm .
we propose that the spin fluctuations arising from the magnetic frustration along the @xmath12-axis destroy the 3d afm long - range order , leading to random stacking of cu - v layers .
+ we acknowledge x. h. chen provides us with original experimental data .
this work was supported by the national sciences foundation of china under grant no .
11074257 and 11104274 , and knowledge innovation program of the chinese academy of sciences .
numerical calculations were performed at the center for computational science of caships .
p. blaha , k. schwarz , g. madsen , d. kvasnicka , and j. luitz : `` computer code wien2k , an augmented plane wave plus local orbitals program for calculating crystal properties '' , karlheinz schwarz , technische universit@xmath75 t wien , austria , ( 2001 ) .
f. rodolakis , p. hansmann , j .-
rueff , a. toschi , m. w. haverkort , g. sangiovanni , a. tanaka , t. saha - dasgupta , o. k. andersen , k. held , m. sikora , i. alliot , j .-
iti@xmath76 , f. baudelet , p. wzietek , p. metcalf , and m. marsi , phys .
lett . * 104 * , 047401 ( 2010 ) .
|
we present electronic and magnetic properties of a honeycomb compound in@xmath0cu@xmath1vo@xmath2 in this paper .
we find that the parent phase is a charge transfer insulator with an energy gap of about 1.5 ev . singly occupied d@xmath3 electrons of copper ions contribute an @xmath4 = 1/2 spin , while vanadium ions show nonmagnetism .
oxygen 2@xmath5 orbitals hybridizing with a small fraction of cu 3@xmath6 orbitals dominate the density of states near @xmath7 .
the planar nearest - neighbor , next - nearest - neighbor and interplane superexchange couplings of cu spins are @xmath8 @xmath9 16.2 mev , @xmath10 @xmath9 0.3 mev and @xmath11 @xmath9 1.2 mev , suggesting a low - dimensional antiferromagnet @xcite .
we propose that the magnetic frustration along the @xmath12-axis leads to a quantum spin disorder in in@xmath0cu@xmath1vo@xmath2 , in accordance with the recent experiments .
300 pt
|
the study of a motion with a constant proper acceleration is a classical exercise of special relativity that can be found in many textbooks @xcite . with its analytical solution ,
it is possible to show that the limit speed of light is asymptotically reached despite the constant proper acceleration .
the very prominent notion of event horizon can be introduced in a simple context and the problem of the twin paradox can also be analysed . in many articles of popularisation
, it is sometimes stated that the point of view of an observer with a constant proper acceleration can not be treated within the theory of special relativity and that theory of general relativity is absolutely necessary .
actually , this is not true .
the point of view of an uniformly accelerated observer has been studied , for instance , in refs .
@xcite . in this paper ,
some particular topics are summed up and developed .
the metric for an observer with a proper constant acceleration ( coming from infinity and coming back towards infinity ) is built in sec .
[ sec : metric ] by considering a series of inertial frames instantaneously at rest with this observer .
the notions of local time and local velocity for such an observer are introduced in secs .
[ sec : ltime ] and [ sec : lspeed ] .
in particular , the equation of motion of a freely moving object is studied in sec . [
sec : emffobj ] and its local velocity calculated in sec .
[ sec : vffobj ] .
the problem of the distance existing between two accelerated observers with the same constant proper acceleration @xcite is examined in sec .
[ sec : distobs ] , in terms of the metric associated with one of the accelerated observer .
at last , the metric of the accelerated observer is compared with the metric due to a gravitational field in sec . [
sec : mgravpot ]
. the equivalence principle of the general theory of relativity states that a gravitational field and an acceleration field are locally equivalent .
this means that , in a sufficiently small region of space and for a sufficiently small duration , the gravitational field can be cancelled in a suitable accelerated frame .
conversely , a gravitational field can be simulated by an accelerated frame .
thus , this paper can be considered as a first approach of the general theory of relativity by studying the point of view of an accelerated observer .
the equation of motion , along a straight line , of an accelerated observer with a constant proper acceleration can be found in a great number of textbooks @xcite ( in the following , the coordinates perpendicular to the velocity of the observer do not play any role and they are ignored ) . if @xmath0 is the modulus of the proper acceleration , @xmath1 the velocity of the observer and @xmath2 its acceleration along the @xmath3-axis , both measured in an inertial frame @xmath4 , the one - dimensional equation of motion is given by @xmath5 with the initial conditions @xmath6 and @xmath7 at @xmath8 , the integration of eq .
( [ 2.125 ] ) gives @xmath9.\end{aligned}\ ] ] this motion is also called hyperbolic motion because the last equation can be recast into the form @xmath10 which is the equation of a branch of hyperbola in spacetime ( see fig .
[ fig : oa1 ] ) .
the asymptotes of this curve are the two straight lines with equations @xmath11 .
the velocity of each physical object is such that @xmath12 .
this implies that the angle between the tangent at each point of the world line of an object and the time - axis in fig .
[ fig : oa1 ] is always comprised between 0 and @xmath13 .
consequently , the asymptotes of the world line of the accelerated observer defines two event horizons : the events above " the future horizon can not send any information to the accelerated observer , who can not send any information to events located below " the past horizon .
these two event horizons will be discussed below .
the well known relation between infinitesimal intervals of proper time and of coordinate time , @xmath14 can be integrated to obtain the relation between the elapsed proper time @xmath15 for the accelerated observer and the elapsed time @xmath16 for a stationary observer in the inertial frame @xmath17 clocks for the observers are synchronised such that @xmath18 when @xmath8 . with relations ( [ 2.126 ] ) , ( [ 2.126b ] ) and ( [ 2.129 ] ) , velocity and position for the accelerated observer can be computed as a function of the proper time @xmath19.\end{aligned}\ ] ] a system of local coordinates for the accelerated observer can be built by considering a series of inertial frames instantaneously at rest with this observer . a particular event on the world line of the accelerated observer
is noted @xmath20 .
this observer occupies a position @xmath21 at a time @xmath22 in the inertial frame @xmath4 and is characterized by a reduced velocity @xmath23 in this frame .
a new inertial frame @xmath24 can be built with its spacetime origin on @xmath20 and with a reduced velocity @xmath25 in @xmath4 ( see fig . [
fig : oa2 ] ) . the lorentz transformation between the two inertial frames @xmath4 and @xmath24
can be written @xmath26 equation ( [ 2.126 ] ) implies that @xmath27 at the event @xmath20 , the spacetime coordinates of the accelerated observer in the inertial frame @xmath24 are obviously @xmath28 and @xmath29 . by expressing @xmath21 as a function of @xmath22 ,
the system ( [ metobs1 ] ) becomes @xmath30 the accelerated observer can now use the inertial frame @xmath24 to build a proper system of coordinates . if an event @xmath31 occurs at a time @xmath32 in the frame @xmath24 , it is natural for the accelerated observer to consider that this event occurs at the time @xmath33 indicated by a clock of the observer , since the time coordinate of the observer is also @xmath32 in the frame @xmath24 ( see fig . [
fig : oa2 ] ) .
this time @xmath33 is then identical to its proper time @xmath15 .
it is also natural to assign at this event @xmath31 a position @xmath3 in the proper frame of the observer which is identical to the position @xmath34 of @xmath31 in the frame @xmath24 .
finally , the event @xmath31 with spacetime coordinates @xmath35 in the frame @xmath24 has spacetime coordinates @xmath36 in the proper frame of the accelerated observer .
the relation ( [ 2.129 ] ) indicates that the proper time of the observer at time @xmath22 is given by @xmath37 .
consequently , the factor @xmath38 is equal to @xmath39 and the system ( [ metobs2 ] ) can be rewritten @xmath40 the metric associated with the accelerated observer can now be determined by computing the invariant @xmath41 , which gives @xmath42 it is worth mentioning that a spacetime with such a metric has no curvature , since this metric is obtained from a flat metric with a change of coordinates .
it can be verified that the calculation of the curvature tensor for the metric ( [ etc5 ] ) gives a null scalar curvature , as expected @xcite . in the proper frame of the accelerated observer , called @xmath43 here ,
each object with a position @xmath3 positive ( negative ) is located above ( below ) the observer , since the acceleration defines a privileged vertical direction along the @xmath3-axis .
it is worth noting that the metric ( [ etc5 ] ) is not defined for @xmath44 . to understand why this position is so particular ,
it is necessary to study the spacetime structure of the surroundings of the accelerated observer . with relations ( [ apacc6 ] ) ,
it is possible to determine the equations of the coordinates lines of the frame @xmath43 in the inertial frame @xmath4 . in this last frame ,
the equation of a coordinate line @xmath33 with @xmath45 constant is @xmath46 this curve is a branch of hyperbola whose asymptotes are the two event horizons mentioned above .
these horizons are located on the degenerate asymptotes obtained with @xmath47 in eq .
( [ lctemps ] ) .
obviously , the world line of the accelerated observer in the frame @xmath4 is given by @xmath48 .
the equation of a coordinate line @xmath3 with @xmath49 constant is @xmath50 this is a straight line containing the event @xmath51 , that is to say the intersection of the two event horizons .
when @xmath52 , the straight line is parallel to the bisector of the axis @xmath53 and @xmath54 .
some coordinate lines are drawn in fig .
[ fig : oa3 ] .
it can be seen that the future horizon and the past horizon correspond respectively to the time coordinate lines @xmath55 and @xmath56 .
both horizons form also the space coordinate line @xmath57 .
the spacetime region with @xmath44 is then behind the event horizons .
the event horizons are then always located at a distance @xmath58 downward for the accelerated observer . in order to understand the nature of these horizons ,
let us assume that a stationary observer in the inertial frame @xmath4 sends photons to the accelerated observer : the emission starts in order that the first photons reach the accelerated observer at @xmath59 , for instance ; it stops when the stationary observer crosses the future horizon ( photons emitted later can not be received ) .
it is clear from fig .
[ fig : oa4 ] that all photons are received after an infinite time , despite the fact that they are emitted during a finite time @xmath60 .
let us consider a clock at rest at position @xmath3 in the proper frame @xmath43 of the accelerated observer .
an interval of proper time @xmath61 for this clock corresponding to an interval of coordinate time @xmath62 is given by @xmath63 since @xmath64 .
let us remark that the coordinate time is also the proper time at the level of the accelerated observer ( @xmath65 ) .
the time flows with the same rate for all clocks with the same altitude " in @xmath43 ( same value of @xmath3 ) . on the contrary , clocks located at different altitudes measure different intervals of proper time for a same interval of coordinate time .
the time flows more slowly for all clocks located below a reference clock .
in particular , the interval of proper time @xmath61 vanishes at the level of event horizons . for the accelerated observer , the time
freezes " at the coordinate @xmath57 ( see fig . [
fig : oa4 ] ) .
it is worth noting that , for a value of @xmath0 close to the terrestrial gravitational acceleration ( around 10 m / s@xmath66 ) , @xmath58 is around one light year ( 10@xmath67 m ) . with this value of @xmath0 ,
two clocks located 1 m apart will be get out of synchronisation by about 1 s every 10@xmath67 s ( around @xmath68 yr ) .
an interval of local time can be defined , @xmath69 , with which it is possible to build in the vicinity of a particular event a local metric similar to the usual minkowski metric @xmath70 this metric can be considered as the metric of an inertial frame instantaneously at rest with the proper frame of the accelerated observer .
but , in this frame , the intervals of time considered must be small enough in order that the accelerated frame do not move appreciably .
an interval of local time corresponds then to an interval of time measured by a clock at rest in this instantaneous inertial frame .
the metric ( [ etc6 ] ) allows only a local description of an infinitesimal region of spacetime like a minkowski spacetime with local coordinates @xmath71 .
let us consider two events @xmath72 and @xmath73 on the world line of an object that moves in the non inertial frame @xmath43 .
the coordinate time interval between these events is @xmath62 and the coordinate distance is @xmath74 .
it is possible to define three types of velocity for this object : * the coordinate velocity @xmath75 . * the proper velocity @xmath76 , which is the spatial part of the world velocity . * the local velocity @xmath77 , computed with the interval of local time @xmath78 .
hence , we have @xmath79 and @xmath80 the local velocity @xmath81 is the coordinate velocity of the object in the inertial frame that is instantaneously at rest relative to the non inertial frame of the accelerated observer ( see previous section ) . this velocity can not thus exceed the speed of light .
this is not the case for the coordinate velocity @xmath82 since , according to eq .
( [ etc8 ] ) , @xmath82 could exceed @xmath83 if @xmath84 is large enough .
if the object is a photon , its local velocity @xmath81 must be the invariant @xmath83 , but its coordinate velocity varies as a function of its position .
the interval of proper time elapsed between two events infinitesimally close on the world line of a moving object in the non inertial frame is given by @xmath85 where @xmath82 is the coordinate velocity of the object .
the finite duration of proper time between two events @xmath72 and @xmath73 on this world line is calculated by integration @xmath86 the principle of equivalence requires that the world line of a freely moving object that passes through two fixed events is such that the elapsed proper time between these two events is maximum @xcite . if the integral ( [ etc11 ] )
is written @xmath87 then the calculus of variations shows that the functional @xmath88 must satisfy the following condition @xmath89 this is the euler - lagrange equation , which ensures that the quantity @xmath90 is extremum .
let us mention that this problem is formally identical to the computation of the minimal length between two points on a curved space described by a spatial metric .
the curve satisfying this condition is called a geodesic . here , the curve in space is replaced by a world line in the spacetime and the variation produces a maximum in the lapse of proper time instead of a minimum of spatial distance .
substituting @xmath88 by its value in eq .
( [ etc13 ] ) gives @xmath91 by using @xmath92 .
the quantity @xmath93 is the coordinate acceleration of the moving object in the non inertial frame of the accelerated observer .
let us remark that @xmath94 , as expected , in the case of an inertial frame for which @xmath95 . with the function @xmath84 given by eq .
( [ etc5 ] ) , the coordinate acceleration is written @xmath96.\ ] ] when the object moves slowly ( @xmath97 ) near the accelerated observer ( @xmath98 ) , this equation of motion reduces to @xmath99 the object accelerates downward at the rate @xmath0 , as expected .
this corresponds to a free fall in a constant gravitational field ( see sec .
[ sec : mgravpot ] ) .
equation ( [ etc14 ] ) can be integrated to yield the velocity of the freely moving object as a function of its position in the non inertial frame . multiplying both sides of this equation by @xmath74 and using the relations @xmath100
the acceleration equation becomes @xmath101 with the notation @xmath102 , this equation can be put into the form @xmath103 this differential equation can be solved by using the separation of variables method , if one introduces the new function @xmath104 .
but it is simpler to try a solution of the form @xmath105 it is then easy to see that the coefficients @xmath106 are such that @xmath107 , @xmath108 , @xmath109 is arbitrary and @xmath110 for @xmath111 .
so , we can write @xmath112 where @xmath113 is the constant of integration ( @xmath114 in eq .
( [ etc17ter ] ) ) . using eq .
( [ etc8 ] ) , the local velocity of the freely moving object is given by @xmath115 with this equation , it is possible to prove that the local velocity of an object can not exceed the speed of light if its local velocity is below @xmath83 at one point of its world line .
since @xmath84 is positive , @xmath81 can exceed @xmath83 only if @xmath113 is negative .
( [ etc19 ] ) implies that @xmath116 if we assume that @xmath117 at one point of its world line , then @xmath118 .
since @xmath113 is a constant , it is always positive and @xmath81 is always less than @xmath83 . alike , if the local velocity of a photon is equal to @xmath83 at one point of its world line , then @xmath119 .
consequently , @xmath113 is null everywhere on the world line and the local velocity of the photon is always equal to @xmath83 .
these results are important since they show that the local velocity of an object can not exceed the speed of light and that the local velocity of a photon is always @xmath83 .
the local velocity of a freely moving object has been calculated in the proper frame of the accelerated observer .
it is possible to compute this speed from the equation of motion of the object given in the inertial frame @xmath4 .
the two approaches must obviously give the same result .
with eqs .
( [ apacc6 ] ) , the function @xmath84 is given by @xmath120 by differentiating both eqs .
( [ apacc6 ] ) and by noting @xmath121 and @xmath75 , the relation between the two velocities in the two frames is given by @xmath122 where @xmath123 is a constant since the object is moving freely .
the quantity @xmath124 can be rewritten as a function of coordinates @xmath53 and @xmath16 using eqs .
( [ apacc6 ] ) .
after inversion of eq .
( [ oaplus2 ] ) and using the definition ( [ etc8 ] ) of the local speed , it is found that @xmath125 when @xmath8 , we have @xmath126 , as expected since the proper frame of the accelerated observer coincides instantaneously with the inertial frame @xmath4 at this time . since the object is freely moving , its equation of motion can be written in the inertial frame @xmath4 as @xmath127 where @xmath128 and @xmath129 are constants .
the introduction of this relation in eq .
( [ oaplus3 ] ) gives the local velocity @xmath81 as a function of time @xmath16 of the inertial frame , but it is more interesting to modify eq .
( [ oaplus3 ] ) in order to recover eq .
( [ etc19 ] ) . with the notation @xmath130 and with the use of eqs .
( [ oaplus1 ] ) and ( [ oaplus4 ] ) , the function @xmath131 for the freely moving object can be written @xmath132 with some calculations , it can then be shown , with eqs .
( [ oaplus3 ] ) , ( [ oaplus4 ] ) and ( [ oaplus6 ] ) , that the local velocity of this object is given by @xmath133 this equation is identical to eq .
( [ etc19 ] ) and the value of @xmath113 is then @xmath134 a simpler procedure exists to find the value of @xmath113 .
since it is a constant , it can be evaluated at any time , for instance at @xmath8 .
this constant is then given by formula ( [ etc20 ] ) in which @xmath135 , as explained above , and @xmath136 .
so the formula ( [ oaplus7bis ] ) is obtained directly . from these formulas , it is clear that @xmath137 for a photon , as expected . on the event horizons ( @xmath57 ) , the function @xmath131 is vanishing and the velocity of the object tends towards the speed of light . from the point of view of the accelerated observer , he is at infinity when the object reaches a horizon ( see sec . [
sec : metric ] and fig .
[ fig : oa3 ] ) , that is to say when its velocity tends towards the speed of light in the inertial frame @xmath4 .
the function @xmath131 in eq .
( [ oaplus6 ] ) is a quadratic form in @xmath16 which possesses a maximum equal to @xmath138 . in this case
, the local velocity is vanishing . the object and the accelerated observer
have then the same velocity in the inertial frame @xmath4 .
finally , the relation ( [ oaplus3 ] ) can be rewritten into the form @xmath139 since @xmath81 is only defined for @xmath140 and @xmath141 ( see fig .
[ fig : oa3 ] ) , then @xmath142 .
equation ( [ oaplus3b ] ) is thus the relativistic addition of two velocities below the speed of light .
it is shown again that the local velocity @xmath81 can not exceed @xmath83 .
it is possible to give an interpretation of the speed @xmath143 .
the equalities @xmath144 , where @xmath22 is the time coordinate of the accelerated observer in the inertial frame @xmath4 ( see sec .
[ sec : metric ] ) , imply that the first equation of the system ( [ apacc6 ] ) can be rewritten @xmath145 using the eqs .
( [ oaplus1 ] ) and ( [ oaplus8 ] ) in the formula ( [ oaplus3b ] ) for @xmath143 , one can find @xmath146 .
the local speed @xmath81 of the freely moving object in the proper frame of the accelerated observer at coordinate time @xmath33 is then the relative speed between the constant velocity @xmath147 of the object in the inertial frame @xmath4 and the instantaneous velocity @xmath148 of the accelerated observer in the same inertial frame at time @xmath22 , with @xmath149 .
let us consider two spaceships with the same proper acceleration @xmath0 moving in an inertial frame @xmath4 .
their world lines , plotted in fig . [ fig : oa5 ] , are such that , at @xmath8 , they are a distance @xmath150 apart , at rest in @xmath4 . at each time in the inertial frame , the ships possess the same velocity and are the same distance @xmath150 apart .
a rod of length @xmath150 is fixed between the two spaceships when they are at rest . as the speed of the ships increases
, the lorentz - fitzgerald contraction will occur from the point of view of stationary observers in @xmath4 .
this rod will then tend to be lengthened by an increasing stress since the distance between the two ships is constant in @xmath4 .
this classical problem is studied in many papers ( see for instance refs .
@xcite ) .
it is interesting to analyse this problem from the point of view of an observer on board a ship .
the parametric equations of motion of the first ship , as a function of its proper time @xmath15 , in the inertial frame @xmath4 are ( see eqs .
( [ 2.129 ] ) and ( [ 2.130b ] ) ) @xmath151.\ ] ] the ones for the second ship in the same frame , as a function of its proper time @xmath152 , are @xmath153.\ ] ] clocks in all frames are synchronised : @xmath154 .
let us note @xmath155 the position of the second spaceship in the proper frame of the first one .
the relations between the coordinates in the frame @xmath4 and the coordinates in the proper frame of the first spaceship are given by eqs .
( [ apacc6 ] ) . by introducing the coordinates ( [ 2mob2 ] ) of the second ship in eqs .
( [ apacc6 ] ) , the following relations between the proper times @xmath15 and @xmath152 are found @xmath156 with the identification @xmath157 in eqs .
( [ 2mob3 ] ) . the position @xmath155 can be obtained by eliminating the proper time @xmath152 in these equations .
this yields a quadratic equation in @xmath155 whose the only physical solution is @xmath158 with @xmath159 as expected . when the proper time @xmath160 , eq .
( [ 2mob4 ] ) reduces to @xmath161 from the point of view of an observer in the first spaceship , the distance increases between the two ships .
this in agreement with the reasoning sustained above about the rod .
this phenomenon is due to the fact that the spaces of simultaneity for an observer on board the first space ship are different from the spaces of simultaneity for a stationary observer in the inertial frame .
this can be seen on an example in fig .
[ fig : oa5 ] where @xmath150 is fixed at 0.5 @xmath58 : at the time @xmath162 , @xmath163 is clearly greater then @xmath150 .
in the newtonian theory of gravity , the gravitational field is derivable from a function @xmath164 call the gravitational potential . in a one - dimensional space ,
the acceleration @xmath93 of a freely falling object in an arbitrary frame is given by @xmath165 in the vicinity of an accelerated observer , objects seems to undergo the effects of a gravitational field ( see eq .
( [ etc14c ] ) ) . one can ask what kind of relation could exist between this pseudo gravitational field and the metric of the accelerated observer .
( [ etc21 ] ) has a form similar to eq .
( [ etc14 ] ) , which relates the acceleration with the spacetime metric . at the classical limit , low velocity ( @xmath97 ) and near minkowski metric ( @xmath166 ) , eq .
( [ etc14 ] ) reduces to @xmath167 with this approximation , we have @xmath168 by integration , it is found that @xmath169 where @xmath170 is a constant of integration allowing to fix the value of the potential at a particular point .
equation ( [ etc14c ] ) , obtained at the same classical limit , implies that the acceleration is constant and equal to @xmath171 . in this case
, the potential is @xmath172 ( a constant @xmath173 can be absorbed into the constant @xmath170 ) , and eq .
( [ etc14c ] ) gives @xmath174 the constant is fixed at @xmath175 , in order that the metric is a minkowski metric at the level of the accelerated observer ( @xmath65 ) .
in the general case , the function @xmath84 is actually given by @xmath176 the real metric and the metric at the classical limit coincide when @xmath98 , that is to say for small accelerations or for short distances from the accelerated observer , as expected . to conclude
let us show with a realistic example that a gravitational field is locally equivalent to an acceleration field .
if an observer is far form a source of gravitation and travels only on short distances , he can consider the gravitational field as uniform and can define a constant @xmath177 which measures the value of the local acceleration .
let us consider the schwarzschild metric for a mass @xmath20 with spherical symmetry , located at the origin . in the usual spherical coordinates , it is written @xmath178 where @xmath179 is the universal gravitational constant .
for an observer at rest ( @xmath180 ) located at altitude @xmath181 , the interval of proper time @xmath61 is related to an interval of coordinate time @xmath62 by @xmath182 let us assume that this observer is located at a great distance @xmath183 from the origin ( @xmath184 ) .
he can define a local constant value @xmath177 of the acceleration by the standard formula @xmath185 moreover , this observer makes experiments only for values of @xmath181 close to @xmath183 .
thus , he considers values of @xmath186 with @xmath187 . using eq .
( [ etc27 ] ) and eq .
( [ etc28 ] ) , he can find the relation between an interval of proper time @xmath188 at altitude @xmath189 and an interval of proper time @xmath190 at altitude @xmath183 @xmath191 since @xmath192 and @xmath187 , the first order expansion of formula ( [ etc29 ] ) gives @xmath193 let us now look at the case of an observer with a constant proper acceleration .
the formula ( [ etc5bis ] ) gives the interval of proper time @xmath194 at altitude @xmath3 as a function of an interval of proper time @xmath62 at altitude @xmath65 .
for small values of @xmath3 or for weak acceleration @xmath0 , this formula can be written @xmath195 this relation is formally identical to relation ( [ etc30 ] ) .
locally , the two metrics , the one for the gravitational field and the one for the acceleration field , can not be discriminated .
the author ( fnrs research associate ) would like to thank the fnrs for financial support .
aa sears f w and brehme r w 1968 _ introduction to the theory of relativity _ ( london : addison - wesley ) misner c w , thorne k s and wheeler j a 1973 _ gravitation _ ( san francisco : freeman ) semay c and silvestre - brac b 2005 _ relativit restreinte .
bases et applications _ ( paris : dunod ) desloge e a and philpott r j 1987 uniformly accelerated reference frames in special relativity _ am .
j. phys . _
* 55 * 252 - 261 boughn s p 1989 the case of the identically accelerated twins _ am .
j. phys . _
* 57 * 791 - 793 evett a a 1972 a relativistic rocket discussion problem _ am .
_ * 40 * 1170 - 1171 bell j s 1987 _ how to teach special relativity _ in _ speakable and unspeakable in quantum mechanics _ ( cambridge university press ) 67 - 80 tartaglia a and ruggiero m l 2003 lorentz contraction and accelerated systems _ eur .
_ * 24 * 215 - 220
|
relying on the equivalence principle , a first approach of the general theory of relativity is presented using the spacetime metric of an observer with a constant proper acceleration . within this non inertial frame , the equation of motion of a freely moving object
is studied and the equation of motion of a second accelerated observer with the same proper acceleration is examined . a comparison of the metric of the accelerated observer with the metric due to a gravitational field is also performed .
|
the bakry - emery method allows one to establish convex sobolev inequalities and to compute exponential decay rates towards equilibrium for solutions to diffusion equations @xcite .
the key idea of bakry and emery is to differentiate a so - called entropy functional twice with respect to time and to relate the second - order derivative to the entropy production .
our aim is to develop a discrete version of this technique , and in this paper , we present a step forward in this direction .
the study of discrete bakry - emery methods and related topics is rather recent .
caputo et al .
@xcite computed exponential decay rates for time - continuous markov processes , using the bochner - bakry - emery method . given a stochastic process with density @xmath0 and the entropy functional @xmath1 , the core of the bakry - emery approach is to find a constant @xmath2 such that the inequality @xmath3 holds for all time . integrating this inequality ,
one may show that @xmath4 which implies that @xmath5 for all @xmath6 , i.e. , the entropy decays exponentially fast along @xmath0 .
the relation between @xmath7 and @xmath8 is achieved in @xcite by employing a discrete bochner - type identity which replaces the bochner identity of the continuous case .
the bochner - bakry - emery method was extended by fathi and maas in @xcite in the context of ricci curvature bounds and used by the authors of @xcite to derive discrete beckner inequalities .
another approach has been suggested by mielke @xcite .
he investigated geodesic convexity properties of nonlocal transportation distances on probability spaces such that continuous - time markov chains can be formulated as gradient flows .
very related results have been obtained independently by chow et al .
@xcite and maas @xcite .
the geodesic convexity property implies exponential decay rates @xcite .
mielke showed that the inequality @xmath3 is equivalent to the positive semi - definiteness of a certain matrix such that matrix algebra can be applied .
this idea was extended recently to certain nonlinear fokker - planck equations @xcite .
all these examples involve spatial semi - discretizations of diffusion equations .
temporal semi - discretizations often employ the implicit euler scheme since it gives entropy dissipation , @xmath9 , under rather general conditions ; see , e.g. , the implicit euler finite - volume approximations in @xcite .
entropy - dissipating higher - order semi - discretizations have been analyzed in @xcite .
however , there seem to be no results for fully discrete schemes using the bakry - emery approach . in this paper
, we make a first step to fill this gap . in order to understand the mathematical difficulty in fully discrete schemes ,
consider the abstract cauchy problem @xmath10 where @xmath11 is a ( nonlinear ) operator defined on its domain @xmath12 of the banach space @xmath13 with dual @xmath14 .
if the dual product @xmath15 is nonnegative , where @xmath16 is the ( frchet ) derivative of the entropy and @xmath0 a solution to , then @xmath17 showing entropy dissipation .
next , consider the implicit euler scheme @xmath18 where @xmath19 is an approximation of @xmath20 and @xmath21 is an approximation of @xmath22 still satisfying @xmath23 .
here , @xmath24 is the discrete entropy , which is supposed to be convex .
then entropy dissipation is preserved by the scheme since @xmath25 the problem is to estimate the discrete analog of @xmath7 .
it turns out that the inequality in is too weak , we need an equation ; see section [ sec.cbe ] for details .
we overcome this difficulty by developing two ideas .
the _ first idea _ is to identify the elements which are necessary to build an abstract discrete bakry - emery method . unlike in the continuous case , we distinguish between the discrete entropy production @xmath26 and the fisher information @xmath27 .
we explain this difference in section [ sec.abstract ] .
the bakry - emery method relies on an estimate of @xmath28 , which approximates @xmath7 .
for this estimate , discrete versions of suitable integrations by parts and chain rules are necessary .
our _ second idea _ is to `` translate '' a nonlinear integration - by - parts formula to the discrete case , using the systematic integration by parts method of @xcite .
this leads to a new inequality for numerical three - point schemes as explained next .
again , consider first the continuous case .
we show in lemma [ lem.ineq ] that for all @xmath29 and all smooth positive functions @xmath30 , @xmath31 where the constant @xmath32 depends on @xmath22 and @xmath33 ; see below . the proof is based on systematic integration by parts @xcite .
the discrete counterpart is the following inequality : for any @xmath34 , there exists a region @xmath35 of admissible values @xmath36 , containing the line @xmath37 , such that for all @xmath38 with @xmath39 , @xmath40 , @xmath41 where @xmath42 ; see lemma [ lem.ineqd ] .
interestingly , the inequality is not true for each term but only with the sum .
the admissible set @xmath35 for is generally smaller than @xmath43 ; see section [ sec.ibp ] .
we conjecture that @xmath44 for @xmath45 .
inequality is the first nonlinear summation - by - parts formula derived from a systematic method .
we believe that this idea will lead to a whole family of new finite - difference inequalities useful in numerical analysis , and we will explore this in a future work .
we apply the abstract discrete bakry - emery method in section [ sec.pme ] to an implicit euler finite - difference approximation of the porous - medium equation @xmath46 where @xmath47 and @xmath48 is the one - dimensional torus .
we assume , for simplicity , that @xmath49 and identify @xmath48 with @xmath50 $ ] .
the entropy functional is @xmath51 , where @xmath52 and @xmath53 is the constant steady state .
we show in proposition [ prop.ent ] that @xmath1 decays exponentially fast to zero for all @xmath54 , where @xmath55 with a decay rate depending on @xmath56 and @xmath57 .
to overcome the difficulty with the entropy production inequality , we introduce the new variable @xmath58 and write the porous - medium equation in the form @xmath59 the advantage of this formulation is that the entropy becomes _ linear _ in the variable @xmath60 . thus avoiding inequality .
we discretize by an implicit euler finite - difference scheme .
let @xmath61 be the time step , @xmath62 the space step , and let @xmath63 be an approximation of @xmath64 , @xmath65 .
the iterative scheme reads as @xmath66 where @xmath65 , @xmath67 , and @xmath68 , @xmath69 .
we show in lemma [ lem.ex ] the existence of solutions to as well as the preservation of nonnegativity .
however , the total mass @xmath70 is not conserved , which is the price that we have to pay for the estimation of the entropy production .
we discuss this point in section [ sec.num ] .
our main result reads as follows .
[ thm.main ] let @xmath71 be a nonnegative solution to and set @xmath72 .
let @xmath73 . then there exist a region @xmath74 , containing the line @xmath75 , and a number @xmath76 such that all @xmath77 with @xmath78 and @xmath79
, it holds that @xmath80 where @xmath81 is the discrete ( relative ) entropy , @xmath82 and @xmath83 the the discrete poincar constant .
moreover , the total mass @xmath84 is increasing in @xmath85 and converges to @xmath86 as @xmath87 .
the exponential decay rate depends on the minimum of the solution , which is not surprising . indeed , because of the degeneracy
, we can not generally expect exponential decay ; an example is the barenblatt solution . algebraic decay rates for implicit euler finite - volume schemes have been derived in , e.g. , @xcite . when the minimum is positive ,
the equation is no longer degenerate , and exponential decay follows .
unfortunately , the theorem does not apply to the shannon entropy @xmath88 , corresponding to @xmath89 , since @xmath90 as @xmath89 .
the reason is that for @xmath89 , the entropy production @xmath91 can not be bounded from above by the fisher information @xmath92 and so , assumption a1 of our abstract bakry - emery method does not hold ; see section [ sec.dbe ] .
erbar and maas @xcite showed that the gradient flow of the shannon entropy with respect to a nonlocal transportation measure equals the discrete porous - medium equation in one space dimension .
the porous - medium equation in several space dimensions was solved by benamou et al .
@xcite by providing a spatial discretization of this equation as a convex optimization problem . in both references ,
no decay rates have been derived .
the set @xmath93 is illustrated in figure [ fig.alphabeta ] for two different values of @xmath94 .
numerical computations indicate that @xmath93 converges ( in the set theoretical sense ) to the set @xmath95 defined in if @xmath96 but for fixed @xmath97 , @xmath93 is strictly contained in @xmath95 . for @xmath98 ( left ) and @xmath99 ( right ) .
the set @xmath95 , defined by @xmath100 , is shown in light blue for comparison ; it contains the dark blue region @xmath93.,title="fig:",width=302 ] for @xmath98 ( left ) and @xmath99 ( right ) .
the set @xmath95 , defined by @xmath100 , is shown in light blue for comparison ; it contains the dark blue region @xmath93.,title="fig:",width=302 ] the paper is organized as follows .
the abstract bakry - emery result is presented in section [ sec.abstract ] , and in section [ sec.ibp ] , inequality is verified .
theorem [ thm.main ] is proved in section [ sec.pme ] .
numerical examples are presented in section [ sec.num ] , and some auxiliary inequalities are recalled in the appendix .
in this section , we present our abstract result . in order to identify the key ingredients of the bakry - emery method , we recall the basic ideas for continuous evolution equations .
let us first consider the abstract cauchy problem @xmath101 the nonlinear operator @xmath22 is defined on some domain @xmath102 of a banach space @xmath13 .
we do not specify the properties of @xmath22 nor its domain since they are not needed in the following . as mentioned in the introduction ,
the idea of the bakry - emery method is to differentiate the entropy functional @xmath103 twice with respect to time along solutions to .
we define the entropy production @xmath104 .
if @xmath105 holds for all @xmath106 ( @xmath107 is the dual product in @xmath13 ) then @xmath108 i.e. , the entropy production is nonnegative and the entropy is nonincreasing along solutions to .
we call @xmath109 the generalized fisher information since if @xmath110 on @xmath111 and @xmath112 , we obtain the fisher information functional @xmath113 . clearly , @xmath114 along solutions @xmath0 to .
differentiating @xmath115 gives @xmath116,h_c'(u)\rangle + \langle a(u),h_c''(u)\pa_t u\rangle \\ & = -\langle a'(u)[a(u)],h_c'(u)\rangle - \langle a(u),h_c''(u)a(u)\rangle,\end{aligned}\ ] ] where @xmath117 is the frchet derivative of @xmath22 at @xmath118 .
if the functional inequality @xmath119,h_c'(u)\rangle + \langle a(u),h_c''(u)a(u)\rangle \ge \lambda_c\langle a(u),h_c'(u)\rangle\ ] ] holds for some @xmath120 then @xmath121 and we conclude exponential decay of @xmath122 with rate @xmath120 . in particular , @xmath123 .
then , integrating the previous inequality over @xmath124 , it follows that @xmath125 assuming that also @xmath126 we conclude that @xmath127 and by gronwall s lemma , @xmath128 converges exponentially fast to zero with rate @xmath129 .
we see that two assumptions are essential : the functional inequality and the limit . on the discrete level
, we need to distinguish between the ( discrete ) entropy production and the ( discrete ) fisher information since @xmath8 and @xmath15 may differ on the discrete level .
we assume that both functionals can be estimated by each other . instead of the functional inequality
we assume a discrete version of inequality .
finally , a discrete version of is required .
let two functions @xmath130 and @xmath131 be given and define @xmath132 , where @xmath60 , @xmath133 and @xmath61 .
we call @xmath134 an entropy , @xmath92 the fisher information , and @xmath91 the entropy production
. the following result does not need any reference to the solution of a discrete problem .
[ prop.be ] let @xmath135 be any sequence .
we assume that there exist @xmath136 , @xmath137 such that @xmath138 for all @xmath67 .
there exists @xmath139 such that @xmath140 for all @xmath67 .
@xmath141 .
then @xmath142 where @xmath143 and @xmath144 .
the discrete decay rate @xmath145 is generally smaller than the decay rate @xmath146 of the fisher information , since @xmath147 and we may have @xmath148 . if the entropy production and the fisher information coincide , i.e. @xmath149 , then @xmath150 . by assumption a2
, it follows that @xmath151 . by assumption a2 again and the second inequality in assumption a1 , we have @xmath152 taking the sum from @xmath153 to @xmath154 , we find that @xmath155 passing to the limit @xmath156 , observing that @xmath157 and , by assumption a3 , @xmath158 , we deduce that @xmath159 which holds for all @xmath160 .
it remains to use the first inequality in assumption a1 to conclude that @xmath161 we deduce that @xmath162 , finishing the proof .
to apply the abstract bakry - emery method to the porous - medium equation , we need to verify the assumptions of proposition [ prop.be ] .
the key condition is assumption a2 . to verify it ,
we `` translate '' some integrations by parts to the discrete level .
it is convenient to investigate the continuous situation first in order to formulate the discrete formula that is needed to show assumption a2 .
consider the nonlinear diffusion equation @xmath163 where @xmath164 , and introduce the ( relative ) entropy @xmath165 here , @xmath166 is the constant steady state .
( recall that @xmath49 . )
[ prop.ent ] let @xmath167 , @xmath168 , and @xmath100 .
then , for any positive smooth solution to , @xmath169 where @xmath170 integrating by parts , the time derivatives of @xmath1 become @xmath171 we wish to estimate the second time derivative . to this end , we set @xmath172 , @xmath173 , and @xmath174
. then the derivatives can be written as @xmath175 in lemma [ lem.ineq ] below we show that there exists @xmath32 such that @xmath176 if the assumption @xmath177 holds .
( note that @xmath167 is equivalent to @xmath178 . )
this condition is actually satisfied since @xmath179 and we infer that @xmath180 furthermore , by the poincar inequality applied to @xmath181 ( see lemma [ lem.poin ] ) , @xmath182 and it follows that @xmath183 denoting by @xmath184 the entropy production , this inequality can be formulated as @xmath185 .
gronwall s lemma then implies that @xmath186 for @xmath6 and in particular @xmath187 . integrating over @xmath188 with @xmath189 and passing to the limit @xmath190
, we see that @xmath191 thus , there exists a sequence @xmath192 such that @xmath193 . following the arguments of (
1ii),juengel / publications / pdf / errata05carri.pdf . ]
it follows that @xmath194 , and since @xmath128 is nonincreasing , any sequence converges , @xmath195 .
we integrate over @xmath124 and use @xmath196 : @xmath197 thus , another application of gronwall s lemma gives the conclusion .
it remains to prove lemma [ lem.ineq ] .
set @xmath198 [ lem.ineq ] let @xmath199 .
then for all smooth positive functions @xmath30 , @xmath200 where @xmath201 the idea of the proof is to employ systematic integration by parts @xcite . since @xmath202 we can formulate as the following problem : find @xmath203 and @xmath32 such that for all smooth positive functions @xmath30 , @xmath204 calculating the derivatives and setting @xmath205 , @xmath206 , this inequality is equivalent to @xmath207 the idea is to interpret the integrand as a polynomial in the variables @xmath208 , @xmath209 and to solve the following polynomial decision problem : find @xmath203 and @xmath32 such that for all @xmath210 , @xmath211 this problem can be solved explicitly .
clearly , it must hold that @xmath212 .
we distinguish two cases : @xmath213 and @xmath214 .
first let @xmath215 . then is valid if the discriminant is nonpositive , @xmath216 choosing the minimizing value @xmath217 the discriminant is nonpositive if and only if @xmath218 set @xmath219 for @xmath73 .
then the previous inequality is true if and only if @xmath220 we infer that if @xmath221 then holds .
this implies that @xmath222 and we need to choose @xmath223 and @xmath177 .
next , let @xmath213 .
then the quadratic term in @xmath209 in vanishes and the mixed term must vanish too , i.e. @xmath224 . hence , the coefficient of the remaining term has to be nonnegative , i.e. @xmath225 .
if @xmath37 , inequality becomes trivial .
the set of all @xmath36 such that @xmath223 and @xmath226 is contained in the set of all @xmath36 satisfying @xmath223 and @xmath177 .
this finishes the proof .
we state now a discrete version of inequality .
[ lem.ineqd ] let @xmath227 satisfy @xmath39 , @xmath40 and let @xmath34 .
there exists a region @xmath228 , containing the line @xmath37 , such that for all @xmath229 , @xmath230 where @xmath231 .
the lemma is trivial as stated since clearly holds for @xmath232 with @xmath233 .
figure [ fig.admregion ] ilustrates the numerical admissible regions for @xmath36 for two different values of @xmath94 .
the admissible region @xmath35 is smaller than the region @xmath43 for the continuous case but it approaches the latter region as @xmath234 .
such that @xmath235 for all @xmath13 , @xmath236 using @xmath237 as in with @xmath238 and @xmath239 ( left ) , @xmath240 ( right ) . the set @xmath35 is depicted in dark blue , @xmath241 in light blue.,title="fig:",width=302 ] such that @xmath235 for all @xmath13 , @xmath236 using @xmath237 as in with @xmath238 and @xmath239 ( left ) , @xmath240 ( right ) .
the set @xmath35 is depicted in dark blue , @xmath241 in light blue.,title="fig:",width=302 ] the idea of the proof of is to add the following discrete version of the integration - by - parts formula , namely @xmath242 the sum vanishes because of the periodic boundary conditions . here
@xmath243 is a free parameter , and the function @xmath244 is a symmetric mean value , i.e. , it satisfies @xmath245 for all @xmath246 , @xmath247 , @xmath248 . for the numerical simulations below , we will choose @xmath249 such that the mean function does not need to be specified . then holds
if we can show the following inequality for all admissible @xmath36 and @xmath250 : @xmath251 we verify this inequality pointwise , i.e. setting @xmath252 and @xmath253 , we wish to find @xmath203 , @xmath139 such that for all @xmath13 , @xmath254 , @xmath255 the first term @xmath256 becomes negative in certain regions ; see figure [ fig.levelset ] .
it is compensated by the second term ( shift term ) on the right - hand side of if we choose the constant @xmath237 according to with @xmath257 as in . and
@xmath258 for @xmath259 , @xmath260 ( left ) and @xmath261 , @xmath262 ( right ) .
we have chosen @xmath263 and @xmath237 as in .,title="fig:",width=283 ] and @xmath258 for @xmath259 , @xmath260 ( left ) and @xmath261 , @xmath262 ( right ) .
we have chosen @xmath263 and @xmath237 as in .,title="fig:",width=283 ] unfortunately , it seems to be difficult to prove analytically in full generality .
note that polynomial quantifier elimination does not apply if @xmath22 and @xmath33 are not integers , since the function @xmath264 generally is _ not _ a polynomial . instead , we verify analytically for all @xmath199 and all @xmath265 in some neighborhood of @xmath266 .
[ lem.t ] let @xmath267 be given by and let @xmath199 , where @xmath43 is defined in .
then there exists a neighborhood @xmath268 of @xmath266 such that for all @xmath269 , @xmath270 holds for @xmath237 as in and with @xmath238 as in .
if the step size @xmath62 is small enough , we expect that the quotients @xmath271 are close to one for all @xmath272 .
this means that @xmath265 lies in a neighborhood of @xmath266 , and the lemma applies .
we use the local coordinates @xmath273 and @xmath274 , which correspond to ( central ) second - order and first - order derivatives . then @xmath275 and @xmath276 .
we develop @xmath267 as a function of @xmath277 at @xmath278 . for this
, we observe that @xmath279 and @xmath280 . indeed , we infer from the properties that @xmath281 which implies that @xmath282 . calculating the taylor series of the terms in @xmath267 with respect to @xmath277 at @xmath278 leads to @xmath283 in particular , as expected ,
the explicit choices of both @xmath284 and @xmath244 do not change the behavior of the shift term locally around the equilibrium @xmath285 or @xmath278 .
moreover , @xmath286 and @xmath287 . combining these expressions gives @xmath288 the polynomial @xmath289 is the same as in .
the proof of lemma [ lem.ineq ] shows that it is nonnegative for all @xmath199 with @xmath237 as in and @xmath290 as in .
we deduce that @xmath235 holds for all @xmath199 if @xmath291 is sufficiently small .
this proves the lemma .
we apply the abstract bakry - emery method to a finite - difference approximation of the porous - medium equation , i.e. , we choose @xmath292 on @xmath48 for suitable functions @xmath118 .
let @xmath61 be the time step and @xmath62 the space step
. a natural scheme would be @xmath293 for all @xmath65 , @xmath67 , and @xmath294 , @xmath295 .
the corresponding discrete entropy is @xmath296 and @xmath297 is the constant steady state .
we choose @xmath78 and @xmath47 .
unfortunately , the abstract bakry - emery method can not be applied to this scheme .
the problem is the second inequality in assumption a1 . indeed , using the inequality @xmath298 for all @xmath247 , @xmath299 , which follows from the convexity of @xmath300 for @xmath78 , inserting the numerical scheme and then summing by parts , we find that @xmath301 this expression can not be estimated further ; it may even have the wrong sign .
we need a scheme that avoids the use of the inequality @xmath298 .
we stress the fact that this problem does not occur in the semi - discrete scheme @xmath302 since then @xmath303 and this expression is nonpositive ( since @xmath78 ) .
our idea is to make the entropy production _
linear _ in its argument . for this
, we introduce the new variable @xmath63 . in the ( continuous ) variable @xmath58
, the evolution equation transforms to @xmath304 , which inspires the numerical scheme @xmath305 for all @xmath65 , @xmath67 , and @xmath68 , @xmath69 .
the discrete entropy and fisher information become @xmath306 where @xmath307 has to be determined and @xmath308 .
the entropy production can be estimated , using summation by parts , as @xmath309 according to lemma [ lem.xy ] , the entropy production can be estimated from below _ and _ above in terms of the fisher information .
after this motivation , we prove the existence of solutions to .
[ lem.ex ] for given @xmath310 , @xmath65 , there exists a solution @xmath311 , @xmath65 , to .
we give only a sketch of the proof since the existence of solutions follows from a standard fixed - point theorem .
we only provide the a priori estimates needed for this argument .
first multiply by @xmath312 and sum over @xmath65 . since @xmath313
, we obtain @xmath314 by summation by parts , this becomes @xmath315 since @xmath316 is nondecreasing .
we infer that @xmath317 and hence @xmath311 .
next , by , @xmath318 and this is the desired a priori estimate .
next , we turn to the proof of our main result .
we verify the assumptions of proposition [ prop.be ] . for this
, we continue our estimates for @xmath91 . applying lemma [ lem.xy ] with @xmath319 and @xmath320 to
, we obtain the inequalities @xmath321 thus , assumption a1 is satisfied with @xmath322 and @xmath323 .
next , we estimate the difference @xmath324 . to this end
, we set @xmath325 , @xmath326 , @xmath327 and write @xmath328 the idea of the proof is to expand @xmath329 around zero : @xmath330 for some @xmath331 .
we show that the right - hand side can be bounded from above by @xmath332 for some @xmath333 , which verifies assumption a2 .
this idea has been first employed in @xcite . clearly , we have @xmath334 .
the first derivatives of @xmath335 equal @xmath336 first , we claim that @xmath337 for any @xmath61 . indeed
, we replace @xmath338 by @xmath339 and obtain @xmath340 it holds that @xmath341 , since this inequality is equivalent to @xmath342 , and this is true for @xmath343 ( which is equivalent to @xmath79 and @xmath344 ) .
moreover , the discriminant @xmath345 is equivalent to @xmath346 which also holds true for @xmath343 .
this shows that @xmath337 and consequently , @xmath347 it remains to compute @xmath348 . inserting the definition of @xmath349 , we find that @xmath350 since @xmath351 .
we apply lemma [ lem.ineqd ] with @xmath352 , @xmath173 and @xmath174 and infer that @xmath353 by the discrete poincar - wirtinger inequality ( lemma [ lem.pwi ] ) , applied with @xmath354 , it follows that @xmath355 and hence , with @xmath356 , @xmath357 this shows assumption a2 of proposition [ prop.be ] and , in particular , after applying gronwall s lemma , @xmath151 .
it remains to prove that assumption a3 , i.e. @xmath141 , holds .
we know that @xmath358 so , for any fixed @xmath65 , @xmath359 is bounded . therefore , there exists a sequence @xmath360 such that @xmath361 for some @xmath362 . by the discrete poincar - wirtinger inequality ( lemma [ lem.pwi ] ) , applied to @xmath363 , where @xmath364 , it follows that @xmath365 since @xmath151 , we deduce that @xmath366 and @xmath367 have the same limit , say @xmath368 .
set @xmath369 .
this defines the entropy @xmath370 for @xmath371 .
it holds that @xmath372 as @xmath360 .
but @xmath373 is nonincreasing , from which we deduce that @xmath374 for any sequence @xmath87 .
according to proposition [ prop.be ] , the discrete entropy converges exponentially with decay rate @xmath375 next , we claim that the total mass @xmath84 is nondecreasing in @xmath85 .
indeed , by the concavity of @xmath376 ( recall that @xmath377 ) , we have @xmath378 for all @xmath247 , @xmath299 and hence , @xmath379 inserting scheme , we find that @xmath380 since @xmath381 satisfies periodic boundary conditions .
this shows the claim .
the monotonicity of the total mass and the convergence property @xmath382 as @xmath360 imply that @xmath383 for @xmath87 , and the convergence is monotone .
this finishes the proof .
we present some numerical results for the porous - medium equation discretized in the previous section . as initial datum
we choose the barenblatt profile @xmath384 where @xmath385 .
we consider two cases .
for the slow diffusion case @xmath386 , we choose @xmath387 , @xmath388 , and @xmath389 the profile will hit the boundary of @xmath390 at time @xmath391 .
for the fast diffusion case @xmath392 , we take @xmath387 , @xmath393 , and @xmath394 such that the maximum of the initial profile equals 1 .
figure [ fig.massincrease ] illustrates the evolution of the total mass for @xmath395 , @xmath392 ( left ) and @xmath396 , @xmath386 ( right ) .
as predicted in theorem [ thm.main ] , the total mass is indeed increasing in time .
the mass defect scales well with both the time step @xmath397 and the grid size @xmath277 , where the influence of @xmath397 is more prevalent . ,
@xmath392 , right : @xmath396 , @xmath386).,title="fig:",width=302 ] , @xmath392 , right : @xmath396 , @xmath386).,title="fig:",width=302 ] the time decay of the ( relative ) entropy @xmath398 is shown in figure [ fig.relentropy.admissible ] for various space and time steps .
we observe that the decay is indeed exponential . here , the steady state @xmath399 ( which is needed to define the relative entropy ) is given by @xmath400 , where @xmath401 is the final time step .
this choice clearly depends on the scheme since the mass is not conserved .
the relative entropy converges exponentially even when @xmath56 is chosen outside of the admissible region ; see figure [ fig.relentropy.not.admissible ] .
, @xmath392 , right : @xmath396 , @xmath386).,title="fig:",width=302 ] , @xmath392 , right : @xmath396 , @xmath386).,title="fig:",width=302 ] outside of the admissible region.,width=340 ]
[ lem.poin ] let @xmath402 .
it holds for all @xmath403 satisfying @xmath404 that @xmath405 and the constant @xmath406 is sharp .
the second inequality is proven in ( * ? ? ?
* lemma a.3 ) .
for the proof of the first inequality , we divide it by @xmath417 and set @xmath418 .
then the inequality is equivalent to @xmath419 which after expansion can be equivalently written as @xmath420 , and this is true .
j. a. carrillo , j. dolbeault , i. gentil , and a. jngel .
entropy - energy inequalities and improved convergence rates for nonlinear parabolic equations . _ discrete contin
b _ 6 ( 2006 ) , 1027 - 1050 . c. chainais - hillairet , a. jngel , and s. schuchnigg .
entropy - dissipative discretization of nonlinear diffusion equations and discrete beckner inequalities .
_ math . model .
50 ( 2016 ) , 135 - 162 .
|
the exponential decay of the relative entropy associated to a fully discrete porous - medium equation in one space dimension is shown by means of a discrete bakry - emery approach .
the first ingredient of the proof is an abstract discrete bakry - emery method , which states conditions on a sequence under which the exponential decay of the discrete entropy follows .
the second ingredient is a new nonlinear summation - by - parts formula which is inspired by systematic integration by parts developed by matthes and the first author .
numerical simulations illustrate the exponential decay of the entropy for various time and space step sizes .
|
the high energy collisions are usually connected with production of large number of secondaries ( mostly @xmath0 and @xmath1 mesons ) . the strong interactions involved here
make their detail descriprion _ from first principles _ impossible and one is forced to turn to phenomenological models of various kinds .
the statistical models were the first successful approaches to the multiparticle production processes since the beginning of the subject almost half century ago @xcite and they remain still very much alive today , especially in all analysis of multiparticle data performed from the point of view of the possible formation of the new state of matter - the quark gluon plasma ( qgp ) @xcite .
they are all based on the boltzmann - gibbs form of entropy , which is identical to the so called shannon entropy used in the information theory approach .
+ we would like to stress at this point that information theory can be also applied to hadronic processes @xcite .
its importance there is best visualized by the following example .
suppose that some experiment provides us with data on a limited set of @xmath2 observables : @xmath3 .
this triggers the theory , which follows with a variety of models , usually completely different ( if not contradictory ) but each claiming good agreement with these data .
the models provide therefore apparently different views of the same physical reality ( data ) leading to confusion on what is actually going on .
it happens this way because experimental data contain only _ limited amount of information _ to which theoretical models add their own specific assumptions born not in data but in our minds .
to quantify this situation one must find the way to answer ( in a model independent way ) the following question : given physical assumptions of a model ( plus phase space and conservation laws ) , what are their _ most trivial _ consequences ?
this can be done only be by resorting to information theory which defines _ triviality _ as the _ lack of information _ ( the less information the more trivial ) .
the measure of information provided by shannon entropy ( or some other kind of entropy , in our talk this will be tsallis entropy @xcite ) allows us then to quantify the whole problem .
let us elaborate this point a bit further .
experiment gives us usually set of points to which one tries to fit some distribution function @xmath4 , which can be thought to be some probability distribution . in this case shannon entropy corresponding to it is @xmath5 .
\label{eq : shannon}\ ] ] we are looking for the most trivial @xmath4 , i.e. , for one containing the least information , and it means for such , which informational entropy is maximal under given circumstances ( constraints ) ) assumes that @xmath4 is measured against some other distribution @xmath6 corresponding to equal _ a priori _ probability being assigned to equal volumes of phase space .
this is sometimes called `` the first principle of statistical mechanics '' @xcite . the boltzmann constant , which normally appears in ( [ eq : shannon ] ) has been henceforth assumed unity , @xmath7 . ] .
what are those constraints ?
at first , being a probability distribution @xmath4 must be non - negative and normalized , i.e. , @xmath8 further , @xmath4 has to reproduce the experimental results given in the form of expectation values @xmath9 of measured set of observables
@xmath10 , i.e. , @xmath11 using eqs.([eq : norm ] ) and ( [ eq : r ] ) as constraints one then maximalizes information entropy ( [ eq : shannon ] ) and as results one gets the seek for distribution @xmath4 , @xmath12 , \label{eq : findis}\ ] ] where @xmath13 is calculated from the normalization condition .
this distribution reproduces known information ( i.e. , it tells us _ the truth , the whole truth _ about experiment ) and contains the least information ( i.e. , it tells us _ nothing but the truth _ ) . in this sense
it is the unique , most plausible ( least biased ) , model independent probability distribution describing the outcomes of our experiment .
the lagrange multipliers @xmath14 are given as solutions of the constraint equations and @xmath15 .
+ now we can understand why apparently completely disparate models can describe sucessfully the same experiment .
it is because they share ( usually in an implicite than explicite way ) the same set of basic common assumptions ( i.e. , constraints @xmath16 ) and , at the same time , they differ completely in the rest of them , i.e. , in those @xmath16 , which contain assumptions particular for a given model ( and which are completely not relevant to its agreement with experiment ) cut - off ) and that not the whole initial energy @xmath17 of reaction is used for the particle production ( i.e. , the existence of the inelasticity @xmath1 or of the leading particle effect ) . ] .
+ this example tells us that statistical approach is not necessarily connected with any thermodynamical model but has , in fact , much broader range of applications . in our example above the formula ( [ eq : findis ] ) is identical with formulas used in thermal models @xcite but parameters @xmath14 can take any values depending on the energy and multiplicity of the event under considerations @xcite .
it is worth to remind at this point the very old observation @xcite that exponential form ( [ eq : findis ] ) of so many distributions observed experimentally can have its origin in the fact that the actual measurements extend only to subsystems of a total system .
the summation over the unmeasured ( i.e. , essentially overaged over ) degrees of freedom introduces the randomnization , which can be most economically described as a kind of thermal bath leading to thermal - like exponential distributions with some effective `` temperature ' ' @xmath18 .
+ in this approach dynamics lies in our ability to reproduce the known information ( i.e. , is hidden in @xmath14 ) .
however , the big advantage of this method is that it prevents us from smuggling in anything additional , not being present in data .
the outcome depends on the funtional form of the entropy .
the shannon form ( [ eq : shannon ] ) is the most popular one because it coincides in thermodynamical limit with the boltzmann - gibbs ( bg ) form allowing for contact with standard thermodynamics .
however , as newtonian mechanics is not eternal and universal ( in the sense that in the limit of large velocities it must be replaced by special relativity , in the limit of vanishingly small masses quantum mechanical effects must be accounted for and in the limit of extremaly large masses ( stars , galaxies , ... ) it must be replaced by general relativity ) the same can be said about the classical bg statistical mechanics .
it is fully satisfactory only if * effective microscopic interactions are short - ranged ( i.e. , we have close spacial connections ) ; * effective microscopic memory is short - ranged ( i.e , we have close time connections ) ; * boundary conditions are non ( multi)fractal ( i.e. , the relevant space - time and/or phase space is non ( multi)fractal ) . as a matter of fact despite the popularity of bg it seems that in majority of situations we are dealing with , at least one of these conditions is not satisfied .
therefore it is justified to look for the possible generalization of bg which would provide a compactifications of all possible deviations from the ideal bg statistics into a minimal - possible only one - additional parameters . in this way
we arrive at the notion of tsallis statistics based on tsallis entropy @xcite : @xmath19 we shall not dwell too much on it here as this is the subject of separate lecture by prof .
tsallis himself @xcite .
the only points we would like to stress here for our purposes are : * for @xmath20 one recovers the usual shannon entropy form of ( [ eq : shannon ] ) ; * with the normalization condition imposed on @xmath4 as before and with the new definition of the mean values , @xmath21 , one replaces formula ( [ eq : findis ] ) with @xmath22^{\frac{1}{1-q } } ; \label{eq : findisq}\ ] ] ( notice that , again , for @xmath23 ( [ eq : findisq ] ) goes to ( [ eq : findis ] ) . as before
, this distribution reproduces known information and conveys least information but this time information is measured in a different way , which is paramatrized by the parameter @xmath24)@xcite .
however , it is known @xcite that such an approach is different from the present nonextensive formalism since the tsallis entropy expressed in terms of the escort probability distribution has a difficulty with the property of concavity . ] ; * the resultant staistics is non - extensive in the sense that for two subsets @xmath25 and @xmath26 such that probabilities @xmath27 the entropy is not additive but instead @xmath28 and additivity is recovered only in the limit @xmath23 ; * the whole legendre structure of thermodynamics is preserved for @xmath29 @xcite .
although the nonextensive statistics were used for a long time in many branches of physics @xcite , their possible imprints in multiparticle production are very fresh . in this talk
we shall concentrate on some cosmic ray experiments and accelerator experiments , which provided us with examples of a nonextensive behaviour .
both will demonstrate traces of nonextensivity seen as nonexponentiality of some distributions , i.e. , where @xmath30^{1/(1-q)}$ ] . at the end we shall
, however , list all other works in which one can find similar conclusions @xcite . +
part of the cosmic ray detector system consists of extrathick lead emulsion chambers .
such chamber is build with a number of sandwiches of x - ray sensitive film , plate of lead and layer of emulsion ( first allows to localize the point at which external particle of cosmic rays enters , second forces it to interact and to start the electromagnetic cascade process , and third allows to calculate deposited energy ) . in some experiments
one observes that distribution of the points of first interaction ( cascade starting points ) does not follow the expected simple exponential rule is the number of cascades originating at depth @xmath18 expressed in the so called cascade units where @xmath31 c.u .
= @xmath32 g/@xmath33 = @xmath34 cm of pb and @xmath35 is the mean free path for the p - pb interacions.]@xcite : @xmath36 but deviates from it noticeably for larger @xmath18 .
this phenomenon acquired even its own name : _ long flying component _
( because original hadrons tend to fly longer without interaction than naively expected ) .
+ ( 25.,16.5 ) * depth distribution of the starting points , @xmath37 , of cascades in pamir lead chamber .
notice the non - exponential behaviour of data points ( for their origin cf .
@xcite ) which can be fitted by tsallis distribution ( [ eq : lfcq ] ) with @xmath38 .
( this figure is reproduced from fig . 1 of @xcite ) .
in @xcite we have argued that the observed effect can be just another manifestation of the fluctuation of the corresponding hadronic cross section @xmath39 ( where @xmath25 denotes the mass number of the target and @xmath40 is the mass of the nucleon , such a possibility is widely discussed in the literature and observed in diffraction dissociation experiments on accelerators , cf .
@xcite for details and references ) .
as was shown in @xcite such fluctuations of cross section ( i.e. , in effect @xmath41 ) with relative variance @xmath42 allow to describe the observed effect .
however , it turns out @xcite that the same data can be fitted by the nonextensive formula @xmath43^{\frac{1}{1-q } } \label{eq : lfcq}\ ] ] with parameter @xmath38 ( in both cases @xmath44 in @xmath45 defined above ) , cf .
+ this example provides us hint on the possible physical meaning of the nonextensvity parameter @xmath24 . comparing both explanations one can suspect that perhaps , at least in this case , it has something to do with the fluctuations of the seemingly constant parameter @xmath41 in the exponential formula ( [ eq : lfc ] ) .
we shall argue later that this is indeed the case but before let us demonstrate another example of this kind , now from the high energy multiparticle production processes domain .
heavy ion collisions are of particular interest at present because they are the only place were one expects the possible formation of the qgp - a new state of matter @xcite .
the termal and/or all kind of statistical models are playing very important role here because one of the crucial parameter is the temperature @xmath18 of reaction deduced usualy from the transverse momentum distributions , @xmath46 .
this is , however , valid procedure only when @xmath46 is of exponential form .
therefore any deviation from such behaviour are always under detailed scrutiny in which one is searching for the possible causes .
in @xcite it was suggested that the extreme conditions of high density and temperature occuring in ultrarelativistic heavy ion collisions can lead to memory effects and long - range colour interactions and to the presence of non - markovian processes in the corresponding kinetic equations @xcite .
actually , as has been shown in @xcite , @xmath47 are best described , cf . , fig .
2 , by a slightly nonexponential function of the type @xmath48^{\frac{1}{1-q}}\qquad \stackrel{q\rightarrow 1}{\longrightarrow}\qquad { \rm const}\cdot \exp\left ( - \frac{\sqrt{m^2 + p^2_t}}{kt}\right ) .
\label{eq : pt}\ ] ] here @xmath49 is the mass of produced particle , @xmath50 is the boltzmann constant ( which we shall , in what follows , put equal unity ) and @xmath18 is , for the @xmath51 case , the _ temperature _ of the reaction considered ( or , rather , the temperature of the hadronic system produced ) .
the deviation from the exponential form one finds is very small , on the level @xmath52 .
however , as we shall demonstrate later it can lead to a quite dramatic effects .
it was also shown in @xcite that to first order in @xmath53 the generalized slope becomes the quantity @xmath54 with @xmath18 being temperature of _ a purely thermal source_. this should be contrasted with the empirical relation for the slope parameter @xmath18 , from which the freeze - out temperature ( at which hadrons are created from the qgp ) @xmath55 is then deduced , @xmath56 ( 25.,16.5 ) * the results for @xmath57 distribution @xmath47 : notice that @xmath52 results describes also the tail of distribution not fitted by the conventional exponent ( i.e. , @xmath51 ) .
this figure is reproduced from fig .
3 of @xcite .
the @xmath58 is a fit parameter usually identified with the average collective ( transverse ) flow velocity of the hadrons being produced . in ( [ eq : tq ] ) one has , instead , _ a purely thermal source _ experiencing a kind of blue shift at high @xmath59 ( actually increasing with @xmath59 ) .
the nonextensivity parameter @xmath24 accounts here for all possibilities one can find in @xcite and could therefore be regarded as a new way of presenting experimental results with @xmath60 , signaling that there is something going on in the collision that prevents it from being exactly thermal - like in the ordinary sense mentioned above .
+ to the same cathegory belongs also analysis of the equilibrium distribution of heavy quarks in fokker - planck dynamics expected in heavy ion collisions @xcite .
not going into details in this case , we shall only mention that it was found that thermalization of charmed quarks in a qgp surroundings proceeding via collisions with light quarks and gluons results in a spectral shape , which can be described only in @xmath24 statistics with q=1.114 ( for @xmath61 gev and temperature of thermal gluons @xmath62 gev ) .
guided by the cosmic ray example we would like to discuss now the hypothesis that the examples of nonextensivity in which exponential distribution becomes a power - like ( lvy type ) distributions is caused by the fluctuations in the system which were not accounted for and which make @xmath29 @xcite . in other words , @xmath24 summarizes the action of averaging over @xmath41 : @xmath63\right\rangle \longrightarrow \left [ 1 \ , -\ , ( 1 - q)\frac{1}{\lambda_0}\varepsilon \right]^{\frac{1}{1-q } } , \label{eq : average}\ ] ] where @xmath64 .
this connection , which is clear in the cosmic ray example , should also be true for the heavy ion collision .
actually , this later case is even more important and interesting because of the long and still vivid discussion on the possible dynamics of temperature fluctuations @xcite and because of its connection with the problem of qgp production in heavy ion collisions @xcite .
actually the conjecture ( [ eq : average ] ) is known in the literature as hilhorst integral formula @xcite but was never used in the present context .
it says that ( for @xmath65 case , where @xmath66 ) @xmath67 .
\label{eq : gf}\ ] ] where @xmath68 . changing variables under the integral to @xmath69 ,
one obtains @xmath70 where @xmath71 is given by the following gamma distribution : @xmath72 with @xmath73 and with mean value and variation in the form : @xmath74 notice that , with increasing @xmath75 the variance ( [ eq : meanvar ] ) decreases and asymptotically ( for @xmath76 , i.e , for @xmath23 ) the gamma distribution ( [ eq : f ] ) becomes a delta function , @xmath77 . the relative variance for this distribution
is given by @xmath78 for the @xmath79 case @xmath80 is limited to @xmath81 $ ] . proceeding in the same way as before , i.e. , making use of the following representation of the euler gamma function ( where @xmath82 ) @xmath83^{\alpha'}\ , = \ , \left(\frac{\alpha ' \lambda_0}{\alpha ' \lambda_0 - \varepsilon}\right)^{-\alpha'}\ , = \
, \frac{1}{\gamma(\alpha')}\ , \int^{\infty}_0\ , d\eta\ , \eta^{\alpha ' - 1}\ , \exp\left [ - \eta\ , \left(1\ , + \ , \frac{\varepsilon}{\alpha ' \lambda_0 - \varepsilon}\right)\right ] , \label{eq : egf}\ ] ] and changing variables under the integral to @xmath84 , we obtain @xmath85 in the form of eq .
( [ eq : def ] ) but with @xmath86 and with the respective @xmath87 given now by the same gamma distribution as in ( [ eq : f ] ) but this time with @xmath88 and @xmath89 .
contrary to the @xmath65 case , this time the fluctuations depend on the value of the variable in question , i.e. , the mean value and variance are now both @xmath80-dependent : @xmath90 however , the relative variance @xmath91 remains @xmath80-independent and depends only on the parameter @xmath24 .
as above the resulting gamma distribution becomes a delta function , @xmath92 , for @xmath93 , i.e. , for @xmath23 .
+ summarizing : one can indeed say that the nonextensivity parameter @xmath24 in the @xmath94 distributions can be expressed by the relative variance @xmath95 of fluctuations of the parameter @xmath41 in the distribution @xmath96 : @xmath97 actually the above result ( [ eq : qg1 ] ) is derived for the particular ( gamma - like ) shape of the fluctuation distribution function . how can it be realized ? to answer this question let us notice @xcite that in the case when stochastic variable @xmath35 is decribed by the usual langevin equation @xmath98\ , \lambda\ , = \ , \phi\ , = \ , { \rm const}\ , > \ , 0 , \label{eq : le}\ ] ] ( with damping constant @xmath99 and source term @xmath100 or @xmath101 ) , then for the stochastic processes defined by the _ white gaussian noise _ form of @xmath102 and correlator ( for sufficiently fast changes ) @xmath103 .
constants @xmath99 and @xmath104 define , respectively , the mean time for changes and their variance by means of the following conditions : @xmath105 and @xmath106 .
thermodynamical equilibrium is assumed here ( i.e. , @xmath107 , in which case the influence of the initial condition @xmath108 vanishes and the mean squared of @xmath35 has value corresponding to the state of equilibrium ) .
] one obtains the following fokker - plank equation @xcite for the distribution function of the variable @xmath35 : @xmath109 with intensity coefficients @xmath110 defined by eq.([eq : le ] ) and equal to ( cf . , @xcite )
: @xmath111 its stationary solution has precisely the gamma - like form we are looking for : @xmath112 \ , = \ ,
\frac{1}{\gamma(\alpha)}\ , \mu\ , \left(\frac{\mu}{\lambda}\right)^{\alpha-1}\ , \exp\left ( -\ , \frac{\mu}{\lambda } \right ) , \label{eq : fres}\ ] ] with the constant @xmath113 defined by the normalization condition , @xmath114 .
it depends on two parameters : @xmath115 with @xmath116 and @xmath117 for , respectively , @xmath65 and @xmath79 .
therefore eq .
( [ eq : qg1 ] ) with @xmath118 is , indeed , a sound possibility with ( in the case discussed above ) parameter of nonextensivity @xmath24 given by the parameter @xmath104 and by the damping constant @xmath99 describing the _
white noise_. +
with the possibility of such interpretation of parameter @xmath24 in mind we can now came back to the example of heavy ion collisions and see what are its consequences in this case .
it is interesting to notice that the relatively small departure of @xmath24 from unity , @xmath119 @xcite , if interpreted in terms of the previous section , indicates that rather large relative fluctuations of temperature , of the order of @xmath120 , exist in nuclear collisions .
it could mean therefore that we are dealing here with some fluctuations existing in small parts of the system in respect to the whole system ( according to interpretation of @xcite ) rather than with fluctuations of the event - by - event type in which , for large multiplicity @xmath121 , fluctuations @xmath122 should be negligibly small @xcite .
this controversy could be , in principle , settled by detailed analyses of the event - by - event type . already at present energies and nuclear targets ( and
the more so at the new accelerators for heavy ions like rhic at brookhaven , now commisioned , and lhc at cern scheduled to be operational in the year 2006 ) one should be able to check whether the power - like @xmath57 distribution @xmath47 occurs already at every event or only after averaging over all events . in the former case we would have a clear signal of thermal fluctuations of the type mentioned above . in the latter case one would have for each event a fixed @xmath18 value which would fluctuate from one event to another ( most probably because different initial conditions are encountered in a given event ) .
+ one point must be clarified , however .
the above conjecture rests on the stochastic equation ( [ eq : le ] ) .
can one expect such equation to govern the @xmath18 fluctuations ? to ansewer this question let us turn once more to the fluctuations of temperature @xcite discussed before , i.e. , to @xmath123 .
suppose that we have a thermodynamic system , in a small ( mentally separated ) part of which the temperature fluctuates with @xmath124 .
let @xmath125 describe stochastic changes of the temperature in time .
if the mean temperature of the system is @xmath126 then , as result of fluctuations in some small selected region , the actual temperature equals @xmath127 .
the inevitable exchange of heat between this selected region and the rest of the system leads to the equilibration of the temperature and this process is described by the following equation @xcite : @xmath128 which is , indeed , of the type of eq .
( [ eq : le ] ) ( here @xmath129 and @xmath130 ) .
this proves the plausability of what was said above and makes the event - by - even measurements of such phenomenon very interesting one .
the above discussed examples do not exhaust the list of the possible imprints of nonextensivity in multiparticle production know ( or thought of ) at present . as it is impossible to review all of them here ,
we shall therefore only mention them .
+ in @xcite the possible effects of nonextensivity on the mean occupation numbers @xmath131 and on the event - by - event fluctuation phenomena have been discussed .
it turns out , for example , that some charateristics of fluctuations are extremaly sensitive to even small departures of @xmath24 from unity .
such departure can easily mimick the existing correlations or the influence of resonances .
+ in @xcite an interesting attempt was presented to fit the energy spectra in both the longitudinal and transverse momenta of particles produced in the @xmath132 annihilation processes at high energies using @xmath24-statistical model . in this way one can have energy independent temperature @xmath18 and nonextensivity parameter @xmath24 rising quickly with energy from @xmath51 to @xmath133 and reflecting long range correlations in the phase space arising in the hadronization process in which quarks and gluons combine together forming hadrons ( actually , this observation has general validity and applies to all production processes discussed here as well ) . in similar spirit
is the work @xcite which attemts to generalized the so called hagedorn model of multiparticle production to @xmath24-statistics . + to the extend to which self - organized criticality ( soc )
is connected with nonextensivity @xcite one should also mention here a very innovative ( from the point of view of high energy collision ) application of the concept of soc to such processes @xcite .
+ the other two examples do not refer to tsallis thermostatistics directly , nevertheless they are connected to it .
first is the attempt to study , by using the formalism of quantum groups , the bose - einstein correlations between identical particles observed in multiparticle reactions @xcite , second are works on intermittency and multiparticle distributions using the so called lvy stable distributions @xcite .
they belong , in some sense , to the domain of nonextensivity because , as was shown in @xcite , there is close correspondence between the deformation parameter of quantum groups used in @xcite and the nonextensivity parameter @xmath24 of tsallis statistics and there is also connection between tsallis statistics and lvy stable distributions @xcite .
some traces of the possible nonextensive evolution of cascade type hadronization processes were also searched for in @xcite . the quantum group approach @xcite
could probably be a useful tool when studying delicate problem of interplay between qgp and hadrons produced from it .
it is plausible that description in terms of @xmath24-deformed bosons ( or the use of some kind of interpolating statistics ) would lead to more general results than the simple use of nonextensive mean occupation numbers @xmath134 discussed above ( for which the only known practical description is limited to small deviations from nonextensivity only ) .
+ one should mention also attempts to use lvy - type distributions to fit the development of the cosmic ray cascades to learn of how many descendat they contain @xcite or the explanation of the feynman scaling violation observed in multiparticle distributions in terms of the @xmath24 parameter @xcite .
to summarize , evidence is growing in favour of the view that the standart statistical model can be enlarged towards the nonstandart statistics and that by including one new parameter @xmath24 it allows to repoduce much broader set of data than it was done so far .
it was demonstrated that @xmath24 is probably connected with the intrinsic fluctuations existing in the system under consideration which were previously not considered at all - statistics only .
a good example is work @xcite showing how fluctuations of the vacuum can change the theoretically expected gaussian spectra in string models into the observed exponential one with a kind of effective `` temperature '' . ] .
let us close with the remarks that , as was shown in @xcite , one can also try to use power - like distributions discussed here to the new formulation of the quantum field theory ( for example , in terms of lorenzian rather than gaussian path integrals , accounting in this manner for some intrinsic long range correlations impossible to dispose of and extremaly difficult to account for by other methods ) . + acknowledgements : one of us ( g.w . ) would like to thank the organizers of the ranp2000 for support making his participation in this workshop possible .
authors are grateful to prof .
s.abe for his valuable comments .
+ l.d .
landau and s.z .
bilenkij , _
nuovo cim .
suppl . _ * 3 * ( 1956 ) 15 and _ collected papers of l.d .
d. ter haar , pergamon press ( 1965 ) , p. 569 .
this model is still in use even today for quick and straightforward estimations of some quantities of interest , cf . , m.gadzicki , _ z. phys .
_ * c66 * ( 1995 ) 659 .
c.tsallis , _ j. stat .
phys . _ * 52 * ( 1988 ) 479 ; cf . also c.tsallis , cond - mat/0010150 ( to appear in _ chaos , solitons and fractals _ ( 2001 ) . for
updated bibliography on this subject see http://tsallis.cat.cbpf.br/biblio.htm .
its recent summary is provided in the special issue of _ braz .
j. phys . _
* 29 * ( no 1 ) ( 1999 ) ( available also at http://sbf.if.usp.br/www_pages/journals/bjp/vol29/num1/index.htm ) .
s.gavin , _ nucl .
* b351 * ( 1991 ) 561 ; j.rau , _ phys . rev . _ * d50 * ( 1994 ) 6911 ; s.schmidt et al . , _ phys . rev . _
* d59 * ( 1999 ) 094005 ; h.heiselberg and x .- n.wang , _ phys . rev . _ * d53 * ( 1996 ) 1892 ; v.k.magas et al . , _ phys
. lett . _ * b459 * ( 1999 ) 33 ; cs.anderlik et al .
* c59 * ( 1999 ) 388 , 3309 ; f.grassi et al .
rev . _ * c62 * ( 2000 ) 044904 ; j.hormuzdiar et al . , _ particle multiplicities and thermalization in high energy collisions _ , arxiv : nucl - th/0001044 .
t.c.p.chui , d.r.swanson , m.j.adriaans , j.a.nissen and j.a.lipa , _ phys .
* 69 * ( 1992 ) 3005 ; c.kittel , _ physics today _ * 5 * ( 1988 ) 93 ; b.b.mandelbrot , _ physics today _ * 1 * ( 1989 ) 71 ; h.b.prosper , _ am .
_ * 61 * ( 1993 ) 54 ; g.d.j.phillires , _ am .
_ * 52 * ( 1984 ) 629 .
l.d.landau and i.m.lifschitz , _ course of theoretical physics : hydrodynamics _ , pergamon press , new york 1958 or _
course of theoretical physics : mechanics of continuos media _ , pergamon press , oxford 1981 .
m.rybczyski , z.wodarczyk and g.wilk , _ self - organized criticality in atmospheric cascades _ , presented at the _ xi isvhecri _ , unicamp - campinas - brazil , july 17 - 21 , 2000 ; hep - ph/0005216 ; to be published in _ nucl .
_ * b * _ ( proc .
suppl.)_. f.s.navarra , o.v.utyuzh , g.wilk and z.wodarczyk , _ violation of the feynman scaling law as a manifestation of nonextensivity _ , presented at chacaltaya meeting on cosmic ray physics , july 23 - 27 , 2000 , la paz , bolivia , to be published in _
nuovo cim .
_ * c * ; hep - ph/0009165 .
* discussion * + question by k.werner : _ concerning your example @xmath46 : this seems to be quite under control , large @xmath57 is perturbative qcd ( power - like ) , small @xmath135 is soft , so you have a superposition of these two processes .
there seems to be no need of these additional fluctuations . _
+ answer : your proposition is just one of many around us .
you are right that adding two known ( in their domain of applicability ) mechanisms will also fit data .
but what we are saying here is that , perhaps , there is mechanism which would show itself in the intermediate region and which can be described by the nonextensive parameter @xmath24 .
and it can be experimentally checked in event - by - event type of analysis , as was presented above .
this would answer the question is there a need for something or not .
besides , it should be stressed that using only one new parameter @xmath24 one can fit quite large ( in @xcite the whole ) region of @xmath57 .
+ question by f.grassi : _ did you try to fit @xmath57 distributions for other type of particles than charged ? _ + answer : no , at least not yet .
+ question by t.kodama : _ if @xmath24 represents the measure of fluctuations of ensemble , then it would also reflect in multiplicity distributions . however , value of @xmath24 is different for the @xmath57 distribution and multiplicity distribution .
how do you interpret this ? _ + answer : it is difficult for me to comment because so far i have not seen such analysis of the multiplicity data .
but if things are really as you say then , assuming that everything was done correctly , i would argue that multiplicity is global characteristic whereas @xmath57 is a local one .
therefore they are sensitive to different aspects of the underlying dynamics and resulting parameters @xmath24 could _ a priori _ be different .
one can also say that for @xmath57 distribution the situation is more clear as we are simple replacing here @xmath136 by @xmath137 .
multiplicity distributions depend on @xmath24 in indirect way only .
|
the statistical methods based on the classical boltzmann - gibbs ( bg ) approach are at heart of essentially all descriptions of multiparticle production processes . in many cases
, however , one observes some deviations from the expected behaviour .
it is also known that conditions necessary for the bg statistics to apply are usually satisfied only approximately .
two attitudes are possible in such situations : either to abandon statistical approach trying some other model or to generalise it to the so called nonextensive statistics ( widely used in the similar circumstances in many other branches of physics ) .
we shall provide here an overview of possible imprints of non - extensitivity existing both in high energy cosmic ray physics and in multiparticle production processes in hadronic collisions , in particular in heavy ion collisions .
+ pacs numbers : 05.40.fb 24.60.-k 05.10.gg + _ keywords : _ high energy multiparticle production , nonextensive statistics , thermal models + [ 3ex ] 0.3 cm 0.3 cm -2 cm = 10000
|
although the dark matter has been gravitationally confirmed by astrophysical observations in various way , one has no information on the properties , e.g. , the mass and the coupling . in a various dark matter models ,
a kind of popular model includes a light new boson which mediates between the standard model and a dark sector , e.g. , ref .
such a light particle simultaneously plays an important role in order to solve several problems , for instance , the small scale structure problems @xcite , the lithium problem @xcite , and the muon @xmath1 anomaly @xcite .
recently , a @xmath0be nuclear transition experiment has reported a signal which can be interpreted as an unknown light vector boson @xcite .
the vector boson ( @xmath2 ) is observed as a resonance in @xmath3 pairs whose invariant mass is @xmath4mev .
if one supposes a purely vector interaction between the standard model matter fields and the light vector boson , the consistency in the other experimental results requires the interaction should be protophobic @xcite , which can be written as @xmath5 where @xmath6 the coupling with the neutron @xmath7 , which is defined as @xmath8 , satisfies @xmath9 .
on the other hand , the coupling with proton @xmath10 is restricted as @xmath11 .
we fix the couplings as @xmath12 , @xmath13 , @xmath14 , and @xmath15 in the following discussion .
phenomenology of the @xmath2 boson and their models have been investigated in refs .
@xcite . in this letter
, we assume the light vector boson to be a gauge boson of a broken u(1)@xmath16 gauge symmetry and to be a mediator between the standard model and the dark sector which includes the dark matter , as in fig .
[ fig : schematic ] . using some simple models ,
we investigate experimental constraints on their parameters in the dark sector , and also discuss the compatibility with the thermal relic dark matter scenario .
as the dark sector , we consider a spontaneous symmetry breaking of the u(1)@xmath16 gauge symmetry by a dark higgs @xmath17 , which is a complex scalar boson charged under u(1)@xmath16 .
the lagrangian of the dark sector is @xmath18 where @xmath19 is the field strength tensor of @xmath2 .
the scalar field @xmath17 is expanded as @xmath20 in the unitarity gauge , and @xmath21 .
some terms irrelevant in our computation are suppressed here .
the parameters are defined as @xmath22 the original parameters @xmath23 and @xmath24 can be written by @xmath25 and @xmath26 with the gauge coupling @xmath27 . for simplicity , we consider that interactions between the higgs boson and the dark sector can be neglected @xcite .
first , we study a complex scalar and a dirac fermion dark matter models , where the dark matters are charged under u(1)@xmath16 .
if the dark matter is the complex scalar field @xmath28 , the lagrangian is @xmath29 since the annihilation into @xmath30 and @xmath31 are the s - wave processes , they dominate the thermal relic abundance . in this model ,
the experimental bounds on these two channels are , in addition to the dark matter mass @xmath32 , determined by the couplings @xmath33 and @xmath34 , respectively .
the annihilation cross section at the dark age ( @xmath35 ) is bounded by an observation of the cosmic microwave background ( cmb ) by the planck as @xmath36@xmath37/s / gev @xcite .
hence , the region where the dark matter is lighter than 30 gev is naively excluded by the result .
even if the dark matter is heavier than the value , the large sommerfeld enhancement through the @xmath2 boson excludes the thermal relic scenario @xcite .
the similar bound is obtained by ams-02 for the region @xmath38
@xcite with @xmath39 .
these indirect signals are 1- or 2-step cascades studied in ref . @xcite .
the region is also excluded by the direct detection result of the lux experiment @xcite . in order to see these bounds ,
we have followed the analysis method used in refs .
these results are shown in fig .
[ fig : directcoupling ] . for simplicity
, we consider only the @xmath31 channel in the figure , so that , only @xmath40 is relevant coupling .
the result can be translated into the @xmath30 channel with the replacement of @xmath41 by @xmath42 in their non - relativistic annihilation cross sections .
even if the both of the channels contribute to the annihilation process , the thermal relic dark matter can not be obtained .
note that recently ref .
@xcite has shown that if the dark matter is lighter than the vector boson a certain parameter region can explain the thermal relic abundance , see fig .
[ fig : directcoupling ] . considering the dirac fermion dark matter @xmath43 ,
the lagrangian is @xmath44 where @xmath43 is the dark matter .
the dark matter annihilates through the s - wave processes into @xmath31 and @xmath45 .
the situation of the experimental bounds and the consequence for the thermal relic scenario are the same with @xmath28 , except for the light dark matter window .
the non - relativistic annihilation cross section of @xmath46 is just a half of the cross section of @xmath47 if @xmath48 , while the result of the @xmath45 channel is give by the replacement of @xmath49 with @xmath50 .
since the sommerfeld enhancement factor is the same with @xmath28 , the result is almost the same as fig .
[ fig : directcoupling ] , so that the thermal relic scenario is also excluded .
next , we study the u(1)@xmath16 singlet dark matter models . interactions between the dark matters and the standard model are induced by the mixing with another particle charged under the u(1)@xmath16 gauge symmetry . in the case of the real scalar dark matter
@xmath51 , the dark higgs @xmath52 can be used as the mediator .
after the spontaneous symmetry breaking , the lagrangian is @xmath53 where we impose a @xmath54 symmetry ( @xmath55 ) to stabilize the dark matter .
the coupling @xmath56 is introduced like @xmath33 . in the previous section
, the direct detection excludes the thermal relic scenario if the dark matters is heavier than about 5gev . in this model
, however , the leading contribution to the direct detection comes from the loop induced diagram shown in fig .
[ fig : diagramdirectdetection ] .
hence , the direct detection bound becomes significantly weaker than the previous models .
since the scalar three point interaction is proportional to @xmath25 , the sommerfeld enhancement factor is also suppressed by @xmath57 @xcite , unless @xmath58 .
we also consider the u(1)@xmath16 singlet dark matter model including a majorana fermion @xmath59 .
since the majorana fermion can not interact with @xmath17 alone , we additionally introduce a dirac fermion @xmath60 whose u(1)@xmath16 charge is the same with @xmath17 .
then , the lagrangian is @xmath61 where the yukawa coupling @xmath62 can be chosen as real and positive without loss of generality .
after the spontaneous symmetry breaking , the coupling becomes the source of the fermion mixing .
the mass eigenstates are obtained by a so(3 ) rotation and a chiral rotation to flip the sign of a mass term as @xmath63 where @xmath64 and @xmath65 are , respectively , the left - handed and the charge conjugation of the right - handed components in @xmath60 , namely , @xmath66 .
the mass eigenstates @xmath67 are the four component majorana fermions defined as @xmath68 .
we assign the mass eigenvalues so as to be @xmath69 , i.e. , the dark matter is @xmath70 .
then , the lagrangian is written as follows , @xmath71 where @xmath72 the mass eigenvalues are related as @xmath73 in the numerical analyses below , we chose @xmath74 to evade the complexity of the co - annihilation of the dark fermions .
the leading contribution to the direct detection signal is also the loop diagram given in fig .
[ fig : diagramdirectdetection ] like the real scalar model . since the s - wave annihilation channel is suppressed by @xmath75 ,
the leading annihilation process is the p - wave .
hence , the indirect detection bound does not work to exclude the thermal relic scenario for the majorana dark matter .
even though the indirect detection bounds are too weak , the sommerfeld enhancement factor @xcite and the self - scattering cross section can be large if one takes the region of the heavy dark matter mass @xmath76 and the lighter mediator mass @xmath26 .
in this situation , the large self - interaction can solve the small scale structure problems as shown in the next section .
we show the indirect and the direct detection bounds of the real scalar dark matter model , and that whether they are compatible with the thermal relic abundance or not .
we also investigate future prospects of the direct detection bound .
the dark higgs @xmath52 mediates the annihilation into the @xmath2 pair and the direct detection as fig .
[ fig : diagramdirectdetection ] . in these processes
, amplitude can be written without @xmath27 . since the annihilation to @xmath30 is almost insensitive to @xmath27 , the physics of this model
is described by only @xmath56 and the dark matter mass .
indeed , our results are not changed in @xmath77@xmath78 . as we have shown in fig .
[ fig : realfig1 ] , since the sommerfeld enhancement is suppressed , the cmb bound excludes the thermal relic scenario only if the dark matter is lighter than about 30gev .
the electron - positron flux excludes the scenario up to the mass of about 100gev . in this letter
, we consider the case that the masses of the @xmath2 boson and the dark higgs bosons are the same scale .
note that because the transfer momentum in the dark matter - nucleon scattering in fig .
[ fig : diagramdirectdetection ] is also the same scale @xmath79@xmath80 , therefore the transfer - momentum contribution to the direct detection is not neglected .
in addition to the lux bound in 2013 @xcite , we have also drawn their recent result @xcite , and the prospects of the xenon1 t @xcite and the lz experiments @xcite .
evaluating the hadronic matrix elements , we have used a result of the lattice qcd simulation @xcite .
it is found that the current lux bound is too weak to exclude the thermal relic scenario .
the expected bound by lz can exclude the scenario up to a few hundreds gev , for @xmath81 mev .
[ figreal1 ] the dark higgs mass dependence of the expected direct detection bounds is shown in fig .
[ figreal2 ] .
it is found that the expected bounds by the xenon1 t and the lz experiments can exclude the scenario up to 200 gev and above 1 tev , respectively .
if the dark higgs is lighter than the @xmath2 boson , the scalar decays via two off - shell states . in the case ,
a loop induced decay into @xmath82 becomes the dominant channel . eventually , the life time of the dark higgs becomes larger than one second , so that the observed matter abundance could be suffered if the dark higgs abundance is too large .
the direct detections can reach the higher dark matter mass for the lighter dark higgs . considering the life time of the dark higgs
, the reaches decrease about 100 gev .
[ figreal2 ]
since the majorana dark matter mainly annihilates through the p - wave process , only the direct detection is important to restrict the thermal relic scenario .
the bounds and the prospects are shown in the fig .
[ figmajo ] .
the yukawa coupling to obtain the thermal relic abundance becomes large , when the dark matter is heavier than the mass difference @xmath83 .
below the value , the cross section is determined by the mass difference , so that the coupling is independent of the dark matter mass . recently , the region heavier than about 40 gev has been excluded for @xmath84 . with the projected experiments ,
the thermal relic region is excluded until the lower limit of their sensitivity , i.e. , @xmath85 .
similar to the real scalar dark matter case , these behavior are almost independent of @xmath27 . considering the heavy dark higgs ,
the direct detection bound becomes weaker .
the current bound is not sensitive if @xmath86 , while the prospected sensitivities by the xenon1 t and the lz experiments reach the dark higgs of 200 and 350 mev , respectively .
as mentioned before , the self - scattering cross section is enhanced by @xmath87 .
we have calculated the velocity - averaged transfer cross section @xcite , and found that due to the enhancement , the self - interaction can be large enough to solve the small scale structure puzzles . in this case
, the coupling @xmath27 should be a bit smaller than @xmath88 .
the details are shown in fig .
[ figself ] .
in this letter , we have investigated the dark matter models where the ptotophobic 16.7mev boson is a mediator between the standard model and the dark sector .
because of the severe constraint from the cmb observation , due to the large sommerfeld enhancement , the thermal relic scenarios are almost excluded in the u(1)@xmath16 charged dark matter models .
considering the spontaneous symmetry breaking of the u(1@xmath89 gauge symmetry , it is found that , when the dark matter is the u(1@xmath89 singlet , the dark matter can easily satisfy the observed relic abundance under the current experimental constraints .
some parameter regions can be probed by the future direct detection experiments .
particularly , in the majorana dark matter model , the large self - scattering cross section to solve the small scale structure puzzles can be achieved , while a bit small @xmath27 is required . according to ref .
@xcite , a light and a long - lived particle could solve the lithium problem .
hence , the parameter region where the dark higgs becomes a long - lived particle also attracts our attention @xcite .
the work of y.y . is supported in part by jsps kakenhi grant number 15k05053 . the author y.y .
thanks m. tanaka and r. watanabe for useful comments .
j. l. feng , b. fornal , i. galon , s. gardner , j. smolinsky , t. m. p. tait and p. tanedo , http://dx.doi.org/10.1103/physrevlett.117.071803[phys .
* 117 * ( 2016 ) no.7 , 071803 [ arxiv:1604.07411 [ hep - ph ] ] . ]
p. h. gu and x. g. he , arxiv:1606.05171 [ hep - ph ] ; l. b. chen , y. liang and c. f. qiao , arxiv:1607.03970 [ hep - ph ] ; y. liang , l. b. chen and c. f. qiao , arxiv:1607.08309 [ hep - ph ] . l. b. jia and x. q. li , arxiv:1608.05443 [ hep - ph ] .
the couplings between the dark sector and the higgs boson are introduced as @xmath90 and @xmath91 .
the higgs to the dark higgs pair decay has four electron - positron pairs in the final state . in order to suppress the significant contribution to the total width of the higgs , since the current bound is five times larger than the standard model prediction , the coupling @xmath92 should be smaller than about 0.06 , see ref .
the coupling @xmath93 is constrained by the direct detection and direct measurement of the higgs invisible width as a higgs portal dark matter . roughly speaking
, the coupling should be smaller than about 0.01 to evade any experimental bound , see ref .
@xcite .
j. m. cline , k. kainulainen , p. scott and c. weniger , http://dx.doi.org/10.1103/physrevd.88.055025[phys .
d * 88 * ( 2013 ) 055025 ] erratum : [ http://dx.doi.org/10.1103/physrevd.92.039906[phys .
d * 92 * ( 2015 ) no.3 , 039906 ] ] [ arxiv:1306.4710 [ hep - ph ] ] .
e. aprile [ xenon1 t collaboration ] , http://dx.doi.org/10.1007/978-94-007-7241-0_14[springer proc .
phys . *
148 * , 93 ( 2013 ) [ arxiv:1206.6288 [ astro-ph.im ] ] . ]
d. s. akerib _ et al . _
[ lz collaboration ] , arxiv:1509.02910 [ physics.ins-det ] .
a. abdel - rehim _ et al . _
[ etm collaboration ] , http://dx.doi.org/10.1103/physrevlett.116.252001[phys .
lett . * 116 * , no .
25 , 252001 ( 2016 ) [ arxiv:1601.01624 [ hep - lat ] ] . ]
|
observation of a protophobic 16.7 mev vector boson has been reported by a @xmath0be nuclear transition experiment .
such a new particle could mediate between the standard model and a dark sector which includes the dark matter . in this letter , we show some simple models which satisfy the thermal relic abundance under the current experimental bounds from the direct and the indirect detections . in a model , it is found that an appropriate self - scattering cross section to solve the small scale structure puzzles can be achieved .
|
cost function of generating units in area @xmath0 with form @xmath1 , where @xmath2 is positive definite .
vector of net load forecast for area @xmath0 .
vector of net load forecast for area @xmath0 at time @xmath3 .
vector of dispatch for area @xmath0 .
vector of all interface flows with predetermined directions where the @xmath4th element is denoted by @xmath5 .
vector of all interface flows at iteration @xmath6 .
vector of all interface flows at time @xmath3 .
vector of interface flows associated with area @xmath0 assuming _ outbound _ directions .
vector after removing the @xmath4th entry of @xmath7 .
vector of transmission capacities for area @xmath0 .
distribution of the net load @xmath8 , _ i.e. _ , @xmath9 .
distribution of the net load @xmath10 , _ i.e. _ , @xmath11 .
set of generation constraints for area @xmath0 .
number of separate external interfaces .
number of independently operated areas .
vector of interface capacities .
shift factors of internal buses to lines in area @xmath0 .
shift factors of proxy buses to lines in area @xmath0 . shadow price for area @xmath0 s power balance constraint .
shadow prices for area @xmath0 s transmission constraints .
shadow price for the @xmath4th interface constraint .
vector of locational marginal prices ( lmps ) at proxy buses of area @xmath0 .
the us electric system is partitioned into balancing authority areas connected by tie lines .
the interactions among different balancing authorities are carried out through interchange scheduling . since each balancing authority area , in general , has more than one neighbors , the operator has to specify how much power should flow across each separate scheduling interface .
the scheduling of tie line flows is critical to the overall system performance and market efficiency given the fact that the interchange amounts to a significant fraction of net energy .
for example , the net flow over external ties accounts for 16.5% of the net energy for load in iso new england in 2015@xcite . in the current external transaction system , operators use two sets of market - based offers to determine external interface schedules : ( 1 ) market participants external transaction offers to buy and sell across an interface , cleared against ( 2 ) the real - time generation supply curves in each control area .
inter - regional energy trades are driven by the interface price differentials , facilitated by market participants and administrated by operators .
intuitively , operators can use external tie lines to enhance market efficiencies and operational benefit by transferring excess power from low cost to high cost areas . in practice , however , the state of the art scheduling techniques are not efficient with two common symptoms : ( 1 ) the under utilization of tie lines in transporting power from low cost to high cost areas , and ( 2 ) the presence of counter - intuitive flows from high cost to low cost areas .
the economic loss from inefficient interchange schedules is substantial , estimated at the level of $ 784 million for the new york and new england customers from 2006 to 2010 @xcite .
one of the main causes of inefficient interchange scheduling is the latency between scheduling time and physical delivery .
since market participants submit bids and offers for the external transactions at least 75 minutes in advance , the information used for interchange scheduling may not reflect the actual system conditions at the time of physical delivery . with the increasing level of renewable integration , this situation is likely to be exacerbated .
techniques aimed at improving interchange efficiency can be classified into two categories .
the first aims to optimize the overall interconnected system in a decentralized fashion .
in particular , the optimal interchange schedules are obtained from the multi - area optimal power flow ( opf ) problem @xcite . among existing prior work
, the authors of @xcite consider the multi - area economic dispatch under wind uncertainty . in @xcite ,
a two - stage stochastic market clearing model is formulated for the multi - area energy and reserve dispatch problem whose solution is obtained based on scenario enumerations . in @xcite ,
the day - ahead tie - flow scheduling is formulated as a two - stage adaptive robust optimization minimizing the cost of the worst - case wind production . in @xcite , an adjustable interval robust scheduling of wind power for day - ahead multi - area energy and reserve market clearing
is proposed .
the uncertainty of wind farms is represented by predefined intervals and the clearing model is formulated as a mixed integer quadratic programming problem . for deterministic multi - area economic dispatch approaches , see @xcite and reference therein .
the main issue of this category approach is the elimination of arbitrage opportunities for the external market participants .
since operators can not trade with each other directly , market participants facilitate trades between control areas .
the multi - area economic dispatch approach thus can not be implemented under the current regulation .
the second category includes the current industrial practices based on the so - called proxy bus approximation @xcite .
the proxy bus is a trading location at which market participants can buy and sell electricity . in @xcite ,
a coordinated interchange scheduling scheme is proposed for the co - optimization of energy and ancillary services .
the proposal of coordinated transaction scheduling ( cts ) in @xcite is a state - of - the - art scheduling technique based on the economic argument using supply and demand functions exchanged by the neighboring operators .
when there is only a single interface in a two - area system , such functions can be succinctly characterized , and the exchange is only made once ; the need of iterations among operators is eliminated . built upon the idea of cts , a stochastic cts for the two - area single - interface scheduling problem is proposed in @xcite .
a shortcoming of existing techniques based on proxy bus approximations is the difficulty of generalizing it for multi - area interconnected systems where multiple scheduling interfaces have to be optimized simultaneously .
the challenge arises from the fact that the interfaces can not be succinctly characterized by a pair of expected demand and supply functions an essential property underlying the approach in @xcite for the single interface scheduling . when multiple interfaces are involved , the simple idea of equating expected demand and supply functions is not applicable and there is no simple notion that the intersection of demand and supply curves gives the social welfare optimizing interchange .
the main contribution of this paper is twofold .
first , we generalize the single interface scheduling problem to the case involving a network of operating areas , each having multiple interfaces with its neighbors .
we consider two types of interface scheduling : ( i ) the synchronous scheduling and ( ii ) the asynchronous scheduling .
the former requires all areas operated under the same scheduling clock whereas the latter allows every pair of operating areas setting their interfaces independently of others . to our best knowledge
, there is no existing results on this problem in the open literature .
second , we present a new scheduling technique based on the classical idea of coordinate descent method . the main idea is to iterate over all interfaces , one at a time , minimizing the overall system cost under uncertainty . in each iteration , a single interface optimization problem is solved by projecting the current solution to a particular coordinate representing a particular interface flow .
the optimality and convergence are guaranteed under nominal assumptions for both synchronous and asynchronous algorithms .
a proxy bus is a physical or virtual location where market participants can trade energy between areas . for interchange scheduling ,
a proxy bus is the location to which generation in the neighboring area is _ assumed _ be dispatched up and down in response to the change of interchange schedule . in this paper , we focus on the multi - proxy bus system where more than one proxy buses are used to represent the transmission network of the adjacent systems .
we use an example to illustrate the multi - proxy bus system . fig .
[ fig : multiproxy ] depicts a multi - proxy bus representation for a 3-area 2-interface interconnected system where each square box indicates a proxy bus .
the interchange vector @xmath7 consists of two interface flows , @xmath12 and @xmath13 , with fixed directions indicated by the arrows in fig .
[ fig : multiproxy ] .
area @xmath0 maintains a regional interchange vector @xmath14 that describes its own interface flows . for convenience ,
we assume the direction of each interface flow in the regional interchange vector being _ outbound_. in this example , the interchange vector for area 1 is @xmath15 , area 2 @xmath16 and area 3 @xmath17 .
the interchange scheduling problem is to optimize the interchange vector @xmath7 for which the expected overall operation cost is minimized _ under the proxy bus model_. this should be distinguished with the problem of multi - area economic dispatch in which the optimal tie line flows and regional generation dispatch are optimized _ without network approximation_.
in this section , we consider the single interface scheduling between area 1 and area 2 .
if the two areas are also physically connected with other areas , we assume that the tie line flows on those areas are fixed , thus not part of the decision process .
the single interface scheduling is a two - stage stochastic optimization : the first stage optimizes the interface flow @xmath18 from area 1 to area 2 as in ( [ opt : s1 ] ) ; the second stage dispatches internal resources @xmath19 and @xmath20 in the least cost manner to balance the interchange @xmath18 and internal net loads @xmath21 and @xmath22 as in ( [ opt : d3]-[opt : d4 ] ) .
the first stage optimization is given by @xmath23\ ] ] where the expectation is taking over all randomness of internal net loads @xmath24 and @xmath25 , and @xmath26 is the transfer capacity of the interface between area 1 and area 2 . given the schedule @xmath18 , and the realizations @xmath21 and @xmath22 , the optimal dispatch @xmath19 and @xmath20 are obtained from the second stage problems : @xmath27 and @xmath28 where @xmath29 , @xmath30 , is the shift factor vector of all transmission lines in area @xmath0 with respect to the single proxy bus in its neighbor , @xmath31 is a vector of ones with the compatible dimension , and the superscript `` @xmath32 '' denotes the transpose operation . from the regional dispatch problem , the demand and supply price for the interchange @xmath18 can be defined by the cost increase for the individual area with respect to the interchange quantity increase . by the envelop theorem ,
the lmp @xmath33 at proxy bus in area @xmath0 is given by @xmath34 where @xmath35 and @xmath36 are the lagrangian multipliers associated with @xmath37 given @xmath18 and @xmath8 for area @xmath0 , @xmath30 .
in general , the two - stage stochastic problem is intractable using standard optimization techniques when @xmath21 and @xmath22 follow continuous distributions .
fortunately , for this particular problem ( [ opt : s1]-[opt : d4 ] ) , there is an indirect approach developed in @xcite by solving the following stochastic social welfare maximization problem : @xmath38dx\ ] ] where @xmath39 $ ] is the expected lmp a function of the net interchange @xmath18at the proxy bus of area @xmath0 , @xmath30 .
the solution of the single interface stochastic optimization problem is given in the following theorem : [ thm : equivalence ] if ( [ opt : d3 ] ) and ( [ opt : d4 ] ) are not degenerate for all @xmath21 and @xmath22 , then problems ( [ opt : s1 ] ) and ( [ opt : s2 ] ) have the same optimizer @xmath40 satisfying @xmath41 if @xmath42 and @xmath26 otherwise .
theorem [ thm : equivalence ] generalizes the tie optimization solution in @xcite to the stochastic setting .
its significance lies in that the optimal interchange is the intersection of expected demand and supply functions rather than the expectation of the intersections of demand and supply curves or the intersection of demand and supply functions using expected generations and demands . in the following ,
we generalize the single interface scheduling algorithm to the multi - area system setting where multiple interface flows are involved .
in particular , the proposed interface - by - interface scheduling ( ibis ) algorithm is specialized for the synchronous scenario in section [ sec : syn ] and the asynchronous scenario in section [ sec : asyn ] .
the first scenario we consider is the synchronous interchange scheduling in which all operators in the interconnected system have a unified timetable , _
i.e. _ , all interface flows are optimized simultaneously at each scheduling time .
consider an interconnected system with @xmath43 independently operated areas ( of an arbitrary network topology ) and @xmath44 separate scheduling interfaces .
the multi - area synchronous interchange scheduling problem , analogous to the single interface scheduling problem , is also a two - stage stochastic optimization : the first stage is to set the values of all interface flows by minimizing the expected overall cost ; the second stage is to minimize the cost of individual areas given the fixed interchange and realized random generation and demand . as a generalization of ( [ opt : s1 ] ) for the single interface scheduling problem , the first stage optimization for the multi - area system is given by @xmath45\ ] ] where @xmath7 is a real vector in dimension @xmath44 , @xmath46 is the expected overall system cost , and @xmath47 is the optimal regional dispatch in area @xmath0 , given the interchange level @xmath14 over the interfaces associated with area @xmath0 and the realized net load @xmath8 . in the second stage
, each operator dispatches the internal resource to meet the interchange schedule @xmath14 and the internal net load @xmath8 in the least cost manner subject to the generation and transmission constraints . the optimization problem for area @xmath0 , @xmath48 ,
is specified as @xmath49 given the first stage decision @xmath14 and the realization of net load @xmath8 , the second stage problems are naturally decoupled and can be solved by their own operators .
the lmp vector @xmath50 at the proxy buses for area @xmath0 is calculated from the lagrangian multipliers of ( [ p : d2 ] ) @xmath51 where @xmath52 and @xmath53 are functions of the realization @xmath8 and the first stage decision @xmath14 .
the expected multi - dimensional lmp function @xmath54 for area @xmath0 is therefore defined as @xmath55\ ] ] where the expectation is taking over all randomness of the net load @xmath56 . the idea of the proposed scheduling algorithm is to iteratively optimize the interchange vector , one interface at a time , until the termination criterion satisfied .
specifically , at iteration @xmath6 , the @xmath4th interface flow is given by @xmath57 where @xmath58 by theorem [ thm : equivalence ] , the optimal solution @xmath59 can be obtained by searching the intersection of the expected supply and demand function defined in ( [ def : bar(pi ) ] ) for @xmath5 with fixed @xmath60 and check if the interface capacity @xmath61 is satisfied .
the detailed synchronous interface - by - interface scheduling ( sibis ) algorithm is given in algorithm [ alg : syn ] .
in practice , a positive value is chosen for @xmath62 to ensure a finite termination of sibis .
when @xmath62 is set to zero , the optimality and convergence behavior of sibis can be proved ( the proof of theorem [ thm : convergence ] is given in appendix [ app : thm1 ] ) .
[ thm : convergence ] let @xmath63 be the sequence generated by algorithm [ alg : syn ] with @xmath64 .
then , every limit point of @xmath63 is optimal to ( [ p : d1 ] ) . it should be noted that sibis is a form of cyclic coordinate descent method in which one cyclically iterates through the directions , one at a time , minimizing the objective function with respect to each coordinate direction at a time
. the early study of the coordinate descent method dates back to 1950s @xcite .
the convergence of the method has been extensively studied in the literature @xcite under various assumptions .
given the strict convex assumption of the regional cost function @xmath65 , the objective function @xmath46 is a continuously differentiable convex function , as shown in the proof of theorem [ thm : convergence ] .
if @xmath46 has local strict convexity in the feasible region of ( [ p : d1 ] ) , linear rate of convergence can be established as the case in @xcite .
the second scenario we consider is the asynchronous scheduling in which an operator with multiple interfaces determines interface flow one at a time . for such cases ,
the multi - interface scheduling problem is effectively reduced to a sequential single interface flow optimization .
mathematically , given the interface flow vector @xmath66 at time @xmath67 , the schedule of the @xmath4th interface @xmath68 at time @xmath3 is given by @xmath69\ ] ] where the expectation is taking over the randomness of net load @xmath10 with respect to the distribution @xmath70 at time @xmath3 , and the regional dispatch @xmath71 for area @xmath0 is the optimal solution to ( [ p : d2 ] ) , given the interchange schedule @xmath14 and the realization of net load @xmath10 .
since the interface flow @xmath5 is the only decision , the objective function in ( [ p : asyn_single ] ) only involves two areas connected by the @xmath4th interface flow .
therefore , the optimal interface flow @xmath72 at time @xmath3 can be obtained by intersecting the expected supply and demand function of @xmath5 given the distribution @xmath70 of the random net load @xmath10 .
note that the expected lmp function @xmath73 , similarly defined in ( [ def : bar(pi ) ] ) , depends on time through the distribution @xmath70 .
the distinction between synchronous and asynchronous scheduling lies in the decision at each scheduling time . for synchronous scheduling ,
the entire interchange vector is optimized via the iterative process given in algorithm [ alg : syn ] at each scheduling time @xmath3 . for the asynchronous scheduling ,
on the other hand , only one element of the interchange vector is optimized at time @xmath3 . therefore ,
the solution of the asynchronous scheduling algorithm at time @xmath3 is suboptimal in terms of minimizing the expected overall system cost .
below is the description of the asynchronous interface - by - interface scheduling ( aibis ) algorithm where the iterative process is carried out over time which should be distinguished with that in algorithm [ alg : syn ] .
we note that if the net load processes @xmath74 are independent in time , the optimal interchange depends only on the marginal distribution of the random load at the time of delivery . if in addition , the process is stationary , _
i.e. _ , @xmath74 is independent and identically distributed ( i.i.d . )
, then the optimal interchange is constant . in this case , aibis is essentially the classical cyclic coordinate decent spread over time . in comparison with sibis , sibis achieves optimal interchange for at every time whereas aibis achieves the optimality over time .
the following theorem , whose proof is given in appedix [ app : thm2 ] , formalize this argument .
[ thm : convergence2 ] let @xmath75 be the sequence generated by algorithm [ alg : asyn ] with @xmath76 .
if the net load @xmath77 for all @xmath0 , then every limit point of @xmath75 is optimal to ( [ p : d1 ] ) . when @xmath74 is not i.i.d .
, the interchange sequence generated by aibis does not converge to that by sibis .
the performance ( averaged over time ) of aibis algorithm does not in general converge to that of sibis ; the lack of synchronization translates to a performance loss . when the load process is a finite state markov chain , however , a modification of aibis that separately adapts the interchange for different load state will have the same time averaged performance as that of sibis .
in this section , we present numerical results of the proposed scheduling algorithms on the ieee 118-bus system .
the performance of the proposed sibis algorithm is compared with the certainty equivalence ( ce ) technique , .
the only difference lies in the price functions to obtain the single interface flow . instead of using the expected supply / demand function @xmath78 $ ]
, the ce method uses the supply / demand function @xmath79)$ ] by substituting the random variable by its the expected value . ] which uses the mean value of the random variable to schedule the interchange .
the comparison of sibis and aibis algorithms is then presented for time varying random processes .
the topology and area partition of the ieee 118-bus system are given in fig .
[ fig:118bus ] .
the interchange vector @xmath7 of this 3-area 2-interface system includes two interface flows where @xmath12 and @xmath13 are the flows to area 2 from area 1 and area 3 , respectively . the load profile , generator capacities , cost functions , and line and bus labels were defined in `` case118 '' in @xcite .
we imposed the maximum capacity of @xmath80 mw on transmission line 8 , 126 and 155 , and @xmath81 mw on the two interfaces .
bus 31 , 66 and 92 were selected as proxy buses for area 1 , 2 , and 3 respectively .
the system uncertainty arose from 12 wind generators ( roughly @xmath82 of buses ) located at bus 25 , 26 , 90 , 91 , 100 , 103 , 104 , 105 , 107 , 110 , 111 , and 112 , as indicated by red boxes in [ fig:118bus ] .
the selection of these locations were intended to simulate two wind farms ; the small one has 2 wind generators in area 1 , and the large one has 10 wind generators in area 3 concentrated on a few neighboring buses .
all wind generators were assumed to be identical and follow a two - mode gaussian mixture distribution whose probability density function is given by @xmath83 where the gaussian distribution @xmath84 represents the high wind scenario and @xmath85 the low wind scenario .
the lmp forecasting technique proposed in @xcite was used to compute the expected lmp . the key idea in @xcite
is to solve a multiparametric program to partition the net load space into critical regions . by the theory of multiparametric quadratic programming , within each critical region ,
the lmp is an affine function of the parameter vector .
since the partition of parameter space is independent of realizations of stochastic generation and demand , the computation of critical regions and functions that map the net load to lmp can be obtained ahead of time .
therefore , the expected lmp @xmath54 can be computed from the conditional distribution of the net load @xmath8 at each level of interchange @xmath14 .
since the optimal interface flow can be obtained by intersecting the expected supply and demand functions , we use the binary search algorithm . specifically , we did a line search within the interface limits to minimize the expected ( sample averaged ) price difference . in this section ,
we show the optimality and convergence behaviour of the proposed synchronized scheduling algorithm and the two most common symptoms of inefficient schedule .
the distribution ( [ dist : mixgaussian ] ) was used as the probabilistic wind production forecast with parameter values @xmath86 ,
@xmath87 , @xmath88 , and @xmath89 .
note that the difference between the proposed algorithm and the benchmark technique is the use of wind production forecast .
the synchronized scheduling algorithm uses the forecasted distribution while the certainty equivalence only uses the mean value @xmath90 .
the initial interchange vector was set at @xmath91 for both methods .
termination rule given in algorithm [ alg : syn ] was used for both methods with tolerance @xmath92 . in this simulation , we examine the utilization of tie lines .
ideally , tie lie should be utilized fully to the extent that power flows from a lower price proxy to a high price proxy .
all bus loads were set at the default values given in `` case118 '' @xcite . from the results shown in table [ table : iuu ] , we observed that the interface flows scheduled by ce and sibis had the same directions but different volumes .
the ce schedule of transferring @xmath93 mw from area 3 to area 2 resulted in the expected price disparity where the exporting area was $ @xmath94 while the importing area was $ @xmath95 .
note that there was adequate transmission capacity , there was still extra transmission capacity and overall cost could have been reduced .
this phenomenon is called interface under - utilization , which means transferring more power across the interface can further reduce the overall system cost .
the economic benefit from sibis can be observed in the expected overall system cost reduction in the sibis schedule . by increasing the interface flow from area 3 to area 2 to @xmath96 mw
, the expected supply and demand prices converged to $ @xmath97 . because there was no interface congestion ,
the expected prices in all three areas converged implying that the sibis schedule was efficient .
.performance comparison in the interface under - utilization scenario [ cols="^,^,^,^",options="header " , ] the convergence behavior was similar to that in simulation 1 as shown in fig .
[ fig : cif]a . the optimality was also verified .
+ we have established theoretical results for aibis in theorem [ thm : convergence2 ] under the independent and stationary assumption for the net load process . in reality , however , the operating condition of the power grid is constantly changing .
so we first verify the optimality of aibis in the i.i.d .
case and compare the performance of sibis and aibis in a more general setting . to verify the optimality of aibis for the i.i.d .
wind process , the generation @xmath98 of wind farm @xmath4 at time @xmath3 was assumed to follow @xmath99 for all @xmath4 and @xmath3 .
the scheduling horizon @xmath100 was set to @xmath101 for aibis and the tolerance @xmath62 was set to @xmath102 for sibis .
the initial interface flows were set at @xmath103 for both algorithms . as shown in fig .
[ fig : asyn_iid ] , sibis achieves optimal interchange for at every time whereas aibis achieves the optimality over time . specifically , the interface flows and the expected cost of sibis are optimal and remain constant over time .
aibis generated a sequence that converged to the optimal interchange and expected cost at time 7 .
it should be noted that the convergence rate is highly dependent on the initial values .
the convergence time of aibis is simply the adaptation of the cyclic coordinate descent method .
+ + the behavior of the proposed aibis algorithm was then investigated using a time varying process of wind generation .
specifically , we varied the mean value of wind generation starting from @xmath80 mw and ending at @xmath104 mw with a constant increment @xmath105 mw . except for the varying mean
, the rest of the setting remained the same .
+ + from fig . [
fig : asyn]a , the difference between sibis and aibis was clearly observed in the scheduled interface flows . since there were only two interfaces , the interface flows scheduled by aibis were alternatively constant during a scheduling time slot . with time increasing , the scheduling difference decreased and so did the expected overall cost shown in fig .
[ fig : asyn]b .
this paper presents a new interchange scheduling technique with the consideration of load and generation uncertainties in the context of multiple proxy bus system .
built upon the idea of coordinate descent method , the interchange vector is iteratively determined , one at a time , to minimize the expected overall system cost .
we first give the properties , implied by lemma [ lem : mp_affine ] , of the objective function @xmath46 in ( [ p : d1 ] ) and establish the convergence of the algorithm .
[ lem : mp_affine ] if problem ( [ p : d2 ] ) is neither primal nor dual degenerate for all @xmath8 , then the optimal value function @xmath106 is convex , continuously differentiable and piece - wise quadratic in both @xmath14 and @xmath8 . let @xmath107 where @xmath60 is defined in ( [ def : qk - i ] ) . using the updating rule ( [ p : s_cmin ] ) ,
we have @xmath108 let @xmath109 be a limit point of the sequence @xmath110 .
note that @xmath111 because @xmath112 is closed .
the monotonicity ( [ eqn : mono ] ) implies that the sequence @xmath113 converges to @xmath114 .
let @xmath115 , where @xmath116 is an index set , be a subsequence of @xmath110 that converges to @xmath109 . from the updating rule ( [ p : s_cmin ] ) and the monotonic property ( [ eqn : mono ] ) , for any interface @xmath4 , we have @xmath117 since @xmath46 is continuous , implied by lemma [ lem : mp_affine ] , taking the limit as @xmath118 tends to infinity on both sides , we have @xmath119 which means @xmath120 is an optimal solution of the following optimization @xmath121 therefore , @xmath120 satisfies the karush - kuhn - tucker ( kkt ) conditions for ( [ opt : q(i ) ] ) , _ i.e. _ @xmath122 where @xmath123 is the partial derivative with respective to @xmath5 and @xmath124 the associated lagrangian multiplier . since @xmath126 for all area @xmath0 , @xmath72 defined in ( [ p : asyn_single ] ) can be obtained by @xmath127 which implies @xmath128 let @xmath129 be a limit point of the sequence @xmath130 .
the monotonicity of @xmath131 implies that the sequence @xmath132 converges to @xmath133 .
let @xmath134 , where @xmath135 is an index set , be a subsequence of @xmath130 that converges to @xmath129 . from the updating rule ( [ eqn : qt(i ) ] ) and the monotonicity of @xmath131 , we have @xmath136 since @xmath46 is continuous , implied by lemma [ lem : mp_affine ] , taking the limit as @xmath137 tends to infinity on both sides , we have @xmath138 which means @xmath139 is an optimal solution of the following optimization @xmath140 therefore , @xmath139 satisfies the kkt conditions ( [ eqn : kkt1]-[eqn : kkt3 ] ) for [ opt : qt(i ) ] at @xmath129 . since ( [ eqn : kkt1]-[eqn : kkt3 ] ) hold for all @xmath4 at @xmath129 , @xmath129 satisfies the kkt conditions ( [ eqn : kkt21]-[eqn : kkt23 ] ) . by the convexity of @xmath46 , the kkt conditions are sufficient and necessary for optimality .
therefore , @xmath129 is optimal to ( [ p : d1 ] ) .
|
the problem of multi - area interchange scheduling under system uncertainty is considered .
a new scheduling technique is proposed for a multi - proxy bus system based on stochastic optimization that captures uncertainty in renewable generation and stochastic load . in particular , the proposed algorithm iteratively optimizes the interface flows using a multidimensional demand and supply functions .
optimality and convergence are guaranteed for both synchronous and asynchronous scheduling under nominal assumptions .
interchange scheduling , decentralized optimization , multi - area system , multiple proxy bus , seam issue , stochastic optimization .
|
the predicted x - ray luminosity and temperature relation from self - similarity ( @xmath1 is shallower than the observed one ( @xmath2 ) .
the steepening of the relation implies that low temperature clusters are less luminous than they would have been if self - similarity were to hold .
a physical explanation of this effect is that as the luminosity is roughly proportional to the gas density squared , there must exist processes which lower the density preferentially in low temperature clusters , hence the steepening of the relation .
such processes are for example preheating and radiative cooling of the gas .
ponman , cannon , & navarro ( 1999 ) have shown that the observed core entropies of low temperature clusters are in excess of what can be achieved by gravitational collapse alone .
they suggest that preheating can raise the entropy before the infall and that level of entropy is maintained during gravitational collapse . as the entropy scales as @xmath3 , where @xmath4 is the electron gas density , and the temperature of the gas is fixed by the virial temperature of the halo , the excess entropy implies a decrease in density , hence a decrease in luminosity .
energy sources which can inject energy into the intergalactic medium ( igm ) are for example agn , quasars , hypernova explosions , supernova explosions and so on . however , it is still debatable whether the energy injection efficiency of the heating sources is sufficient to provide the required entropy level .
on the other hand , radiative cooling is a process which is also important throughout the history of cluster formation .
low entropy gas , i.e. dense and cold gas , is removed from the icm , leaving behind high entropy material .
pearce et al ( 2000 ) have shown that when radiative cooling is included , the gas is slightly heated and the luminosity is greatly reduced thus acting so as to reconcile simulations and observations . in these proceedings , we present results from cosmological simulations of the @xmath0cdm cosmology in which 3 different models are considered . in the first , non - radiative model , non
- adiabatic heating comes about only via shocks following gravitational collapse and the gas is not allowed to cool .
the second cooling model introduces radiative cooling such that 20 percent of the gas in the box has turned into stars by the current day - a value that is higher than suggested by observations ( balogh et al 2001 ) .
finally , the third preheating model has both radiative cooling and an impulsive heating event in which 0.1 kev per particle is injected into the igm at a redshift of 4 .
this lowers the cooled fraction down to 10 percent .
the simulations consist of 160@xmath5 particles each of gas and dark matter within boxes of side 100 @xmath6 mpc .
the cosmological parameters are : density parameter , @xmath7 = 0.35 ; cosmological constant , @xmath8=0.65 , hubble parameter , @xmath9/100 km s@xmath10 mpc@xmath10 = 0.71 and baryon density parameter , @xmath11 = 0.019 . the metallicities are varied with time in the form of @xmath12 , where @xmath13 is the age of the universe in units of the current time . the objective of this is to mimic the gradual enrichment of the icm by stars .
cluster catalogues are constructed at redshift zero .
the catalogues are complete down to a cluster mass of @xmath14 m@xmath15 which corresponds to 500 particles of each species .
details of cluster selection criteria can be found in muanwong et al ( 2001 ) .
in this section , cluster properties are compared to those observed .
we calculate the x - ray properties ( temperature and luminosity ) in a soft x - ray band ( 0.3 - 1.5 kev ) by using the cooling tables of raymond & smith ( 1977 ) . the luminosity is then corrected to a bolometric one by assuming that the gas is isothermal .
the main effect of using band - limited rather than bolometric fluxes is to raise the emission - weighted temperature of low - mass clusters slightly .
we compare the derived @xmath16 relations of the simulated catalogues with a compilation of observations made by horner , mushotzky , & scharf ( 1999 ) . by using mass estimates from galaxy velocity dispersions , x - ray temperature profiles , the isothermal @xmath17-model and the surface brightness deprojection method
, they obtain different scaling laws shown as dashed lines in figure 1 .
the normalisation of the cooling simulation ( triangles ) is greater than that of the non - radiative simulation ( squares ) by about a factor of 2 at the mass of 10@xmath18 m@xmath15 . for clarity ,
we omit the preheating results but they lie roughly between the two .
most of the radiative cooling and preheating clusters , particularly the small systems , are hotter than their virial temperatures ( solid line ) , in accord with expectation .
non - radiative cooling clusters are cooler compared to the virial temperatures due to emission - weighting of cold and dense gas remaining in the cluster core .
the slopes of all 3 simulated catalogues are roughly 0.54 .
we find a good agreement between the relations from the radiative cooling model and observations , given the spread of the 4 estimates from observations . -1.5truecm
-1.5truecm -0.5truecm the effect of radiative cooling is demonstrated dramatically in terms of luminosity .
the @xmath19 relations of the non - radiative cooling , radiative cooling and preheating simulations are shown as squares , triangles and circles respectively in figure 2 . the large decrease in density as a result of radiative energy loss of the gas manifests in a significant reduction in luminosity .
at a temperature of 1 kev , the luminosity is reduced by a factor of 100 .
the radiative cooling model again is found to have a good agreement with observations as shown as solid lines ( xue & wu 200 ) .
the different solid lines are different fits to their subsamples , i.e. groups ( roughly below 1 kev ) , clusters ( above 1 kev ) and a mixture of the two subsamples .
the preheating model produces a relation which lies between that of the other two models .
preheating still results in an increase in entropy , causing the normalisation of the relation to be a factor of 10 smaller than that of the non - radiative cooling simulation , but the agreement with observations is not as good as the radiative cooling simulation .
the non - radiative clusters are consistent with self - similarity .
the relations in the other two models are significantly steeper than 2 , even more so in low temperature clusters . -1.5truecm
-1.5truecm -0.5truecm
we have demonstrated the effect of heating and cooling on x - ray properties of clusters in cosmological simulations .
the radiative cooling model can reproduce both the temperature - mass and luminosity - temperature relations in good agreement with observations .
however , the fraction of cooled gas is considerably higher than observed . in contrast , the preheating model results in a lower cooled fraction but gives a poor match to the observed scaling relations . as a possible resolution , we are currently undertaking a simulation which includes energy feedback from supernovae .
i am grateful for the financial support from the thai government .
i wish to thank peter thomas and scott kay for reading this manuscript and their feedbacks .
simulations presented here were carried out at edinburgh parallel computing centre as part of the virgo supercomputing consortium .
balogh , m. l. , pearce , f. r. , bower , r. g. , & kay , s. t. 2001 , , in press ( astro - ph/0104041 ) horner , d. j. , mushotzky , r. f. , & scharf , c. a. 1999 , , 520 , 78 muanwong , o. , thomas , p. a. , kay , s. t. , pearce , f. r. , & couchman , h. m. p. 2001
, apjl , 552 , l27 pearce , f. r. , thomas , p. a. , couchman , h. m. p. , & edge , a. c. 2000 , , 317 , 1029 ponman , t. j. , cannon , d. b. , & navarro , j. f. 1999 , nature , 397 , 135 raymond , j. c. , & smith , b. w. 1997 , , 35 , 419 xue , y .- j . &
2000 , , 538,65
|
we present results of including heating and cooling in cosmological simulations of the @xmath0cdm cosmology and demonstrate their effects on scaling laws of galaxy clusters .
the scaling relations when radiative cooling is included are in good agreement with observations but the fraction of cooled gas is on the upper limit allowed by observations . on the contrary
, the preheating model has a more realistic cooled fraction but the scaling relations are less well reproduced
. -1.0truecm # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
|
the group @xmath0 is a favourite candidate for grand unified theories ( guts ) @xcite because its 16-dimensional irreducible representation ( irrep ) , the spinor representation , contains all chiral fermions included in a standard model ( sm ) family plus an additional neutrino sm gauge singlet .
moreover , such theories allow for type i @xcite and type ii @xcite seesaw mechanisms ( see also @xcite ) for the light neutrino masses . in the construction of @xmath0 theories
, there are two options @xcite , either using low - dimensional scalar irreps but accepting non - renormalizable terms in the lagrangian , or one sticks to renormalizable terms , then one has to accept high - dimensional scalar irreps according to @xcite @xmath2 where the subscripts `` s '' and `` as '' denote , respectively , the symmetric and antisymmetric parts of the tensor product . in this paper , we deal with the second option .
a special renormalizable model is the so - called `` minimal susy @xmath0 gut '' ( msgut ) @xcite , which uses , for the yukawa couplings , one scalar in the @xmath3 and one in the @xmath4 irrep in order to account for all fermion masses and mixings ; it contains , in addition , one @xmath5 and one @xmath6 scalar irrep , in order to perform the suitable symmetry breakings .
this model has built - in the gauge - coupling unification of the minimal susy extension of the standard model ( mssm ) .
detailed studies of this minimal theory have been performed @xcite , also with small effects of the 120-plet @xcite .
though the msgut works very well in the fermion sector , there is a tension between the scale of the light neutrino masses and the gut scale .
the reason is that the natural order of the neutrino masses in guts is @xmath7 ev , where we have used @xmath8 gev for the electroweak scale and a gut scale of @xmath9 gev .
this neutrino mass scale is too low because @xmath10 ev where @xmath11 is the atmospheric neutrino mass - squared difference for reviews on the status of neutrino masses and mixing see @xcite .
studies of the heavy scalar states @xcite , together with studies of the fermion mass spectrum , have shown that the msgut is too constrained @xcite , and the tension between the scale of light neutrino masses and the gut scale can not be overcome : if one has a good fit of the fermion masses , which requires a seesaw scale below @xmath12 , then the gauge coupling unification of the mssm @xcite is spoiled .
a natural step for supplying additional degrees of freedom to the msgut is to add the 120-plet of scalars @xcite which appears anyway in eq .
( [ tensor])for early works in this direction see @xcite .
alone does not give a good fit in the charged fermion sector @xcite . ]
the disadvantage is that this step adds a considerable number of parameters and reduces the predictability of the theory . adding the @xmath13 leads to a resurgence of the type i seesaw mechanism @xcite , as a consequence of the collapse of the seesaw scale with the gut scale , because type i seesaw allows to enhance the neutrino masses through small yukawa couplings of the @xmath14 @xcite ; without the @xmath13 , i.e. in the msgut , this process leads to the contradictions mentioned above .
the @xmath13 has electrically neutral components only in its four doublets with respect to the sm gauge group .
these contribute to the higgs doublets @xmath15 , @xmath16 of the mssm , which are assumed to be the only light scalar degrees of freedom and the only ones which acquire vevs at the electroweak scale .
thus the msgut enlarged by the @xmath13 inherits , from the msgut , the scalar fields responsible for spontaneous symmetry breaking above the electroweak scale . in @xcite we took this into account by explicitly making the identification @xmath17 where @xmath18 , which defines the seesaw scale , is the vacuum expectation value ( vev ) of @xmath19 , with the usual notation for multiplets of the pati
salam subgroup @xcite @xmath20 of @xmath0 .
furthermore , in that paper we reduced the number of parameters in the fermion mass matrices by assuming a horizontal @xmath21 symmetry and spontaneous cp violation , i.e. real yukawa couplings , with cp violation stemming from the phases of the vevs .
we showed numerically that this scenario can excellently reproduce the known fermion masses and mixings .
recently , the role of spontaneous cp violation has been upgraded .
a `` new msgut '' ( nmsgut ) was proposed @xcite , defined by extending the msgut by the @xmath13 and spontaneous cp violation .
it was shown that the requirement of spontaneous cp violation not only has the virtue of reducing the number of parameters of the theory but it has an important impact , via threshold effects , on the unification scale as well ; it tends to raise the unification scale and with it the masses of all heavy multiplets , thereby suppressing baryon decay . in the present paper we retain eq .
( [ onescale ] ) as a reference point .
we do not employ any horizontal symmetry but we again motivate real yukawa coupling matrices by spontaneous cp violation .
however , we assume that it is of a very specific kind : cp is solely violated by imaginary vevs of the @xmath13 ; the vevs of the @xmath22 and @xmath14 are assumed to be real . in this way
, the mass matrices of the down - quarks , up - quarks , charged leptons and the neutrino dirac - mass matrix are hermitian .
this scenario was originally proposed in @xcite , its compatibility with sufficiently slow proton decay shown in @xcite . however , in @xcite it was assumed that the type ii seesaw mechanism is dominating .
since this is incompatible with having only one large scale , we have in the present paper type i dominance and neglect possible small contributions of type ii , suppressed by @xmath23 .
our scenario gives an excellent fit to all known fermion masses , mixings and the ckm phase @xmath24 , as good as the one in @xcite , though it is of a rather different type .
this shows that the fermion data do not fix the enlarged msgut in a unique way and there is considerable freedom in reducing the number of parameters in this theory .
the paper is organized as follows . in section [ cpviolation ]
we discuss cp - invariant yukawa couplings and lay out our scenario . the method and results of our numerical analysis
are discussed in section [ numerical ] . in section [ concl ]
we present the conclusions .
appendix [ so10 ] contains a small collection of formulas for the @xmath0 spinor representation , which is helpful for section [ cpviolation ] .
let us a define a transformation @xmath25 w^{pq}_\mu(x ) & \to & \varepsilon(\mu ) \eta_{pq } w^{pq}_\mu(\hat x ) , \end{array}\ ] ] where @xmath26 is a fermionic 16-plet , @xmath27 is the charge - conjugation matrix , the 45 gauge fields are denoted by @xmath28 ( @xmath29 ) , @xmath30 for @xmath31 and @xmath32 for @xmath33 , @xmath34 , and the @xmath35 are signs .
no summation is implied in eq .
( [ cpgauge ] ) . the 45 ( hermitian )
generators of the gauge group @xmath0 in the fermionic @xmath36 are given by @xmath37 for a representation of the operators @xmath38 ( @xmath39 of the clifford algebra and useful formulas concerning the spinor irrep @xmath36 see appendix [ so10 ] .
one can easily check that the gauge interaction of the fermionic 16-plet is invariant under the transformation ( [ cpgauge ] ) if @xcite @xmath40 since @xmath41 one finds @xmath42 denoting the generators of the lie algebra @xmath43 by @xmath44 , we mention that @xmath45 is the automorphism associated with root reflection , which is the canonical automorphism associated with cp .
such an automorphism exists for all compact lie groups and is the reason why any gauge lagrangian , whether for fermions or scalars , is cp - invariant @xcite .
now we transfer the cp transformation to the yukawa couplings given by the lagrangian @xmath46 the indices @xmath47 , @xmath48 denote the family indices , @xmath49 are @xmath0 indices , and @xmath27 is the charge - conjugation matrix . summation over family and @xmath0 indices
is implied in eq .
( [ yukawal ] ) .
the matrix @xmath50 , which ensures @xmath0 invariance , is defined in eq .
( [ b ] ) .
the yukawa coupling matrices have the properties @xmath51 we define cp transformations ( no summations implied ) @xmath52 for the scalar fields of the irreps @xmath22 , @xmath13 and @xmath14 , respectively ; the latter two are totally antisymmetric tensor fields , @xmath53 is self - dual in addition .
now we require invariance of the lagrangian ( [ yukawal ] ) under the cp transformation given by eqs .
( [ cpgauge ] ) , ( [ cp10 ] ) , ( [ cp120 ] ) and ( [ cp126 ] ) . as an example we take the @xmath22 and obtain @xmath54 the equality sign on the right - hand side of the arrow defines the condition of cp invariance : the cp - transformed yukawa lagrangian must be identical with its hermitian conjugate .
evaluating eq .
( [ cph ] ) with the help of eq .
( [ b ] ) , we find a hermitian yukawa coupling matrix . performing an analogous computation for the @xmath13 and @xmath14 we arrive at the conclusion that the cp transformation requires @xmath55 together with eq .
( [ yprop ] ) this means that _ all
_ yukawa coupling matrices are real . in order to obtain a non - trivial ckm phase @xmath24
it is necessary to break cp invariance .
the scenario we envisage was originally proposed in @xcite . in the context discussed here , we assume that * the vevs of the @xmath22 and @xmath14 are _ real _ , * cp is spontaneously broken by the vevs of the @xmath13 , * this breaking is maximal , i.e. , the vevs of the @xmath13 are imaginary . thus , the mass matrices of the charged fermions and the neutrino dirac - mass matrix are given , respectively , by @xmath56 with @xmath57 the mass matrices ( [ md])([md ] ) are hermitian .
the light neutrino mass matrix is given by @xmath58 with scalar triplet vevs @xmath59 and @xmath18 .
the mass lagrangian of the `` light '' fermions reads @xmath60 we finish this section with some remarks .
@xmath0 models have included the so - called d - parity @xcite , which is a specific involutory @xmath0 transformation which uses the branching rule @xmath61 under the pati salam group @xcite and exchanges the two pati salam irreps .
one can combine cp with d - parity and interpret such a transformation as parity in the usual sense @xcite .
requiring invariance of the theory under this parity , gives the same restrictions on the yukawa coupling matrices as cp alone , since the theory is invariant under d - parity anyway . in the cp transformation of the @xmath13 , eq .
( [ cp120 ] ) , we could put a minus sign .
then the yukawa coupling matrix would be hermitian and antisymmetric or , equivalently , antisymmetric and imaginary .
in that case , real vevs of the @xmath13 break cp maximally .
cc @xmath62 & @xmath63 + @xmath64 & @xmath65 + @xmath66 & @xmath67 \scriptstyle -}086.5 { \end{array}}$ ] + @xmath68 & @xmath69 + @xmath70 & @xmath71 \scriptstyle -}21.2264 { \end{array}}$ ] + @xmath72 & @xmath73 \scriptstyle -}14768.6 { \end{array}}$ ] + @xmath74 & @xmath75 + @xmath76 & @xmath77 + @xmath78 & @xmath79 + @xmath24 & @xmath80 + & @xmath81 & @xmath82 \scriptstyle -}0.0003 { \end{array}}$ ] + @xmath83 & @xmath84 \scriptstyle -}0.0501 { \end{array}}$ ] + @xmath85 & @xmath86 \scriptstyle -}1.2 { \end{array}}$ ] + @xmath87 & @xmath88 + @xmath11 & @xmath89 \scriptstyle -}0.25 { \end{array}}\big ) \times 10^{-3}$ ] + @xmath90 & @xmath91 + @xmath92 & @xmath93 + @xmath94 & @xmath95 + as argued in the introduction , with only one large scale , the gut scale , in the theory , we can neglect the type ii seesaw contribution in eq .
( [ mnu ] ) .
then a possible phase of @xmath18 is irrelevant .
with @xmath96 we rewrite the mass matrices as @xmath97 without loss of generality we assume @xmath98 to be diagonal
. then all redundant parameters are removed and we arrive at 12 real parameters in @xmath98 , @xmath99 , @xmath100 and six real ratios of vevs .
thus our scenario has 18 independent parameters for 18 observables : nine charged - fermion masses , three mixing angles and the cp phase @xmath24 in the ckm matrix , the atmospheric and solar neutrino mass - squared differences @xmath11 and @xmath87 , and three lepton mixing angles .
( [ md])([mnu ] ) are amenable to a numerical analysis , which will , in particular , yield values for @xmath101 and @xmath102 .
if we fix the triplet vev @xmath18 , e.g. by identifying it with the gut scale see eq .
( [ onescale ] ) , this analysis will also yield definite values for @xmath103 and @xmath104 because @xmath105 a reasonable condition on these vevs is given by @xcite @xmath106 this inequality certainly holds at the electroweak scale . assuming that it holds approximately at the gut scale as well
, we will subject our fit results to this consistency check . to find a numerical solution for the parameters in eqs .
( [ md])([mnu ] ) , we build as usual @xcite a @xmath107-function for the 18 observables , @xmath108 whose input values @xmath109 are given in table [ input ] ; these values refer to an mssm parameter @xmath110 .
the letter @xmath111 symbolizes the set of 18 parameters , i.e. the yukawa couplings and ratios of vevs . the functions @xmath112 express our theoretical predictions , as functions of the parameter set @xmath111 , for the observables , obtained from eqs .
( [ md])([mnu ] ) . as convention for the quark and lepton mixing matrix we use that of the review of particle properties @xcite .
the @xmath107-function is minimized analytically with respect to @xmath102 . in this way
we obtain a @xmath107-function of the remaining 17 parameters , which is minimized numerically by employing the downhill simplex method @xcite . in the following
we will also consider @xmath107 functions where a specific quantity is pinned down to a given value for previous uses of such a method see e.g. @xcite .
if we want to pin down a quantity @xmath113 , which is independent of the 18 observables , to a value @xmath114 , we add @xmath115 in eq .
( [ chi2 ] ) and minimize the thus obtained @xmath116 .
if @xmath117 coincides with one of the observables @xmath118 , the term above is added but at the same time the term @xmath119 has to be removed from the @xmath107 of eq .
( [ chi2 ] ) . in that way
, we can study the sensitivity of our scenario to a variation of a quantity @xmath117 .
.the values of the vev ratios appearing in the mass matrices ( [ md])([mnu ] ) obtained in the fits with normal and inverted neutrino mass ordering . [ ratios ] [ cols=">,^,^,^,^,^,^",options="header " , ] [ [ a - fit - in - the - case - of - normal - neutrino - mass - ordering ] ] a fit in the case of normal neutrino mass ordering : +
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we search for a solution in the case of the normal ordering @xmath120 of the neutrino masses ( @xmath121 , @xmath122 ) . in that case
we find an excellent fit with the following properties : @xmath123 m_1 = 1.81\times 10^{-3 } \ : \mbox{ev } , \quad m_2 = 0.907 \times 10^{-2 } \ : \mbox{ev } , \quad m_3 = 5.006 \times 10^{-2 } \ : \mbox{ev}. \end{array}\ ] ] in the second relation in this equation we have used @xmath124see eq . ( [ onescale ] ) .
the corresponding values of the matrix elements of @xmath98 , @xmath99 , @xmath100 are given by @xmath125 where all numerical values are in units of mev .
the fit values of the vev ratios are listed in table [ ratios ] .
the @xmath107 of the fit practically comes only from two observables : the pull of @xmath126 ( leptonic mixing angle ) is 0.45 and the pull of @xmath66 is 0.31 .
we can ask the question if our scenario makes some predictions . for the best fit we find @xmath127 .
however , this small value misleading , because pinning @xmath128 in @xmath107 shows that in the large range @xmath129 the fit is still very good , with @xmath130 .
the worst @xmath107 is about 15 and occurs around @xmath131 , where for instance @xmath66 is not well reproduced and the leptonic @xmath126 becomes too large . as for @xmath126 , fig .
[ fig13 ] shows that the preferred value is about 0.01 .
however , we can not consider this as a prediction since in a wide range around this value the @xmath107 is still acceptable . only at very small
@xmath126 the fit becomes bad , mainly because of @xmath62 , @xmath66 and the atmospheric mixing angle @xmath132 .
the quantity @xmath133 measures how hierarchical a normal neutrino mass spectrum is .
the @xmath107 as a function of @xmath134 is depicted in fig .
we read off that @xmath135 is preferred and the @xmath107 quickly becomes bad for larger @xmath134 , mainly owing to @xmath62 , @xmath66 and the leptonic @xmath126 . also for very small @xmath134 the fit worsens , for similar reasons as for large @xmath134 , however , not in a dramatic way . on the other hand , in the msgut
there is a preferred range @xmath136 and there is a genuine lower bound on @xmath134 as well @xcite . with eq .
( [ vud ] ) and the upper bound on @xmath137 in eq .
( [ fit - normal ] ) we see that we are allowed to raise @xmath18 to @xmath138 , without violating the inequality ( [ test ] ) .
it has to be checked that our numerical solution given by eq .
( [ solution - normal ] ) and table [ ratios ] respects the perturbative regime in the yukawa sector .
since the procedure has been explained in detail in @xcite , we confine ourselves to the essentials .
the two higgs doublets of the mssm , @xmath15 and @xmath16 , have hypercharges @xmath139 and @xmath140 and vevs @xmath141 and @xmath142 , respectively .
the corresponding yukawa coupling matrices are given by @xmath143 , etc .
it turns out that the largest yukawa couplings are @xmath144 , where the largest contribution comes from @xmath145 .
using @xmath110 , it is given by @xmath146 .
this confirms that the yukawa couplings are safely in the perturbative regime .
finally , we want to estimate the size of type ii seesaw contributions to @xmath147 .
the corresponding mass matrix is given by @xmath148see eqs .
( [ onescale ] ) , ( [ mnu ] ) and ( [ vud ] ) .
the largest element in @xmath100 is the 33-element . with the value of this element from eq .
( [ solution - normal ] ) and @xmath102 from table [ ratios ] , the type ii mass matrix contributes at most @xmath149 . since we expect @xmath150 ev , we find the announced suppression with respect to type i seesaw contributions .
[ [ a - fit - for - the - inverted - neutrino - mass - spectrum ] ] a fit for the inverted neutrino mass spectrum : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + searching for a fit by imposing the inverted ordering @xmath151 of the neutrino masses ( @xmath121 as for the normal spectrum , but @xmath152 ) , we find a solution which is even better as in the case of normal ordering .
it has the following properties : @xmath153 m_1 = 4.920 \times 10^{-2 } \ : \mbox{ev } , \quad m_2 = 5.000 \times 10^{-2 } \ : \mbox{ev } , \quad m_3 = 2.18 \times 10^{-4 } \ : \mbox{ev } , \end{array}\ ] ] with the matrices @xmath154 where all numerical values are in units of mev , and the vev ratios are displayed in table [ ratios ] . for all practical purposes the fit is perfect and
there is no need to give any pull values .
now we come to the predictions of our scenario in the case of the inverted neutrino mass spectrum . concerning cp violation in neutrino oscillations , our best fit gives @xmath155 .
however , this value has no meaning because @xmath107 as a function of @xmath128 is flat for all practical purposes .
the same is true for the leptonic quantities @xmath156 and @xmath126 in the physically relevant ranges .
however , there is a definite prediction for the neutrino mass spectrum : hierarchy is strongly preferred
see fig .
[ figr ] . when the quantity @xmath157 becomes large the fit turns bad
; however , there is no clear - cut reason , it is mostly the down - quark masses and the top - quark mass which are not well reproduced and the fit value of the leptonic @xmath126 is around its experimental upper bound . in the second relation of eq .
( [ fit - inverted ] ) we have again used our reference value ( [ onescale ] ) .
now inequality ( [ test ] ) is respected for @xmath158 , i.e. there is more freedom for @xmath18 than in the normal case . as before ,
large yukawa couplings in @xmath159 and @xmath160 are induced by @xmath145 .
but now , because @xmath161 is so big , a slightly larger coupling is @xmath162 , still in the perturbative regime .
the discussion of the smallness of type ii seesaw contributions to @xmath147 proceeds as for the normal spectrum . in eq .
( [ solution - inverted ] ) the elements @xmath163 , @xmath164 and @xmath165 are rather small .
this might suggest to set them zero , which is achieved by the horizontal @xmath21 symmetry @xmath166 , @xmath167 .
however , this is untenable because it would lead to vanishing @xmath24 .
the reason is that this horizontal @xmath21 can be combined with the cp transformation of section [ cpviolation ] to a new symmetry cp@xmath168 , under which the vacuum state of our scenario is invariant . , the vevs @xmath169 do not change sign ! ] consequently , in that case there is no cp violation @xcite and one can show as it must be in such a case that @xmath170 @xcite .
in this paper we have investigated fermion masses and mixings in an susy @xmath0 scenario , originally proposed in @xcite , where the yukawa coupling matrices of the scalars in the irreps @xmath22 , @xmath13 and @xmath14 are real and cp violation is induced only by imaginary vevs of the @xmath13 .
this gives a scenario with 18 real parameters in the fermion mass sector .
recent results from the msgut require a single heavy @xmath0 breaking scale , which is then cogent for the msgut extended by the @xmath13 as well . there are the following differences between ref .
@xcite and the present paper : firstly , our @xmath147 is induced by the type i seesaw mechanism and type ii is negligible , whereas in @xcite it was assumed that type ii dominates .
secondly , we use a purely numerical method , employing the minimization of a @xmath107 function , whereas ref .
@xcite uses an approximate semianalytical method .
we have found excellent fits to fermion masses and mixings for both types of neutrino mass spectra . we want to emphasize this in particular for the inverted mass spectrum , for which the system of fermion mass matrices in the msgut which has no @xmath13does not allow an acceptable fit @xcite , though complex yukawa couplings and vevs and contributions to @xmath147 from both seesaw types are admitted .
the fits presented in this paper have the following features .
the diagonal yukawa coupling matrix @xmath171 of the @xmath22 is strongly hierarchical and is responsible , in the charged - fermion mass spectra , for the hierarchy between 2nd and 3rd families .
the correct size of the neutrino masses is reproduced by a cooperation of two effects : rather large contributions @xmath172 to the neutrino dirac - mass matrix @xmath173 from the couplings of the @xmath22 and @xmath13 , where @xmath174 is the yukawa coupling matrix of the @xmath13 , and a moderately small coupling matrix @xmath175 of the @xmath14 , which enters with its inverse in the type i seesaw formula .
the contribution of the @xmath13 to the charged - fermion masses and to the ckm matrix is rather small , whereas @xmath176 in @xmath173 introduces large leptonic mixing angles .
similar features were found in the previous sample fit of @xcite , though the assumptions concerning the fermion mass matrices in that paper are quite different from those in the present paper , apart from the use of spontaneous cp violation in both scenarios . unfortunately , our scenario is not very predictive .
however , it does have one clear - cut prediction , namely hierarchy for both the normal and inverted neutrino mass spectrum .
this is quantified by the observable @xmath134 in fig .
[ figr ] from where we read off @xmath177 .
apparently , extending the msgut by the @xmath13 leads to an ambiguous situation concerning fermion mass matrices : quite different assumptions can result in excellent fits . whether these fits are compatible with the nmsgut @xcite , where the vevs are subject to certain relations , remains to be checked .
one aspect seems to emerge : spontaneous cp violation plays in important role in both the fermion mass matrices @xcite and the spontaneous breaking @xcite of @xmath0 .
* acknowledgment : * w.g . is grateful to l. lavoura for useful discussions .
a possible representation on the space @xmath178 of the fivefold tensor product of @xmath179for the clifford algebra associated with the lie algebra @xmath43 is given by @xcite @xmath180 where a superscript @xmath181 denotes the @xmath182-fold tensor product .
the @xmath183 matrices @xmath184 ( @xmath185 ) are the pauli matrices and @xmath186 denotes the @xmath183 unit matrix .
it is easy to check that the @xmath38 ( @xmath187 ) fulfill @xmath188 it is well known that the matrices @xmath189 have precisely the same commutation relations as the basis elements @xmath190 of @xmath43 .
the @xmath191 generate the spinor irrep @xmath36 of @xmath43 on the 16-dimensional space @xmath192 note that @xmath193 anticommutes with all @xmath38 ( @xmath194 ) .
p. minkowski , phys .
b 67 ( 1977 ) 421 ; + t.yanagida , in _ proceedings of the workshop on unified theory and baryon number in the universe _ , o. sawata and a. sugamoto eds .
, kek report 79 - 18 , tsukuba , japan , 1979 ; + s.l .
glashow , in _ quarks and leptons , proceedings of the advanced study institute ( cargse , corsica , 1979 ) _
basdevant et al .
, plenum , new york , 1981 ; + m. gell - mann , p. ramond , and r. slansky , in _ supergravity _ ,
d.z . freedman and f.
van nieuwenhuizen eds . ,
north holland , amsterdam , 1979 ; + r.n .
mohapatra , g. senjanovi , phys .
44 ( 1980 ) 912 .
g. lazarides , q. shafi , c. wetterich , nucl . phys .
b 181 ( 1981 ) 287 ; + r.n .
mohapatra , g. senjanovi , phys .
d 23 ( 1981 ) 165 ; + r.n .
mohapatra , p. pal , _ massive neutrinos in physics and astrophysics _ , world scientific , singapore , 1991 , p. 127 .
j. schechter , j.w.f .
valle , phys .
d 22 ( 1980 ) 2227 ; + s.m .
bilenky , j. hoek , s.t .
petcov , phys .
b 94 ( 1980 ) 495 ; + i.yu .
kobzarev , b.v .
martemyanov , l.b .
okun , m.g .
shchepkin , yad .
32 ( 1980 ) 1590 [ sov
. j. nucl .
32 ( 1981 ) 823 ] ; + j. schechter , j.w.f .
valle , phys .
d 25 ( 1982 ) 774 .
g. senjanovi , in _ proceedings of seesaw25 : international conference on the seesaw mechanism and the neutrino mass , 1011 june 2004 , paris , france _ , j. orloff , s. lavignac , and m. cribier eds . , world scientific , singapore , 2005 , p. 45
[ hep - ph/0501244 ] .
aulakh , r.n .
mohapatra , phys .
d 28 ( 1983 ) 217 ; + t.e .
clark , t.k .
kuo , n. nakagawa , phys .
lett 115b ( 1982 ) 26 ; + k.s .
babu , r.n .
mohapatra , phys .
70 ( 1993 ) 2845 [ hep - ph/9209215 ] ; + c.s .
aulakh , b. bajc , a. melfo , g. senjanovi , f. vissani , phys .
b 588 ( 2004 ) 196 [ hep - ph/0306242 ] .
k. matsuda , y. koide , t. fukuyama , phys .
d 64 ( 2001 ) 053015 [ hep - ph/0010026 ] ; + k. matsuda , y. koide , t. fukuyama , h. nishiura , phys .
d 65 ( 2002 ) 033008 ( erratum - ibid .
d 65 ( 2002 ) 079904 ) [ hep - ph/0108202 ] ; + t. fukuyama , n. okada , jhep 11 ( 2002 ) 011 [ hep - ph/0205066 ] ; + b. bajc , g. senjanovi ' c , f. vissani , phys .
( 2003 ) 051802 [ hep - ph/0210207 ] ; + h.s .
goh , r.n .
mohapatra , s.p .
ng , phys .
b 570 ( 2003 ) 215 [ hep - ph/0303055 ] ; + h.s .
goh , r.n .
mohapatra , s.p .
ng , phys . rev .
d 68 ( 2003 ) 115008 [ hep - ph/0308197 ] ; + b. bajc , g. senjanovi , f. vissani , phys . rev .
d 70 ( 2004 ) 093002 [ hep - ph/0402140 ] .
k. matsuda , t. fukuyama , h. nishiura , phys .
d 61 ( 2000 ) 053001 [ hep - ph/9906433 ] ; + s. bertolini , m. frigerio , m. malinsk , phys .
d 70 ( 2004 ) 095002 [ hep - ph/0406117 ] ; + s. bertolini , m. malinsk , phys .
d 72 ( 2005 ) 055021 [ hep - ph/0504241 ] .
aulakh , a. girdar , int . j. mod .
phys . a 20 ( 2005 ) 865 [ hep - ph/0204097 ] ; + t. fukuyama , a. ilakovac , t. kikuchi , s. meljanac , n. okada , eur .
j. c 42 ( 2005 ) 191 [ hep - ph/0401213 ] ; + b. bajc , a. melfo , g. senjanovi , f. vissani , phys . rev .
d 70 ( 2004 ) 035007 [ hep - ph/0402122 ] ; + c.s .
aulakh , a. girdar , nucl .
b 711 ( 2005 ) 275 [ hep - ph/0405074 ] ; + t. fukuyama , a. ilakovac , t. kikuchi , s. meljanac , n. okada , j. math
46 ( 2005 ) 033505 [ hep - ph/0405300 ] ; + t. fukuyama , a. ilakovac , t. kikuchi , s. meljanac , n. okada , phys .
d 72 ( 2005 ) 051701 [ hep - ph/0412348 ] ; + c.s .
aulakh , phys .
d 72 ( 2005 ) 051702 [ hep - ph/0501025 ] .
aulakh , expanded version of the plenary talks at the _ workshop series on theoretical high energy physics _
, iit roorkee , uttaranchal , india , march 1620 , 2005 , and at the _
8th european meeting `` from the planck scale to the electroweak scale '' ( planck05 ) _ , ictp , trieste , italy , may 2328 , 2005 , hep - ph/0506291 ; + b. bajc , a. melfo , g. senjanovi ' c , f. vissani , phys .
b 634 ( 2006 ) 272 [ hep - ph/0511352 ] .
|
we consider fermion masses and mixings in a renormalizable susy @xmath0 gut with yukawa couplings of scalar fields in the representation @xmath1 . we investigate a scenario defined by the following assumptions : i ) a single large scale in the theory , the gut scale .
ii ) small neutrino masses generated by the type i seesaw mechanism with negligible type ii contributions .
iii ) a suitable form of spontaneous cp breaking which induces hermitian mass matrices for all fermion mass terms of the dirac type .
our assumptions define an 18-parameter scenario for the fermion mass matrices for 18 experimentally known observables . performing a numerical analysis
, we find excellent fits to all observables in the case of both the normal and inverted neutrino mass spectrum .
|
photons are promising candidates for quantum information processing @xcite , due to their weak interaction to environment and easy single - qubit operations . in photonic quantum information processing ,
preparing high - quality entangled states of photons plays a key role .
the reason is that photons lack of interactions , while photonic quantum information protocols usually can be realized with off - line - generated entangled photons , such as the one - way quantum computation @xcite , and the quantum teleportation @xcite . today
, biphoton entanglement generation has become general in laboratory .
the most convenient way arises from the spontaneous parametric down - conversion ( spdc ) processes in nonlinear crystals .
extremely high - fidelity biphoton bell states have been prepared via spdc processes recently @xcite ( up to @xmath1 in ref .
also , high - quality entangled photon pairs have been used for the loophole - free test of local realism @xcite . however , the state fidelity of three or more entangled photons is still at a much lower level @xcite , mainly due to the difficulty of entangling photons from independent sources .
entangling independent photons is also the key ingredient in various quantum information tasks , such as entanglement swapping @xcite and teleportation .
thus , moving from biphoton to multiphoton entanglement means much more than a simple expansion . in this paper
, we experimentally generate an ultrahigh - fidelity four - photon greenberger - horne - zeilinger ( ghz ) state , with only 20 mw pump power , using our new sandwichlike " einstein - podolsky - rosen ( epr ) source @xcite .
this epr source is based on the beamlike type - ii phase matching @xcite , and is rather suitable for multiphoton entanglement generation .
it achieves high brightness , high fidelity and high collection efficiency at the same time . the high collection efficiency and high - fidelity enables high - performance multiphoton experiments . and the high brightness greatly reduces the power requirement for the pump laser employed in multiphoton experiments . with the high - fidelity ghz state
, we further demonstrate a high - performance entanglement swapping operation .
we also test a demanding hardy - like inequality in the nonsignaling senario , which can detect genuine n - partite nonlocality using only 2n local measurement settings , but require rather high state fidelity .
spdc in nonlinear crystals provides a convenient way to produce entangled photon pairs . according to the phase - matching type
, there are two kinds of widely used entanglement sources .
the type - i entanglement source @xcite uses a two - crystal geometry ( closely placed , relatively thin , identically cut ) .
the down - converted photons from the two crystals almost emit as one cone centered at the pump beam due to the small thicknesses of the two crystals .
the optic axes of the two crystals lie in the vertical and horizontal plane , respectively .
so , a @xmath2-polarized photon in the pump laser will be down - converted equally in either crystal .
and the down - converted photon pairs from different crystals have orthogonal polarizations .
when choosing two emitting directions from the cone symmetrically ( about the pump beam ) , the collected photon pairs are in the polarization entangled state . in the type - ii entanglement source
@xcite , the down - converted photons emit into two crossed cones , and have orthogonal polarizations .
so , when photon pairs are collected along the two intersecting directions , they are in the bell state .
however , in both cases , most of the down - conversion fields are wasted , so the brightness of these sources are greatly limited . to overcome this shortcoming ,
the beamlike type - ii phase matching is designed and implemented @xcite , in which the down - conversion fields shrink into two separate beams .
this is accomplished by decreasing the phase - matching angle from the situation of two crossed cones .
each beam has a gaussian - like intensity distribution and a small divergence angle , thus all down - conversion fields can be easily collected . with a two - crystal geometry ,
beamlike entangled photon pairs source has been designed @xcite and implemented @xcite . in a most recent work @xcite , we further design and implement a new sandwichlike " epr source ( see appendix ) under the beamlike phase matching condition .
the key idea is to make all ordinary ( o ) photons emitting in one path , while all extraordinary ( e ) photons emitting in the other path , as shown in fig . 1 . and the nonlinear crystals used for spdc processes here are @xmath3-barium borate ( bbo ) crystals . due to the high brightness and high quality of this source ,
we choose it to generate our four - photon ghz state . experimental setup for our sandwichlike " epr source .
the thicknesses of the linbo@xmath4 crystals are different in the two paths , which is due to the walk - off effect of the pump beam and the down - converted beams in the bbo crystals .
the half - wave plates ( hwps ) before the fiber couplers are used for polarization alignment.,scaledwidth=40.0% ] next , we show how to prepare the high - fidelity four - photon ghz state from two sandwichlike epr sources .
the key requirement is the ability to entangle photons from independent sources . due to the lack of interaction between photons ,
the general method is using the measurement induced nonlinearity @xcite , e.g. , using a projection operator to project the product state onto the entangled state with a certain probability . the seminal work @xcite gives us a feasible scheme to make it into practice .
the key idea is to use pulsed pump laser and relatively narrow - band filters for interfering photons .
here we follow their scheme .
the experimental setup is shown in fig .
the ultraviolet pump laser for each epr source is generated in the same way as described in the appendix , except that we tune the pulse duration of the mode - locked ti : sapphire laser to 90 fs for better multiphoton interference visibility . and
two sandwichlike bbo strategies are employed to generate the entangled photon pairs 1 , 2 and 3 , 4 .
thus the input state is @xmath5 denotes the horizontal ( vertical ) polarization state of the photons .
all photons are coupled into single - mode fibers for spatial filtering .
the two e - photons ( 2 and 3 ) are directed to the pbs for the hong - ou - mandel ( hom ) interference .
the pbs acts as a parity check gate @xmath6 , such that the two interfering photons can emit into two different output ports only if they have the same polarization .
thus , when we postselect the cases that there is one and only one photon in each output port ( 1 , @xmath7 , @xmath8 , 4 ) , the input state is projected onto the four - photon ghz state @xmath9 .
then , each photon passes through a narrow - band filter for spectral filtering . here , to achieve a good trade - off between better collection efficiency and higher multiphoton interference visibility , we choose 2 nm and 3 nm fwhm filters for e- and o - photons , respectively . experimental setup to generate the four - photon ghz state .
the abbreviations of the components are hwp , half - wave plate ; qwp , quarter - wave plate ; if , interference filter ; pbs , polarizing beam splitter ; pas , polarization analyzing system ; apd , avalanche photodiode detector .
the two epr sources generated two pairs of entangled photons 1 , 2 and 3 , 4 .
the hwps after the fiber couplers are used for polarization alignment .
the two e - photons are directed to overlap on the central pbs , which can project the input state onto the four - photon ghz state .
one of the fiber couplers is mounted on a translation stage to finely adjust the arriving time of the interfering photons .
the tiltable qwp after the pbs is used for tuning the relative phase between the two terms in the ghz state .
each photon passes though a narrow - band filter for spectral selection and then enters the final pas .
the inset shows the details of the pas , which consists of one motorized qwp , one motorized hwp , one pbs and two apds .
, scaledwidth=40.0% ] to characterize the observed state , we use both the entanglement witness and quantum state tomography ( qst ) . in the measurement part
, each photon is detected by a polarization analyzing system ( pas ) , which can measure it in any desired polarization basis with a high alignment accuracy .
the entanglement witness for the four - photon ghz state @xcite is @xmath10 where @xmath11 and @xmath12^{\otimes4 } , ( k=0,1,2,3)$ ] are local measurement operators . in order to suppress the higher - order photon pair emission noise
, we use a very low pump power of 20 mw and still obtain a moderate fourfold coincidence counting rate of 0.42 per second .
the measurement result is shown in fig .
3 , which yields @xmath13 , thus the state fidelity can be deduced as @xmath14 , where @xmath15 is the density matrix of the experimentally prepared state . on the other hand , we use the over - complete qst @xcite to estimate the observed state , which needs 81 joint measurement settings .
the total measurement time is over 6 hours .
then we use the maximum likelihood approach to reconstruct the density matrix @xmath15 .
figure 4 shows the real and imaginary parts of @xmath15 .
the state fidelity is calculated to be @xmath16 , which agrees with the above result very well .
measurement result of the witness .
( a ) fourfold coincidence counts measured in the h / v basis .
the data collection time is 10,000 s. the expectation value of a is just equal to the proportion of the two correct terms , which is @xmath17 .
( b ) the expectation values of @xmath18 , which yield an average value of @xmath19 .
the data collection time is 2000 s for each measurement settings .
, scaledwidth=50.0% ] the real ( left ) and imaginary ( right ) parts of the density matrix @xmath15 , which is reconstructed by the maximum likelihood method from the recorded data .
the height of the largest incorrect bar in the two pictures is only 0.03.,scaledwidth=50.0% ]
our setup can also be used to demonstrate entanglement swapping , the realization of which has rather profound implications .
first , it is the key process in the quantum repeater @xcite , which requires exchanging entanglement between different photon pairs .
second , in the distributed quantum computation @xcite , entanglement swapping allows us to connect distant nodes of a large network .
furthermore , it can also be viewed as teleportation of a genuinely unknown state @xcite .
the principle of entanglement swapping can be understood by rewriting the product state in eq .
( [ eq : inputstate ] ) in the following way @xmath20 , @xmath21 are bell states .
when we project the two independent photons 2 and 3 onto one of the four bell states , the other two photons 1 and 4 will simultaneously collapse to the same bell state , although they have never been able to interact with each other . in practice , the bell state measurement can be realized by a pbs with two pass in the output ports .
the pass are set to measure polarizations in the @xmath22 basis . for @xmath23 ,
the measurement result will be @xmath24 or @xmath25 ; for @xmath26 , the measurement result will be @xmath27 or @xmath28 .
for @xmath29 and @xmath30 , the two photons end up at the same detector and with identical polarizations due to the hom interference .
thus , we can only discriminate @xmath23 and @xmath26 here .
if we want to discriminate the other two bell states , we can insert a @xmath2 half - wave plate ( hwp ) in one of the input port of the pbs .
since we have used the over - complete set of bases in qst to characterize the ghz state as described hereinbefore , we can also use these data to demonstrate entanglement swapping , e.g. , by picking out the data that photons 2 and 3 are measured in the + /-
figure 5 shows the tomographic result for photons 1 and 4 when the photons 2 and 3 are projected onto @xmath23 or @xmath26 .
we can calculate the average swapping fidelity to be @xmath31 , which , to our best knowledge , is the highest value ever reported .
here we just assume that the input epr states are perfect , and calculate the fidelity between the state of photons 1 , 4 and the corresponding bell state .
real ( left ) and imaginary ( right ) parts of the reconstructed density matrix of photons 1 , 4 .
( a ) , ( b ) the state of photons 2 , 3 are projected on @xmath23 , the fidelity of the state of photons 1 , 4 corresponding to @xmath23 is @xmath32 .
( c ) , ( d ) the photons 2 and 3 are projected on @xmath26 , the fidelity of the state of photons 1 , 4 corresponding to @xmath26 is @xmath33 .
, scaledwidth=50.0% ]
high - fidelity multipartite entangled states are not only important for various quantum information protocols , but also useful in the fundamental test of quantum mechanics . here , using the observed four - photon ghz state , we experimentally test a novel hardy - like inequality introduced in @xcite , which can detect genuine multipartite nonlocality in the nonsignaling scenario .
the inequality has some significant features .
first , compared with the standard svetlichny inequality @xcite , it needs only 2n joint measurement settings for n parties , which reduces much experimental effort ( the svetlichny inequality requires @xmath34 joint measurement settings ) .
second , it can prove that all entangled permutation symmetric pure states are genuine multipartite nonlocal . in the following
, we can see that a high state - fidelity is necessary to violate the inequality .
.[table:1 ] measurement results of the hardy - like inequality .
the middle column lists the theoretical expectation values of the ideal ghz state , for each measurement setting . in order to obtain a small error bar , the data collection time is 8 hours for each of the first two settings and 4 hours for each of the other settings , and we get totally 58,000 fourfold coincidence counts .
the final result shows a violation of the inequality by more than 4 standard deviations .
note that all the raw data in each measurement settings have been corrected for the different detection efficiencies of the two apds in each pas .
the error bars are deduced from the raw data and poisson counting statistics . [ cols="^,^,^",options="header " , ] for @xmath35 case , the inequality has the following form : @xmath36 here @xmath37 and @xmath38 are binary valued observables ( with outcomes 0 and 1 ) for the i - th party , and @xmath39 has opposite outcomes compared with @xmath38 . in our case , to demonstrate the genuine multipartite nonlocality of the ghz state by violating this inequality , we need to find suitable dichotomous projection operators @xmath37 and @xmath38 for the i - th party , and @xmath40 is orthogonal to @xmath38 ( @xmath41 ) . to do this ,
we look for the projection measurements by numerical search . for simplicity
, we confine all the projection measurements in the x - z plane of the bloch sphere and assume @xmath42 , @xmath43 due to the symmetry of the inequality .
the projection angles of the operators are @xmath44 e.g. , @xmath45 and @xmath46 . using such measurement settings
, we can calculate that the fidelity of the four - photon ghz state should be higher than @xmath47 in order to violate the inequality , assuming the noise to be white noise .
the measurement result is shown in table i. the overall data collection time is over 40 hours .
our result shows a violation of the inequality by more than 4 standard deviations , which demonstrates the genuine multipartite nonlocality of the observed state using only 8 joint measurement settings .
there are three main imperfections in our system which limit the quality of the observed ghz state .
one is the imperfection of our epr source , the fidelity of which is about 0.99 .
the second is the higher - order photon pairs emission noise in the spdc process , which can be characterized by the measurement in the h / v basis ( fig .
3(a ) ) , because this measurement has nothing to do with the interference visibilities in our system , either the hom interference or the biphoton entanglement interference . by using a rather low pump power ( 20 mw ) to generate the ghz state , the error rate of the h / v measurement is reduced to @xmath48 .
note that all polarization elements in our system ( including pbss , wave plates , polarization - maintaining of the fibers , etc . )
have rather high extinction ratios @xmath49 , which is negligible comparing with other noises .
the last noise is the distinguishability between two independent photons , especially the temporal mismatch induced by the pump pulse duration . as depicted in @xcite , the time jitter of the interfering photons
should be much shorter than their coherence length , which can be easily understood because the larger the overlapped area of the photon wave packets is , the higher interference visibility we will observe . in order to increase the visibility , on one hand , we use narrow - band filters ( 2 nm ) to stretch the coherence length of the interfering photons ; on the other hand , we finely adjust the mode - locked laser to minimize the pump pulse duration to 90 fs . with these measures ,
we observe a four - photon hom interference with a visibility of @xmath50 . in summary
, we experimentally generate an ultrahigh - fidelity four - photon ghz state , and demonstrate a high performance entanglement swapping operation with the same setup .
we further demonstrate the genuine multipartite nonlocality of the ghz state in the nonsignaling senario , by testing a hard - to - violate hardy - like inequality .
these results greatly reduce the large fidelity - gap between previous biphoton and multiphoton experiments
. our setup would also be useful in implementing other multi - particle quantum information tasks with high quality .
national natural science foundation of china ( no . 61327901 , no .
61490711 , no . 11274289 , no .
11325419 , no . 61225025 , no .
11474268 , no . 11374288 , no .
11504253 ) ; strategic priority research program ( b ) of the chinese academy of sciences ( grant no .
xdb01030300 ) ; national youth top talent support program of national high - level personnel of special support program ( no .
bb2470000005 ) ; fundamental research funds for the central universities ( no .
wk2470000018 , no .
wk2470000022 ) . here
we show how to build up the sandwichlike epr source .
this new epr source has a sandwichlike structure : two beamlike - phase - matching type - ii bbo crystals with a true - zero - order half - wave plate ( thwp ) between them .
the two bbo crystals have the same cutting angles for beamlike spdc emission and are placed in the same way , so they both produce the product state @xmath51 .
the thwp in the middle works at the wavelength of the down - conversion photons and is rotated to @xmath2 , so the polarization states of the idler and signal photons from the first bbo are exchanged through it . when the two possible ways of generating photon pairs ( through the first or the second bbo ) are further made indistinguishable by spatial and temporal compensations ,
the photon pairs are prepared in the polarization entangled state @xmath52 .
the main advantage of the sandwichlike structure compared to the previous beamlike source @xcite is that the photons in the same path have the same ordinary ( extraordinary ) spectral property .
thus it has the same spirit with the entanglement concentration " scheme @xcite , which can erase the additional timing information presented in the ultrafast pump pulse and lead to an entanglement source independent of spectral filtering , crystal thickness , and pump bandwidth .
furthermore , our source does nt need the hom interferometer , which is critical in the original concentration scheme .
the experimental setup of the sandwichlike source is shown in fig .
1 . the pump laser comes from the second - harmonic generation ( shg ) of a mode - locked ti : sapphire laser ( with a duration of 140 fs , a repetition rate of 76mhz , and a central wavelength of 780 nm ) , which is not shown in fig .
1 , for simplicity .
the shg is done by focusing the 780 nm ti : sapphire laser beam onto a 1.5-mm - thick type - i collinear phase - matching bbo crystal , with a 150-mm - focal - length plano - convex lens .
the generated 390 nm laser beam is then collimated by a plano - convex lens with 75 mm focal length .
the lens l1 of 150 mm focal length is used to focus the collimated pump beam onto the sandwichlike strategy for down - conversion .
the two bbo crystals are both 1-mm - thick and cut at @xmath53 , @xmath54 .
the thwp is inserted between the two bbos , and sticked with them , using very thin ( 0.05 [email protected] mm thick ) and narrow glass stripes and optic glue between them . the glass stripes ( together with optic glue ) are placed along the edges of the bbo , i.e. , outside their clear aperture , to avoid optical damage by strong ultraviolet pump laser .
finally , the gap between the thwp and each bbo is estimated to be in the range from 0.04 mm to 0.08 mm , due to the uncertain thickness of the optical glue .
the two l2 lenses of 125 mm focal length are used to collimate the down - converted beams . note that the optimal collimating positions of the l2 lenses are not the same for the two bbo crystals due to the gap between them . to make the gap as small as possible we use the thwp and bound the three crystals together with very thin glass stripes , which can benefit the collection efficiency . even with these measures ,
the difference positions for l2 still exist and in general we choose the intermediate position , which will lead to a @xmath56 decrease of the collection efficiency for both bbo crystals .
the down - converted photons are finally coupled into single - mode fibers with aspheric lenses ( thorlabs , model : f220fc - b ) .
the coupling efficiency is about @xmath57 in our system , and the total photon pair collection efficiency is @xmath58 , when 3 nm and 8 nm fwhm filters are used for e- and o - photons respectively .
the different filtering bandwidths for e- and o - photons are due to the asymmetric generation bandwidth of them .
the 3 nm bandwidth is the usually chosen bandwidth for multi - photon interferences .
and we determine the 8 nm bandwidth by combining two different bandpass filters ( semrock , ld01 - 785/10 and tbp01 - 800/12 ) and tilting them to form a bandwidth - tunable filter , then watching the corresponding collection efficiency to get a best value .
compared with the traditional type - i and type - ii entanglement sources , the only additional requirement of our source is the spatial compensation , which is realized by two linbo@xmath4 crystals .
they are cut with the optic axes lying in the horizontal plane and being @xmath2 away from the incident beam , thus working as a beam - displacer .
the thicknesses of them are 1 mm and 3.2 mm for o- and e - photons , respectively . to check whether the compensation is accomplished , we can translate the fiber coupler ( together with the fiber ) along the horizontal direction perpendicular to the incident down - converted beams , to see whether there is only one peak value of the collected photon counting rates .
we can also use this method to determine the thickness of the linbo3 crystal , because the translation distance is proportional to the thickness of the crystal when the lenses are removed .
note that the linbo@xmath4 crystals should be placed before the collimating lenses l2 .
we use yvo@xmath59 crystals for temporal compensations .
the thicknesses are 0.6 mm and 0.42 mm for o- and e - photons respectively . in practice , due to the standard errors of thicknesses and distances in manufacturing the sandwichlike structures ,
the thicknesses of the yvo@xmath59 crystals may be slightly different in different cases .
but the biggest differences we observed are only about 0.03 mm , which corresponds to a @xmath60-order quartz compensation plate ( @xmath61780 nm ) .
29ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _
_ ( , ) link:\doibase 10.1103/physrevlett.86.5188 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.86.910 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.70.1895 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.180408 [ * * , ( ) ] link:\doibase 10.1103/physrevx.5.041052 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.250401 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.250402 [ * * , ( ) ] http://dx.doi.org/10.1038/nphoton.2014.218 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.86.4435 [ * * , ( ) ] @noop * * , ( ) http://dx.doi.org/10.1038/nphys700 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.260402 [ * * , ( ) ] link:\doibase 10.1364/ol.26.000843 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physreva.60.r773 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.75.4337 [ * * , ( ) ] link:\doibase 10.1103/physreva.68.013804 [ * * , ( ) ] link:\doibase 10.1364/ol.33.000968 [ * * , ( ) ] http://dx.doi.org/10.1038/35051009 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physreva.76.030305 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys1777 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.210504 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.83.33 [ * * , ( ) ] @noop ( ) link:\doibase 10.1103/physrevlett.112.140404 [ * * , ( ) ] link:\doibase 10.1103/physrevd.35.3066 [ * * , ( ) ] link:\doibase 10.1103/physreva.67.010301 [ * * , ( ) ]
|
high - quality entangled photon pairs generated via spontaneous parametric down - conversion have made great contributions to the modern quantum information science and the fundamental tests of quantum mechanics .
however , the quality of the entangled states decreases sharply when moving from biphoton to multiphoton experiments , mainly due to the lack of interactions between photons . here ,
for the first time , we generate a four - photon greenberger - horne - zeilinger state with a fidelity of @xmath0 , which is even comparable to the best fidelity of biphoton entangled states .
thus , it enables us to demonstrate an ultrahigh - fidelity entanglement swapping the key ingredient in various quantum information tasks .
our results push the fidelity of multiphoton entanglement generation to a new level and would be useful in some demanding tasks , e.g. , we successfully demonstrate the genuine multipartite nonlocality of the observed state in the nonsignaling scenario by violating a novel hardy - like inequality , which requires very high state - fidelity .
|
the picture emerging from many x - ray studies over the last couple of decades of active galactic nuclei ( agns ) is that emission features in the x - ray spectra present in low - luminosity objects ( seyfert 1 galaxies ) become scarce in agns with 210 kev intrinsic luminosity exceeding @xmath7 @xmath8 ( synonymously , quasars ) .
these trends are discussed at length in nandra _
( 1997a , 1998 ) and reeves _ et al . _
( 1997 ; and references therein ) .
the transition to a featureless x - ray power - law continuum ( except for possible line - of - sight absorption at low energies ) in the high luminosity agns , especially radio loud sources , is not fully understood but may be related to the complete ionization of matter responsible for emission - line features and/or beaming of the x - ray continuum swamping out emission - line features .
the radio - loud quasar pks 0637@xmath0752 ( @xmath9 , hunstead , murdoch & shobbrook 1978 ) has been observed by the _ einstein _
ipc ( wilkes & elvis 1987 ; zamorani _ et al .
_ 1981 , 1984 ; elvis and fabbiano 1984 ) , _ exosat _ ( lawson _ et al .
_ 1992 ; comastri _ et al .
_ 1992 ; saxton _ et al .
_ 1993 ) , and _ ginga _ ( williams _ et al .
_ 1992 ; lawson and turner 1997 ) .
these x - ray observations had poor energy resolution and yielded only continuum parameters for a simple power - law plus neutral absorber model . the most tightly constrained measurements of the slope were from _ ginga _ with a photon index of @xmath10 .
the column density , @xmath11 , has not been well - constrained , although lawson and turner ( 1997 ) claim a change from @xmath12 to @xmath13 ( intrinsic ) between two _ ginga _ observations .
the _ einstein _
observations were under - exposed and so fluxes were uncertain due to the poor spectral constraints .
nevertheless , the non - detection of pks 0637@xmath0752 in one mpc observation and a clear detection in another mpc observation is a clear indicator of a variable continuum ( elvis and fabbiano 1984 ) .
_ ginga _ still provided the most reliable 210 kev flux , @xmath14 corresponding to a luminosity of @xmath15 ( @xmath16 and @xmath17 throughout ) . in one _
ginga _ observation the detection of an fe - k emission line at @xmath18 kev in the quasar frame has been reported with equivalent width @xmath19 ev but the result was marginal ( see williams _ et al .
_ 1992 ; lawson and turner 1997 ) . in this _
letter _ we report on the results of a new x - ray observation of pks 0637@xmath0752 with _
( _ advanced satellite for cosmology and astrophysics _ ; see tanaka , inoue & holt , 1994 ) . with some surprise we have discovered an emission - line in the object centered at @xmath5 kev in the quasar frame .
what is even more surprising is that an emission - line at this energy has not been observed in _ any _ agn at all , irrespective of luminosity .
_ asca _ observed pks 0637@xmath0752 in 1996 november 1011 for a duration of @xmath20 ks .
since it is important to demonstrate that our main observational result in this paper , namely the discovery of an unidentified emission line , is not an instrumental artefact , we describe the data analysis in some detail .
the reader is referred to tanaka _
( 1994 ) for details of the instrumentation aboard _
asca_. the two solid state imaging spectrometers ( sis ) , hereafter sis0 and sis1 , with a bandpass of @xmath21 kev , were operated in 1-ccd faint and bright modes .
the two gas imaging scintillators ( gis ) , hereafter gis2 and gis3 , with a bandpass of @xmath22 kev , were operated in standard ph mode . the sis faint and bright mode data were combined and the sis energy scale was fixed using a prescription based on measurements of cas a ( dotani _ et al . _ 1997 ) .
this corrects for the continuing decline in the cti of the ccds with time and is based on interpolation or extrapolation of the cas a measurements .
the last available cas a measurements were made just three months before the pks 0637@xmath0752 observation and since the cti changes very slowly with time , the extrapolation results in a systematic uncertainty in the gain which is much less than @xmath23 at 1 kev ( dotani _ et al .
version 1.1 of the sis response matrix generator was used in the analysis .
however , no corrections were made for any possible offsets , fluctuations and distortion of the ccd dark level distribution ( the latter also known as the ` rdd ' effect ) .
such corrections ( which themselves are subject to uncertainty ) can only be applied to faint mode data and for 1-ccd mode their effects will only be noticeable for very high statistical quality data . considering the signal - to - noise of our data these effects can safely be neglected , since there are not enough photons to utilize the full energy resolution of the sis and the overall uncertainty in the energy scale is less than @xmath24 ev ( dotani _ et al .
more importantly , as a calibration control , we will examine data for two other agn observed by _
asca _ shortly before and shortly after pks 0637@xmath0752 ( 4 ) . for the purpose of assessing the spectral results below
, we note that spectral fits to gis data from the crab have systematics @xmath25 over the range 0.710 kev with the current calibrations ( fukuzawa , ishida , and ebisawa 1997 ) .
below @xmath26 kev the gis efficiency is greatly reduced and the calibration uncertain . in the analysis below we include the gis data down to 0.5 kev for the sake of continuity but this does not affect our results because the data are dominated by statistical errors at low energies . from @xmath26 kev to @xmath27 kev the sis calibration is generally consistent with the gis , and _ bepposax lecs _ and _ mecs _ to @xmath28 or better ( see grandi _ et al . _ 1997 ; orr _ et al .
below @xmath26 kev the uncertainty is larger , especially between @xmath29 kev .
however , our result is not sensitive to the absolute calibration at these energies and , in any case , we will compare the sis data for pks 0637@xmath0752 with data from other sources , taken shortly before and after the pks 0637@xmath0752 observation .
data were screened so that accepted events satisfied the following criteria : ( i ) data were taken outside the saa ; ( ii ) the time since or before passage through the saa or a satellite day / night transition is @xmath30 s ; ( iii ) the elevation angle to earth is @xmath31 ; ( iv ) the magnetic cut - off rigidity ( cor ) is @xmath32 ; the deviation of the satellite from the nominal pointing position was @xmath33 ; ( v ) the sis parameters measuring active ccd pixels registered @xmath34 active pixels per second , and ( vi ) the radiation - belt monitor registered @xmath35 ct / s .
hot and flickering pixels in the sis were removed .
screening resulted in net ` good ' exposure times in the range @xmath3649.1 ks for the four instruments .
images were accumulated from the screened data .
the _ asca _ images , as well as a _ pspc _ image obtained from archival data ( a 1.3 ks observation made in 1992 may 26 ) , were analysed for possible nearby contaminating sources .
no such sources were detected .
source events were extracted from circular regions with radii of 4 for the sis and 6 for the gis .
background events were extracted from off - source regions . as a check , background spectra were also made from a 1-ccd blank - sky observation which resulted from the non - detection of the quasar iras 15307 + 3252 ( observed 1994 july ) .
these background data were subject to identical selection criteria and used to check the invariance of the spectral results for pks 0637@xmath0752 described below .
the _ asca _ lightcurves do not show significant variability , the highest excess variance ( e.g. see nandra _ et al . _
1997b ) we obtain is from sis0 and has the value @xmath37 for 128 s bins .
this is to be compared with @xmath38 we obtain for the highly variable seyfert 1 mcg @xmath06@xmath030@xmath015 from a 4 day _ asca _ observation .
the count rates in the accumulated background - subtracted spectra ranged from @xmath390.41 ct / s .
spectra were binned to have a minimum of 20 counts per bin in order to utilize @xmath40 as the fit statistic .
we fitted spectra from the four _ asca _ instruments simultaneously in the range 0.59.5 kev with a simple power - law plus cold , neutral absorber model .
a total of two interesting parameters were involved ( the photon index , @xmath41 , and column density , @xmath11 ) , plus four independent instrument normalizations .
( the deviation of any of the four normalizations from their mean was less than 7% ) .
the results are shown in table 1 .
the best - fitting photon index , @xmath3 , is consistent with typical historical values ( 1 ) .
the column density , @xmath11 , is consistent with the galactic value of @xmath42 obtained from dickey and lockman ( 1990 ) .
the instrument - averaged _ observed _ 0.5 - 2 kev and 210 kev fluxes are 1.3 and @xmath43 respectively .
the 0.510 kev luminosity in the source frame is @xmath44 .
thus the source has dimmed by almost a factor of 3 relative to the _ ginga _ observations made in 1998 july 4 and 1990 august 23 ( lawson and turner 1997 )
. however , the flux is consistent ( within systematic uncertainty ) with an _ exosat _ observation made in 1984 september 8 ( comastri _ et al .
_ 1992 ) .
although the above simple continuum model provides an adequate fit , inspection of the ratios of data to model ( figure 1 ) reveal a statistically significant ` hump ' at @xmath45 kev ( observed frame ) in both sis0 and sis1 .
the gis data are ambiguous since the effective area is falling rapidly and the energy resolution is worse than the sis .
we repeated the four - instrument fits with the addition of a simple gaussian to model the apparent emission - line feature .
three extra parameters were involved , namely the line center energy , @xmath46 ; the gaussian intrinsic width , @xmath47 , and the line intensity , @xmath48 . _ note that we will refer to all three parameters , plus the equivalent width ( ew ) , in the quasar frame_. initially we allowed all three parameters to float and the results are shown in table 1 .
the effect on the continuum parameters is negligible ( @xmath49 in both @xmath41 and @xmath11 ) .
it can be seen that the line is not resolved .
the results from repeating the fit with @xmath47 fixed at 0.01 kev ( narrow line ) are also given ; the difference in @xmath40 is completely negligible ( 0.1 ) so we use this fit to derive bounds on @xmath46 and ew .
note that the decrease in @xmath40 compared to the model without an emission line is @xmath50 for the addition of two free parameters so the line feature is detected at a confidence level much greater than 99% .
it is appropriate here to use a @xmath51 criterion to derive 90% confidence errors on the best - fitting line parameters since we want to know the bounds on @xmath46 and ew , independently of @xmath41 and @xmath11 ( e.g. see yaqoob 1998 ) .
we obtain @xmath52 kev and @xmath53 ev .
we also searched for fe - k line emission by adding a gaussian with center energy fixed at various values in the range 6.46.97 kev ( with the low - energy line removed ) .
neither a narrow ( @xmath54 ) or broad ( @xmath55 ) gaussian component was statistically required , the 90% one - parameter upper limits on the ew being 41 and 82 ev respectively .
all the above spectral results were repeated with the alternative background spectra made from a blank field , as described in 2 , and the results were confirmed to be robust to the background used .
since an emission - line feature like that reported above has never before been observed in a quasar ( or indeed any active galaxy ) , we sought explanations which did not require the line to be intrinsic to the quasar . to check whether the sis calibration is responsible we examined sis data for 3c 371 ( a quasar with @xmath56 , observed five days before pks 0637@xmath0752 ) and for pictor a ( a radio galaxy with @xmath57 , observed fifteen days after @xmath58 see also eracleous and halpern , 1998 ) .
figure 2 shows the ratios of data to a simple power - law plus absorber model ( both sis0 and sis1 ) for all three sources fitted over the 0.59.5 kev range , but zoomed in the range 0.51.5 kev ( observed frame ) .
it can be seen that the emission - line feature in pks 0637@xmath0752 is not present in 3c 371 or pictor a. we have already stated in 2 that there are no contaminating sources in the _ asca _ or _ pspc _ images . the emission line can not be a background feature ( for example from galactic diffuse emission ) because the background from nearby regions in the same field and at the observed energy of the emission line in the sis lies at least a factor 12 below the on - source spectrum .
it is unlikely that the emission line is produced in the intervening galaxy at @xmath59 ( elvis and fabbiano 1984 ) since the galaxy would have to be extremely luminous ( almost as luminous as the quasar is thought to be ) .
we also examined the _
pspc _ spectra .
these data show residuals to a power - law model which suggest the presence of a line like that observed by _
asca_. however , the pspc data do not allow the presence of the line to be statistically confirmed , as the energy resolution is extremely poor and signal - to - noise is very low because the exposure time is only 1.3 ks .
we conclude that the emission line at an energy of 1.60 kev in the rest - frame of pks 0637@xmath0752 is not an instrumental artifact and is almost certainly associated with the quasar itself .
we have reported the results of a new x - ray observation of pks 0637@xmath0752 with _
asca_. although the luminosity had diminished by a factor of @xmath2 compared to previous _
ginga _ observations ( six and eight years earlier ) , the _ asca _ data are the most sensitive to date in the 0.510 kev band ( @xmath60 kev in the quasar frame ) . however , the _ asca _ luminosity is consistent with an _ exosat _ observation made twelve years earlier .
it is possible that the larger field of view of _ ginga _ was confused with another source which may have contributed to the higher luminosity so the apparent variability should be treated with caution .
we find the continuum to be well modeled by a simple power - law with @xmath3 and galactic absorption .
we find no evidence for fe - k emission although the data allow a line with moderate equivalent width ( 90% confidence quasar - frame upper limits in the range 4182 ev , corresponding to line intrinsic widths in the range 00.5 kev respectively ) . our most important result from the _ asca _ observation is the discovery of a narrow emission line at an energy @xmath5 kev .
we have presented strong evidence that this is not an instrumental artefact and is likely to be associated with pks 0637@xmath0752 itself . with conservative assumptions for systematic errors ,
we find the line center to lie in the range 1.481.72 kev and to have an ew of @xmath6 ev in the quasar frame .
line emission in this energy band has not been observed in any other agn to date , and its identification is not at all obvious .
the allowed range of energies encompasses transitions from mgxi
xii , fexxii xxvi and nixix xxviii ( assuming no bulk motion ) .
highly ionized gas is commonly seen in absorption in low luminosity seyfert 1 galaxies ( e.g. reynolds 1997 ; george _ et al .
_ 1998 ) and in emission in some seyfert 1 ( below ) and many seyfert 2 galaxies ( e.g. turner _ et al .
. however theoretical models of both optically - thick and optically - thin gas predict the stronger lines elsewhere in the spectrum , particularly due to a blend of o , ne and fe lines @xmath61 kev ( e.g. see zycki _
_ 1994 ; kallman _ et al .
_ 1996 ; netzer 1996 , and references therein ) . indeed
the presence of emission lines at such energies was suggested in _ einstein _
sss observations ( e.g. turner _ et al .
_ 1991 ) and have been detected in the _ asca _ spectra of ngc 4151 ( figure 1 in yaqoob _
1995 ) , pg 1244 + 026 ( blend at @xmath45 kev , fiore _
1998 ) , and ton s 180 ( blend at @xmath62 kev , turner , george , nandra 1998 ) .
the detection of ovii(0.57 kev ) emission has been claimed in the case the _ asca _ data from ngc 3783 ( george , turner , netzer 1995 ) , mkn 290 ( turner _ et al .
_ 1996 ) , and ic 4329a ( cappi _ et al .
_ 1996 ) , although uncertainties in the calibration @xmath63 kev make their reality less compelling . however , none of these lines are consistent with the observational result for pks 0637@xmath0752 without an assumption of bulk in / outflow . the emission feature closest in energy to that reported in pks 0637@xmath0752 and which has also been observed in _ asca _ data from both seyfert 2 galaxies ( turner _ et al .
_ 1997 ) and starburst galaxies ( e.g. ptak _ et al .
_ 1997 ) is the sixiii(1.841.86 kev ) blend .
however , the identification of the line observed in pks 0637@xmath0752 with such emission not only requires an anomalous si abundance relative to o , ne , s and fe , but redshifting of the new emission - line energy by @xmath640.30 ( i.e. the line - emitting matter must be _ in_falling at velocities @xmath65 ) with respect to the optical emission line gas which has @xmath9 ( hunstead _ et al .
future observations of this intriguing x - ray emission - line feature at high spectral resolution with _ axaf _ , _ xmm _ and _ astro - e _ , will clearly help to shed light on its identity and origin .
the authors thank the _ asca _ mission operations team at isas , japan , and all the instrument teams for their dedication and hard work in making these _ asca _ observations possible .
this research made use of the heasarc archives at the laboratory for high energy astrophysics , nasa / gsfc .
lccc @xmath41 & @xmath66 & @xmath67 & @xmath67 @xmath11 ( @xmath68 ) & @xmath69 & @xmath70 & @xmath71 @xmath47 ( kev ) & & @xmath72 & 0.01 ( fixed ) @xmath46 ( kev ) & & @xmath73 & @xmath74 @xmath48 & & @xmath75 & @xmath76 ew ( ev ) & & @xmath77 & @xmath78 @xmath40 & 568.5 & 553.8 & 553.9 degrees of freedom & 530 & 527 & 528 cappi , m. , mihara , t. , matsuoka , m. , hayashida , k. , weaver , k.a .
, & otani , c. 1996 , , 458 , 149 comastri , a. , setti , g. , zamorani , g. , elvis , m. , giommi , p. , wilkes , b. , & mcdowell , j. 1992 , , 384 , 62 dickey , j.m .
, & lockman , f.j .
1990 , ann . rev
dotani , t. , _ et al .
_ 1997 , _ ascanews _ , 5 , 14 eracleous , m. , & halpern , j. 1998 , , in press elvis , m. , & fabbiano , g. 1984 , , 280 , 91 fiore , f. , _ et al . _ 1998 , , 298 , 103 fukazawa , y. , ishida , m. , & ebisawa , k. 1997 , _ ascanews _ , 5 , 3 george , i.m . ,
turner , t.j . , & netzer , h. 1995 , , 438 , l67 george , i.m . ,
turner , t.j .
, netzer , h. , nandra , k. , mushotzky , r.f . , & yaqoob , t. 1998 , , 114 , 73 grandi , p. , _ et al .
_ 1997 , , 325 , l17 hunstead , r.w .
, murdoch , h.s . , & shobbrook , r.r .
1978 , , 185 , 149 kallamn , t.r . , liedahl , d. , osterheld , a. , goldstein , w. , & kahn , s. 1996 , , 465 , 994 lawson , a.j . ,
turner , m.j.l . ,
williams , o.r . , stewart , g.c . , & saxton , r.d .
1992 , , 259 , 743 lawson , a.j . , & turner , m.j.l .
1997 , , 288 , 920 nandra , k. , mushotzky , r.f . ,
george , i.m . ,
turner , t.j . , & yaqoob , t. 1997a , , 488 , 91 nandra , k. , george , i.m . , mushotzky , r.f . ,
turner , t.j . , &
yaqoob , t. 1997b , , 476 , 70 nandra , k. , george , i.m . ,
mushotzky , r.f . ,
turner , t.j .
, & yaqoob , t. 1998 , in preparation netzer , h. 1996 , , 473 , 781 orr , a. , molendi , s. , fiore , f. , grandi , p. , parmar , a.n . , & owens , a. 1997 , , 324 , l77 ptak , a. , serlemitsos , p. , yaqoob , t. , mushotzky , r.f . , & tsuru , t. 1997 , , 113 , 1286 reeves , j.n . ,
turner , m.j.l . , ohashi , t. , & kii , t. 1997 , , 292 , 468 reynolds , c.s .
1997 , , 286 , 513 saxton , r.d . ,
turner , m.j.l . ,
williams , o.r . , stewart , g.c . , ohashi , t. , & kii , t. 1993 , , 262 , 635 tanaka , y , inoue , h. , & holt , s.s .
1994 , , 46 , l37 turner , t.j . ,
george , i.m .
, nandra , k. 1991 , , in press turner , t.j . , weaver , k.a . , mushotzky , r.f . ,
holt , s.s . , & madejski , g.m .
1991 , , 381 , 85 turner , t.j . ,
george , i.m .
, kallman , t. , yaqoob , t. , & zycki , p.t .
1996 , , 472 , 571 turner , t.j . ,
george , i.m . ,
nandra , k. , & mushotzky , r.f .
1997 , , 113 , 23 wilkes , b.j . , & elvis , m. 1987 , , 323 , 243 williams , o.r . , _
_ 1992 , , 259 , 743 yaqoob , t. , edelson , r. , weaver , k. a. , warwick , r. s. , mushotzky , r. f. , serlemitsos , p. j. , & holt , s. s. 1995 , apj , 453 , l81 yaqoob , t. 1998 , , 500 , 893 zamorani , g. , _ et al .
_ 1981 , , 245 , 357 zamorani , g. , giommi , p. , maccacaro , t. , & tanabaum , h. 1984 , , 278 , 28 zycki , p.t . ,
krolik , j.h . ,
zdziarski , a.a . , & kallman , t.r .
1994 , , 437 , 597
* figure 1 * + ratio of the _ asca _ data for pks 0637@xmath0752 ( @xmath9 ) to the best - fitting power - law plus absorber model .
notice the significant residuals in the sis0 and sis1 data , corresponding to @xmath79 kev in the quasar frame ( see text ) .
* figure 2 * + ratio of sis data to best - fitting power - law plus absorber models for 3c 371 ( @xmath56 ) , pks 0637@xmath0752 ( @xmath9 ) and pictor a ( @xmath57 ) . the 3c 371 were taken only five days before the observation of pks 0637@xmath0752 while the pictor a data were taken fifteen days after pks 0637@xmath0752 .
thus calibration uncertainties in the sis can not account for the emission - line feature seen in pks 0637@xmath0752 .
note that the best - fitting parameters for 3c 371 are @xmath80 and @xmath81 and for pictor a they are @xmath82 and @xmath83 ( errors correspond to @xmath51 ) .
|
we report the results from an _ asca _ observation of the high - luminosity , radio - loud quasar pks 0637@xmath0752 ( redshift 0.654 ) , covering the 0.815 kev band in the quasar - frame .
we find the source to have a luminosity @xmath1 in the 210 kev band , a factor of @xmath2 lower than during a previous _ ginga _ observation .
the continuum appears to be well modeled by a simple power - law with @xmath3 , with no evidence for absorption by material intrinsic to the quasar , or fe - k emission ( with an equivalent width @xmath4 ev at 90% confidence ) .
however we do find evidence for a narrow emission line at an energy @xmath5 kev and equivalent width @xmath6 ev ( both in the quasar frame ) .
line emission at these energies has not been observed in any other active galaxy or quasar to date .
we reject the possibility that this line is the result of instrumental artifacts , and briefly explore possible identifications . _ accepted july 1998 for publication in astrophysical journal letters _
|
the first generation of kilometer - scale gravitational wave ( gw ) laser interferometric detectors with sensitivity in the frequency band 10 hz to few khz and ultra cryogenic bar detectors sensitive at frequencies around 1 khz will start collecting data soon .
the tama 300 ( tsubona 1995 ) has already done the first large scale data acquisition ( tagoshi et al .
2001 ) , while ligo ( abramovici et al . 1992 ) and geo600 ( danzmann 1995 ) have recently carried out their first science observations .
virgo ( bradaschia et al . 1991 ) may become operational in couple of years . also , an eighty meter research interferometer aciga ( mccleland et al . 2000 ) near perth , australia is under construction , hoping that it may be possible to extend it to multi - kilometer scale in the future . at present
, majority of searches are focussed in the detection of _ chirp _ and burst signals .
however interest for the search of continuous gravitational wave ( cgw ) signals from the output of detectors is growing ( jaranwoski , et .
al 1998 , jaranowski & krlak 1999 , 2000 ; astone et al . 2002 ; brady et .
1998 , brady & creighton 2000 ) due to the possibility of enhancing the signal - to - noise ratio ( s / n ) by square root of observation time @xmath1 . an optimistic estimate suggests that the earth based laser interferometric detectors may detect such signals with an observation time of 1-yr .
the strength of the cgw largely depend on the degree of long - lived asymmetry in the source .
there are several mechanism for producing such an asymmetry ( pandharipande et al . 1976 ; bonazzola & gourgoulhon 1996 ; zimmermann & szedenits 1979 ; zimmermann 1980 ) .
the estimates of the asymmetry in neutron stars shows that the amplitude of cgw may be @xmath2 .
hence , long integration will be required to get the signature of signal .
but this in turn induces several other problems viz .
doppler modulation and non - stationarity of the detector noise .
consequently data analysis becomes more harder .
however , doppler modulation will provide the information of position of the source in the sky .
it has been realized that with present computing power , coherent all - sky full frequency search in the bandwidth ( bw ) of the detector of a month long data is computationally prohibitive .
some computationally efficient alternative approaches have been suggested viz .
_ tracking _ and _ stacking _ ( brady and creighton 2000 ; schutz 1998 ) . _
tracking _ involves the tracking of lines in the time - frequency plane built from the fourier transform ( ft ) of one day long stretches of data while _ stacking _ involves dividing the data into day long stretches , searching each stretch for signals , and enhancing the detectability by incoherently summing the ft of data stretches .
also , accurate modeling of gw form , optimal data processing and efficient programming are an integral part of all sky - search . the basic method to analyze the detector output to get the signature of gw signals rely on how efficiently one
can fourier analyze the data .
fourier analysis of the data has the advantage of incorporating the interferometers noise spectral density .
the problem for the search of cgw largely depend on how accurately one can take into account the translatory motion of the detector acquired from the motions of the earth in solar system barycentre ( ssb ) frame .
it has been shown ( srivastava and sahay 2002a , b ) that amplitude modulation will only redistribute the power of the frequency modulated ( fm ) signal in five frequency bands @xmath3 , @xmath4 , @xmath5 , where @xmath4 and @xmath6 are the frequencies of the fm signal and the rotational frequency of the earth respectively .
hence it is sufficient to consider only fm signal for the analysis of the matching of signals from different locations in the sky .
the study of the matching of signals from locations in the sky that differ in their co - latitude and longitude by @xmath0 radians has been made for 1-d data set ( srivastava and sahay 2002c ) .
they observed symmetries in the sky locations . in this paper
we report that the observed symmetries are not generic .
we study this problem for longer observation time and the other parameters which affect the symmetries . in the next section
we briefly review the ft of fm signal . in section 3 , using the concept of fitting factor ( ff ) , we investigate the matching of signals in an all sky - search with reference to the earth based laser interferometric detector by considering the source location as the parameter of signal manifold and templates corresponding to different source locations for longer observation time .
we present our conclusions in section 4 .
the time dependence of the phase of a monochromatic cgw signal of frequency @xmath7 observed at detector location is given as ( srivastava and sahay 2002a ) @xmath8\nonumber \\ & = & 2\pi f_o t + { \cal z}\cos ( a\xi_{rot } - \phi ) + { \cal n}\cos ( \xi_{rot } - \delta ) -
{ \cal r } - { \cal q } \label{eq : phit}\end{aligned}\ ] ] where @xmath9 @xmath10 where @xmath11 , @xmath12 , @xmath13 and @xmath14 represent respectively the earth s radius , the average distance between the centre of earth from the origin of ssb frame , the rotational and the orbital angular velocity of the earth . @xmath15 and
@xmath16 represent the obliquity of the ecliptic and the velocity of light .
@xmath17 is the colatitude of the detector .
here @xmath18 represents the time in s elapsed from the instant the sun is at the vernal equinox and @xmath19 is the local sidereal time at that instant , expressed in radians .
@xmath20 and @xmath21 denote the celestial colatitude and celestial longitude of the source .
these coordinates are related to the right ascension , @xmath22 and the declination , @xmath23 of the source via @xmath24 the two polarisation states of the signal can be taken as @xmath25 \label{eq : hpt}\ ] ] @xmath26 \label{eq : hct}\ ] ] @xmath27 , @xmath28 are the time independent amplitude of @xmath29 , and @xmath30 respectively . + for the analysis of the matching of signals from different locations in the sky , it is sufficient to consider either of the polarisation states of fm signal .
we consider the @xmath31 polarisation of the signal of unit amplitude .
therefore , the ft for a data of @xmath32 observation time may be expressed as ( srivastava and sahay 2002b ) @xmath33e^{-i2\pi ft}dt \nonumber \\ & & \simeq \frac{\nu}{2 w_{rot } } \sum_{k = - \infty}^{k = \infty } \sum_{m = - \infty}^{m = \infty } e^ { i { \cal a}}{\cal b } [ \tilde{{\cal c } } - i\tilde{{\cal d } } ] \ ; ; \ ; \label{eq : hf}\end{aligned}\ ] ] where @xmath34 j stands for the bessel function of first kind . the computational strain to compute @xmath35 from equation ( [ eq : hf ] ) can be reduced by @xmath36 by using the symmetrical property of the bessel functions , given as @xmath37\ ; + \nonumber \\ & & j_o ( { \cal z})\sum_{m = 1}^{m = \infty } { j_m({\cal n})\over \nu^2 - m^2 } \left [ ( { \cal y } { \cal u } - { \cal x } { \cal v } ) - i ( { \cal x } { \cal u } + { \cal y } { \cal v } ) \right]\ ; + \nonumber \\ & & \left.\sum_{k = 1 } ^{k = \infty}\sum_{m = - \infty}^{m = \infty } e^ { i { \cal a}}{\cal b}\left ( \tilde{{\cal c } } - i\tilde{{\cal d } } \right)\right]\ ; ; \label{eq : fm_code}\end{aligned}\ ] ] @xmath38 equation ( [ eq : fm_code ] ) contains double infinite series of bessel function .
however , we know that the value of bessel function decreases rapidly as its order exceeds the argument . from equations ( [ eq : pq ] ) , it can be shown that the order of bessel function required to compute @xmath35 in the infinite series are @xmath39 @xmath40
matched filtering is the most suitable technique for the detection of signals whose wave form is known .
the wave forms are used to construct a bank of templates , which represent the expected signal wave form with all possible ranges of its parameters ( source location , ellipticity , etc . ) . in an all sky search by match filtering ,
srivastava and sahay 2002c observed symmetries under the following transformations @xmath41 @xmath42 @xmath43 the above observations are made by studying few cases for 1-d data set .
hence , it will be important to understand and check the generic nature of the matching of signals under the above transformations . to study this problem we use the formula for ff ( apostolatos 1995 ) which quantitatively describes the closeness of two signals , given as @xmath44 where @xmath45 and @xmath46 represent respectively the fts of the actual signal wave form and the templates .
the inner product of two waveform @xmath47 and @xmath48 is defined as @xmath49 where @xmath50 denotes complex conjugation , @xmath51 denotes the ft of the quantity underneath @xmath52 and @xmath53 is the spectral density of the detector noise . in our analysis
we assume the noise to be stationary and gaussian .
it is remarked that to compute the inner product of the signals one would require the bw of the doppler modulated signal . from the doppler shift ( srivastava and sahay 2002a ) ,
the bw of doppler modulated signal may be given as @xmath54 let us consider that geo600 detector ( the position and orientation of the detectors can be found in jaranowski et al .
1998 ) receive a cgw signal of frequency @xmath55 hz ( unreasonably low frequency has been chosen for illustrative purposes limited by available computational power ) from a source located at @xmath56 . in order to evaluate the matching of the signals in colatitude , we first maximize ff over @xmath21 by choosing @xmath57 .
now , we wish to check the symmetries in colatitude represented by equation ( [ eq : sym_theta ] ) for the data set @xmath58 d. for the purpose we maximize the ff over @xmath20 by varying @xmath59 in discrete steps over entire range i.e. @xmath60 to @xmath61 . for the present case it is sufficient to take the ranges of @xmath62 and @xmath63 as @xmath64 to @xmath65 and @xmath66 to @xmath67 respectively and @xmath68 .
the results so obtained are shown in figure ( [ fig : theta_var ] ) . to establish the observed symmetries ,
we similarly compute the ff for @xmath69 d and observe that the matching of signals remains almost same . however ,
due to obliquity of the ecliptic , the variation in the matching of signals will depend on source frequency , colatitude and detector position and orientation .
we check the dependence of the ff on this parameters .
the result so obtained for different earth based laser interferometric detectors are shown in the tables ( [ table : theta_var_ff ] ) , & ( [ table : coeff_theta ] ) and figures ( [ fig : freq_beta0 ] ) & ( [ fig : freq_beta90 ] ) .
@xmath70 & @xmath71 & ff & ff & & & @xmath70 & @xmath71 & ff & ff + & & ( @xmath72)&(@xmath73 ) & & & & & ( @xmath72)&(@xmath73 ) + 0.5 & 179.5 & 0.9999 & 0.9970 & & & 45 & 135 & 0.9992&0.9939 + 1 & 179 & 0.9999 & 0.9970 & & & 50 & 130 & 0.9993 & 0.9933 + 5 & 175 & 0.9992 & 0.9970 & & & 55 & 125 & 0.9995 & 0.9985 + 10 & 170 & 0.9985 & 0.9968 & & & 60 & 120 & 0.9996 & 0.9988 + 15 & 165 & 0.9986 & 0.9966 & & & 65 & 115 & 0.9997&0.9992 + 20 & 160 & 0.9987 & 0.9963 & & & 70 & 110 & 0.9998&0.9994 + 25 & 155 & 0.9987 & 0.9959 & & & 75 & 105 & 0.9998 & 0.9997 + 30 & 150 & 0.9988 & 0.9954 & & & 80 & 100 & 0.9999 & 0.9998 + 35 & 145 & 0.9990 & 0.9949 & & & 85 & 95 & 0.9999 & 0.9999 + 40 & 140 & 0.9991 & 0.9944 & & & 89 & 91 & 0.9999 & 0.9999 + .coefficients of the fall of ff with @xmath7 under the transformation represented by equation ( [ eq : sym_theta ] ) for @xmath72 and @xmath74 . [ cols="<,^,^,^,^ " , ]
in view of blind all sky search for cgw , we have studied the matching of signals from different source locations assuming the noise to be stationary and gaussian . for fixed @xmath7
, we observed that the matching of signals from locations in the sky that differ in their colatitude by @xmath0 radians is independent of @xmath32 , and @xmath21 .
however , it falls with @xmath7 .
but in longitude the matching of signals falls with @xmath32 , @xmath20 , @xmath21 , @xmath7 .
we believe that the matching of signals will increase for the real data .
this is due to the resolution provided by the fast fourier transform ( fft ) .
however , it may not be relevant to account this symmetries for the search of cgw from the output of ligo - i , geo600 and tama detectors
. this analysis will be more relevant if one performs hierarchical search ( mohanty & dhurandhar 1996 ; mohanty 1998 ) .
this search is basically a two step search , in first step the detection threshold is kept low and in the second step a higher threshold is used .
the higher threshold is used for those templates which exceed the first step threshold .
the study of the matching of signals in different source locations will be also relevant for laser interferometer space antenna ( lisa ) ( hough j. , 1995 ) .
the work has been initiated and may be useful for the data analysis of cgw .
i am thankful to iucaa for providing hospitality where major part of the work was carried out .
i am thankful to prof .
dhurandhar , iucaa , pune for his useful discussions and mr .
a. sengupta , iucaa , pune for reading the paper .
abramovici a. , althouse w.e . ,
drever r.w.p . , g@xmath75rsel y. , kanwamura s. , raab f.j .
shoemaker d. , sievers l. , spero r.e . ,
thorne k.s . ,
vogt r.e . , weiss r. , whitcomb s.e . , and zucker z.e . , sciences * 256 * , 325 , ( 1992 ) bradaschia c. , calloni e. , cobal m. , fabbro r.d . ,
virgilio a.d .
, giazotto a. , holloway l.e . , kautzky h. , michelozzi b. , montelatici v. , passuello d. , and velloso w. , in gravitation 1990 , proceedings of the banff summer institute , edited by mann r. and wesson p. ( world scientific , singapore , 1991 )
|
we investigate the matching of continuous gravitational wave ( cgw ) signals in an all sky search with reference to earth based laser interferometric detectors .
we consider the source location as the parameters of the signal manifold and templates corresponding to different source locations .
it has been found that the matching of signals from locations in the sky that differ in their co - latitude and longitude by @xmath0 radians decreases with source frequency .
we have also made an analysis with the other parameters affecting the symmetries .
we observe that it may not be relevant to take care of the symmetries in the sky locations for the search of cgw from the output of ligo - i , geo600 and tama detectors .
* keywords : * gravitational wave methods : data analysis pulsars : general .
|
language is a defining trait and a prominent way of communication for the human species @xcite . on the other hand ,
languages are social constructs with an entity of their own @xcite .
these emergent objects , the shared codes that allow communication , are our concern throughout this paper .
we talk about contact situations when speakers of different tongues come together .
they decide what language to adopt when talking to each other , and this leads to certain dynamics of use and knowledge across populations .
these dynamics reflect an _ interaction _ between languages .
diverse phenomenology exists associated to such contact situations @xcite , but we are interested in the coarse grained dynamics of the number of speakers of different ( usually two ) languages within a population .
contact situations happen for social constructs other than languages @xcite .
opinions and political views @xcite are shared and discussed while economic agents engage in fusions , competitions , etc , that make them rise and fall @xcite .
these interactions are resolved by the choices made by individuals , who are free to take sides or change their minds over time .
this determines the fate of the emergent social objects preventing or giving rise to homogeneous opinions , political shifts , strong economic holdings , or unified communities of speakers .
a legitimate question is whether there exist mathematical rules behind these interactions and whether some universal regularities affect every human society alike .
a premise behind this question is that , while we can not track each individual , aggregating large masses might dissolve the most skewed positions revealing clear - cut rules for the average trends @xcite .
this picture is fully exploited by the branch of physics known as statistical mechanics . to solve the dynamics going on in a glass of water we need to keep track of the location , velocity , and energy of every water molecule .
these numbers vary greatly as atoms bump into each other and exchange energy and momentum i.e. as they interact so the task readily becomes impossible .
but statistics ensures that if we average millions of molecules together those large fluctuations cancel out while the averages behave according to smooth mathematical rules .
thus , the mean position , velocity , and energy become volume , pressure , and temperature ; and the laws of thermodynamics tell us how these quantities relate to each other . similarly , consider humans as a sort of _ social atom _ @xcite endowed with free will to fluctuate around some given options , or to interact with each other interchanging opinions , trading , through coercion , etc .
looking at small groups we often observe an untameable diversity .
as we amass large numbers of people , trends emerge and the rate of change between options may become more regular .
mathematical laws might arise , not because humans follow some unfathomable law , but as a statistical property out of their free choices .
maximally random particles yield some of the better established physical laws those of thermodynamics . given an ensemble of not - so - random agents , why would they not similarly obey some elegant mathematical laws ?
interestingly , it was the development of these ideas in social sciences what brought statistics to physics @xcite .
we are , somehow , returning to the origins .
early works on this field include studies of ethnic and cultural dynamics .
schelling ( @xcite ; see also @xcite and @xcite ) proposed a simple model of citizens choosing their neighborhood . consider a checkerboard with its squares distributed randomly ( figure [ fig:1]*a * ) .
each square represents a house whose occupant opts for one of two options ( black or white ) .
every day there is a probability that each individual tries to move out into a random location .
she does so if at the new location she has more similar neighbors than before .
schelling showed that , no matter how low the probability of moving out is , the checkerboard tends to maximally segregated communities .
we rarely observe fully isolated races or social classes ( which correspond to the black and white options of the model ) , but actual scenarios are close enough ( figure [ fig:1]*b * , @xcite ) . in another celebrated work ,
the political scientist robert axelrod proposed a model of how cultures survive in an interconnected world @xcite .
according to this model we tend to a unified dominant culture if the level of human interaction is above a certain threshold .
disperse and segregated cultures survive if that interaction is kept low
i.e. if cultures are effectively isolated .
despite these interesting results , it must always be noted the difference between the real world and what is possible within a model . along this paper
we will deal with _
idealized _ languages whose properties may often contradict the intuition of a linguist . we never claim that the results outlined apply to all languages , or that we are delivering unchangeable truths this is against scientific thought itself . rather , we will try to convey in written words what the solutions to a series of equations mean for these idealized languages .
this suggests constraints to the real world , given our mathematical knowledge and , sometimes , empirical data .
all this must be challenged as new evidence mounts up .
whenever some notion contradicts well - established knowledge in linguistics , this is suggesting us how our models shall be improved in the future , thus allowing a greater prediction power .
take as examples the idealized situations described by schelling and axelrod .
they establish limiting cases that we expect to find in some social systems _ unless something else is going on_. these limits send clear messages within a model of idealized cultures : keep segregated or risk absorption by a uniforming option .
because the real world is often far from such homogeneous scenarios , we can be quite sure that _ something else is going on _ that has not been taken into account by the equations , and this suggests how to improve them .
language contact situations have theoretical relevance to understand universal laws of social interaction . at the same time
these theoretical advances have implications for _ minority languages _ and make suggestions to help promote them . as minority languages we understand those with low prestige whose speakers are not necessarily a minority when the contact with a _ dominating language _ ( larger prestige ) begins .
we review recent advances from the perspective of physics , which has given some theoretical approaches to language contact situations .
we are concerned with the survival of minority languages , so we focus on works that have addressed specific aspects that seem critical for this issue .
works developing more technical and theoretical details are discussed in the abundant , up - to - date reviews about the topic @xcite .
two relevant factors have been identified that affect the survival of minority languages : geography and bilingualism . in section
ii we review models that gradually incorporate these elements , with emphasis on the latter .
we discuss the prospects of minority languages given the theoretical results .
these , as for the schilling and axelrod models , refer to ideal mathematical situations which are difficult to bring together with real data .
we attempt to do that in section iii fitting the dynamics of galician - spanish coexistence in northwest spain .
obvious limitations show up , also concerning the kind of data available which is often scarce , qualitative , and difficult to link to current theories @xcite .
this is so because we are dealing with systems with many relevant facets that affect each other in nontrivial ways , another reason why theoretical results must be taken with extreme care .
these and other issues are discussed in the concluding part , section iv .
in 2003 , abrams and strogatz introduced a concise , yet insightful model that captured the dynamics of dying languages @xcite .
earlier approaches existed @xcite , but abrams and strogatz s contribution was simple and their equations had clear interpretations that explained a wealth of data parsimoniously .
this sparked a new wave of research that expanded their methods and lasts until today .
assume a population with @xmath0 speakers of language @xmath1 and @xmath2 speakers of language @xmath3 .
consider this population to be constant and normalized to unity ( @xmath4 ) .
let , @xmath5 be the probability that a speaker decides to abandon language @xmath3 for language @xmath1 .
similarly , @xmath6 for @xmath1-speakers switching to @xmath3 .
this can be written mathematically in the form @xmath7 and @xmath8 , where @xmath9 and @xmath10 introduce the prestiges .
the parameter @xmath11 was termed volatility by castell et al .
@xcite , who associate it to a tendency to switch language use .
the constant @xmath12 is related to a global rate at which the language shift takes place : if @xmath12 is larger , the flow of speakers from one language to another occurs faster ; if it is smaller , the dynamics take more time to trace the same relative evolution .
the approach of the literature reviewed here often focuses on the distributions of speakers when a long time has elapsed .
this is not affected by @xmath12 , so this parameter has been largely ignored .
it would be very interesting , though , if some future works would relate empirically the parameter from this model to actual situations of language coexistence .
we take the prestiges to be normalized @xmath13 ( hence @xmath14 and @xmath15 ) .
the symbol @xmath16 here stems from the original work by abrams and strogatz @xcite , in which the word _ status _ was used instead of prestige .
we adopt the latter term in accordance to the linguistics literature .
the probabilities @xmath6 and @xmath5 of switching languages build up a flux of speakers that is captured by a set of differential equations ( see @xcite for details ) .
the solutions to these equations ( @xmath17 and @xmath18 ) tell us the fraction of @xmath1- and @xmath3-speakers over time ( figure [ fig:2 ] ) . as time tends to infinity
, the system relaxes to the fixed fractions @xmath19 and @xmath20 at which the flux of speakers ceases .
this is the stable state of the system .
what equilibrium configuration is reached depends on the parameters ( @xmath11 and @xmath16 , but not @xmath12 ) and the number of speakers of each tongue when the contact process began . for this model the only stable states for @xmath21 imply the extinction of one of the languages , usually the one with lowest prestige even if it has a large initial supporting population . for @xmath22 ,
the survival of both languages is possible @xcite . a detailed explanation of how this volatility parameter affects the dynamics in this and other models can be found in @xcite .
the power of this model is revealed when its equations are confronted with real data . to do this , recorded time evolutions of fractions of speakers
are plotted against time and the equations of the model are fit to the data ( figure [ fig:2 ] ) . as a result of this fit
we obtain the most likely parameters associated to the data
e.g. we derive the prestige parameter @xmath16 for each pair of language under research .
different combinations of parameters could account for the decay of up to @xmath23 pairs of minority languages in quite different contexts @xcite .
it was consistently found that @xmath21 ( indeed @xmath24 , far from the @xmath22 values that allow coexistence ) ; so that this model together with the data used to test it are bad news for minority languages : they will likely be driven to extinction unless they do not engage in competition . these results were rapidly challenged by more realistic equations : `` how does geography affect the competition ? '' `` what if bilingualism is considered ? '' but the abrams - strogatz ( as ) model remains a remarkable contribution for two reasons : i ) it primed the use of new tools in language population dynamics and ii ) it established clear limiting cases similarly to how schelling and axelrod did for ethnic groups or cultures .
patriarca and leppnen @xcite introduced the geographical factor in the as model through a population spread over a plane with two adjacent regions .
each region ascribed a larger prestige to one of the tongues .
locally , the languages interact following the as model .
both languages survive within their preferred region , but this segregated coexistence is due to the odd assignment of prestige . indeed
, prestige should arise from the use of language , not from properties of the land .
if we substitute _ all _ speakers anywhere by speakers with a foreign language , the new people will not switch spontaneously to the old tongue . seeking more realism , patriarca and heinsalu @xcite
study another important feature related to space : human mobility .
in @xcite speakers are allowed to move randomly while the prestiges of the languages do not depend on the physical location of the speakers ( as they did in @xcite ) .
figure [ fig:3 ] shows how dense communities of minority speakers might reverse an unfavorable situation @xcite .
@xmath1-speakers ( dominant ) are distributed equally in figures [ fig:3]*a1 * and [ fig:3]*c1 * , @xmath3-speakers are more concentrated in the initial distribution of figure [ fig:3]*b1 * than in [ fig:3]*d1*. the more compact community nucleates a resistance against @xmath1 that eventually turns over the situation , driving @xmath1 to extinction ( figure [ fig:3]*a5 * , [ fig:3]*b5 * ) .
the more diffused @xmath3-speakers can not avoid the extinction of their language ( figure [ fig:3]*c5 * , [ fig:3]*d5 * ) .
survival of both languages in @xcite is also possible , and it is associated to landforms such as mountains or islands .
this does not depend on artificially larger prestiges ( as in @xcite ) . instead , human mobility is the key due to the low mixture imposed by the physical barriers .
a clear message is sent : survival of different languages within a geographic space is possible , but both tongues must remain relatively segregated . this might be fine for the dominating language , but isolation for minority tongues seems harmful ( e.g. in economic terms ) or impossible , especially in our hyper - communicated world . note that this is a result valid within the hypothesis of this model . again , while real situations will likely be more complicated , the model establishes clear frontiers and mechanisms for the survival of coexisting tongues .
an exploration of the model from patriarca and heinsalu @xcite ( and variations upon it ) with a stress on an ecological interpretation is made in @xcite : @xmath1 and @xmath3 become species competing for resources ( speakers ) .
other models not based on the as equations incorporate geometry as well @xcite and yet other models study the diffusion of linguistic traits across space @xcite .
the latter focuses on the broad distributions of languages spoken in our planet , the number of tongues within a family , and the number of speakers of a tongue in average .
these works tell us little about a specific contact situation , but they interpret minority languages in a systemic sense : it is possible that the distribution of languages ( including endangered ones ) across the planet @xcite arises as a trait of a large system in overall equilibrium .
finally , note that geometry is a way of constraining the interactions between speakers ( modeled as abstract agents when simulating the equations computationally ) . in the models reviewed so far ,
these agents change their language depending on other agents nearby , unless some landforms prevent that .
hence , speakers interact with their neighbors in a two dimensional grid .
other kinds of constraints affect everyday real interactions
e.g. belonging to a community such as a university or a trade union , practicing sports , etc .
these interactions bias how and which agents come together , thus composing a _ social network _ that effectively alters the two dimensional geometric structure presented so far .
several works study how the underlying social network affects the dynamics of the abrams - strogatz models @xcite .
these works also deal with bilingualism and important insights will be discussed later .
regarding monolingual models alone , these social structures seem not to alter notably the dynamics studied so far .
bilingualism is a crucial element of language contact .
it could be dismissed before ( as the good results in @xcite indicate ) because the pairs of tongues involved are fairly different and bilingualism was relatively marginal . ` relatively marginal ' does not mean that it is not important or that it does not exist .
it means that it has been possible to develop a theory that satisfactorily explains some salient dynamics but that does not take bilingualism into account .
when we explore language dynamics in deeper detail , evidence readily suggests that it is necessary to incorporate a notion of bilingualism into our models @xcite .
when dealing with empirical data , bilingualism is defined through some qualitative traits
e.g. competence or usage of both languages as will be exemplified later .
but now bilingualism is an abstract concept whose effective definition depends only on what terms it brings into the equations , and qualitative issues are less important .
mira and colleagues @xcite introduced bilingualism in the as model through a group of bilinguals ( @xmath25 , as opposed to the previous @xmath1 and @xmath3 monolinguals ) .
a fraction @xmath26 of speakers belongs to this group .
population size remains constant and normalized ( @xmath27 ) .
they considered an _ attracting population _ that includes the bilinguals , so that the urge of @xmath3-speakers to learn a new language is proportional to @xmath28 .
the term _ attracting population _ was introduced in @xcite and reflects what members of a society might induce a speaker to switch languages .
we come back to it later . from all @xmath3-speakers changing options ,
a fraction @xmath29 retain their language while @xmath30 become full @xmath1-monolinguals .
this is their working definition of bilingualism : an abstract group @xmath25 able to attract population away from @xmath1 and @xmath3 and capable of retaining a fraction @xmath29 of the shifting population .
the larger @xmath29 , the easier it is for speakers to join @xmath25 .
hence , when the model is matched to real data and @xmath25-speakers correspond to a qualitative notion of bilinguals , @xmath29 becomes a measure of how easy or convenient it is to master both languages i.e. of the perceived similarity between @xmath1 and @xmath3 .
this parameter was termed _
interlinguistic similarity_. we can write down all probabilities of switching groups as we did for the as model ( @xmath31 , @xmath32 , @xmath33 , @xmath34 )
. these again establish fluxes that are captured by a set of differential equations @xcite . in figure
[ fig:4]*a1-a3 * we plot possible dynamics that these equations can generate .
what evolutions we see for a pair of languages depends on their parameters ( @xmath11 , @xmath12 , @xmath29 , @xmath16 ) and initial conditions .
periodic oscillations are not possible @xcite so every contact situation evolves towards stable fractions of speakers ( @xmath19 , @xmath20 , @xmath35 ) .
some of the possible equilibriums of this model present monolingual populations of both tongues always along a bilingual community .
hence , persistent bilingualism is possible and it enables the steady coexistence of both languages without spatial segregation @xcite .
bilingualism acts as a contact surface between the opposed groups that softens the competition . without bilingualism , the models most faithful to data lead either to the extinction or isolation of the minority language , as seen before .
bilingualism opens intermediate scenarios with varying ( even large ) presence of the minority language .
the distribution of speakers at equilibrium ( @xmath19 , @xmath20 , @xmath35 ) depends on @xmath11 , @xmath29 , and @xmath16 ( but not @xmath12 ) and the initial fractions in each group when the contact situation started . in figure
[ fig:4]*b * we plot these parameters in three axes .
the surfaces enclose those @xmath36 combinations for which language coexistence is possible .
figure [ fig:4]*c * is a cross section of that surface .
empirical measurements from galician - spanish coexistence are plotted to assess the fate of these contact situations : according to the model , red crosses belong to systems that will lose the endangered language , while black plus symbols might present stable coexistence .
alternatively we can extrapolate the equations with the measured parameters and see if bilingualism declines or persist over time ( figure [ fig:4]*d * ) .
earlier attempts to model bilingualism by baggs and freedman @xcite in the 1990 s had already advanced the possibility of stable language coexistence .
their models were fairly abstract and required further assumptions ( as in @xcite ) before they could reproduce realistic scenarios .
the baggs - freedman models broke the symmetry between @xmath1 and @xmath3 , explicitly favoring a dominating tongue .
since coexistence is still possible even so , bilingualism again means good news for minority languages . in a remarkable model , zhang and
gong @xcite rely solely on biological and physical principles to deduce equations that , with minimal empirical input , faithfully reconstruct several real situations .
they incorporate bilingualism naturally . failing to do so would not yield good reconstructions of english - welsh and english - gaelic competitions , the authors find , thus stressing the relevance of bilingualism .
this is a promising approach that derives relevant parameters from first principles and that shall apply where little data is available .
the parameters involved have clear interpretations in terms of influence between languages and population dynamics .
those equations are also ready to adopt more languages and geography appears naturally .
we look forward to more applications of this model in relevant , ongoing contact situations .
minett and wang @xcite also include bilingualism in the as model .
they only allow @xmath37 ( not @xmath38 ) flows and take monolinguals as the only attracting population .
the stable distributions in their model imply the death of one tongue .
minett and wang suggest what strategies might be applied to sustain a language , but this discussion remains highly abstract . other variations with bilingualism ( many of them inspired by this last model ) incorporate geography and social networks , as discussed above @xcite . a great review of the role of social networks oriented to linguists can be found in @xcite .
two important insights are summarized there : first , in some cases the introduction of bilingual agents tends to accelerate the extinction of one of the languages .
this happens in so - called _ small - world _ networks , in which the social structure ensures that all speakers are relatively close to each other i.e. options or opinions spread easily across the network .
this seems at odds with the results just discussed @xcite , in which bilingualism favors language coexistence .
these differences will be solved in the next section .
the second important message is that , when the network presents a strong community structure ( with social clusters relatively isolated from each other but well connected within themselves ) , coexistence of both languages is possible if monolinguals are confined into homogeneous clusters and bilinguals occupy the interface between the communities .
this , again , implies a kind of survival by segregation as the one discussed in section ii.b due to geography .
heinsalu , patriarca and lonard @xcite consider the role of the attracting population .
when nothing exists but @xmath1- and @xmath3-speakers , only they can encourage you to switch groups .
when bilingual speakers enter the game , they might influence your decision too . in the previous section we saw two extremes : for mira and colleagues @xcite , bilinguals are as influential as monolinguals . for minett and wang @xcite only monolinguals affect you despite the existence of a bilingual group . based on this latter model , heinsalu et al .
@xcite consider attracting populations @xmath39 and @xmath40 so that transition probabilities @xmath41 and @xmath42 take the form @xmath43 and @xmath44 , with @xmath45 $ ] . only monolinguals can attract speakers away from the bilingual group . the minett - wang @xcite model did not contain stable bilingual configurations .
this is also the model adopted by @xcite and @xcite to simulate agents in social networks , in which the presence of bilinguals accelerated the extinction of a language when the network was well connected hence hindering language coexistence . with the modifications by @xcite ,
depending on the relative prestiges of both languages and other parameters that control the flow of speakers , stable bilingual configurations arise if @xmath46 and @xmath47 .
this indicates that , for both languages to coexist , their speakers must perceive bilinguals as representatives of the opposite tongue .
this perception has to be above some threshold ( @xmath46 , @xmath47 ) , which in turn depends on the prestiges .
another requirement of the stable bilingual distributions is that @xmath48 transitions must be rare ( @xmath25 eventually agglutinates all speakers in this model if it is stable ) .
this is interpreted in @xcite as a requisite for bilinguals to have their own intrinsic advantages .
note , finally , that @xmath1- and @xmath3-monolinguals might perceive the bilingual community differently because @xmath49 and @xmath50 might differ .
this is interesting to model real data , e.g. , if an asymmetrical situation forces speakers of a tongue ( but not of the other ) to become bilinguals .
unfortunately , @xcite is the only existing study about attracting populations . that research was performed for @xmath51 , which already presents all features of the minett - wang model .
we tested informally how introducing similar attracting populations ( @xmath52 and @xmath53 ) in mira et al.s model affects its dynamics .
the numerical , preliminary results confirm heinsalu et al.s observations : decreasing @xmath49 , @xmath50 ( i.e. when bilinguals are not perceived as representatives of the alternative option ) hardens the conditions for stable coexistence .
the form of the attracting population is an important issue given the controversy around bilingualism .
this affects both theoretical ( `` how must bilinguals enter the equations ? '' ) and empirical studies .
most surveys rely on speakers self - assessing their use of each tongue .
methodological discrepancies exist even for studies of a same situation carried out by one same institution .
future research must clarify how the definition of bilingualism in real data affects the explanatory power of the theoretical models .
galician is a romance language evolved in the northwest of the iberian peninsula after the collapse of the roman empire .
galician - portuguese first constituted a unified tongue that split in two after the political division of the portuguese and galician kingdoms in the xvi century .
galicia was integrated in what nowadays is spain , whose ruling elite spoke castillian ( spanish ) . towards the end of the xix century
the contact between galician and spanish intensified reaching most demographic substrates , notably in the cities .
this is reflected in a lasting decline @xcite starting at the turn of the xx century .
this suggests that both languages were effectively segregated until the @xmath54s .
similar processes happened to basque , catalan , and other spanish tongues . since the early @xmath55s the real academia galega ( galician royal academy ) periodically commissions surveys to assess the condition of the galician language resulting in exhaustive sociodemographic reports that link galician or spanish usage to geography , age , or economic substrate .
although the same institution developed all these works , methodological differences exist between successive versions , precisely because human language is such a nontrivial feature and intense procedural and theoretical debate still exists .
we focus on the age segregation of galician and spanish usage from the 2008 _ mapa sociolingstico de galicia _ ( galician sociolinguistic map ) .
informants were asked to classify themselves as : i ) only speak galician , ii ) mostly speak galician , iii ) speak both equally , iv ) mostly speak spanish , v ) only speak spanish .
we aggregated the three central groups as bilinguals to build up data for the variable @xmath26 discussed above , hence bilingualism is defined through _ self - assessment of language usage _ for this academic exercise .
the extreme options make up @xmath0 and @xmath2 respectively .
we will compare these variables to mira et al.s model @xcite .
these data ( figure [ fig:4]*d * ) give us noisy snapshots of the real dynamical process of language shift going on .
we can take the model equations @xcite , try different values for their parameters ( @xmath11 , @xmath12 , @xmath29 , @xmath16 ) , and select those that are in good agreement with the data
i.e. @xmath56-combinations whose dynamics follow closer the empirical fractions of speakers over time .
data from the mapa sociolingstico de galicia @xcite separates speakers by geographic subdivisions , so we could repeat this process for several such sub - units and also for the aggregated data .
plotting the parameters found for each region we can assess where is language coexistence _ not _ possible ( red crosses in figure [ fig:4]*c * as indicated above ) .
figure [ fig:4]*d * shows extrapolations for the aggregated galician data for which bilingualism is predicted to be a stable trait .
both monolingual groups are predicted to survive , with the spanish monolingual group likely dominating
however , the data is consistent with a great variability concerning the final fractions of monolingual speakers .
note the importance of having _ quantitative _ and _ objectively reproducible _
measurements for similarity and prestige .
as we announced before , when the model in @xcite is matched to data , @xmath29 can be interpreted as a similarity between the languages involved .
more specifically , @xmath29 reflects the _
interlinguistic similarity _ as perceived by the speakers of both languages in the precise situation under research .
figure [ fig:4]*c * shows how galician and spanish are perceived with an almost constant similarity across different areas ( @xmath57 with a standard deviation of @xmath58 ) .
this highlights the robustness of @xmath29 as an empirical measure of distance despite the great variation of the prestige parameter , which reflects very different sociolinguistic scenarios .
other models have been tested against real data , including the original as - model @xcite .
language shift in the british isles is an empirically well documented case with rich data ( see @xcite and references therein , where a model with local birth , death , and migration is also successfully tested ) .
zhang and gong s model @xcite , as remarked above , also reproduced data with minimal empirical input .
all these are predictions made by different models with the data available .
this is not an indisputable truth , but rather a guideline that must be subjected to revision and that allows us to falsify the equations .
it was shown @xcite that these models are sensible to how we account for bilingualism .
besides , empirical data are updated regularly , also due to methodological advances .
statements about real - life situations must be taken with extreme care and be continuously refined .
extreme social and economic events shall invalidate our current conclusions , simply because they violate the hypotheses of the models .
locating such game - changers in real data ( compared to usual statistical fluctuations ) can give us a hint of the magnitude that a policy must have to support minority languages .
this review began with theoretical considerations beyond language contact which dealt with human interactions in general : do laws emerge for social dynamics as in physics ? in
what circumstances are they valid ? are there , indeed , fundamental laws ?
the contact scenarios reviewed are fueled by interactions between people : observing other speakers might compel us to change our tongue . hence , what is the value of human interaction ?
can we quantify it in general as we quantify the mass and charge of an electron ?
powerful mathematical and empirical reasons suggest positive answers for some of these questions .
this position is compatible with delicate matters about free will : intrinsic laws that guide individual actions do not exist . instead ,
unavoidable regularities emerge ( even for the most random systems , as thermodynamics shows ) whose average trends and shifting rates become clear - cut as we aggregate larger numbers of people .
we focus here on practical applications to language contact scenarios .
this is a testing bench for the logic just outlined , and a relevant one with deep social and economic implications .
the last decade has seen interesting developments , mainly hindered by the multiple factors involved and the difficult interpretation of real data @xcite .
recent advances single out important factors to assess the fate of minority languages , which we summarize : * * extinction or segregated survival as limiting cases * : this is supported by the original as model @xcite and also by shelling s @xcite and axelrod s @xcite account of ethnic and cultural dynamics .
when interaction ( hence competition ) is intense , the tendency is to homogenize the cultural space . only through weak interactions
can different cultures survive , but weak interactions mean isolation . * * the geographic distribution of speakers matters * : as shown in @xcite , a dense stronghold of speakers of a minority language might reverse an adverse situation . but this still implies the extinction of one of the tongues .
both languages can survive if human mobility is reduced by physical barriers
i.e. , again , if languages are effectively segregated .
* * bilingualism is an asset * : a naive view might suggest that a language loses speakers through the bilingual group . but
some theoretical models indicate that bilinguals ( when successful and salient ) soften the competition and enable possibilities that were forbidden in simpler models @xcite .
put otherwise : pairs of languages for which a stable bilingual group can not be successfully established are deemed to either disappear or live isolated . *
* bilingualism demands two conditions to enable a steady coexistence @xcite * : i ) for the speaker of a language , bilinguals must be representative of the opposite option
i.e. bilinguals must contribute to the attracting population .
ii ) the transitions away from bilingualism must be rarer than the transitions towards bilingualism , suggesting that being bilingual has attached some extra social value .
this is not far fetched : bilinguals not only can engage with more people , but also bring together large , linguistically isolated groups .
a note of caution is in order .
while testing the models gave positive results , agreement between equations and data does not mean that all the theoretical conclusions must hold always .
these models are valid when some strict conditions are met , but social systems are prone to unpredictable events impossible to account for . localizing drastic shifts in the recorded data ( e.g. due to wars or migrations ) can be an interesting exercise .
such events should notably alter the smooth evolutions reproduced by the equations .
these could give us a quantitative reference about the magnitude that a policy must have to positively affect the fate of minority languages .
more theoretical developments are necessary to incorporate such rare events in the theory while more empirical tests are necessary to determine the limits of the current models .
new threats and opportunities arise for minority languages as the socio - economic stage changes , prominently thanks to the internet and the levels of globalization attained in the last decades .
there is a chance now to test models of microscopic situations against real data gathered online .
also , modeling multilingual scenarios , studying their stability , and comparing them with real trends seems within reach .
san miguel , maxi , vctor m. eguluz , raul toral , and konstantin klemm .
`` binary and multivariate stochastic models of consensus formation . '' _ computing in science and engineering _ 7 , no . 6 ( 2005 ) : 67 - 73 .
vzquez , federico , xavier castell , and maxi san miguel .
`` agent based models of language competition : macroscopic descriptions and order - disorder transitions . ''
_ journal of statistical mechanics : theory and experiment _ 2010 , no .
04 ( 2010 ) : p04007 .
hadzibeganovic , tarik , dietrich stauffer , and christian schulze .
`` boundary effects in a three - state modified voter model for languages . ''
_ physica a : statistical mechanics and its applications _ 387 , no .
13 ( 2008 ) : 3242 - 3252 .
de oliveira , viviane m. , marcelo af gomes , and ing ren tsang .
`` theoretical model for the evolution of the linguistic diversity . ''
_ physica a : statistical mechanics and its applications _ 361 , no . 1 ( 2006 )
: 361 - 370 .
de oliveira , viviane m. , paulo ra campos , marcelo af gomes , and ing ren tsang .
`` bounded fitness landscapes and the evolution of the linguistic diversity . ''
_ physica a : statistical mechanics and its applications _ 368 , no . 1 ( 2006 )
: 257 - 261 .
stauffer , dietrich , christian schulze , f. welington s. lima , sren wichmann , and sorin solomon .
`` non - equilibrium and irreversible simulation of competition among languages . ''
_ physica a : statistical mechanics and its applications _ 371 , no . 2 ( 2006 )
: 719 - 724 .
castell , xavier , luca loureiro - porto , vctor m. eguluz , and maxi san miguel .
`` the fate of bilingualism in a model of language competition . '' in _ advancing social simulation : the first world congress _ , pp .
springer japan , 2007 .
castell , xavier , riitta toivonen , vctor m. eguluz , and maxi san miguel .
`` modelling bilingualism in language competition : the effects of complex social structure . '' in _ proceedings of the 4th conference of the european social simulation association_. irit editions , 581 - 584 ( 2007 ) .
stauffer , dietrich , xavier castell , vctor m. eguluz , and maxi san miguel .
`` microscopic abrams - strogatz model of language competition . ''
_ physica a : statistical mechanics and its applications _ 374 , no . 2 ( 2007 )
: 835 - 842 .
kandler , anne , roman unger , and james steele .
`` language shift , bilingualism and the future of britain s celtic languages . ''
_ philosophical transactions of the royal society b : biological sciences _ 365 , no .
1559 ( 2010 ) : 3855 - 3864 .
otero - espinar , m. v. , lus f. seoane , juan j. nieto , and jorge mira .
`` an analytic solution of a model of language competition with bilingualism and interlinguistic similarity . '' _ physica d : nonlinear phenomena _ 264 ( 2013 ) : 17 - 26 .
|
recent contributions address the problem of language coexistence as that of two species competing to aggregate speakers , thus focusing on the dynamics of linguistic traits across populations .
they draw inspiration from physics and biology and share some underlying ideas e. g. the search for minimal schemes to explain complex situations or the notion that languages are extant entities in a societal context and , accordingly , that objective , mathematical laws emerge driving the aforementioned dynamics .
different proposals pay attention to distinct aspects of such systems : some of them emphasize the distribution of the population in geographical space , others research exhaustively the role of bilinguals in idealized situations ( e. g. isolated populations ) , and yet others rely extremely on equations taken unchanged from physics or biology and whose parameters bear actual geometrical meaning . despite the sources of these models
so unrelated to linguistics
sound results begin to surface that establish conditions and make testable predictions regarding language survival within populations of speakers , with a decisive role reserved to bilingualism . here
we review the most recent works and their interesting outcomes stressing their physical theoretical basis , and discuss the relevance and meaning of the abstract mathematical findings for real - life situations .
|
the first observational evidence of @xmath1f stellar nucleosynthesis was reported by jorissen , smith & lambert ( 1992 , hereafter jsl ) .
these authors derived f enhancements up to a factor 50 solar in a sample of galactic agb stars , and found a correlation between this enhancement and the c / o ratio . since the c / o is expected to increase as a consequence of third dredge up ( tdu ) episodes during the agb phase ( e.g. busso et al . 1999 ) , this was interpreted as a clear evidence of f production in these stars .
further observational evidence of such a production exists from studies of post - agb stars ( werner , rauch & kruk 2005 ) and planetary nebulae ( otsuka et al . 2008 ) .
other sites for f production have also been proposed : wolf - rayet stars ( meynet & arnould 2000 ) and neutrino spallation in core - collapse supernovae ( woosley & haxton 1988 ) . however
, the role of these sources in the f budget is still uncertain ( cunha et al . 2003 ; palacios , arnould & meynet 2005 ) . nevertheless , from galactic chemical evolution models , renda et al .
( 2004 ) concluded that all three of these three sources are required to explain the observed galactic evolution of f , as deduced from abundance determinations in field stars ( cunha & smith 2005 ; cunha , smith & gibson 2008 ) , although , only in agb stars there exist an observational confirmation that f production is an ongoing process .
fluorine can be produced in agb stars from the nuclear chain @xmath2 ; where protons are mainly provided by @xmath3 and neutrons by @xmath4 .
however , the theoretical attempts made to explain the jsl results are unsatisfactory .
the problem is that current agb models fail to explain the highest f enhancements found in the c - rich objects of the jsl sample ( goriely & mowlavi 2000 ; lugaro et al .
this discrepancy has led to a deep revision of the uncertainties in the nuclear reaction rates involved in the synthesis of f in agb stars ( lugaro et al .
2004 ; stancliffe et al .
2005 ) , to argue for alternative nuclear chains and/or to propose the existence of non - standard mixing / burning processes , similar to those commonly used to explain some of the isotopic anomalies found in dust grains formed in the envelopes of agb stars ( e.g. nollett et al .
2003 ; busso et al .
however , no solution has been found up to date , leaving the subject of the origin of this element open . very recently , abia et al .
( 2009 , hereafter paper i ) derived f abundances from vlt spectra in three galactic agb c - stars in common with the jsl sample , namely : aq sgr , tx psc ( n - type ) and r scl ( j - type ) , by using mainly the hf r9 line at @xmath5 m . in that work
, we showed that this hf line is almost free of blends in agb c - stars and thus is probably the best tool for f abundance determinations in these stars ( see also uttenthaler et al .
for these three stars , we derived f abundances @xmath6 dex lower in average than those obtained by jsl .
we ascribed this difference to molecular blends ( mainly of c@xmath7 and cn ) .
these new f abundances are in better agreement with the most recent theoretical nucleosynthesis predictions in low - mass ( @xmath8 m@xmath9 ) tp - agb models ( cristallo et al .
motivated by this result , we have reanalysed the f abundances in the whole c - stars sample studied by jsl using the same tools as in paper i. in this letter , we confirm the results found in paper i and show that the new f abundances nicely agree with the theoretical predictions of low - mass tp - agb models .
this apparently solves the long standing problem of f enhancements in agb stars without requiring the existence of any extra mixing / burning process .
our data consist of high - resolution spectra * from @xmath10 to 2.5 @xmath11 m * obtained with the 4 m telescope at kitt peak observatory and a fourier transform spectrometer , which have been used in several abundance analysis of agb stars ( see lambert et al .
1986 and jsl for details ) .
typical resolutions ranged from 0.05 to 0.14 @xmath12 ( r@xmath13 to 30000 ) .
these spectra contain many hf lines of the vibrational 1 - 0 band although we used only the lines r9 to r23 .
high resolution optical spectra ( r@xmath14 ) * in the range @xmath15 * obtained at the 3.5 m tng at la palma observatory with sarg echelle spectrograph were also analysed .
these spectra served us to derive the @xmath16element content in * the stars bl ori , st cam , y tau , cy cyg , rz peg and gp ori using the same procedure as in abia et al .
( 2002)*. all program stars were ratioed with hot stars to remove telluric spectral features by using the task _ telluric _ within the iraf package .
the typical s / n ratio of the spectra is larger than 100 .
we adopted the stellar parameters ( t@xmath17 , log @xmath18 , [ fe / h ] , @xmath19 , * cno abundances * , c / o and @xmath20c/@xmath21c ratios ) derived by previous studies : abia et al .
( 2002 ) and lambert et al .
( 1986 ) for n - type stars , abia & isern ( 2000 ) for j - type stars and zamora ( 2009 ) for stars of sc - type ( see these works for details ) .
in addition , we also adopted the @xmath22 ratios from the literature when available ( harris et al .
1987 ) , although this has a minimal impact on the derived f abundance .
the method of analysis involves the comparison of theoretical lte spectra ( computed with the _ turbospectrum _ code ) with the observed ones .
we used the latest generation of c - rich atmosphere models for giant stars ( gustafsson et al .
2008 ) and up - to - date molecular and atomic line lists in the spectral ranges studied ( see paper i for details ) .
figure 1 shows an example of comparison between the observed and theoretical spectrum for the c - star uu aur ( n - type ) in two different spectral ranges .
in particular , top panel shows the fit to the r15 and r16 hf lines , the sole lines used by jsl to derive the f abundance in the c - rich objects .
it can be seen that we can reproduce reasonably well the observed spectrum and that the synthetic spectrum computed with the f abundance obtained by jsl ( lower dashed line ) , would overestimate the f abundance in this star . our best estimation in this star ( log @xmath23 , continuous line in fig .
1 ) fits quite well the six hf lines used , with a dispersion of only @xmath24 dex ( see also table 1 ) .
similar quality fits were obtained for all the stars except for v hya , which was excluded from the analysis since its effective temperature is outside the range covered by our grid of models .
table 1 lists the final f abundance derived in each star , assigning double weight to the r9 line when estimating the mean f abundance because this is the less blended among the available hf lines ( see paper i for a discussion ) . for most of the stars ,
more than three different hf lines were used .
the agreement between the different lines is in general good with a typical mean dispersion of @xmath25 dex .
the formal error in the f abundance due to uncertainties in the stellar parameters ranges from @xmath26 dex , depending on the particular hf line . considering the abundance dispersion and the uncertainty in the continuum position as additional sources of error
, we estimate an total uncertainty of @xmath27 dex in the [ f / h ] ratio ( see paper i )
. however , the abundance ratio between f and any other element typically has a smaller error since some of the above uncertainties cancel out when deriving the abundance ratio : for the [ f / fe ] ratio we estimate a total uncertainty of @xmath28 dex .
the third column in table 1 shows the difference found between the f abundance derived by jsl and this work .
the differences range from 0.3 dex up to more than 1 dex , the average value being @xmath29log @xmath30(f)@xmath31 , confirming the main result of paper i. as noted there , differences in the stellar parameters adopted in the analysis can not explain this discrepancy .
actually , the major differences are found for the t@xmath17 values , which for the n- and j - type stars in the sample , do not exceed 150 k. in the case of sc - type stars , differences in adopted t@xmath17 may reach up to 300 k , but in any case , this can not explain the differences of more than @xmath32 dex in the f abundance found in some of these stars .
differences in the atomic data are also discarded since we used exactly the same hf line list than jsl .
finally , differences between the grid of model atmospheres might account up to a maximum of 0.15 dex , for the n- and j - type stars . for the sc - type stars
, we found the largest abundance differences ( see table 1 ) . for these stars , atmosphere models indeed play a significant role .
jsl used for the analysis of sc stars johnson ( 1982 ) models , while we use those from gustafsson et al .
* for instance , considering the typical stellar parameters of a sc star in the sample ( t@xmath17/log g/[fe / h]@xmath33 and c / o@xmath34 ) , we find differences in the f abundance from @xmath35 dex ( r9 , r15 and r16 lines ) to @xmath36 dex ( r23 line ) * when using johnson models instead of gustafsson et al .
ones , * but it may reach up to @xmath37 dex for lower t@xmath17*. in any case , for the n- and j - type stars , we conclude ( as in paper i ) that the difference between the molecular line list used is the main cause of the discrepancy with respect to the analysis of jsl * ( see table 2)*. for the sc - type stars , it is a combination of that , the use of a different grid of model atmospheres and , in the case of cy cyg and gp ori , the adoption here of a t@xmath38 k cooler . it should be mentioned that the structure of the c - rich atmosphere models change dramatically with a tiny variation of the c / o ratio when it is very close to 1 , as it is the case for sc stars . therefore , the abundance analysis in these stars are affected by systematic differences between the model atmosphere and the real star and/or the treatment of the molecular equilibrium in the computation of the atmosphere models . in consequence , for sc - type stars
the uncertainty in [ f / h ] is certainly larger than @xmath39 dex .
the main consequence of the new f abundances is that the large [ f / fe ] ( or [ f / o ] ) ratios ( up to 1.8 dex ) found by jsl in galactic agb c - stars are systematically reduced .
the largest f enhancements are now close to @xmath40 dex .
these enhancements can be accounted for by current low - mass tp - agb nucleosynthesis models of solar metallicity , as we will show below . as noted in 1 , during the ascension along the agb , fresh carbon
is mixed within the envelope due to tdu episodes .
eventually , an o - rich agb star becomes a c - star when the c / o ratio in the envelope exceeds 1 .
similarly , f is also expected to increase in the envelope during the agb phase , thus a fluorine vs. carbon correlation should exist .
figure 2 shows the observed relationship derived in this study .
we have also included in this figure the intrinsictc ( @xmath41 yr ) in their envelopes ( smith & lambert 1990 ) , revealing their intrinsic nature . ] o - rich agb stars studied by jsl ( not analysed here , open circles ) . excluding the j - type stars ( triangles ) , a clear increase of the f abundance with the c abundance can be seen .
this behaviour is well reproduced by theoretical agb models .
lines in fig . 2 show the predicted f and c content in the envelope ( cristallo et al .
2010 ) for a 1.5 m@xmath9 with different metallicities ( continuous lines ) and 2 m@xmath9 model ( dashed line ) with solar metallicity .
these metallicities match those of the stellar sample ( @xmath42[fe / h]@xmath43 ) .
the theoretical curves start with an envelope composition as determined by the first dredge - up and end at the last tdu episode . on the other hand , fig .
2 seems to indicate that o - rich stars and sc stars ( squares ) have f abundances systematically larger than n - type stars ( dots ) for a given c abundance .
similarly to the intrinsic c - stars , also in the case of the o - rich stars this may be simply due to a systematic error affecting the analysis : in fact , for stars of spectral types k and m ( with [ fe / h]@xmath44 ) , jsl derived an average f abundance @xmath45 dex higher ( 4.69 ) than the solar f abundance adopted here ( 4.56 ) .
these k and m stars are sub - giants or rgb stars , thus they are not expected to present f enhancements . excluding that the sun might be anomalous with respect to nearby stars of similar metallicity ( f determinations in unevolved dwarf stars are needed in this sense )
, blends may also play a role in the analysis of f lines in o - rich agb stars . decreasing the f abundances by 0.13 dex in them
, the agreement with theoretical predictions would be definitely better .
however , it does not apply to the case of the star bd@xmath46 ( at log @xmath30(f)@xmath47 in fig .
2 ) ; the large f enhancement in this star can be only obtained in metal - poor agb models , but this star has [ fe / h]@xmath48 .
something similar seems to occur with sc - type stars .
sc stars are agb stars with c / o@xmath49 . according to the accepted chemical ( spectral ) evolution along the agb ( m@xmath50ms@xmath50s@xmath50sc@xmath50n ) , these stars should present equal - or - slightly lower f enhancements than n - type stars . however , some of the studied sc stars clearly deviate from this sequence , showing larger f abundances with respect to n stars .
this is reinforced by wz cas ( [ fe / h]@xmath44 , at the right upper corner of fig .
2 ) , which shows a huge f enhancement .
we remind , however , that this star is indeed peculiar : it is one of the few super li - rich c - stars known , also presenting very low @xmath20c/@xmath21c ( 4.5 ) and @xmath51o/@xmath52o ( 400 ) ratios .
the fact that most of the sc - stars studied here show low or no @xmath16element enhancements ( see below ) , makes the evolutionary status of these stars very uncertain .
the connection between the f and the @xmath53-process is less straightforward .
there are two different contributions to the f production in low mass agb stars .
the first comes from the @xmath54n production in the radiative @xmath21c pocket , the site where the s - process main component is built up : when , during the interpulse period , the @xmath4 is activated within the pocket , some protons are released by the main poisoning reaction , @xmath3 , thus producing @xmath54n via @xmath55 .
the second contribution involves the same reaction chain , but it is activated when the @xmath21c left by the advancing h - burning shell is engulfed into the convective zone generated by a thermal pulse . in the latter case , the correlation with the @xmath16process is less stringent , because the resulting neutron flux is not enough to give rise to a sizable production of elements beyond iron . at nearly solar metallicity ,
the latter contribution accounts for @xmath5670% of the total f production * ( e.g .
cristallo 2006)*. nevertheless , a correlation between the f and the @xmath16element overabundance is in any case expected if a large enough @xmath21c pocket forms after each dredge up episode .
figure 3 shows the new [ f / fe ] ratios in c - stars vs. the observed average @xmath16element enhancement * ( from abia et al .
2002 ; zamora 2009 or derived in this work , see 1 ) * .
j - type stars are not shown in this figure as they do not have @xmath16element enhancements ( see table 1 ; abia & isern 2000 ) . similarly to fig .
2 , we have included the intrinsic o - rich stars analysed by jsl taken their @xmath16element content from smith & lambert ( 1990 ) .
it is clear that the f and @xmath16nuclei overabundances correlate and that this correlation is nicely reproduced by theoretical tp - agb stellar models ( cristallo et al . 2010 ) .
the theoretical predictions in fig . 3 correspond to 1.5 , 2 and 3 m@xmath9 tp - agb models with z@xmath57 ( [ fe / h]@xmath58 , the average metallicity of the stars analysed here ) starting from the 1@xmath59 tp pulse .
the number of tps calculated until the occurrence of the last tdu is indicated in the figure for each model .
models with a metallicity within @xmath60 dex of this value do not significantly differ from those in figure 3 .
in addition , with the combination of models with different stellar mass and metallicity , it is possible not only to reproduce the new observed [ f / fe ] vs. @xmath16element relationship , but also the detailed f and @xmath16element abundance pattern observed in a particular star ( see e.g. fig . 3 in paper i ) . note that this is not possible for most of the stars when adopting the f determinations of jsl
. as already noted , sc - type stars ( except gp ori ) show larger f abundances than predicted by low - mass agb models with the same @xmath16element enhancement .
a possible explanation could be that the sc stars are more massive ( in average ) than the bulk of the n - type c - stars . from agb models with mass of about 3 - 5 m@xmath9
, we expect smaller @xmath21c pockets and weaker @xmath16element surface enrichments , even if the number of tps and tdu episodes is larger .
it occurs because the larger the core mass is , the smaller the he - rich intershell is with a steeper pressure gradient .
notwithstanding , the @xmath21c left in the ashes of the h - shell still provide a substantial contribution to the fluorine production .
recent studies of the luminosity function of galactic c - stars ( guandalini 2008 ) support such a hypothesis .
they indicate that the sc - type stars are among the most luminous agb c - stars .
once again , the case of wz cas is particularly extreme .
this star shows a huge f abundance ( [ f / fe]@xmath61 ) with almost no @xmath16element enhancement .
its large li abundance might be interpreted as a consequence of the hot bottom burning ( hbb ) mechanism which operates only for m@xmath62 m@xmath9 .
* however , the operation of the hbb would be at odds with the presence of f in the envelope since @xmath63o might destroy @xmath1f*. detailed abundance studies ( including f ) in a larger sample of sc - stars is needed to confirm this hypothesis .
fluorine abundances in galactic agb carbon stars are determined from high resolution infrared spectra using state - of - the - art c - rich spherical model atmospheres , updated atomic and molecular line lists and lte spectral synthesis .
we derive f abundances that are in average @xmath6 dex lower than those derived in previous studies . a clear correlation between the f and @xmath16element enhancements is found .
the new fluorine enhancements can be explained by standard low - mass agb stellar models .
f abundance determinations appear as a valuable tool to test theoretical agb models .
studies of the f abundance in metal poor unevolved and agb stars are mandatory to understand the galactic abundance evolution of this element and to evaluate the role of agb stars in its origin , respectively .
part of this work was supported by the spanish grants aya2008 - 04211-c02 - 02 and fpa2008 - 03908 from the mec .
p. de laverny and a. recio - blanco acknowledge the financial support of programme national de physique stellaire ( pnps ) of cnrs / insu , france .
ke gratefully acknowledges support from the swedish research council .
we are thankful to b. plez for providing us molecular line lists in the observed infrared domain .
abia , c. , & isern , j. 2000 , apj , 536 , 438 abia , c. , et al .
2002 , apj , 578 , 817 abia , c. , et al .
2009 , apj , 694 , 971 ( paper i ) busso , m. , gallino , r. , & wasserburg , g.j .
1999 , ara&a , 37 , 239 busso , m. , wasserburg , g.j . , nollett , k.m .
, & calandra , a. 2007 , apj , 671 , 802 cristallo , s. 2006 , pasp , 118 , 1360 cristallo , s. , et al .
2009 , apj , 696 , 797 cristallo , s. et al .
2010 , in preparation cunha , k. , smith , v.v . ,
lambert , d.l . , & hinkle , k.h .
2003 , aj , 126 , 1305 cunha , k. , & smith , v.v .
2005 , apj , 626 , 425 cunha , k. , smith , v.v . , &
gibson , b. 2008 , apj , 679 , l17 goriely , s. , & mowlavi , n. 2000 , a&a , 362 , 599 grevesse , n. , asplund , m. , & sauval , a.j .
2007 , space science reviews , 130 , 105 guandalini , r. 2008 , aip conf .
1001 , 339 gustafsson , b. et al .
2008 , a&a , 486 , 951 harris , m.j . ,
1987 , apj , 316 , 294 johnson , h.r . 1982 , apj , 260 , 254 jorissen , a. , smith , v.v . , & lambert , d.l .
1992 , a&a , 261 , 164 ( jsl ) lambert , d.l . ,
gustafsson , b. , eriksson , k , hinkle , k.h .
1986 , apjs , 62 , 373 lugaro , m. et al .
2004 , apj , 615 , 934 meynet , g. , & arnould , m. 2000 , a&a , 335 , 176 nollett , k.m .
, busso , m. , & wasserburg , g.j .
2003 , apj , 582 , 1036 otsuka , m. , izumiura , h. , tajitsu , a. , & hyung , s. 2008 , apj , 682 , l108 palacios , a. , arnould , m. , & meynet , g. 2005 , a&a , 443 , 243 renda , a. , et al .
2004 , mnras , 354 , 575 smith , v.v . , &
lambert , d.l .
1990 , apjs , 72 , 387 stancliffe , r.j . ,
2005 , mnras , 360 , 375 uttenthaler , s. et al .
2008 , apj , 682 , 509 werner , k. , rauch , t. , & kruk , j.w .
2005 , a&a , 433 , 641 woosley , s.e . , & haxton , w.c .
1988 , nature , 334 , 5 zamora , o. 2009 , ph .
d. thesis , university of granada lccccc @xmath64 & & & & & + aq sgr & 4.63@xmath65 ( 4 ) & 0.85 & 1.03 & 0.10 & + bl ori & 4.92@xmath66 ( 4 ) & 0.57 & 1.04 & 0.37 & 1.00 + r lep & 4.45@xmath67 ( 2 ) & 0.47 & 1.03 & 0.34 & + rt cap & 4.95@xmath68 ( 5 ) & 0.36 & 1.10 & 0.40 & + rv cyg & 4.70@xmath69 ( 4 ) & 0.63 & 1.20 & 0.15 & 0.07 + s sct & 4.62@xmath70 ( 4 ) & 1.08 & 1.07 & 0.10 & 0.40 + ss vir & 4.55@xmath71 ( 2 ) & & 1.05 & 0.00 & 0.30 + st cam & 4.72@xmath72 ( 4 ) & 0.94 & 1.14 & 0.20 & 0.45 + tu gem & 4.92@xmath71 ( 5 ) & 0.88 & 1.12 & 0.36 & + tw oph & 4.61@xmath73 ( 3 ) & 0.30 & 1.20 & 0.46 &
+ tx psc & 4.82@xmath69 ( 6 ) & 0.73 & 1.03 & 0.65 & 1.10 + u cam & 4.58@xmath71 ( 4 ) & 0.54 & 1.30 & 0.12 & 0.55 + u hya & 5.09@xmath74 ( 5 ) & & 1.05 & 0.61 & 1.10 + uu aur & 4.88@xmath24 ( 5 ) & 0.76 & 1.06 & 0.26 & 0.45 + ux dra & 4.85@xmath75 ( 4 ) & 0.64 & 1.05 & 0.45 & 0.65 + v460 cyg&4.65@xmath69 ( 4 ) & 0.68 & 1.06 & 0.14 & 0.80 + v aql & 4.62@xmath72 ( 5 ) & 0.54 & 1.25 & 0.00 & 0.20 + vy uma & 4.70@xmath74 ( 3 ) & 1.02 & 1.06 & 0.24 & 0.90 + w cma & 4.95@xmath74 ( 6 ) & 0.56 & 1.05 & 0.20 & 0.60 + w ori & 4.42@xmath76 ( 4 ) & 0.65 & 1.16 & -0.11 & 0.25 + x cnc & 4.93@xmath69 ( 3 ) & 0.76 & 1.14 & 0.67 & + y hya & 4.90@xmath65 ( 5 ) & 0.33 & 1.52 & 0.44 & + y tau & 4.57@xmath77 ( 6 ) & 0.49 & 1.04 & 0.01 & 0.05 + z psc & 4.90@xmath78 ( 4 ) & 0.42 & 1.01 & 0.40 & 0.97 + @xmath79 & & & & & + cy cyg & 5.05@xmath76 ( 4 ) & 0.98 & 1.04 & 0.39 & 0.10 + fu mon & 5.15@xmath69 ( 5 ) & 1.35&1.00 & 0.59 & 0.40 + gp ori & 5.26@xmath74 ( 6 ) & 0.94&1.00 & 0.70 & 1.25 + rz peg & 5.00@xmath69 ( 6 ) & 1.06&1.00 & 0.44 & 0.00 + wz cas & 5.80@xmath66 ( 6 ) & 0.40&1.01 & 1.15 & 0.16 + @xmath80 & & & & & + r scl & 4.18@xmath70 ( 3 ) & 1.22&1.34 & 0.09 & + ry dra & 4.48@xmath73 ( 3 ) & 0.44 & 1.18 & -0.08 & @xmath810.20 + t lyr & 4.35@xmath82 ( 2 ) & 0.45&1.29 & -0.21 & + vx and & 4.35 ( 1 ) & & 1.76 & -0.20 & 0.2 + y cvn & 4.40 ( 1 ) & 0.50&1.09 & -0.10 & @xmath810.2 + lcc r9 & & @xmath83 + r15 & @xmath20c@xmath84n , @xmath85 & @xmath86 + r16 & @xmath20c@xmath20c , @xmath87 & @xmath88 + r17 & @xmath20c@xmath84n , @xmath89 & @xmath90 + r22 & @xmath21c@xmath84n , @xmath91 & @xmath92 + r23 & @xmath20c@xmath84n , @xmath93 & @xmath94 +
|
an analysis of the fluorine abundance in galactic agb carbon stars * ( 24 n - type , 5 sc - type and 5 j - type ) * is presented .
this study uses the state - of - the - art carbon rich atmosphere models and improved atomic and molecular line lists in the @xmath0 m region .
f abundances significantly lower are obtained in comparison to previous study in the literature .
the main reason of this difference is due to molecular blends . in the case of carbon stars of sc - type ,
differences in the model atmospheres are also relevant .
the new f enhancements are now in agreement with the most recent theoretical nucleosynthesis models in low - mass agb stars , solving the long standing problem of f in galactic agb stars .
nevertheless , some sc - type carbon stars still show larger f abundances than predicted by stellar models .
the possibility that these stars are of larger mass is briefly discussed .
|
the increasing number of large surveys carried out over the last decade has revolutionized the fields of observational cosmology and astrophysics but , at the same time , it has posed several problems .
first of all , despite former surveys , which were primarily focused on censing and determining the properties of a given source class in a given band , recent surveys have been truly characterized by an observational multi - wavelength approach , whose final goal is to estimate the physical source properties over the entire electromagnetic spectrum and investigate how different selection criteria at different wavelengths are able to provide a complete census of such sources .
although an unbiased source selection is hard to achieve , current surveys in e.g. the obscured active galactic nuclei ( agn ) research field have allowed us to assess how biases work in source selection at different wavelengths , e.g. in the mid - ir vs. x - rays ( e.g. , fiore et al . 2008 , 2009 ; donley et al . 2008 and references therein ) .
similar conclusions can be easily extended to other source classes and bands . if , on the one hand , this approach is clearly consuming in terms of telescope / facility time and man - power , on the other hand it allows for a proper characterization of the intrinsic properties of a given source population .
although the science goals which motivate surveys at different wavelengths may be different , there are some common aspects which will be discussed in the following in order to understand the complex processes behind the build - up of source catalogs and the achievement of the published results .
motivated by the experience gained over the last decade in the field of surveys by members of our collaboration , hereafter some practical examples and problems related to the process of source identification will be presented . in particular
, we will mention some issues concerning the recent surveys carried out with and in the cosmic evolution survey ( cosmos ) field , keeping in mind that some problems can be generalized to other surveys and wavelengths .
the cosmos survey ( scoville et al .
2007a ) , an treasury project , is a deep and wide extra - galactic survey over a contiguous @xmath0 2 deg@xmath1 field designed to have sufficient area to limit the effects of cosmic variance in the investigation of e.g. large - scale structures ( lss ; scoville et al . 2007b ; gilli et al . 2009 ) , and depth to study the high - redshift galaxy and agn populations . among the wealth of multi - wavelength data ( not all covering the entire area
; see the review talk by mccracken , these proceedings ) in this region of the sky , the observations carried out with ( 1.55 ms ) and ( 1.8 ms ) the so - called xmm - cosmos and c - cosmos surveys ( hasinger et al .
2007 ; elvis et al .
2009 ) cover the entire 2 deg@xmath1 of the cosmos field and the innermost 0.9 deg@xmath1 , respectively . while observations are able to achieve a flux limit over 90% of the area of @xmath2 @xmath3 ( @xmath4 @xmath3 ; cappelluti et al .
2009 ) , making spectral analysis for a significant fraction of sources feasible ( e.g. , mainieri et al . 2007 ) , allows for a factor @xmath0 10 deeper x - ray flux detections in both the soft and hard bands .
these depths are well suited for both agn and normal / starburst galaxy studies in the band ( ranalli et al . , in preparation ) and , given the typically low background in exposures , allow detection of faint emission using stacking analysis techniques ( see fiore et al .
2009 for a recent application ) .
the final xmm- and c - cosmos catalogs comprise 1822 and 1761 point - like sources above a detection likelihood of 10 and 10.8 , respectively ( cappelluti et al .
2009 ; elvis et al .
2009 ) . despite the ultra - deep exposures in the deep fields ( north and south ; see hasinger & brandt 2005 for a review , and luo et al .
2008 for recent updates ) , which are focused on observing the same portion of the sky ( i.e. , centered on the same aimpoint ) repetitively , the mosaics in most of the current surveys are based on observations of neighbouring regions with large overlaps among the different observations ( but , for a different observing strategy , see aegis ; laird et al .
both xmm - cosmos and c - cosmos surveys are based on a tiling scheme which , in the latter survey , allows us to achieve a remarkably uniform exposure over the inner 0.45 deg@xmath1 ( for details , see elvis et al .
2009 and puccetti et al .
, in preparation ) .
this translates directly into a sharply defined flux limit .
in such a complex tiling strategy the same source is detected in more than one pointing with largely different point spread functions ( psfs ) , possibly causing a worse source position reconstruction , hence a larger positional error in the final catalog , despite the potentially sharp psf ( fwhm@xmath0 0.5 ) for on - axis source detection .
the first drawback of having a few arcsec `` error box '' associated to each source , depending on the source position across the field - of - view and the observed number of counts , is related to the match of the sources to the counterparts observed at other wavelengths .
this has strong consequences in the process of source identification and classification via optical and near - ir spectroscopy , making the already difficult task of spectroscopy ( because of the faint magnitudes of many counterparts ) quite challenging , since a larger number of sources should be observed and identified . having these issues in mind , the source catalogs in both xmm - cosmos and c - cosmos surveys
have been matched to catalogs at other wavelengths using the deep and good - quality i , k , and 3.6 data ( brusa et al .
2007 ; brusa et al . , in preparation ; civano et al .
, in preparation ) . in particular
, using near - ir data allows for a more complete source association , since a fraction of sources , especially at the very faint flux limits probed by , comprises obscured , optically faint ( because obscured and at high redshift ) agn , as expected from the agn synthesis models of background ( xrb ; e.g. , gilli , comastri & hasinger 2007 ) . the adopted source matching method consists of the extensively used likelihood ratio ( lr ) technique ( sutherland & saunders 1992 ; for some recent applications of this method , see ciliegi et al . 2003 ; brusa et al .
2005 ; laird et al .
2009 ; motch , these proceedings ) . for
a given optical / near - ir counterpart , lr is defined as the ratio between the probability that the source is the correct identification and the corresponding probability for a background , unrelated object . the chosen threshold for lr should therefore be a a careful compromise between completeness ( i.e. , most of the sources are finally recovered ) and reliability ( i.e. , the number of possible spurious associations must be kept as low as possible ) . without entering in too many details about this procedure ,
we note that this method has produced reliable source associations for the sources in both xmm- and c - cosmos surveys .
for instance , @xmath0 80% of the xmm - cosmos sources had already a counterpart using the i and k - band catalogs , and this fraction increased by @xmath0 14% using sharper positions .
the irac 3.6 data increased the number of associations by a further 9% and 5% for the xmm- and c - cosmos sources , respectively .
this means that the final fraction of sources without an optical / near - ir match ( i.e. , cases where either there are multiple sources in the `` error box '' above the adopted lr threshold or at least one possible counterpart below the lr threshold ) is 11% in xmm - cosmos ( brusa et al .
, in preparation ) and only 1% in c - cosmos ( civano et al . , in preparation ) , thus allowing for statistically reliable studies of the multi - wavelength properties of the sources , including those with faint optical counterparts .
besides the overall good matching results , we note that the adoption of the lr technique requires significant efforts , mostly aimed at verifying the reliability of source associations and identifications beyond their `` statistical '' validity .
the complexities and subtleties of such procedure have been faced at least in the cosmos survey using visual inspection of the candidate counterparts on a source - by - source basis , requiring significant man - power on month timescales , given the large number of sources available and the continuous updates in the reference catalogs , which are extensively used before their final releases . in this context
, virtual observatory ( vo ) can be of primary interest for people working in the field of galactic and extra - galactic surveys , provided that the available multi - wavelength catalogs are as uniform as possible ( e.g. , for photometric - redshift analyses and spectral energy distribution studies , magnitudes within the same extraction region should be homogeneously selected ) and rigorously checked . if such conditions were satisfied , a further implementation to current vo matching tools could be the selection of one or more matching methods within vo itself , and running tests of source reliability , possibly using simulations to reassure about the obtained results . given the way people are organized in present surveys ( i.e. , each group is mostly involved in reducing and analyzing data at a given wavelength , and finally in building up a source catalog ) , it must be said that implementing source - matching tasks in vo could be of limited use .
on the other hand , given the large amount of data that will be available in the near future using the next generation of ground- and space - based facilities , this issue will become more and more important and time - consuming , thus invoking for a significantly different approach in the way we currently deal with observational data .
in the following , some of the main next - generation facilities in the survey field will be outlined , along with potential problems in data handling and future catalog construction .
these facilities will provide powerful tools for source identification on relatively short timescales .
[ [ erosita ] ] over the next few years , one of the major advances in the field of surveys will be provided by the mission ( predehl et al .
2007 ) which is scheduled for launch in 2012 .
this mission is focused on surveys , being the natural continuation of the former all - sky survey performed in the soft band by the satellite in early 90 ( rass ; voges et al .
1999 ) . with an estimated psf half energy width ( hew ) of @xmath0 15 for on - axis observations ( and average psf hew of @xmath0 25over the entire field - of - view of 0.83 deg@xmath1 ) ,
is primarily intended to perform a 4-yr all sky survey , reaching a flux limit of @xmath5 ( @xmath6 ) @xmath3 in the band , with typical exposures of @xmath0 2 ks , and a 0.5-yr deep survey over 400 deg@xmath1 at flux limits about a factor @xmath0 5 and @xmath0 13 fainter in the two bands , respectively ( with an average exposure of @xmath0 27 ks ) . for what concerns the study of the agn physics and evolution
, the planned all - sky survey will allow for detections of @xmath0 4 million agn , 25000 ( 2100 ) of which being above redshifts @xmath7=3 ( 4 ) .
these numbers derive from the most up - to - date xrb synthesis models ( gilli et al .
2007 ) and calibrations from the team ; in particular , the on - flight background ( cosmic plus particle ) has been simulated for a l2 orbit . in survey mode , where source statistics is photon limited ,
the sensitivity has been estimated by imposing that a source can be detected if at least 9 net counts are present and by assuming a tipical power - law spectrum with photon index @xmath8=1.8 and galactic column density . then the source counts have been converted into a flux using the response matrices available at the time of writing . in deep survey mode
, the background counts have been accumulated in 30 circular apertures , and limiting counts have been estimated by imposing to have a 3@xmath9 detection above the background .
counts were then converted into fluxes by using the same spectral model as above .
this approach will represent a major improvement in the field of statistical studies of agn ; for comparison , less than one thousand agn are detected in the ultra - deep exposures in the cdf - n and cdf - s , and of the order of a couple of thousands in cosmos ( see @xmath10[xray_matches ] ) .
the deep survey , although on a limited region , will allow for detections of faint agn at high redshifts , which are currently poorly sampled by current surveys ( e.g. , see fig . 3 of vignali et al .
2005 ) .
the large amount of sources detected by will require , at some level , multiple campaigns of photometric and spectroscopic follow - up observations . just to give an idea of the numbers involved in these follow - ups , at the conservative flux limit of @xmath3 for the deep survey
, we find @xmath0 570 sources in xmm - cosmos , with the faintest sources having an i - band ( 3.6 ) magnitude of @xmath0 24 ( 22 ; both magnitudes are in the ab system ) .
covering 400 deg@xmath1 at such depth is currently hard to accomplish , and the whole spectroscopic coverage will be even more complex and time - consuming . given the psf , the source identification will not be an easy process and will take advantage of follow - up observations , especially in the near - ir , where the planned and surveys will provide a significant contribution to source identification , hence to observational cosmology .
these and other planned facilities that will be available to the community over the coming years are briefly reviewed in the following .
[ [ wise ] ] the wide - field infrared survey explorer ( ) is a nasa mission ( launch scheduled for mid 2009 ) aimed at surveying the entire sky in four bands ( 3.3 , 4.7 , 12 , and 23 ) , with sensitivities of 0.12 , 0.16 , 0.65 and 2.6 mjy ( after eight repetitions of 8.8s each ) , respectively .
such sensitivities are worse than those achieved by the instruments onboard the satellite , but are a factor @xmath0 1000 better than those obtained with the iras surveys ( mainzer et al .
2006 ) . furthermore , although the resolutions allowed by are typically worse ( fwhm of 6 and 12 in the near - ir and mid - ir , respectively ) than , the large field - of - view of ( 47 arcmin@xmath1 ) will allow to perform a survey of the entire sky in just six months .
this means that studies like those regarding obscured high - redshift agn by means of mid - ir to optical flux ratio selection , as those carried out recently by several authors ( e.g. , daddi et al .
2007 ; fiore et al .
2008 , 2009 ; dey et al . 2008 ; lanzuisi et al .
2009 ) , will be possible over much larger sky regions ( although at higher flux limits ) , provided that coverage at other wavelengths ( to select and properly characterize the sources ) will be granted . similarly ,
detection of obscured agn from shallow and moderate - depth surveys up to high - redshifts will be possible , as shown in fig .
[ scuba_sed_persp ] , where the optical , infrared and sub - mm spectral energy distribution ( sed ) of a luminous type 2 quasar at @xmath7=1.957 detected by the hellas2xmm survey ( fiore et al .
2003 ) is presented . the intense star formation derived for this source ( @xmath0 1500 )
is suggestive of an agn passing through a complex phase , being caught presumably before the process of expelling the obscuring gas ( thus quenching the star formation ) and becoming a broad - line unobscured quasar has started .
agn with similar properties do not seem to be uncommon at high redshifts , as shown in recent work ( e.g. , mainieri et al . 2005 ; polletta et al .
2008 ) . through proper
broad - band sed fitting , and will be able to constrain the individual components ( stellar , agn , and starburst ) and estimate the nuclear bolometric luminosity ( see vignali et al .
2009 for details ) .
[ [ vista ] ] the visible and infrared survey telescope for astronomy ( ) is a 4-m wide - field ( 1.65 deg diameter ) survey telescope equipped with a near - ir camera facility ( at least at the beginning of the operations , although it is also capable of operating in the visible with a suitable camera ) , covering the 0.852.3 wavalength range with five broad - band filters ( @xmath11 , @xmath12 , @xmath13 , @xmath14 , and @xmath15 ) and a narrow - band filter centered at 1.18 .
located at the eso cerro paranal observatory in chile , the planned capabilities are such to suitably match the needs of the 8 m - class telescopes of vlt . the large field - of - view betrays that is mainly designed for carrying out large - area surveys in the near - ir domain . in this regard , it is properly suited for follow - up identification programs ( e.g. , see @xmath10 [ erosita ] ) .
one of the proposed surveys is the hemisphere survey that will image @xmath0 20000 deg@xmath1 in the @xmath13 and @xmath15 bands . about 40% of this survey , dedicated to a deep observational program searching for the dark energy ( dark energy survey des )
, will cover @xmath0 5000 deg@xmath1 also in the @xmath14 band over initially 125 nights down to 10@xmath9 ab magnitude limits in the @xmath13 , @xmath14 , and @xmath15 bands of 20.4 , 20.0 , and 19.4 , respectively ( banerji et al .
furthermore , the ultravista survey is aimed at imaging the cosmos field down to unprecedented depths , taking advantage also of the deep 3.64.5 coverage with the planned extended deep survey ( seds ) `` warm - mission '' project .
the ultravista survey will use all of the near - ir filters , including the narrow - band one to search for ly@xmath16 emitters at very high redshifts ( @xmath17 ) . finally , we mention also the deep extragalactic observations survey ( video ) , which will cover 15 deg@xmath1 of the sky ( some of which already included in swire ) with the @xmath11 , @xmath12 , @xmath13 , @xmath14 and @xmath15 filters , jointly with the recently approved extragalactic representative volume survey ( servs ) project .
the panoramic survey telescope and rapid response system ( ps . ) consists of four telescopes of 1.8 m each located in hawaii ( the prototype ps1 being already on site ) .
its innovative design for wide - field imaging , which comprises a 64@xmath1864 ccd array ( 600@xmath18600 pixel each , for a total of 1.4 billion pixels ) , is such to allow observations of the available sky ( @xmath0 30000 deg@xmath1 ) several times each month , down to @xmath19 ( 5@xmath9 ) . although the immediate goals of ps .
are related to discover and characterize earth - approaching objects , the huge amount of images guaranteed by this facility will allow for fast photometric observations of large sky regions ( up to @xmath0 3000 - 6000 deg@xmath1 in one night ) .
furthermore , the wavelength range of the available filters ( @xmath20 , @xmath21 , @xmath22 , @xmath7 , and @xmath23 , covering the @xmath0 470010500 interval ) is suited to provide a first - order characterization of the sources , including an estimate of their redshift via photo-@xmath7 techniques . [ [ lsst ] ] the large synoptic survey telescope ( ) represents one of the most revolutionary ideas for the next decade .
aimed at probing dark energy and dark matter , mapping the solar system and the milky way , and exploring transient object , will offer unique opportunities for large - area surveys .
it consists of a large , wide - field ground - based system designed to obtain multiple images covering the @xmath0 30000 deg@xmath1 of sky which are visible from cerro pachn in chile .
the current baseline design is characterized by an 8.4 m ( 6.5 m effective ) primary mirror with a 9.6 deg@xmath1 field - of - view and a 3.2-billion pixel camera ( gilmore et al .
2008 ) , and allows observations of about 10000 deg@xmath1 of sky using pairs of 15s exposures in two of the six photometric bands ( @xmath24 , @xmath20 , @xmath21 , @xmath22 , @xmath7 , and @xmath23 ) every three nights on average , with typical 5@xmath9 depth for point - like sources of @xmath25 ( ivezic et al .
the `` deep - wide - fast '' survey mode will be assigned 90% of the time , thus allowing for observations of @xmath0 20000 deg@xmath26 in the six bands ( over the ten years of operations ) and reaching a final depth of @xmath27 .
the remaining 10% of the time will be allocated to special programs such as the `` very deep and fast time domain survey '' , reaching @xmath28 .
three billion galaxies are expected to have reliable photometric redshifts at the end of the observations , and 2080 million agn should be detected ( i.e. , two orders of magnitude larger than the current agn census ) ; a fraction of agn ( depending on their absolute magnitudes , redshifts , and observed time intervals ) will be recognized as such on the basis of their variability , thanks to the multiple observing - epoch strategy . while the previous facilities will focus on imaging observations of large sky regions , here we present a couple of new facilities which will be available to the community to carry out spectroscopic programs .
the first is the hetdex ( hobby - eberly telescope dark energy experiment)/virus ( visible integrated - field replicable unit spectrographs ) project , which consists on an innovative set of spectrographs ( for a total number of @xmath0 34000 fibers ) to observe thousands of galaxies each night .
the hetdex / virus survey , with the primary goal of exploring the issues related to the dark energy through a three - dimensional map of a large volume of space , is scheduled for 20102013 and is going to provide redshifts for about 1 million galaxies . over the 20092014 period , the
sloan digital sky survey iii ( sdss - iii ) will carry out a program of four surveys on three scientific themes : dark energy , dynamics and chemical evolution of the milky way , and studies of planetary systems .
in particular , the baryon oscillation spectroscopic survey ( boss ) project , in order to map the spatial distribution of luminous galaxies and quasars , will survey 10000 deg@xmath1 and provide redshifts for @xmath0 1.5 million galaxies up to @xmath7=0.7 , and 160000 quasars at @[email protected] , using a 1000-fiber spectrograph with resolution r@xmath02000 .
given the large amount of data that will be available to the community in the future , it seems plausible that vo can have an effective role in data handling and related tasks , as actually presented by many authors during this workshop . as pointed out in @xmath10[xray_matches ] , large multi - wavelength datasets require homogeneous data selection and validation , and quality checks . having said this , it is imperative for all of the people releasing reduced data and corresponding source catalogs to carry out high - level quality checks and exhaustive accompanying information . from an perspective
, we have shown that the process of building up and cross - correlating catalogs using proper matching criteria ( e.g. , the likelihood ratio technique ) is time - consuming and requires , at some level , visual inspection to support the statistical reliability of the matched sources . if , on the one hand , some checks are clearly beyond the possibilities of vo , on the other hand data homogenization and automatization of some matching procedures ( and corresponding checks )
would be highly appreciated by the community .
such implementations would allow for a quicker , more extensive and exhaustive exploitation of the available datasets .
similarly , automated tasks to perform stacking analyses at different wavelengths and estimate photometric redshifts using broad - band matched catalogs would greatly enhance the use of vo , especially by people not directly ( or deeply ) involved in all the issues and problems related to surveys and data handling .
synergy with vo is undoubtedly a challenging issue over the coming years when billion - object catalogs will be available and the necessity of optimizing the many - sided scientific outputs on relatively short timescales will become more and more pressing .
cv acknowledges eso financial support ; the authors would like to thank the members of the hellas2xmm , elais - s1 , xmm - cosmos , and c - cosmos collaborations .
banerji m. , abdalla f. b. , lahav o. , lin h. 2008 , 386 , 1219 brandt w.n . &
hasinger g. 2005 , 43 , 827 brusa m. , comastri a. , daddi e. , et al .
2005 , 432 , 69 brusa m. , zamorani g. , comastri a. , et al .
2007 , 172 , 353 cappelluti n. , brusa m. , hasinger g. , et al .
2009 , , in press ( arxiv:0901.2347 ) ciliegi p. , zamorani g. , hasinger g. , lehmann i. , szokoly g. , wilson g. 2003 , 398 , 901 daddi e. , alexander d. m. , dickinson m. , et al .
2007 , 670 , 173 dey a. , soifer b.t . , desai v. , et al .
2008 , 677 , 943 donley j.l .
, rieke g.h .
, prez - gonzlez p.g . , barro g. 2008 , 687 , 111 elvis m. , civano f. , vignali c. , et al .
2009 , , submitted fiore f. , brusa m. , cocchia f. , et al .
2003 , 409 , 79 fiore f. , grazian a. , santini p. , et al .
2008 , 672 , 94 fiore f. , puccetti s. , brusa m. , et al .
2009 , , in press ( arxiv:0810.0720 ) gilli r. , comastri a. , hasinger g. , 2007 , 463 , 79 gilli r. , zamorani g. , miyaji t. , et al .
2009 , , in press ( arxiv:0810.4769 ) gilmore k. , kahn s. , nordby m. , et al .
2008 , 7014 hasinger g. , cappelluti n. , brunner h. , et al .
2007 , 172 , 29 ivezic z. , tyson j.a . , allsman r. , et al .
2008 , arxiv:0805.2366 laird e.s .
, nandra k. , georgakakis a. , et al .
2009 , 180 , 102 lanzuisi g. , piconcelli e. , fiore f. , feruglio c. , vignali c. , salvato m. , gruppioni c. , 2009 , , submitted luo b. , bauer f.e .
, brandt w.n . , et al . 2008 , 179 , 19 mainieri v. , rigopoulou d. , lehmann i. , et al .
2005 , 356 , 1571 mainieri v. , hasinger g. , cappelluti n. , et al .
2007 , 172 , 368 mainzer a.k . , eisenhardt p. , wright e.l .
, liu f .- c .
, irace w. , heinrichsen i. , cutri r. , duval v. 2006 , 6265 mccracken h.j .
2009 , these proceedings motch c. 2009 , these proceedings polletta m. , omont a. , berta s. , et al .
2008 , 492 , 81 predehl p. , andritschke r. , bornemann w. , et al .
2007 , 6686 scoville n. , aussel h. , brusa m. , et al .
2007a , 172 , 1 scoville n. , aussel h. , benson a. , et al .
2007b , 172 , 150 sutherland w. & saunders w. 1992 , 259 , 413 vignali c. , brandt w.n . , schneider d.p . ,
kaspi s. 2005 , 129 , 2519 vignali c. , pozzi f. , fritz j. , et al . 2009 , , submitted voges w. , aschenbach b. boller th . , et al .
1999 , 349 , 389
|
here we review some of the main issues related to multi - wavelength source identification and characterization , with particular emphasis on the field of surveys carried out over the last years .
this complex and time - consuming process is going to represent one of the main difficulties over the coming years , when significantly larger surveys , both in area and depth , will be carried out with the new generations of space- and ground - based facilities like e.g. , , , ps . , and .
the virtual observatory can offer a reliable way to approach to a new concept of data handling and multi - wavelength source characterization , provided that uniform and rigorous data analyses and extensive quality checks are performed .
|
in the past couple decades , tremendous effort has been put forth towards incorporating thin - film magnetic cores into integrated passive circuit elements , including inductors and transformers @xcite , for a variety of applications .
the traditional operating frequency and flux - amplification properties of integrated magnetic films have made them ideal for power management applications up to now @xcite .
however , the cost of depositing such films thick enough for power electronics devices has limited their usage in integrated systems to date . more recently , as mobile electronics have begun to demand higher inductances and quality factors in smaller form factors than previously achievable by spiral air - core inductors in that frequency range @xcite , magnetic cores are being redesigned to compete to meet the new need @xcite .
, of the device compared to its air - core equivalent .
the inductance enhancement was evaluated within the frequency band where the inductance is flat . of the four inductors clustered at high frequencies ,
two belong to this work , while the other two belong to @xcite .
the solid circles mark this work , embodying a combined improvement towards higher q factor and operating frequency .
, width=336 ] while the high permeability of magnetic materials increases the dc value of inductance , the challenge often faced is to extend that permeability and corresponding inductance enhancement to higher frequencies of use to mobile applications ( namely in the 1 - 5 ghz range ) . at the intrinsic ferromagnetic resonance ( fmr )
frequency of magnetic materials ( typically between 1 - 2 ghz for large blanket films ) , the relative permeability drops to unity and magnetic loss tangent peaks , therefore making the inductance enhancement due to the material negligible and , instead , the losses due to it dominant . nevertheless , through various means of deposition and patterning the frequency response of the magnetic permeability can be improved and extended over a larger bandwidth .
figure [ review ] summarizes the performance of various high frequency magnetic - core inductors from literature .
this work provides a detailed examination of the optimization choices involved in maximizing the inductance enhancement and bandwidth , while simultaneously attempting to maximize the quality factor of magnetic - core inductors .
this paper discusses the most crucial considerations regarding material selection and fabrication process as well as magnetic - core and device design .
such optimizations were then applied , yielding a fabricated device with the highest combined performance for an integrated magnetic - core inductor at high frequencies ; the device offers a magnetic enhancement over air core extending beyond 6 ghz , a low frequency inductance of 1nh , and a peak quality factor of 4 at approximately 3 ghz .
high frequency performance of magnetic inductors depends heavily upon the magnetic - core material and design .
high permeability magnetic materials provide greater inductance enhancements over their air - core equivalents , but simultaneously suffer from higher eddy currents and lower fmr frequencies .
elimination of eddy currents by lamination - breaking up the conductive magnetic core with insulating spacers to limit the flow of eddy currents - is one component of improving the high frequency behavior , however the largest contribution to the improved frequency response of the inductors comes from increasing the fmr frequency and thereby reducing magnetic losses in the frequency range of interest . extending the bandwidth of magnetic materials by increasing the fmr frequency
can be done in a couple of ways involving using the shape of the magnetic material to change its uniaxial anisotropy and resultant permeability spectrum .
uniaxial anisotropy keeps the magnetization of the inductor core in the plane of the film with the ac permeability measured perpendicular to the magnetization direction corresponding to the rotation of magnetization in a small signal applied magnetic field .
the first method of increasing the fmr frequency is to pattern the dimensions of the core so as to maximize the length of the magnetic core parallel to its in - plane uniaxial magnetization .
this practice has been shown to increase the fmr frequency while consequently also reducing the permeability as a fundamental trade - off @xcite .
the second method builds on this and utilizes ultra - thin laminations in the thickness direction to also increase the fmr frequency @xcite .
both of these methods were combined in the design of the magnetic core inductors described in this paper .
the core was patterned into a 250 @xmath1 m wide x 1000 @xmath1 m long rectangular bar with the magnetization along the length direction .
figure [ bhloops ] shows the overlapped magnetization hysteresis curves for the blanket and patterned films , indicating an increase in anisotropy field and a decrease in permeability for the patterned film .
the anisotropy , marked by the field required for saturation , was seen to increase by a factor of 2@xmath2 .
0.45 0.45 a comparison of the permeability spectra of the blanket ( measured ) and patterned film ( calculated ) in fig .
[ permeability ] confirms that there is a corresponding decrease in the low frequency permeability value .
the shape - dependent permeability spectrum of the magnetic core was calculated according to eqn .
[ analyticalpermeability ] , where the material parameters were measured and where the demagnetization factors were simulated using the same method described in @xcite .
@xmath3 where @xmath4 @xmath5 @xmath6 @xmath7 @xmath8 is the gyromagnetic ratio , @xmath9 is the magnetic anisotropy field , @xmath10 is the saturation magnetization , and @xmath11 , @xmath12 , and @xmath13 are the demagnetization factors along the out - of - plane dimension , width dimension , and length dimension ( easy - axis ) , respectively . in the measured film spectra ,
the fmr peak in the blanket film appears suppressed , likely as a result of large eddy currents or interlaminate capacitance @xcite . and approximate demagnetization factors obtained through simulation ( @xmath14 , @xmath15 , and @xmath16 ) .
the fmr frequency of the blanket film is @xmath02.2 ghz , while that of the patterned film is @xmath03.3 ghz ( above the 3 ghz limit of the measurement setup).,width=336 ] the fmr frequency can be specifically calculated by the following kittel equation : @xmath17\left [ h_k+(n_x - n_y ) m_s \right ] } \ ] ] nevertheless , by looking at the imaginary permeability peak location , which represents the fmr frequency and the point at which magnetic losses begin to dominate , a direct comparison can be made of the bandwidth of the magnetic cores . therefore , patterning of the magnetic core effectively increased its bandwidth by more than 1 ghz . increasing the bandwidth
is accompanied by the consequential decrease in the permeability , as seen in fig .
[ permeability ] .
therefore , although pushing the magnetic losses out to higher frequencies potentially improves the quality factor ( q ) , the inductance decrease with permeability somewhat negates the effect . in order to compensate for the reduction in inductance while also minimizing the resistance of the inductor , the structure was extensively simulated for both maximum inductance and maximum q factor .
the greatest two design parameters contributing to increasing the inductance ( with least effect on the resistance ) are reducing the width of the windings and reducing the spacing between windings . both of these reductions work to increase the efficiency of flux linkage from one winding to the next without increasing the length the solenoid coil . the final width and spacing selected
were each 20 @xmath1 m .
although reducing the width of the windings increases inductance , it also increases the resistance of the coil .
therefore , after optimizing for inductance , the resistance had to be optimized to improve the q. increasing the thickness of the coil can accomplish this ; while it has minimal effect on the inductance , it increases the q because it reduces the coil resistance . due to fabrication constraints , however , the coil thickness in this work was limited to 3 @xmath1 m thick copper windings and the magnetic core thickness was limited to 1 @xmath1 m thick cofeb film .
higher q and higher inductance would be expected by increasing the thicknesses of each , respectively . despite the fabrication limitations
, this work demonstrates a major push forward for magnetic inductors operating at high frequencies .
the solenoid inductors shown in fig . [ inductors ] were fabricated to demonstrate a direct comparison between air - core ( left ) and magnetic - core ( right ) inductors of the same design to evaluate the contribution due to the magnetic film .
fabrication of the inductors involved six major phases : isolation , bottom inductor windings , planarization and insulation , magnetic core , planarization and insulation , and vias and top inductor windings . in the first phase ,
40 @xmath1 m of the commercial polymer su-8 2015 was formed on the surface of a 4 in silicon wafer to provide sufficient isolation between the radio frequency ( rf ) inductor and potential substrate parasitics .
bottom inductor windings , top inductor windings , and vias were deposited using a through - mold electroplating process : a ti / cu seed layer was first evaporated onto the wafer , photoresist patterns provided walls for a subsequent electroplating process to build 3 @xmath1 m thick cu windings , and finally , the photoresist was removed and the non - plated seed layer areas were etched away .
a two - layer su-8 2002 polymer process was developed to first fill the gaps between windings for planarization and then provide an additional 1.5 @xmath1 m insulation layer between the windings and the magnetic layer .
the material cofeb was selected for the magnetic core based on its previously demonstrated excellent high frequency properties @xcite .
the magnetic core was laminated to break up the conductive magnetic core with insulating spacers that further improve broadband performance by reducing high frequency losses due to eddy current circulation in the thickness direction .
the magnetic core was deposited in a single vacuum chamber under an applied magnetic field used to induce a uniaxial magnetic anisotropy along the length of the inductor .
rf sputter deposition alternated between 63 nm co~43~fe~43~b~14~ ( at% ) and 6 nm sio~2~ to construct a 16x(cofeb / sio~2~ ) laminated magnetic stack with total thickness of approximately 1 @xmath1 m .
the stack was then patterned using optical photolithography and etched in a 1:1:2 ratio of hf : hno~3~:h~2~o .
proper alignment between the external magnetic field during deposition and the subsequent lithographic patterning of the cores is critical for ensuring the ideal , linear , easy - axis domains in the magnetic core that provide broadband constant inductance @xcite . 0.45 0.45
to directly evaluate the performance enhancement of the inductor due to the magnetic material , similar inductors both with and without the magnetic core were measured .
the inductance , quality factor ( q ) , and resistance results are presented in fig .
[ results ] .
incorporation of the magnetic core is seen to double the inductance of the air - core inductor , providing approximately 1 nh inductance into the ghz range .
decreasing inductance at higher frequencies is likely due to ferromagnetic resonance , as well as a combination of misalignment of the applied magnetic field during deposition of the cores @xcite , parasitic capacitances , and eddy currents along the length of the core .
evidence of misalignment of the magnetic anisotropy can be seen from the hysteresis loops in fig .
[ ealoops ] .
a peak quality factor of 4 was achieved at approximately 3 ghz , representing the best performance of a high - frequency magnetic - core inductor to date .
increase in frequency is attributed to the increase in bandwidth of the magnetic core by pushing the fmr frequency higher .
this understanding is supported by the parallel trends in both the high frequency resistance results and the typical imaginary permeability spectrum ( representing magnetic losses ) .
this direct correspondence indicates that the main losses contributing to the increase in resistance at high frequencies and the decrease in quality factor are due to the magnetic core .
substrate effects are negligible due to the very thick , 40 @xmath1 m , isolation layer separating the inductor from the si substrate . for best comparison , the dc resistances of the inductors should be identical
. however , since the inductors were selected from different locations on the wafer some wafer - level process variations ( coil thickness , etch - back , via - opening , etc . ) , the low - frequency resistances of the inductors differ slightly from each other , leading to undesired reductions in q for the inductors with higher resistance . nevertheless , as expected , the inductor with higher inductance and lower dc resistance yielded the highest quality factor .
compared to previously published air - core @xcite and magnetic - core @xcite inductors , the results suggest a very competitive new advancement of magnetic inductors for rf power amplifiers and other applications . 0.5 0.5 0.5 while previous work has demonstrated an increase in inductor frequency due to similar patterning for higher fmr frequencies , either the inductance enhancements of those designs were very low ( @xmath181.2@xmath2 ) @xcite or the q - factors were low ( @xmath192.5 ) @xcite .
the inductors shown here are the first to demonstrate overall optimization of inductance enhancement , quality factor , and operating frequency , simultaneously . of the previously reported integrated magnetic inductors that had similar inductance or inductance enhancement values , much lower peak quality factor frequencies
were observed @xcite . in some cases ,
the value of the peak q - factor was similar @xcite , but in other cases , the q was in the range of 10 - 20 @xcite .
the slightly lower quality factor in the current design can be attributed to a number of factors including the increased dc inductor winding resistance due to narrower and thinner copper lines ( 20 @xmath1 m wide and 3 @xmath1 m thick in this work compared to 150 @xmath1 m wide and 4.5 @xmath1 m thick in @xcite ) and other process variations .
narrower inductor windings , however , result in better inductive coupling , thereby increasing the inductance and compensating for the decrease in permeability accompanying the narrow patterning of the magnetic inductor . furthermore , with larger copper thickness , the quality factor is expected to increase to comparable or higher values than those reported in other works .
these results in addition to simulations confirm that magnetic - core inductors can , indeed , be used to satisfy the requirements of future rf and mobile applications .
the methodology for design and fabrication of high - frequency magnetic - core inductors was presented .
furthermore , integrated magnetic inductors have been demonstrated with increased frequency bandwidth to meet the need for ghz - range power amplifier applications .
the thin - film magnetic core inductors have an inductance of 1 nh , embodying a @xmath20 inductance enhancement over its air - core counterpart at low frequencies and a @xmath21 enhancement at the frequency of the peak quality factor .
the 3 ghz frequency and peak quality factor of 4 are the highest reported combined performance to date for a magnetic core inductor .
extension of the operating frequency of the inductor is attributed to the use of patterning and laminations to increase shape anisotropy in the magnetic material and increase its ferromagnetic resonance frequency , reducing the magnetic losses within the desired operating frequency range . with continued design optimization of the inductor windings and magnetic core , it is expected that the unmet need for high - q , high - frequency inductors may soon be met with magnetic materials .
research for this project was conducted with government support under fa9550 - 11-c-0028 and awarded by the department of defense , air force office of scientific research , national defense science and engineering graduate ( ndseg ) fellowship , 32 cfr 168a .
part of this work was performed using the stanford nanofabrication facility ( snf ) and at the stanford nano shared facilities ( snsf ) at stanford university .
p. dhagat , s. prabhakaran , and c. r. sullivan , `` comparison of magnetic materials for v - groove inductors in optimized high - frequency , '' _ ieee transactions on magnetics _
40 , no . 4 , pp . 20082010 , 2004 . c. r. sullivan , s. prabhakaran , p. dhagat , and y. sun , `` thin - film inductor designs and materials for high - current low - voltage power , '' _ transactions of the magnetics society of japan _
, vol . 3 , no . 4 ,
pp . 1268 , 2003 .
s. prabhakaran , c. r. sullivan , t. odonnell , m. brunet , and s. roy , `` microfabricated coupled inductors for dc - dc converters for microprocessor power delivery , '' _
pesc record - ieee annual power electronics specialists conference _ , vol . 6 , pp .
44674472 , 2004 .
r. meere , n. wang , t. odonnell , s. kulkarni , s. roy , and s. c. omathuna , `` magnetic - core and air - core inductors on silicon : a performance comparison up to 100 mhz , '' _ ieee transactions on magnetics _ , vol . 47 , no . 10 , pp . 44294432 , 2011 .
n. sturcken , e. j. o. sullivan , n. wang , p. herget , b. c. webb , l. t. romankiw , m. petracca , r. davies , r. e. fontana , g. m. decad , i. johnkymissis , a. v. peterchev , l. p. carloni , w. j. gallagher , and k. l. shepard , `` a 2.5d integrated voltage regulator using coupled - magnetic - core inductors on silicon interposer , '' _ ieee journal of solid - state circuits _ , vol .
48 , no . 1 ,
pp . 244254 , 2013 .
d. w. lee , k .-
hwang , and s. x. wang , `` fabrication and analysis of high - performance integrated solenoid inductor with magnetic core , '' _ ieee transactions on magnetics _
44 , no . 11 , pp . 40894095 , 2008 .
k. ikeda , k. kobayashi , k. ohta , r. kondo , t. suzuki , and m. fujimoto , `` thin - film inductor for gigahertz band with cofesio - sio / sub 2/2 multilayer granular films and its application for power amplifier module , '' _ ieee transactions on magnetics _
39 , no . 5 , pp .
30573061 , sep 2003 .
d. gardner , g. schrom , f. paillet , b. jamieson , t. karnik , and s. borkar , `` review of on - chip inductor structures with magnetic films , '' _ ieee transactions on magnetics _ , vol .
45 , no .
47604766 , oct 2009 .
y. gao , s. z. zardareh , x. yang , t. x. nan , z. y. zhou , m. onabajo , m. liu , a. aronow , k. mahalingam , b. m. howe , g. j. brown , and n. x. sun , `` significantly enhanced inductance and quality factor of ghz integrated magnetic solenoid inductors with fegab / al2o3 multilayer films , '' _ ieee transactions on electron devices _ , vol . 61 , no . 5 , pp . 14701476 , 2014 .
e. yun , m. jung , c. cheon , and h. g. nam , `` microfabrication and characteristics of low - power high - performance magnetic thin - film transformers , '' _ ieee transactions on magnetic _ , vol .
40 , no . 1 , pp . 6570 , 2004 .
f. m. rhen , p. mccloskey , t. odonnell , and s. roy , `` high - frequency permeability of electroplated conife and conife - c alloys , '' _ journal of magnetism and magnetic materials _ , vol .
320 , no . 20 , pp .
e819e822 , oct 2008 . a. m. leary , p. r. ohodnicki , and m. e. mchenry , `` soft magnetic materials in high - frequency , high - power conversion applications , '' _ journal of minerals , metals , and materials _ , vol .
64 , no . 7 , pp . 772781 , 2012 .
s. shen , p. r. ohodnicki , s. j. kernion , a. m. leary , v. keylin , j. f. huth , and m. e. mchenry , `` nanocomposite alloy design for high frequency power conversion applications , '' _ energy technology 2012 : carbon dioxide management and other technologies _ , pp . 275282 , 2012 .
m. yamaguchi , m. baba , k. suezawa , t. moizumi , k. i. arai , y. shimada , a. haga , s. tanabe , and k. ito , `` magnetic rf integrated thin - film inductors , '' in _ microwave symposium digest .
2000 ieee mtt - s international _ , no .
c , 2000 , pp .
205208 .
m. yamaguchi , s. bae , k. h. kim , k. tan , t. kusumi , and k. yamakawa , `` ferromagnetic rf integrated inductor with closed magnetic circuit structure , '' in _ ieee mtt - s international microwave symposium digest _ , 2005 , pp .
351354 .
m. yamaguchi , k. suezawa , y. takahashi , k. i. arai , s. kikuchi , y. shimada , s. tanabe , and k. ito , `` magnetic thin - film inductors for rf - integrated circuits , '' _ journal of magnetism and magnetic materials _
215 , pp . 807810 , 2000 .
m. yamaguchi , k. suezawa , m. baba , k. arai , y. shimada , s. tanabe , and k. itoh , `` application of bi - directional thin - film micro wire array to rf integrated spiral inductors , '' _ ieee transactions on magnetics _ , vol .
36 , no . 5 , pp .
35143517 , 2000 . w. xu , s. sinha , t. dastagir , h. wu , b. bakkaloglu , d. s. gardner , y. cao , and h. yu , `` performance enhancement of on - chip inductors with permalloy magnetic rings , '' _ ieee electron device letters _
32 , no . 1 ,
pp . 6971 , 2011 . w. xu , h. wu , d. s. gardner , s. sinha , t. dastagir , b. bakkaloglu , y. cao , and h. yu , `` sub-100 @xmath1 m scale on - chip inductors with cozrta for ghz applications , '' _ journal of applied physics _ ,
109 , no . 7 , p. 07a316
, 2011 .
g. perrin , j. c. peuzin , and o. acher , `` control of the resonance frequency of soft ferromagnetic amorphous thin films by strip patterning , '' _ journal of applied physics _ ,
81 , no . 8 , p. 5166 , 1997 .
k. kim , m. yamaguchi , and j. kim , `` ferromagnetic resonance behaviors of integrated copdalo magnetic film on coplanar waveguide , '' _ journal of the korean physical society _ ,
51 , no . 6 , pp .
20262030 , 2007 .
m. vroubel , y. zhuang , b. rejaei , j. burghartz , a.m. crawford , and s. wang , `` calculation of shape anisotropy for micropatterned thin fe ni films for on - chip rf applications , '' _ ieee transactions on magnetics _ , vol .
40 , no . 4 ,
pp . 28352837 , jul .
2004 .
r. m. walser , w. win , and p. m. valanju , `` shape - optimized ferromagnetic particles with maximum theoretical microwave susceptibility , '' _ ieee transactions on magnetics _ , vol .
34 , no . 4 , pp . 13901392 , 1998 .
l. h. chen , t. j. klemmer , k. a. ellis , r. b. van dover , and s. jin , `` soft - magnetic properties of fe co b thin films for ultra - high - frequency applications , '' _ journal of applied physics _
87 , no . 9 , pp . 14 , 2000 .
g. a. lebedev , b. viala , t. lafont , d. i. zakharov , o. cugat , and j. delamare , `` converse magnetoelectric effect dependence with cofeb composition in ferromagnetic / piezoelectric composites , '' _ journal of applied physics _
111 , no . 7 , p. 07c725
, 2012 .
a. el - ghazaly , j. m. mullenix , r. m. white , and s. x. wang , `` kerr - imaged edge - curling wall effects of narrow magnetic cores , '' _ ieee transactions on magnetics _ , vol .
49 , no . 7 , pp . 40174020 , jul 2013 . j. lopez - villegas , j. samitier , and c. cane , `` improvement of the quality factor of rf integrated inductors bylayout optimization , '' _ ieee transacations on microwave theory and techniques _ , vol . 48 , no . 1 , pp . 7683 , 1998 .
c. d. meyer , s. s. bedair , b. c. morgan , and d. p. arnold , `` high - inductance - density , air - core , power inductors , and transformers designed for operation at 100 - 500 mhz , '' _ ieee transactions on magnetics _ , vol .
46 , no . 6 , pp .
22362239 , 2010 .
m. raieszadeh , p. monajemi , s .- w .
yoon , j. laskar , and f. ayazi , `` high - q integrated inductors on trenched silicon islands , '' in _ ieee international conference on micro electro mechanical systems _ , 2005 , pp .
199202 .
m. frommberger , c. schmutz , m. tewes , j. mccord , w. hartung , r. losehand , and e. quandt , `` integration of crossed anisotropy magnetic core into toroidal thin - film inductors , '' _ ieee transactions on microwave theory and techniques _ , vol .
53 , no . 6
ii , pp . 20962099 , 2005 .
t. wang , y. peng , w. jiang , y. m. huang , b. m. f. rahman , r. divan , d. rosenmann , and g. wang , `` integrating nanopatterned ferromagnetic and ferroelectric thin films for electrically tunable rf applications , '' _ ieee transacations on microwave theory and techniques _ , 2016 .
|
the demand on mobile electronics to continue to shrink in size while increase in efficiency drives the demand on the internal passive components to do the same .
power amplifiers require inductors with small form factors , high quality factors , and high operating frequency in the single - digit ghz range .
this work explores the use of magnetic materials to satisfy the needs of power amplifier inductor applications .
this paper discusses the optimization choices regarding material selection , device design , and fabrication methodology .
the inductors achieved here present the best performance to date for an integrated magnetic core inductor at high frequencies with a 1 nh inductance and peak quality factor of 4 at @xmath03 ghz .
such compact inductors show potential for efficiently meeting the need of mobile electronics in the future .
|
the sharp rise of the electromagnetic proton structure function , @xmath1 , toward low values of bjorken @xmath2 , discovered at hera @xcite , and the observation of a large fraction of diffractive - like events @xcite are attributed to a large gluon density in the proton at very low @xmath2 values , typically @xmath3 .
the rise of @xmath1 with decreasing @xmath2 can be accommodated by the qcd , dglap evolution equations @xcite in nlo down to momentum transfer squared @xmath4 @xcite .
this suggests that perturbative effects set in at relatively low values of the interaction scale .
however , the validity of the dglap evolution equation is established through fits to data which involve many unknown parameters and therefore it may be doubtful .
exclusive vector meson ( v ) production at high @xmath5 has been proposed as an alternative method to infer the gluon content of the proton @xcite .
high energy elastic v production in deep inelastic scattering ( dis ) may be described , in the rest frame of the proton , by the fallowing sequence of happenings @xcite .
the incoming lepton emits a virtual photon , which subsequently fluctuates into a @xmath6 pair .
the life time of such a quark pair fluctuation is long enough so that it is the pair that elastically scatters off the proton and evolves , long after the interaction , into a v state . the character of the interaction of the @xmath6 pair with the proton depends on the transverse momenta of the pair .
if the transverse momentum is large , the spatial transverse separation between the quarks is small and it forms a color dipole , whose interaction with the proton may be calculated perturbatively @xcite .
the leading process is two gluon exchange .
if the transverse momentum is small , the color dipole is large and perturbative calculations do not apply . in this case
the interaction looks similar to hadron - hadron elastic scattering and the process should proceed through pomeron exchange as expected from regge phenomenology @xcite .
the @xmath6 wave function of the virtual photon depends on the polarization of the virtual photon . for longitudinally polarized photons , small transverse size @xmath6 dominate .
the opposite is true for transversely polarized photons .
the attractive features of elastic v production is that , at high @xmath5 , the longitudinal component of the virtual photon dominates .
the interaction cross section for the latter can be , in principle , fully calculated in perturbative qcd .
moreover , for heavy vector mesons , like the @xmath7 or the @xmath8 , perturbative calculations apply even at @xmath9 , as the smallness of the dipole is guaranteed by the mass of the quarks .
independently of particular calculations @xcite , in the region dominated by perturbative qcd , the following features are predicted : * the total @xmath10 cross section , @xmath11 , exhibits a steep rise with @xmath12 , the photon - proton center - of - mass energy , that can be parameterized as @xmath13 , with @xmath14 increasing with @xmath5 ; * the @xmath5 dependence , which for a longitudinally polarized photon is expected to behave like @xmath15 , becomes milder , more like @xmath16 , due to the sharp increase of the gluon density with @xmath5 ; * the distribution of @xmath17 , the four momentum transfer squared at the proton vertex , becomes universal , with little or no dependence on @xmath12 or @xmath5 ; all these features have been observed at hera @xcite in the exclusive production of @xmath18 , @xmath19 , @xmath20 and @xmath7 mesons . however , the agreement between measurements and theory is far from perfect .
there are many factors that may spoil the agreement .
first and foremost , the measured cross sections contain the possibly soft transverse component of the virtual photon , which has to be modeled .
in addition , the perturbative calculations have the following uncertainties : * the calculation of @xmath21 involves the so - called skewed parton distributions @xcite , which are not yet well tested and involve gluon distributions outside the range which is constrained by global parton density analyses @xcite ; * higher order corrections have not been fully calculated , therefore the overall normalization is uncertain and the scale at which the gluons are probed is not known ; * the fast rise of @xmath11 implies the presence of the real part of the scattering amplitude , which is not fully known ; * the wave functions of the vector mesons are not fully under control . in spite of all these problems , it is generally felt that precise measurements of @xmath11 as a function of @xmath12 , @xmath5 , @xmath17 and the mass of the vector meson , @xmath22 , with a separation into longitudinal and transverse components @xcite , will help to resolve the theoretical uncertainties and ultimately lead to a better understanding of the parton distributions in the proton as well as of the dynamics of high energy interactions in the presence of a large scale .
one of the challenges in confronting perturbative qcd calculations with data is the ability to establish a region where hard interactions dominate over the soft component .
the soft component is believed to be well described by regge phenomenology , according to which at high energy , the pomeron exchange dominates in the production of diffractive states .
the parameters of the pomeron trajectory are known from measurements of total cross sections in hadron - hadron interactions and elastic proton - proton measurements .
it is usually assumed that the pomeron trajectory is linear in @xmath17 and has the following form : @xmath23 the parameter @xmath24 determines the energy behavior of the total cross section , @xmath25 and @xmath26 describes the increase of the exponential slope @xmath27 of the @xmath17 distribution with increasing @xmath28 .
the value of @xmath26 is expected to be inversely proportional to the typical transverse momenta squared of the partons participating in the exchanged trajectory @xcite .
indeed a large value of @xmath26 would suggest the presence of low transverse momenta , typical of soft interactions .
the latest fits of @xmath24 @xcite and @xmath29 @xcite are give : @xmath30 these values for the trajectory parameters describe the so called soft pomeron trajectory .
the non - universality of @xmath24 has been established in dis , where the slope of the rise of the @xmath31 total cross section with @xmath12 has a pronounced @xmath5 dependence @xcite .
the issue of the @xmath26 dependence on the hardness of the interaction may be addressed in the study of exclusive v production at hera .
the value of @xmath26 may be determined from the @xmath12 dependence of @xmath27 , since @xmath27 is expected to behave as @xmath32 the parameter @xmath26 may also be derived from the @xmath12 dependence of the differential cross section @xmath33 at fixed @xmath17 , @xmath34 } \ , , \ ] ] where @xmath35 is a function of @xmath17 only .
the latter approach has the advantage that no assumption needs to be made about the @xmath17 dependence .
the measurements of @xmath36 in exclusive @xmath37 photoproduction show that , while @xmath24 for @xmath37 is larger the soft @xmath24 value , the value of @xmath26 for @xmath37 is smaller @xcite than the soft value .
another way to investigate the contribution of hard and soft components in v production is by projecting out the interactions induced by the longitudinally and transversely polarized virtual photons .
due to the respective wave functions structure , small @xmath6 configurations are dominantly longitudinal while large configurations are dominantly transverse .
if one assumes that exclusive v production proceeds through @xmath28-channel helicity conservation ( schc ) , the separation into the longitudinal and transverse components is possible . at small @xmath17 , the hypothesis of schc has been directly tested in the data , and small deviations have been observed @xcite .
however , the deviations are small enough not to jeopardize the decomposition of the cross sections into the longitudinal and transverse components .
the angular distributions of the decay products of the vector mesons give access to the spin density matrix elements , which are bilinear combinations of the helicity amplitudes @xmath38 where @xmath39 ( @xmath40 ) is the v ( virtual photon ) helicity @xcite .
as @xmath17 increases , a possible change in helicity is more likely due to the change in the v s direction and the transfer of the transverse momentum carried by the gluons .
the following features are expected by pqcd models @xcite in dis for @xmath41 : * a constant ratio of the helicity conserving amplitudes , @xmath42 , with @xmath17 ; * a @xmath43 dependence for the ratio of the single helicity flip to the non - flip amplitudes @xmath44 and @xmath45 ; * a linear @xmath17 dependence for the ratio of the double flip to the non - flip amplitudes @xmath46 ; * the hierarchy @xmath47 the differential cross section , @xmath48 , of v production has an exponentially falling cross section at low @xmath17 and a power like behavior at high @xmath17 . an exponential behavior is associated with a gaussian charge distribution and the slope , @xmath27 , of the exponent is related to the width of the distribution and is a measure of the interaction size . in a naive picture for elastic scattering
, @xmath27 has contributions from the size of the @xmath6 pair which decreases with @xmath5 and the constant size for the proton . at high @xmath5 , for all type of elastic v production
, @xmath27 should reach an asymptotic value equal to the proton size .
the cross section for elastic v photo or electroproduction , with small transfer momentum to the proton , has the form @xcite @xmath49 where @xmath50 and @xmath51 are the form factors of the proton and the v , respectively , that account for the probabilities of their elastic production , and @xmath52 is the amplitude of the constituent interaction .
the cross section of proton dissociation v production does not contain the proton form factor @xmath53 vertex factorization @xcite predicts that the ratio of proton dissociation over elastic differential cross section is @xmath5 independent .
the ratio of equation ( [ qq : ryskin1 ] ) to equation ( [ qq : ryskin2 ] ) is equal to the square of the proton form factor .
the dependence of the cross section @xmath54 on the center of mass energy , @xmath12 , is shown in fig.[fig : w_php_allvm.ps ] for photoproduction of various vector mesons @xcite .
while the @xmath12 dependence of light vector mesons ( @xmath55 @xmath20 @xmath19 ) is @xmath56 , as expected from regge phenomenology , the @xmath37 cross section has a steep rise with @xmath12 , which is a signature of a hard process , as predicted by pqcd .
= 0.3 a change in the @xmath12 dependence is also seen when moving from low to high @xmath5 , as shown in fig.[fig : h199 - 010f23.eps ] , where the @xmath12 dependence of @xmath57 for various @xmath5 values @xcite is plotted . for each @xmath5 value the @xmath12 dependence of @xmath57
is fitted with a form @xmath58 .
the parameter @xmath14 is related to the exchanged trajectory @xmath36 by @xmath59 . to extract @xmath24 from @xmath14 , @xmath60
is taken from measured values @xcite and the value @xmath61 is assigned .
the resulting @xmath24 is plotted as a function of @xmath5 in fig.[fig : h199 - 010f24.eps ] and a marked increase of @xmath24 with @xmath5 is observed .
= 1 = 1 the comparison between the measurements of @xmath62 in photoproduction and various pqcd calculations @xcite , based on different gluon distributions , is shown in fig.[fig : cross_section_theory.eps ] @xcite .
the pqcd calculations are known to suffer from large theoretical uncertainties , therefore no discrimination between the gluon distributions can be made from this comparison . also shown in fig.[fig : cross_section_theory.eps ] is the result of a calculation @xcite based on the golec - biernat wsthoff ( gbw ) model @xcite of the photon , whose parameters were fitted to the measurements of the inclusive structure function .
the gbw model gives a good representation of the data , assuming a double gaussian wave function of the @xmath37 .
= 1 = 1 although the @xmath12 dependence of @xmath37 photoproduction can not be explained by the exchange of a universal soft pomeron , the data can be fitted to the two - pomeron model @xcite as shown in fig.[fig : w_jpsi_fit1.ps ] @xcite .
the parameters of the established soft and the proposed hard pomeron trajectories are @xmath63 and @xmath64 .
the relative contributions of the hard , soft and mixing terms of the model are found to vary between @xmath65 at @xmath66 to @xmath67 at @xmath68 . in fig.[fig
: jpsidis - xsec.eps ] the @xmath12 dependence of @xmath62 is shown for various @xmath5 values @xcite .
the measurements are compared with theoretical predictions of frankfurt et al .
( fks ) @xcite using cteq4 m @xcite , and of martin et al .
( mrt ) @xcite using the cteq5 m @xcite parametrization .
the calculations are consistent with the data within the uncertainty of the measurements . also shown in fig.[fig : disjpsi_delta.eps ] are the values of @xmath14 , obtained by fitting in each @xmath5 , the functional form @xmath69 .
the results may indicate a slight increase in @xmath70 at high @xmath5 .
= 1 = 1 the h1 collaboration @xcite observed that when the cross sections for various vs , weighted by the appropriate @xmath71 factors ( @xmath72 ) were plotted as function of @xmath73 , at @xmath74 , they seem to line up on a universal curve , as shown in fig.[fig : h100 - 070f3.eps ] .
however there was a slight indication that the @xmath37 cross sections lie above the curve ( see the insert in fig.[fig : h100 - 070f3.eps ] ) .
the new precise measurements @xcite of @xmath37 production do not fit this picture , as shown in fig.[fig : vmheavy.eps ] , where the @xmath12 dependence of light and heavy vs , appropriately weighted by the @xmath71 factor , is plotted in bins of @xmath75 .
the @xmath37 data lie far above the light vs at all scales .
it seems , therefore , that the variable @xmath75 is a good variable for comparing the light vs but is not adequate when the @xmath37 is added to the comparison .
= 1 = 1 the @xmath77 trajectory , @xmath76 , can be measured by fitting the @xmath12 dependence of the @xmath78 cross section in bins of @xmath17 , @xmath79 , where @xmath80 .
the parameters of the @xmath77 trajectory were extracted from measurements of @xmath18 @xcite and @xmath37 production @xcite and are summarized in table [ tab : pomtraj ] .
the measurements of @xmath76 and the fitted trajectories are shown in fig.[fig : gl_fit_syst2.eps ] . and in fig.[fig :
alpha_final.eps ]
the measured trajectories change with the @xmath81 and @xmath5 .
.compilation of results obtained for @xmath24 and @xmath26 .
[ cols="<,>,<,>,<",options="header " , ] [ tab : pomtraj ] = 1 = 1
the separation of the @xmath78 cross section into the contribution from longitudinal photons , @xmath82 , and transverse photons , @xmath83 , can be performed by measuring the v density matrix elements , @xmath84 @xcite .
these matrix elements are used to parameterize the angular distribution @xmath85 of the helicity angles .
the density matrix elements , @xmath84 , are related to the transition amplitudes of longitudinal or transverse photons into longitudinal or transverse vs. the transition amplitudes are denoted as @xmath86 , where @xmath87 is a transverse ( longitudinal ) photon going into a @xmath88 transverse ( longitudinal ) v. the relations between the transition amplitudes have been predicted by pqcd models @xcite .
the most extensive measurements @xcite of @xmath90 were preformed for @xmath18 electroproduction .
the ratio @xmath91 of cross sections induced by longitudinal polarized virtual photons to the transversely polarized ones , @xmath92 , is related to the spin density matrix elements of @xmath18 through @xmath93 where @xmath94 is a linear combination of the @xmath18 density matrix elements , and @xmath95 is proportional to the contribution of the sch non conservation amplitude over the total amplitude .
a small breaking of schc has been predicted @xcite and measured @xcite ( @xmath96 @xcite ) .
the @xmath5 and @xmath12 dependence of @xmath91 is plotted in fig.[fig : r.eps ] and fig.[fig : wr.eps ] , respectively
. a continuous rise of @xmath91 with @xmath5 is observed .
the @xmath5 dependence of @xmath91 is well described by a functional form @xmath97 with @xmath98 and @xmath99 @xcite . on the other hand , there seems to be no @xmath12 dependence of @xmath91 for fixed values of @xmath5 .
this indicates that @xmath82 and @xmath83 have the same energy dependence , which is fixed by the @xmath5 value rather than the longitudinal or transverse configuration .
= 1 = 1 the extracted @xcite values of for @xmath18 production are presented in fig.[fig : h1rho_r04t.eps ] , where @xmath100 and @xmath101 .
no significant variation of with @xmath102 is observed .
this observation implies that the slopes of the exponentially falling @xmath17 distributions for the transverse and longitudinal @xmath28-channel helicity conserving amplitudes , @xmath103 and @xmath104 , are very similar .
= 1 = 1 the combination @xmath105 is presented in fig.[fig : h1rho_r5t.eps ] @xcite .
@xmath106 is expected @xcite to be proportional to the product of the dominant non - flip amplitude @xmath103 and the single flip @xmath107 amplitude , @xmath108 .
the @xmath107 amplitude is expected to be the largest helicity flip amplitude .
the @xmath109 matrix element , on the other hand is proportional to the non - dominant amplitudes @xmath110 .
the strong dependence of the combination is thus attributed mainly to the predicted @xcite @xmath111 dependence of the ratio of @xmath107 to the non - flip amplitudes .
the values for are shown in fig.[fig : h1rho_r1t.eps ] .
the matrix element @xmath112 is proportional to the single flip amplitude @xmath107 square , @xmath113 , while the matrix element @xmath114 is expected to have the same @xmath17 dependence as the double flip amplitude @xmath115 , @xmath116 , therefore the dependence of the combination is expected to be linear , up to effects of the single and double - flip amplitudes in the denominator @xmath117 .
the combination is significantly different from zero and negative , which implies violation of schc .
the sign of the combination gives information on the relative strength of the @xmath118 and @xmath119 products of amplitudes , @xmath120 , and therefore @xmath121 .
it confirms that the @xmath107 amplitude is significantly larger than the double flip amplitude in the present kinematic domain .
the differential cross section as function of @xmath17 has been studied for @xmath18 electroproduction .
the values of @xmath122 are extracted from a fit to an exponential form , @xmath123 , for @xmath124 in elastic production @xmath125 and for @xmath126 in proton dissociation @xmath18 production @xmath127 , for different @xmath5 regions .
the data cover a kinematic range of @xmath128 and @xmath129 ( elastic ) and @xmath130 ( proton dissociation ) .
= 1 = 1 the values of @xmath131 and @xmath132 are shown in fig.[fig : pdis_el_bq2.eps ] as a function of @xmath5 .
the @xmath5 dependence of @xmath27 is well described by the following functional forms , @xmath133 one clearly sees the decrease in the value of @xmath131 with @xmath5 , approaching an asymptotic value , from fit @xmath134 .
the slopes @xmath131 and @xmath132 are described with the same @xmath5 dependence , in compliance with the expectations of vertex factorization @xcite .
it is interesting to note that the ad hoc parametrization of @xmath132 , based on the @xmath5 dependence of @xmath91 , describes well the measurements of @xmath132 and @xmath131 .
the parameterization is motivated by the expected @xmath5 dependence of the cross section and the relation between the probability for longitudinal configuration and the size of the @xmath6 pair .
the measurements of @xmath131 , done by h1 @xcite ( fig.[fig : h199 - 010f19.eps ] ) , show the same @xmath5 behavior .
it is also evident that at low @xmath5 , the hera measurements lie systematically above the low energy fixed target results .
this might indicate shrinkage of the diffractive peak , at low @xmath5 , as @xmath12 increases . in order to test the factorization hypothesis at the proton vertex @xcite
, the ratio of @xmath135 of the elastic and proton dissociation reactions is studied as function of @xmath5 , for two fixed values of @xmath17 , as shown in fig.[fig : pdis_el_rat.eps ] .
the ratio is consistent with being @xmath5 independent .
this lends support to the hypothesis that factorization holds at the proton vertex for @xmath18 electroproduction in the region @xmath136 and @xmath137 .
it is evident that the hardness of the process in the elastic production of vector mesons in @xmath0 collisions at high energy , is influenced by both initial ( e.g. @xmath5 ) and final state configurations ( e.g. @xmath81 ) .
this is demonstrated by the change of the energy dependence with @xmath5 and @xmath81 , and by the decrease of the @xmath27 slope with increasing @xmath5 and @xmath81 .
longitudinal and transverse photons seem to have the same energy dependence at fixed @xmath5 , as well as the same @xmath17 dependence .
this indicates that the @xmath138 pairs initiated by a longitudinal or transverse photon , and contributing to the production of a given vector meson , are typically of the same size .
this size is determined both by the virtualty @xmath5 of the fluctuating photon and by the probability to form the vector meson final state .
although the overall picture is consistent with pqcd predictions , the quantitative calculations need further improvements .
i am very grateful to alberto santoro and the lishep organizing committee for their financial support and hospitality during the lishep advanced school on hep and the following workshop on diffractive physics .
zeus collab .
, m. derrick et al .
h1 collab . , i. abt et al . , .
zeus collab .
, m. derrick et al .
h1 collab . , i. abt et al . , .
v.n . gribov and l.n .
lipatov , and 675 ; + g. altarelli and g. parisi , ; + yu.l .
dokshitzer , .
zeus collab . , j. breitweg et al . , ; + h1 collab . , s. aid et al .
, ; + a.d .
martin et al . , ; + m. gluck e. reya and a. vogt .
cteq collab .
lai et al . , .
brodsky , l. frankfurt , j.f .
gunion , a.h .
mueller , and m. strikman , .
h. abramowicz , l. frankfurt and m. strikman , surveys high energy phys . *
11 * ( 1997 ) 51 .
l. frankfurt , m. mcdermott , v. guzey and m. strikman , .
collins , _ an introduction to regge theory and high energy physics _ , cambridge university press , england , ( 1997 ) .
l. frankfurt , w. koepf and m. strikman , ; + m.g .
ryskin , ; + a.d .
martin , m.g .
ryskin and t. teubner , .
d.y . ivanov and r. kirschner , .
zeus collab .
, paper submitted to the 29th international conference on high energy physics , abstract 793 , vancouver , ( 1998 ) ; + zeus collab .
, j. breitweg et al . , ; + zeus collab . , j. breitweg et al . , .
zeus collab .
, paper submitted to the 30th international conference on high energy physics , abstract 880 , osaka , ( 2000 ) .
h1 collab .
, c. adloff et al . , .
zeus collab . , j. breitweg et al . , .
radyushkin , ; + x.ji , .
l. frankfurt , m. mcdermott and m. strikman , journal of high energy phys .
* 103 * ( 2001 ) 45 .
see e.g. m.f .
mcdermott , desy-00 - 126 , ( 2000 ) , and references therein .
gribov v n , k. kang , j.r .
cudell , v.v .
ezhela , s.b .
lugovsky and n.p .
tkachenko , hep - ph/9812429 .
a. donnachie and p.v .
landshoff , ; + a. donnachie and p.v .
landshoff , .
h1 collab .
, c. adloff et al . , .
zeus collab . , j. breitweg et al . , .
a. levy , .
zeus collab .
, paper submitted to the international europhysics conference on high energy physics , abstract 548 , budapest , 2001 .
h1 collab . , c. adloff et al . , .
zeus collab . , j. breitweg et al . , .
k. schilling and g. wolf , .
i. royen and j. cudell , ; + .
e. kuraev , n. nikolaev and b. zakharov , .
ryskin , yu.m .
shabelski and a.g .
shuvaev , .
gribov , d. brown for h1 collab .
, proceeding of dis01 , 27 april - 1 may 2001 , bologna , italy .
a. d. martin , m. g. ryskin and t. teubner , .
zeus collab .
, s. chekanov et al , .
journal c ; doi 10.1007/s10052 - 002 - 0953 - 7 ( online publication ) . a.c caldwell and m.s .
k. golec - biernat and m. wsthoff , ; + k. golec - biernat and m. wsthoff , .
a. donnachie , p.v .
landshoff , .
h1 collab .
, c. adloff et al
. , a. levy for the zeus collaboration , presented at dis02 , 28 april - 3 may 2001 .
l. frankfurt , w. koepf and m. strikman , .
lai et al . , .
zeus collab . ; j. breitweg et al . , .
a. kreisel for zeus collaboration , proceeding of dis01 , 27 april - 1 may 2001 , bologna , italy .
h1 collab .
, c. adloff et al . , accepted by phys lett .
|
an extended study of exclusive vector meson production in @xmath0 interactions has been performed by the h1 and the zeus collaborations at the hera collider .
recent measurements are reported and discussed within the framework of the dipole model and pqcd .
# 1#2#3 _ ann .
phys .
( ny ) _ * # 1 * ( # 2 ) # 3 .
# 1#2#3 _ astrophys .
j. _ * # 1 * ( # 2 ) # 3 .
# 1#2#3 _ astrophys .
j. lett .
_ * # 1 * ( # 2 ) # 3 .
# 1#2#3 _ acta .
phys . pol .
_ * # 1 * ( # 2 ) # 3 .
# 1#2#3 _ ann .
rev .
nucl . part .
sci .
_ * # 1 * ( # 2 ) # 3 . # 1#2#3 _ computer phys .
comm .
_ * # 1 * ( # 2 ) # 3 .
# 1#2#3 _ eur .
phys .
j. _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ erratum _ * # 1 * ( # 2 ) # 3 . # 1#2#3 _ ibid . _ * # 1 * ( # 2 ) # 3 .
# 1#2#3 _ j. math .
phys . _ * # 1 * ( # 2 ) # 3 .
# 1#2#3 _ int .
j. mod .
phys . _
* # 1 * ( # 2 ) # 3 # 1#2#3 _ jetp lett . _
* # 1 * ( # 2 ) # 3 .
# 1#2#3 _ j. phys
. g. _ * # 1 * ( # 2 ) # 3 .
# 1#2#3 _ mod .
phys .
lett .
_ * # 1 * ( # 2 ) # 3 .
# 1#2#3 _ nature ( london ) _ * # 1 * ( # 2 ) # 3 . # 1#2#3 _ nuovo cim . _ * # 1 * ( # 2 ) # 3 . # 1#2#3 _ nucl .
instr .
meth . _
* # 1 * ( # 2 ) # 3 # 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 .
ps .
# 1#2#3 _ physica scripta _ * # 1 * ( # 2 ) # 3 . # 1#2#3 _ rev .
mod .
phys . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ rep .
prog .
phys . _ * # 1 * ( # 2 ) # 3 .
# 1#2#3 _ sov .
j. nucl .
phys . _
* # 1 * ( # 2 ) # 3 # 1#2#3 _ sov .
phys .
jept _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ sov .
phys .-
usp . _
* # 1 * ( # 2 ) # 3 .
# 1#2#3 _ zeit .
phys . _ * # 1 * ( # 2 ) # 3
|
landscapes built of a large number of different fluxes give rise to a vast discretuum of vacua , and so will naturally contain ones that , like our own , have a tiny cosmological constant @xcite . in this paper , we show that transitions between these vacua typically occur by giant leaps , discharging many fluxes in unison ; such transitions dramatically alter the cosmological constant . in a previous paper @xcite
, we found effects that enhance giant leaps , which relate to the presence of a radion . in this paper
, we find a different set of enhancements , which exist even when the radion is fixed . in sec .
[ sec : multiflux ] , we study landscapes built of many genuinely different fluxes , and in sec . [
sec : monkey ] we study the more sophisticated case where these different fluxes arise from a single higher - dimensional flux wrapping many different cycles of the internal manifold .
our paradigmatic example is the bousso - polchinski ( bp ) landscape @xcite .
following bp , we treat the radion as fixed , and work in the thin probe - brane approximation .
transitions in this landscape have received much attention @xcite , but this paper is the first to identify the dominant decay : giant leaps .
we show that the branes that discharge different fluxes attract , so that decay proceeds by whole stacks of branes at once , rather than a single brane as was previously assumed .
though still generically exponentially suppressed , giant leaps are the fastest decay . in @xcite , we show that giant leaps persist away from these approximations by studying decays in an explicit radion stabilization , 6d einstein - maxwell theory with many distinct fluxes .
a collection of @xmath2 different fluxes , whose legs all point down the extra dimensions , makes a contribution to the cosmological constant , @xmath3 , where @xmath4 @xmath5 is the charge of the @xmath6th magnetic quantum , and @xmath7 is the number of units of the @xmath6th magnetic flux .
these fluxes may be discharged by nucleating a brane that forms a bubble in the extended directions . if the brane carries @xmath8 units of charge , then inside the bubble @xmath9 so that the energy density inside is reduced by @xmath10 the surface tension of the bubble is set by the brane tension ; for the black branes , m - branes and d - branes we are interested in , that tension is @xmath11 where @xmath12 is the 4d planck mass .
the rate to nucleate such a bubble is @xmath13 , where in the thin - wall , probe - brane , semiclassical , no-4d - gravity approximation @xcite @xmath14 what is the fastest decay ?
it depends on the @xmath5s and the @xmath15s . for definiteness and simplicity we will focus throughout this paper on the case of identical fluxes @xmath16 .
we will also focus on a special class of states , the ` diagonal ' states that have @xmath17 . these states are privileged because states off the diagonal , with unequal @xmath18 , decay towards them ; the largest @xmath15 discharge first , as these decays most rapidly reduce @xmath19 . for diagonal states , one decay direction ,
` monoflux decays ' , is to drop @xmath20 units from a single flux , while leaving the other fluxes untouched . a second decay direction , `
multiflux decays ' , is to drop just a single unit each from @xmath20 different fluxes .
these definitions coincide for @xmath21 , which has tension @xmath22 and rate exponent @xmath23 . but as @xmath20 increases , there are two effects that make monoflux decays slower than multiflux decays . * @xmath24-effect * : for monoflux decays , @xmath25 . this is linear because these extremal branes do not interact their magnetic repulsion exactly cancels their gravitational attraction .
for multiflux decays , @xmath26 . this is slower than linear because the branes bind together since they are charged under different fluxes , there is now no magnetic repulsion to cancel their gravitational attraction .
smaller @xmath24 means faster decay . * @xmath27-effect * : for monoflux decays , @xmath28 . this is slower than linear because the flux lines repel . for multiflux decays , @xmath29
this is linear because the flux lines are different and do nt interact .
larger @xmath27 means faster decay .
the rate exponent @xmath30 can be evaluated for both decay directions .
monoflux decays get slower with increasing @xmath20 , @xmath31 > b_1 $ ] .
but multiflux decays get faster @xmath32 = \frac{b_1}{n}. \label{eq : multirate}\ ] ] indeed the fastest possible decay of a diagonal state is @xmath33 , discharging one unit of each flux in unison and moving one step down the diagonal .
( that moving many steps down the diagonal is slower follows from the monoflux result , with @xmath34 . ) if the @xmath5 vary , then the analog of the diagonal states are those with @xmath35s as equal as possible consistent with quantization .
decaying down the diagonal now may mean discharging more than a single unit of the fluxes with small @xmath5 .
gravitational corrections are easy to add ; the form of the corrections was calculated by coleman and de luccia @xcite . while @xmath23 receives large gravitational corrections , eq .
[ eq : multirate ] is exact for decays from minkowski . for decays from low and intermediate de sitter ,
it receives only small modifications , so that our parent vacuum is likely a giant leap away . from high de sitter states , however , gravitational modifications are significant and small steps are the fastest decay .
to confirm giant leaps away from the thin - wall , probe - brane approximation requires choosing an explicit flux compactification , as we have done in @xcite .
in the last section we saw that , in landscapes built of many genuinely different fluxes , the most likely decay is a giant leap . in this section
we will investigate whether the same holds true when the ` different ' fluxes are really , from the higher - dimensional viewpoint , the _ same _ flux , just wrapped around different compact cycles . following bp @xcite ,
let s compactify on a flat @xmath36-torus . to preserve macroscopic lorentz symmetry , a @xmath37-form @xmath38 that lives in this ( 3
+ 1+@xmath36)-dimensional spacetime must point legs in _ no _ macroscopic direction ( and so @xmath39 must point legs in _ every _ macroscopic direction ) ; after integrating out the extra dimensions , @xmath38 becomes a scalar ( and @xmath39 a four - form ) .
let s consider an @xmath40 that wraps a single basis @xmath37-cycle [ cols="<,^,^,^,^,^,^,^,^,^,^",options="header " , ] to see why the dual cycle is appropriate , consider integrating the magnetic source equation @xmath41 in a gaussian pillbox across the brane . restricting to the extra dimensions , if this flat brane is spanned by one forms @xmath42 , then @xmath43 and indeed this works for @xmath44 , @xmath45 , etc .
the effective tension of this brane is proportional to the fundamental tension times its volume in the extra dimensions , @xmath46 where @xmath47 ; again this is consistent with @xmath48 .
a general @xmath40 will have components along many of the @xmath49 basis @xmath37-cycles of the @xmath36-torus @xmath50 the cross - terms in @xmath19 all vanish so we recover eq .
[ fsquared ] , and with it the @xmath27-effect .
what of eq .
[ tension ] and the @xmath24-effect ? to discharge many fluxes at once , the brane must wrap many basis co - cycles the appropriate configuration will be the one that does this with minimum possible area , and so minimum possible effective tension . in the remainder of this section ,
we study this constrained optimization problem ; we start with a few simple examples and build up to the general case . for now , we set @xmath51 and the sides of the torus to 1 . for a @xmath52 * -form * on an @xmath53 * -torus * , @xmath54 . to discharge a flux wound round the @xmath55-direction ( @xmath56 )
requires a brane wound round the @xmath57-direction ( @xmath58 ) , and vice versa . to discharge both at once , @xmath59 ,
one possibility is to use two different branes , wrapping the @xmath57- and @xmath55-directions respectively , for a total tension of @xmath60 .
but you can do better .
the brane can instead stretch across the diagonal ( @xmath61 ) , so that it still wraps the same cycle but has lower area , for a tension of @xmath62 : [ fig:5d1each ] this diagonal brane is the shortest brane that achieves this transition ( it s the minimum area element of its homology class , the ` minimal brane ' ) , and the @xmath24-effect is retained .
analogous configurations can be found for any combination of fluxes dropped . to discharge @xmath63 , the minimal brane
is stretched in the normal direction to the flux dropped , @xmath64 , and has length @xmath65 .
for example , [ fig:5dmultieach ] is the minimal brane with @xmath66 and @xmath67 , shown both in the covering space and projected onto the unit cell . in this case , eq .
[ tension ] just amounts to pythagoras theorem . for a @xmath52 * -form * on an @xmath68 * -torus * , @xmath69 .
discharging @xmath56 takes a two - brane stretched along the @xmath57-@xmath70 cycle , and so on . as before , discharging @xmath71 does nt take two branes and total tension 2 , but only a single flat brane with @xmath72 and total tension @xmath62 : [ fig:6dneighbours ] indeed , for any @xmath73 there is a minimal brane with tension @xmath74 ; the normal to the brane points in the direction of the dropped flux . for example
, [ fig:6ddouble ] is the minimal brane that drops @xmath75 .
up to now it has all been straightforward
we could unwrap any set of fluxes in any combination with a single flat brane , and the story has been the same as it was for the genuinely different fluxes of sec . 2 . in higher dimensions , however , this is no longer true .
a @xmath76 * -form * on an @xmath77 * -torus * , which has @xmath78 , has all the complexities of the general case , and we will use it to exemplify them .
the source of these complexities is that you can not always discharge a given flux with a single flat brane .
for example , to discharge @xmath79 takes two branes , one that spans the @xmath70-@xmath80 plane and another that spans the @xmath55-@xmath57 plane .
there is no better way : these two branes only intersect at a point and can not reduce their area by coalescing .
the problem is that this @xmath81 can not be written in the form @xmath82 .
if it could , then by eq .
[ decomposable ] , these @xmath83s span a single flat brane that would discharge this @xmath84 ; the right - hand side of eq .
[ decomposable ] is the volume form of the brane , so its effective tension would be given by eq .
[ tension ] and we would recover the full @xmath24-effect .
the property that @xmath84 is writeable in this form is known as decomposability , and a two - form is decomposable if and only if it has just a single pair of nonzero eigenvalues .
[ one - forms and @xmath85-forms are always decomposable , which is why this behavior first appears at @xmath86 . ]
a direct consequence is that the different effective 4d fluxes are not now interchangeable . in sec . 2 , when any two fluxes discharged together , the brane had tension @xmath87 . here
, it depends _ which _ two fluxes .
neighboring fluxes can be discharged together with tension @xmath87 [ e.g. @xmath88 is decomposable ] whereas the two fluxes of eq .
[ uhoh ] require tension 2 .
the rotational symmetry of the landscape is broken switching a pair of occupation numbers , @xmath89 , can change the decay rate . bizarrer still , the reflection symmetry is broken too switching the sign of one occupation number , @xmath90 ( and so @xmath91 ) , can also change the decay rate . for instance , @xmath92 is decomposable and discharges fast , whereas the same combination with the sign of the @xmath93 term flipped is not decomposable and discharges slowly .
these complexities mean that finding the fastest decay is a more intricate problem than it was for the many genuinely different fluxes of sec . 2 .
let s start , as before , by considering decay from a diagonal state with all the occupation numbers equal @xmath94 ( and continue using @xmath86 , @xmath95 as our prototype ) .
the fastest decay need no longer be to discharge one unit of every type of flux .
such a @xmath84 is not decomposable , so the tension must be larger than @xmath96 , and some of the @xmath24-effect is lost . but some of the @xmath24-effect is preserved : we can certainly do better than 6 separate branes by clustering neighbors and allowing them to coalesce . but
even this is not the ideal configuration
these diagonal branes still intersect along a line and want to pull tight .
the resulting optimal configuration can not be flat , since @xmath84 is not decomposable ; indeed the minimal brane is generally curved @xcite . because of this impediment , decays that would have been subdominant in sec .
2 can now be fastest , so that diagonal states need not decay to diagonal states .
for example , it may be best to discharge fewer than @xmath2 fluxes just so that the minimal brane can be flat and the full @xmath24-effect retained .
if the signs are right , five fluxes may be discharged in unison , @xmath97 , by a single flat brane . on the other hand
, it may be best to discharge _ more _ flux .
one unit each of five fluxes and two units of the sixth can be discharged in unison , @xmath98 , by a single flat brane .
this can be the fastest decay , so that the leaps can be even gianter than in sec .
[ sec : multiflux ] . for higher dimensional tori
, the story follows analogously .
[ one case of special interest is the bp case , of m - theory on a flat @xmath99-torus with a @xmath100-form flux , so @xmath101 .
there , a flat brane can discharge a single unit each of at most 21 of the 35 fluxes , @xmath102 . ] as before , states that are off the diagonal will decay towards it , and if the @xmath5 vary , the fastest decays will stick to the approximate diagonal .
the innovation of the bousso - polchinski model @xcite is to introduce many different kinds of flux , giving a huge discretuum of vacua including many in the habitable zone .
we have shown that , precisely because of the many fluxes , decay proceeds by giant leaps .
giant leaps are mediated by minimal branes which , already for tori , exhibit exotic behavior not visible in the effective theory .
we expect more general manifolds to have yet richer decay phenomenology , with consequences for the stability of stringy de sitter , probabilities in the landscape @xcite , and the cosmological constant @xcite .
thanks to raphael bousso , steve gubser , chris herzog , igor klebanov , silviu pufu , paul steinhardt , herman verlinde and masahito yamazaki for helpful discussions .
a. r. brown and a. dahlen , `` small steps and giant leaps in the landscape , '' phys .
d * 82 * , 083519 ( 2010 ) [ arxiv:1004.3994 [ hep - th ] ] .
t. clifton , s. shenker and n. sivanandam , `` volume weighted measures of eternal inflation in the bousso - polchinski landscape , '' jhep * 0709 * , 034 ( 2007 ) [ arxiv:0706.3201 [ hep - th ] ] ; r. bousso and i. s. yang , `` landscape predictions from cosmological vacuum selection , '' phys .
d * 75 * , 123520 ( 2007 ) [ arxiv : hep - th/0703206 ] ; d. schwartz - perlov , `` anthropic prediction for a large multi - jump landscape , '' jcap * 0810 * , 009 ( 2008 ) [ arxiv:0805.3549 [ hep - th ] ] .
a. r. brown and a. dahlen , `` bubbles of nothing and the fastest decay in the landscape , '' phys .
d * 84 * , 043518 ( 2011 ) [ arxiv:1010.5240 [ hep - th ] ] .
j. garriga and a. megevand , `` coincident brane nucleation and the neutralization of lambda , '' phys .
d * 69 * , 083510 ( 2004 ) s. r. coleman , `` the fate of the false vacuum .
semiclassical theory , '' phys .
d * 15 * , 2929 ( 1977 ) ; j. d. brown and c. teitelboim , `` neutralization of the cosmological constant by membrane creation , '' nucl .
b * 297 * , 787 ( 1988 ) .
s. r. coleman and f. de luccia , phys .
d * 21 * , 3305 ( 1980 ) .
a. hashimoto and w. taylor , `` fluctuation spectra of tilted and intersecting d - branes from the born - infeld action , '' nucl .
b * 503 * ( 1997 ) 193 c. v. johnson , `` d - branes , '' chapter 11 , _ cambridge university press _ ( 2003 ) .
|
there is a standard story about decay in multi - dimensional flux landscapes : that from any state , the fastest decay is to take a small step , discharging one flux unit at a time ; that fluxes with the same coupling constant are interchangeable ; and that states with @xmath0 units of a given flux have the same decay rate as those with @xmath1 .
we show that this standard story is false .
the fastest decay is a giant leap that discharges many different fluxes in unison ; this decay is mediated by a ` minimal ' brane that wraps the internal manifold and exhibits behavior not visible in the effective theory .
we discuss the implications for the cosmological constant .
|
this paper is concerned with the following classical result , known as poincar inequality ( or poincar - friedrichs inequality ) : [ poincare ] let @xmath5 be a bounded domain satisfying the interior cone condition , and @xmath6 .
then , there exists a constant @xmath1 such that @xmath7 the proof is very simple and standard .
assume , reasoning by contradiction , that @xmath8 is a sequence such that @xmath9 by the rellich - kondrachov theorem , one obtains that @xmath10 , where : @xmath11 but if @xmath12 almost everywhere in a domain , then @xmath13 must be constant ( see chapter 9 of @xcite , for instance ) . and this yields the desired contradiction . as one can observe , two main ingredients come into play in the proof : first , the compactness of the @xmath14 embedding .
for that one needs the interior cone condition on @xmath4 and its boundedness .
second , the connectedness of the domain . in this paper
we try to answer the following question : under which conditions on @xmath4 we can assure the existence of @xmath3 independently of the set @xmath4 ?
the above discussion suggests that one should need three kind of assumptions : 1 . a uniform bound on @xmath4 , 2 .
an interior cone condition ( with a fixed cone ) , 3 . and , in a certain sense , a uniform connectedness assumption on @xmath4 .
the goal of this paper is to prove that this is indeed the case .
the uniformity of the poincar constant is useful in some free boundary problems or shape optimization problems , see @xcite .
our motivation comes from the derivation of a uniform local version of the moser - trudinger inequality , which is very useful in the variational study of the mean field equation on compact surfaces ( see @xcite , and also @xcite , @xcite ) .
we hope to use the results of this paper to give a proof for such inequality in a forthcoming work . before going on with our exposition ,
let us mention some previous results .
in the literature there are some results concerning the dependence of the poincar constant on @xmath4 : see for instance @xcite . however , the assumptions required there involve uniform @xmath15 regularity on @xmath4 and other conditions . here
we are interested in a less regular framework . on the contrary , in @xcite
more information is given , related to the solution of the associated neumann eigenvalue problem .
in @xcite the uniformity of the poincar constant is proved for uniformly lipschitz domains .
let us point out that lipschitz domains must lie ( locally ) on one side of the boundary .
the unit ball with a segment removed , for instance , is not lipschitz .
moreover , domains with interior cusps are not covered , for instance . in both cases the poincar inequality holds , which makes
one think that the uniform lipschitz assumption is too restrictive . throughout the paper we shall use the following notation : @xmath16 we will consider the following hypotheses . 1
. [ h1 ] there exists @xmath17 such that @xmath18 .
[ h2 ] there exists a fixed finite cone @xmath19 such that each point @xmath20 is the vertex of a cone @xmath21 congruent to @xmath19 and contained in @xmath4 .
[ h3 ] there exists @xmath22 such that for any @xmath23 , @xmath24 is a connected set .
this last condition is a sort of uniform connectedness assumption " .
we will show in example [ ejemplo ] that a condition in this direction is necessary in our results . roughly speaking ,
condition [ h3 ] avoids the existence of arbitrarily narrow junctions between two regions in @xmath4 .
the main result of this note is the following .
[ main ] for any @xmath6 there exists a constant @xmath25 such that for any domain @xmath4 satisfying [ h1]-[h3 ] , @xmath26 here @xmath3 depends only on @xmath27 , @xmath28 and the cone @xmath19 given in conditions [ h1]-[h3 ]
. we can also give some generalizations of this result , that will be discussed in section 3 .
in particular , we can prove the uniform estimate in the following situation : [ coro ] for any @xmath6 , @xmath17 and @xmath29 there exists a constant @xmath25 such that for any connected set @xmath30 , @xmath31 here @xmath32 denotes the domain @xmath33 .
observe that @xmath3 is independent of the choice of @xmath34 .
the uniformity of the poincar constant on this type of sets was the original motivation of our work .
we think that this corollary could be one of the ingredients of a future local version of the moser - trudinger inequality .
our results improve that of @xcite since a connected and uniformly lipschitz domain satisfy our conditions [ h1]-[h3 ] ( see the final appendix ) .
moreover , we can cover many situations in which the domains are not lipschitz , as explained above .
our proof is also quite different from that of @xcite , which uses different notions of convergence of sets to pass to a limit in an argument by contradiction . here
we give a direct proof , based on two main tools .
the first one is the following property : 1 .
[ q ] for any @xmath35 , there exists @xmath36 independent of @xmath4 such that @xmath37
. let us point out that any domain @xmath4 satisfies that condition with @xmath38 depending on @xmath4 ; this is due to the continuity from above of the lebesgue measure .
we shall see that conditions [ h1 ] and [ h2 ] imply the uniformity of the choice of @xmath38 .
the second ingredient of our arguments is the observation on the uniformity of the constant in the sobolev embedding under our hypotheses .
the rest of the paper is organized as follows . in section 2
we first give some preliminary results , some of them well - known , which are useful later in the proof of theorem [ main ] .
section 3 is devoted to the discussion of some extensions of our results .
in particular , corollary [ coro ] will be proved in section 3 . in a final appendix
we show that uniformly lipschitz domains , bounded and connected , satisfy conditions [ h1]-[h3 ] .
this implies that theorem [ main ] generalizes the result of @xcite .
in this section we first give some preliminary results , which will be of use later .
after that , we shall prove theorem [ main ] .
some notation is in order ; given @xmath39 and @xmath40 , we define : @xmath41 we also need a notion of uniformity for the lipschitz regularity of a domain .
the following definition follows closely that of chenais @xcite .
we first introduce some notation . given @xmath42 , we denote @xmath43 where @xmath44 and @xmath45 give the coordinates of @xmath46 in a given coordinate system .
[ def : lip ] we say that @xmath4 is a uniform lipschitz domain with constants @xmath47 , @xmath48 ( and we denote @xmath49 if the following is satisfied : for all @xmath50 , there exists a local coordinate system satisfying what follows
. there exists a function @xmath51 a lipschitz function with lipschitz constant smaller than @xmath52 such that : @xmath53 here , @xmath54 @xmath55 observe that for uniformly bounded domains , this definition coincides with the definition of strong local lipschitz property of adams ( @xcite , page 66 ) .
this property is equivalent to the condition assumed in @xcite , see section 3 of @xcite .
now let us recall and comment the following result , due to gagliardo @xcite ( alternatively , see adams @xcite , theorem 4.8 ) .
[ teo : ga ] let @xmath4 a bounded domain satisfying the cone condition [ h2 ] .
then there exists a finite collection @xmath56 of open subsets of @xmath4 such that : 1 . @xmath57 .
2 . @xmath58 is a lipschitz domain .
[ rem : ga ] by having a look at the proof of the above result ( @xcite , theorem 4.8 ) , one obtains easily that the lipschitz constants of the graphs depend only on the cone @xmath19 .
moreover , the number of sets @xmath59 in the proof can be bounded in terms of @xmath19 and @xmath60 of condition [ h1 ] .
this implies the following uniform version : under conditions [ h1 ] and [ h2 ] , then @xmath61 , and @xmath62 , @xmath52 , @xmath63 are positive constants depending only on the cone @xmath19 and the constant @xmath60 given by conditions [ h1 ] and [ h2 ] .
in the following proposition we prove property [ q ] , which will be essential in our proof of theorem [ main ] .
[ prop : q ] let us fix @xmath17 and a finite cone @xmath19 .
then , property [ q ] holds for any @xmath4 satisfying [ h1 ] and [ h2 ] referred to @xmath60 and @xmath19 respectively .
we will prove the proposition in two steps : the case @xmath64 .
this is straightforward .
indeed , consider a graph like in definition [ def : lip ] , and take @xmath65 sufficiently small .
define @xmath66 it is clear that for any point @xmath67 , the distance to the boundary of @xmath4 is at most @xmath68 , where @xmath52 is the lipschitz constant of @xmath69 .
therefore , @xmath70 observe also that the number of graphs needed to cover @xmath4 is uniformly bounded .
so it suffices to take @xmath71 sufficiently small , according to @xmath72 .
indeed , it follows that the dependence of @xmath38 on @xmath72 is linear . the general case , with @xmath73 satisfying [ h1 ] , [ h2 ] . by [ teo : ga ] ,
@xmath74 where each @xmath75 .
obviously , @xmath76 .
then , it is easy to show that : @xmath77 then , we can conclude by step 1 .
since @xmath78 depend only on @xmath19 and @xmath60 , we obtain the uniform estimate .
another important tool in our proof is the classical sobolev embedding .
the uniformity of the sobolev constant is important for our purposes and not so well - known , so let us state it in detail . for a proof , see @xcite , lemmas 5.10 and 5.15 , and also theorem 8.25 ( observe the remark at the end of the proof of this last theorem ) .
[ sobolev ] let @xmath5 satisfy [ h1 ] , [ h2 ] , and @xmath6 .
then there exists @xmath79 depending only on @xmath80 and @xmath81 such that : 1 .
if @xmath82 , @xmath83 , for any @xmath84 , @xmath85 $ ] .
2 . if @xmath86 , @xmath83 , for any @xmath84 , @xmath87 .
+ even more , there holds : @xmath88 , where @xmath89 is the orlicz norm associated to @xmath90 .
if @xmath91 , then any function @xmath92 is equal almost everywhere to a continuous function .
moreover , @xmath93 .
so far , we have considered conditions [ h1 ] and [ h2 ] .
condition [ h3 ] is less standard and can be considered as a uniform connectedness condition . in the following example
we show that this condition is indeed necessary for the thesis of the theorem [ main ] .
[ ejemplo ] in this example we consider @xmath94 , and we write the coordinates of a point in @xmath95 as @xmath96 , with @xmath97 and @xmath98 .
let us define the cone : @xmath99 for any @xmath35 , we define the set : @xmath100 where @xmath101 . clearly , @xmath102 is a connected set and the family of sets @xmath102 satisfy [ h1 ] and [ h2 ] , but they do not satisfy [ h3 ] . we shall find a sequence @xmath103 such that : @xmath104 define @xmath105 since @xmath106 is odd with respect to the @xmath107 variable , @xmath108 .
it is also easy to show that : @xmath109 finally , we compute : @xmath110 now we are in conditions to prove our main result : choose @xmath35 arbitrarily small , to be fixed at the end of the proof .
take @xmath111 such that @xmath112 , as given by property [ q ] .
fix also @xmath113 .
consider a covering @xmath114 , where @xmath115 are open balls of radius @xmath116 .
we denote by @xmath117 the class of connected sets formed by unions of such balls ; in other words : @xmath118 clearly each element of @xmath117 satisfies a cone condition and hence each of them has a poincar constant @xmath119 , in virtue of theorem [ poincare ] . in other words ,
@xmath120 since @xmath117 is finite , take @xmath121 the maximum of all those constants .
observe that @xmath121 depends only on @xmath116 , and hence on @xmath72 .
define : @xmath122 observe that under our assumptions @xmath123 .
moreover , it is clear that @xmath124 is connected , and then @xmath125 .
take now @xmath126 with @xmath127 , @xmath128 .
our intention is to find a lower bound for @xmath129 .
let us assume that @xmath130 , otherwise we are done .
we first give an estimate on @xmath131 by using hlder inequality : @xmath132 let us denote by @xmath133 the average of u on @xmath124 .
then , @xmath134 observe also that @xmath135 , where @xmath136 is the volume the cone @xmath19 , which serves as a lower bound for the volume of @xmath4 .
we choose @xmath72 small enough so that @xmath137 .
our intention now is to find a lower estimate for the term @xmath138 we now use again hlder inequality to estimate the term : @xmath139 therefore , @xmath140 it suffices now to estimate @xmath141 .
let us take any @xmath142 so that the sobolev embedding holds , see proposition [ sobolev ] .
since we are assuming that @xmath143 , we have that @xmath144 . by using again hlder inequality ,
we obtain : @xmath145 by putting together and , we have that : latexmath:[\[\int_a |u - u_a|^p\ , dx \geq 1- 2 c_s^p \ , { \varepsilon}^{\frac{q - p}{q } } - c_p \frac{{\varepsilon}^{\frac{p-1}{p}}}{v-{\varepsilon } } ( |\omega|^{\frac 1 p}+ we only need to chose @xmath72 small enough so that the above term is greater than @xmath147
. therefore , @xmath148 as a corollary , we obtain the following result . given @xmath149 a measurable set with nonzero measure , we denote by @xmath150 the average of u in @xmath151 , that is , latexmath:[$u_e = there exists a constant @xmath25 such that for any domain @xmath4 satisfying [ h1]-[h3 ] , 1 . if @xmath82 , @xmath153 2 .
if @xmath86 , @xmath154 3 .
if @xmath91 , @xmath155 moreover , @xmath3 depends only on @xmath27 , @xmath156 and the cone @xmath19 given in conditions [ h1]-[h3 ] .
compared with theorem 1 and corollary 2 of @xcite , here the constant @xmath3 is independent of the set @xmath151 .
the dependence of the estimate on @xmath151 , when @xmath157 is small , is made explicit here .
obviously , it suffices to prove the estimate for @xmath84 with @xmath158 and @xmath159 . by theorem [ main ]
, there exists @xmath25 ( depending only on @xmath60 , @xmath19 and @xmath160 ) such that @xmath161 .
observe that : @xmath162 therefore it suffices to estimate @xmath150 in cases 1 and 2 , and @xmath163 in case 3 .
for that we will use the sobolev embedding ; observe that @xmath164 .
if @xmath82 , we have : @xmath165 depends on @xmath3 and @xmath166 .
but then : @xmath167 if @xmath86 , we need to use orlicz norms : we refer to @xcite , chapter 8 .
let us define @xmath168 , and @xmath169 its complement function ( which is not explicit ) . by using the generalized hlder inequality , @xmath170 here
@xmath171 is the characteristic function of the set @xmath151 .
we now use the definition of the orlicz norm : @xmath172@xmath173 above we have just used the general inequality @xmath174 therefore , @xmath175 if @xmath91 , then @xmath13 is continuous .
moreover , @xmath176 by the uniformity of the constant on the sobolev inequality .
putting together estimates , , , , we finish the proof
. one can extend also the previous results to sets @xmath151 which are hypersufaces , in the spirit of theorem 2 of @xcite .
we leave the details to the interested reader .
in this section we discuss some possible extensions of our results to more general frameworks .
first , we will discuss the possible relaxation of the hypotheses of theorem [ main ] .
in particular , we will prove corollary [ coro ] . finally , we will briefly comment the extension of our results to riemannian manifolds .
the cone condition [ h1 ] has been needed in our proof at two different steps .
first , it is needed to guarantee a sobolev inequality ( see proposition [ sobolev ] ) , together with the uniformity of the constant .
observe that the proof of theorem [ main ] only needs a sobolev inequality for some exponent @xmath142 ; the limiting exponent is not needed here .
second , [ h1 ] is used to derive property [ q ] given in proposition [ prop : q ] , which otherwise should be somehow imposed .
indeed , we can introduce it in a generalized version of condition [ h3 ] .
our result can be generalized as follows .
[ general ] take @xmath6 , and let @xmath177 be a family of domains satisfying the following hypotheses : 1 .
[ f1 ] there exists @xmath17 such that @xmath18 for any @xmath178 .
[ f2 ] there exists @xmath142 and @xmath79 such that for any @xmath178 , @xmath84 , + @xmath179 3 .
[ f3 ] for any @xmath35 there exists @xmath180 such that for any @xmath178 we can choose a connected set @xmath181 such that @xmath182 and @xmath183
. then , there exists @xmath25 such that for any @xmath178 , @xmath7 the proof of theorem [ general ] is basically the same as the proof of theorem [ main ] , with the obvious modifications due to the new hypotheses .
the details are left to the reader .
the literature on sobolev inequalities for domains not satisfying the cone property is huge .
for instance , it is known that the limiting sobolev inequality does not hold for domains containing outward cusps . however , for cusps having power sharpness , a sobolev embedding to some @xmath184 space is possible , see @xcite .
it is to be expected that some uniformity on the assumptions would imply an uniform constant in that inequality , providing [ f2 ] .
many other sobolev embeddings have also been found , and it is not possible to give here a detailed account of them : see for instance @xcite and the references therein . on the other hand ,
condition [ f3 ] puts together condition [ q ] and [ h3 ] in a more general fashion .
condition [ f3 ] does not seem easy to check in general , though .
however , as a particular case , we can prove corollary [ coro ] .
it is not clear to us whether a domain under the conditions of corollary [ coro ] satisfies condition [ h3 ] . in any case ,
corollary [ coro ] is a consequence of theorem [ general ] , as we show below .
we only need to check that under the conditions of corollary [ coro ] , we can apply theorem [ general ] .
let us define : @xmath185 we recall for the definition of @xmath32 . in what follows we can always consider @xmath34 to be a compact set ; otherwise , we would argue on its closure .
clearly , any domain in @xmath177 satisfies the cone property with a cone depending only on @xmath116 .
therefore , conditions [ f1 ] and [ f2 ] are satisfied .
let us check condition [ f3 ] .
given @xmath186 and @xmath187 connected , we define : @xmath188 clearly , @xmath189 is a connected set .
observe now that @xmath190 .
the triangular inequality readily implies that @xmath191 . in order to prove [ f3 ] , it suffices to estimate the measure of @xmath192 independently of the choice of @xmath34 .
let us point out that : @xmath193 the first equality holds just by the definition of @xmath189 ; let us justify briefly the second identity .
the inclusion @xmath194 follows easily from the triangular inequality . for the inclusion @xmath195 ,
take @xmath196 and @xmath197 with @xmath198 .
choose @xmath46 in the segment @xmath199 $ ] such that @xmath200 , @xmath201 .
then , @xmath202 and also @xmath203 .
it suffices to observe that @xmath204 to conclude .
now , observe that @xmath205 satisfies the cone condition with a fixed cone .
now it suffices to recall the arguments of the proof of proposition [ prop : q ] to conclude . because of our original motivation in the study of local versions of the moser - trudinger inequality , we are interested in a uniform poincar inequality in compact riemannian manifolds .
this extension is quite direct , but it deserves a couple of comments .
first of all , we have not been able to find a specific definition of the interior cone condition for a domain in a manifold .
still , we think that it must be written elsewhere , but since we can not cite a reference we give below a definition which is useful for our purposes .
let @xmath206 be a compact manifold .
it is clear that we can consider on @xmath206 a finite family of charts @xmath207 , @xmath208 such that @xmath209 .
we say that @xmath210 satisfies the interior cone condition if for any @xmath211 such that @xmath212 , @xmath213 satisfies the interior cone condition at any point of @xmath214 . by using this finite family of charts
it is easy to generalize proposition [ q ] to the case of sub - domains of a fixed compact riemannian manifold ( instead of [ h1 ] ) .
moreover , also the uniform sobolev embedding holds , as can be easily checked passing to charts .
therefore , the results of theorem [ main ] , theorem [ general ] and corollary [ coro ] follow with basically the same proof .
in theorem [ main ] we have proved the uniformity of the poincar constant for domains satisfying certain properties , namely [ h1 ] , [ h2 ] , [ h3 ] . in this final section
we show that connected uniformly lipschitz domains ( in the sense of definition [ def : lip ] ) satisfy [ h2 ] and [ h3 ] .
so , our result extends that of @xcite to a less regular framework .
property [ h2 ] being straightforward , it suffices to show property [ h3 ] .
the proof we give here is inspired by min - max theory and deformation flows .
it is possible that there is a more direct way to prove this result , but in our opinion the proof itself is interesting in its own right .
observe that with that definition , @xmath217 and @xmath218 ( for @xmath36 ) .
the idea of the proof is to show that those two sets are homotopically equivalent if @xmath219 , for some @xmath220 small independent of @xmath4 . in order to prove that
, we will use a deformation argument .
it is important to observe here that @xmath221 is a lipschitz map , but not @xmath15 , and therefore one can not use a gradient flow as usually .
however , in 1981 , k. c. chang ( @xcite ) was able to apply variational methods to lipschitz maps . for that
, he needed a notion of generalized gradient , due to clarke , that we remind below .
moreover , he was able to prove the existence of a continuous pseudogradient flow for lipschitz maps . with these two main tools , he was able to apply min - max techniques to lipschitz functionals .
here we just use those ideas to find an homotopical equivalence between @xmath4 and @xmath222 . for that , we will need to show that @xmath221 has no critical points in @xmath223 ; it is here that the lipschitz assumption on @xmath4 comes into play .
it can be checked that the function @xmath229 is continuous and convex , and that @xmath230 , where @xmath231 is the lipschitz constant for @xmath225 .
we define the generalized gradient of @xmath225 at @xmath232 , denoted @xmath233 , as the subdifferential of the map @xmath229 at @xmath234 . in other words ,
@xmath235 let @xmath4 be a uniformly lipschitz domain , with the definition [ def : lip ] , and @xmath221 the map defined in .
then , there exists @xmath237 depending only on the constants @xmath52 , @xmath63 such that @xmath221 has no critical points in @xmath238 .
take @xmath239 ; by the definition of lipschitz regularity , in a neighborhood of @xmath232 , @xmath240 is a graph on the last component ( with a conveniently chosen coordinate system ) .
take @xmath241 .
we will show that there exists @xmath242 such that @xmath243 ; this readily implies that @xmath221 has no critical points in @xmath244 . here
@xmath250 is chosen depending on @xmath52 , @xmath63 . by the lipschitz property of the domain , @xmath251 for any @xmath252 .
in other words , @xmath253 ( see ) . an easy geometrical argument
implies the following ( see figure 1 ) : let @xmath257 be a uniformly lipschitz domain , with the definition [ def : lip ] , and @xmath221 the map defined in .
then , there exists @xmath237 depending only on the constants @xmath60 , @xmath52 , @xmath63 such that @xmath222 is homotopically equivalent to @xmath4 for all @xmath258 .
in particular , if @xmath4 is connected , @xmath24 is connected .
the proof is a direct application of the arguments in @xcite . in that paper ,
lemma 3.3 gives a lipschitz pseudogradient vector field and lemma 3.4 uses it to define a convenient deformation , which provides us with the desired homotopy .
observe that the ( ps ) condition holds since we are in a compact framework without critical points .
|
the classical poincar inequality establishes that for any bounded regular domain @xmath0 there exists a constant @xmath1 such that @xmath2 in this note we show that @xmath3 can be taken independently of @xmath4 when @xmath4 is in a certain class of domains .
our result generalizes previous results in this direction .
dedicated to a. ambrosetti , a guide who definitively changed my life
|
the traditional cosmology has been revolutionized by modern sophisticated observation techniques in distant type ia supernovae ( sne ia ) ( e.g. riess et al . 1998 ; schmidt et al . 1998 ; perlmutter et al . 1999 ) , cosmic microwave background ( cmb ) fluctuations ( e.g. bennett et al . 2003 ; spergel et al . 2003 ) , and large - scale structure ( lss ) ( e.g. allen et al .
2003 ; tegmark et al .
each type of cosmological data trends to play an unique role in measuring cosmology . in modern cosmology
, it has been convincingly suggested that the global mass - energy budget of the universe , and thus its dynamics , is dominated by a dark energy component , and that the currently accelerating universe has once been decelerating ( e.g. , riess et al .
the cosmography and the nature of dark energy as well as its evolution with redshift are one of the most important issues in physics and astronomy today .
gamma - ray bursts ( grbs ) are the most intense explosions observed so far .
they are believed to be detectable up to a very high redshift ( lamb & reichart 2000 ; ciardi & loeb 2000 ; bromm & loeb 2002 ; gou et al .
2004 ) , and their high energy photons are almost immune to dust extinction .
these advantages would make grbs an attractive cosmic probe . from the isotropic - equivalent peak luminosity @xmath14variability ( or spectral lag )
relation ( fenimore & ramirez - ruiz 2000 ; norris , marani , & bonnell 2000 ) , the standard energy reservior @xmath1 of grb jets ( frail et al . 2001 ) , the @xmath14 peak energy @xmath2 of the @xmath3 spectrum in the burst frame relation ( lloyd - ronning & petrosian 2002 ; lloyd - ronning & ramirez - ruiz 2002 ; yonetoku et al .
2004 ) , the isotropic - equivalent energy @xmath15 relation ( amati et al .
2002 ) to the beaming - corrected energy @xmath4 relation ( ghirlanda et al .
2004a , ghirlanda relation hereafter ) , grbs are towards more and more standardizable candles
. however , these relations in grbs have not been calibrated by a low-@xmath13 grb sample , so one should look for a method which is different from the `` classical hubble diagram '' method in sne ia .
the luminosity relations with the variability and spectral lag make grbs a distance indicator in the same sense as cepheids and sne ia , in which an observed light - curve property can yield an apparent distance modulus ( dm ) .
schaefer ( 2003 ) considered these two relations for nine bursts with known redshifts and advocated a new cosmographic method ( hereafter method i ) for grbs .
in method i , one first calibrates the two relations with the observed sample for a certain cosmology , and then applies the best - fit relations back to the observed sample to obtain a @xmath16 or a probability @xmath17 for this cosmology .
similar to the brightness of sne ia , the energy reserviors in grb jets are also clustered , but they are not fine enough for precise cosmology ( bloom et al .
amati et al .
( 2002 ) found @xmath18 ( @xmath19 ) relation from 12 _ bepposax _ bursts .
the _ hete-2 _
observations confirm this relation and extend it to x - ray flashes ( sakamoto et al .
2004a ; lamb et al .
in addition , it also holds within a grb ( liang , dai & wu 2004 ) .
the ghirlanda relation is written as @xmath20 , where @xmath21 and @xmath22 are dimensionless parameters .
theoretical explanations of this relation include the standard synchrotron mechanism in relativistic shocks ( zhang & mszros 2002 ; dai & lu 2002 ) together with the afterglow jet model , or the emission from off - axis relativistic jets ( yamazaki , ioka & nakamura 2004 ; eichler & levinson 2004 ; levinson & eichler 2005 ) .
this relation could also be understood due to comptonization of the thermal radiation flux that is advected from the base of an outflow in the dissipative photosphere model ( e.g. rees & mszros 2005 ) .
if these explanations are true , the ghirlanda relation appears to be intrinsic .
thus , dai , liang & xu ( 2004 ; dlx04 ) considered the ghirlanda relation for 12 bursts and proposed another cosmographic method ( hereafter method ii ) for grbs . in method ii , one makes marginalizations over the unknown parameters in the ghirlanda relation to obtain a @xmath16 or a probability @xmath17 for a certain cosmology . following schaefer s method , ghirlanda et al .
( 2004b ; gglf04 ) and friedman & bloom ( 2004 ; fb04 ) also investigated the same issue but used different grb data . recently ,
firmani et al .
( 2005 ; fgga05 ) considered the ghirlanda relation for 15 bursts and proposed a bayesian approach for the cosmological use ( hereafter method iii ) . in method
iii , to obtain the probability for a certain cosmology , one considers contributions of all the possible @xmath4 relations associated with their unequal weights .
the detailed procedures of the three methods are shown in @xmath23 , which indicate that method iii is the optimized one .
as analyzed previously , due to the lack of low-@xmath13 grbs , methods i , ii and iii are different from the `` classic hubble diagram '' method in sne ia . in this paper
, we investigate the constraints on cosmological parameters from the observed 17 grbs with different methods . because the present grb sample is a small one
, it is necessary to use a large simulated sample , which may be established in the _ swift _ era , to discuss the cosmographic capabilities with different methods .
this paper is arranged as follows . in @xmath24 , we describe our analytical methods and data .
the results from the observed grb sample are presented in @xmath25 . in @xmath26 , we perform monte carlo simulations and analyze the results from the simulated grb sample .
conclusions and discussion are presented in @xmath27 .
according to the relativistic fireball model , the emission from a spherically expanding shell and from a jet is similar to each other , if the observer is along the jet s axis and the lorentz factor of the fireball is larger than the inverse of the jet s half - opening angle @xmath28 ; but when the lorentz factor drops below @xmath29 , the jet s afterglow light curve is expected to present a break because of the edge effect and the laterally spreading effect ( rhoads 1999 ; sari , piran & halpern 1999 ) .
therefore , together with the assumptions of the initial fireball emitting a constant fraction @xmath30 of its kinetic energy into the prompt @xmath0-rays and a constant circumburst particle density @xmath31 , the jet s half - opening angle is derived to be @xmath32 where @xmath33 , @xmath34 , @xmath35 .
the `` bolometric '' isotropic - equivalent @xmath0-ray energy of a grb is given by @xmath36 where @xmath37 is the fluence ( in units of erg@xmath38 ) received in an observed bandpass and the quantity @xmath39 is a multiplicative correction of order unity relating the observed bandpass to a standard rest - frame bandpass ( 1-@xmath40 kev in this paper ) ( bloom , frail & sari 2001 ) .
the energy release of a grb jet is thus given by @xmath41 with the fractional uncertainty ( fb04 ) being @xmath42 + c_\theta \nonumber \\ & & \times \left [ { \left ( { \frac{{3\sigma _ { t_j } } } { { t_j } } } \right)^2 + \left ( { \frac{{\sigma _ { n_0 } } } { { n_0 } } } \right)^2 + \left ( { \frac{{\sigma _ { \eta _ \gamma } } } { { \eta _ \gamma- \eta^2 _ \gamma } } } \right)^2 } \right],\end{aligned}\ ] ] where @xmath43 ^ 2.\ ] ] the ghirlanda relation is @xmath44 where @xmath21 and @xmath22 are assumed to have no covariance , and @xmath45 .
combining eqs .
( 1 ) , ( 2 ) , ( 3 ) and ( 6 ) , we derive the apparent luminosity distance with the small angle approximation ( i.e. , @xmath46 ) when @xmath47 rad , and @xmath48 when @xmath49 rad . ] as @xmath50^{2a/3 } } } { { ( ks_\gamma t_{j , d } ) ^{1/2 } ( n_0 \eta _ \gamma ) ^{1/6 } } } { \rm { mpc}},\ ] ] assuming that all the observables are independent of each other and their errors satisfy gaussian distributions , we derive the fractional uncertainty of the apparent luminosity distance without the small angle approximation , @xmath51 + \frac{1}{4}\frac{{1 } } { { ( 1 - \sqrt { c_{\theta } } ) ^2 } } \left [ \left ( { \frac{{\sigma _ c } } { c}}\right)^2 \right . \nonumber \\ & & \left . + \left ( { a\frac{{\sigma _ { e_p^{obs } } } } { { e_p^{obs } } } } \right)^2 + \left ( { a\frac{{\sigma _ a } } { a}\ln \frac{e_p}{100 } } \right)^2 \right]+ \frac{1}{4}\frac{c_{\theta } } { { ( 1 - \sqrt { c_{\theta } } ) ^2 } } \nonumber \\ & & \times \left [ { \left ( { \frac{{3\sigma _ { t_j } } } { { t_j } } } \right)^2 + \left ( { \frac{{\sigma _ { n_0 } } } { { n_0 } } } \right)^2}+ \left ( { \frac{{\sigma _ { \eta _ \gamma } } } { { \eta _ \gamma ( 1 - \eta _ \gamma ) } } } \right)^2 \right].\end{aligned}\ ] ] for simplicity , we consider @xmath52 and @xmath53 throughout this paper ( frail et al . 2001 )
. the apparent dm of a burst can be given by @xmath54 with the uncertainty of @xmath55 on the other hand , the theoretical luminosity distance in @xmath7-models ( carroll , press & turner 1992 ) is given by @xmath56^{-1/2}\},\end{aligned}\ ] ] where @xmath57 , and `` sinn '' is @xmath58 for @xmath59 and @xmath60 for @xmath61 . for @xmath62 , equation ( 11 ) degenerates to be @xmath63 times the integral .
usually , the likelihood for the parameters @xmath64 and @xmath65 can be determined from a @xmath16 statistic , where @xmath66 } ^2,\ ] ] where the dimensionless hubble constant @xmath67 is taken as 0.71 .
if the ghirlanda relation could be calibrated by low-@xmath13 bursts , the above @xmath16 statistic becomes the same as that in sne ia , that is , @xmath68 } ^2,\ ] ] where @xmath69 should be marginalized .
the procedures of method i , ii and iii are as follows : * method i * ( see s03 , gglf04 , fb04 ) the procedure of this method is to ( 1 ) fix @xmath70 , ( 2 ) calculate @xmath71 and @xmath1 for each burst for that cosmology , ( 3 ) best fit the @xmath12 relation to yield @xmath72 and @xmath73 , ( 4 ) substitute @xmath72 and @xmath73 into eqs .
( 7 ) and ( 8) , and thus derive @xmath74 and @xmath75 for each burst for cosmology @xmath76 , ( 5 ) calculate @xmath16 for cosmology @xmath76 by comparing @xmath71 with @xmath74 , @xmath75 , and then convert it to the probability by @xmath77 ( riess et al .
1998 ) , ( 6 ) repeat steps 1@xmath785 from @xmath79 to @xmath80 to obtain the probability for each cosmology . therefore , method i is formulizd by @xmath81 * method ii * ( see dlx04 ) the procedure of this method is to ( 1 ) fix @xmath82 , ( 2 ) calculate @xmath71 and @xmath1 for each burst for that cosmology , ( 3 ) best fit the @xmath12 relation to yield @xmath72 and @xmath73 , ( 4 ) substitute @xmath72 and @xmath73 into eqs .
( 7 ) and ( 8) , and thus derive @xmath74 and @xmath75 for each burst for cosmology @xmath76 , ( 5 ) repeat steps 1@xmath784 from @xmath79 to @xmath83 to obtain @xmath71 , @xmath74 and @xmath75 for each burst for each cosmology ; ( 6 ) re - fix @xmath84 , ( 7 ) calculate @xmath85 by comparing @xmath86 with @xmath87 , @xmath88 , and then convert it to a conditional probability by @xmath89 , ( 8) repeat step 7 from @xmath79 to @xmath90 to obtain the probability for cosmology @xmath84 by @xmath91 , ( 9 ) repeat steps 6@xmath788 from @xmath92 to @xmath93 to obtain the probability for each cosmology .
method ii is described by @xmath94 * method iii * ( see fgga05 ) method iii is an improvement of method ii .
its key idea is to consider unequal weights for different @xmath4 relations , i.e. , unequal weights for different conditional probabilities @xmath95 .
therefore , the first seven steps of this method are the same as those of method ii .
the follow - up procedure is to ( 8) repeat step 7 from @xmath79 to @xmath90 to obtain an iterative probability for cosmology @xmath84 by @xmath96 ( here the initial probability @xmath97 for each cosmology is regarded as equal ; e.g. @xmath98 ) , ( 9 ) repeat steps 6@xmath788 from @xmath92 to @xmath93 to obtain an iterative probability @xmath99 for each cosmology ; ( 10 ) replace @xmath97 on step 8 with @xmath99 on step 9 , then repeat steps 8@xmath789 , and thus reach another set of iterative probabilities for each cosmology , ( 11 ) run the above cycle again and again until probability for each cosmology converges , i.e. @xmath100 after tens of cycles . in this method ,
to calculate the probability @xmath101 for a favored cosmology , we consider contributions of all the possible @xmath4 relations associated with their weights .
the conditional probability @xmath102 denotes the contribution of some certain relation , and @xmath103 weights the likelihood of this relation for its corresponding cosmology .
therefore , the bayesian approach can be formulized by @xmath104 however , fgga05 took different calculations for the conditional probability @xmath105 . making use of the incomplete gamma function
, they transformed @xmath106 into its corresponding conditional probability @xmath102 .
actually , once the parameters @xmath72 and @xmath73 of the ghirlanda relation are calibrated for cosmology @xmath82 , they become `` known '' for the cosmic model @xmath84 .
so herein the meaning of @xmath85 is the same as that in sne ia . in this paper
, we redefine the conditional probability @xmath102 by the formula of @xmath89 .
the great diversity in grb phenomena suggests that the grb population may consist of substantially different subclasses ( e.g. macfadyen & woosley 1999 ; bloom et al .
2003 ; sazonov et al .
2004 ; soderberg et al .
2004 ) . to make grbs a standard candle ,
a homogenous grb sample is required .
the most prominent observational evidence for a grb jet is its temporal break in their afterglow light curves .
for some bursts , e.g. , grb030329 , their temporal breaks are observed in both optical and radio bands .
berger et al . ( 2003 ) argued that these two breaks are caused by the narrow component and wide component of the jet in this burst , respectively , indicating that the physical origins of the breaks in the optical band and in the radio band are different .
in addition , the radio afterglow light curves fluctuate significantly .
for example , in the case of grb970508 , the light curve of its radio afterglow does not present clearly a break .
only a lower limit of @xmath107 days was proposed by frail et al .
furthermore , the light curve of its optical afterglow is proportional to @xmath108 , in which case no break appears ( galama et al .
we thus include only those bursts whose temporal breaks in optical afterglow light curves were well measured in our analysis .
we obtain a sample of 17 grbs excluding grb970508 .
they are listed in table 1 .
we correct the observed fluence in a given bandpass to a `` bolometric '' bandpass of @xmath109 with spectral parameters . the fluence and spectral parameters for a burst fitted by different authors may be affected by different criterions ( or systematical biases ) in their works .
we thus collect a couple of fluence and spectral parameters from the same original literature . for grb 970828 ,
grb 980703 , grb 991216 , and grb 011211 , the fluence and spectral parameters are unavailable in the same original literature , so we choose their fluences measured in the widest energy band available in the other literature . for grb 011211 , we approximately take the high - energy spectral index @xmath110 to be @xmath111 because it is unreported in amati ( 2004 ) . the spectra of _
hete-2_-detected grb020124 , grb020813 , grb021004 , grb030226 and xrf030429 are not fitted by the band function but the cutoff power law model .
however , it is appropriate that their corresponding `` bolometric '' fluences are calculated by the former model with @xmath112 , avoiding the potential systematical bias which is brought by applying different spectral models for one observed sample ( barraud et al .
2003 ) .
the circumburst densities of several bursts in our sample have been obtained from broadband modelling of the afterglow emission ( e.g. , panaitescu & kumar 2002 ) . for the bursts with unknown @xmath31 , we assumed @xmath113 as the median value of the distribution of the measured densities , together with a constant fractional uncertainty of @xmath114 ( ghirlanda et al .
2004a ; dlx04 ) .
we first fit the @xmath4 relation for the cosmology of @xmath11 , @xmath115 , and @xmath116 , and obtain @xmath117 , @xmath118 , @xmath119 and @xmath120 , together with the @xmath121 , using eq .
( 4 ) for the estimation of @xmath122 ( press et al .
1999 ) . substituting the best - fit results into eqs .
( 7 ) and ( 8) , we plot the hubble diagram for the observed grb sample in the concordance model of @xmath11 , which is shown in fig 1 ( filled circles ) . comparatively shown
are the hubble diagrams for the binned gold sample of sne ia ( open circles ; riess et al .
2004 ) , and the theoretical model of @xmath11 and @xmath115 ( solid line ) . the constraints with method i are shown in fig 2 ( solid contours ) .
in the concordance model of @xmath11 , we find a @xmath123 .
we measure @xmath5 ( @xmath6 ) for a flat @xmath7cdm universe .
dlx04 proposed method ii based on the principle that if there are unknown cosmology - independent parameters in the @xmath16 statistic , they are usually marginalized over ( i.e. integrating the parameters according to their probability distribution ) . note that dlx04 let the parameter @xmath21 of the ghirlanda relation intrinsically equal to 1.50 .
for the purpose of universality , in this paper we let it vary freely , similar to the parameter @xmath22 of this relation ( see details in @xmath23 ) . the constraints with method ii are shown in fig 3 ( solid contours ) .
this method provides a more stringent constraint of @xmath8 ( @xmath6 ) for a flat universe . by method
i , we recalibrate the ghirlanda relation for each cosmology , i.e. , @xmath64 and @xmath65 taken from 0 to 1 .
we find the half - opening angles of all the bursts are less than 0.23 rad , and that the mean @xmath124 and @xmath125 are 0.049 and 0.083 . as a result , the typical error terms in eq .
( 8) are @xmath126 , @xmath127 , @xmath128(3\sigma _ { t_j } /t_j )
\sim 0.091 $ ] , @xmath128(\sigma _ { n_0 } /n_0 ) \sim 0.118 $ ] , @xmath129(\sigma _ c /c ) \sim 0.055 $ ] , @xmath130(\sigma _ { e_p^{obs } } /e_p^{obs } ) \sim 0.168 $ ] , and @xmath131\left| { \ln [ e_p^{obs } ( 1 + z)/100 ] } \right|\sim 0.056 $ ] .
these error terms give a typical uncertainty of apparent dm @xmath132 magnitudes , which is a factor of @xmath133 larger than that derived from the sn ia gold sample .
firmani et al .
( 2005 ) proposed the bayesian approach to use grbs as cosmic rulers .
being different from their work , we redefined the conditional probability in their method and called it method iii in this paper .
the constraints on cosmological parameters with this method are shown in fig 4 ( solid contours ) .
compared with previous two methods , method iii does give much more stringent confidence intervals at the same confidence levels ( c.l . ) .
the results are inspiring and demonstrate the advantages of high-@xmath13 distance indicators in constraining cosmological parameters .
the data set is consistent with the concordance model of @xmath11 , yielding a @xmath10 .
we also find @xmath9 ( @xmath6 ) for a flat universe .
as discussed above , the method for the grb cosmology is different from that for the sn cosmology due to the lack of the low-@xmath13 calibration .
the low-@xmath13 calibration shall greatly enhance the cosmographic capability of grbs ( assuming no cosmic evolution for the ghirlanda relation ) .
therefore , one may ask : _ could a large high- @xmath13 grb sample effectively measure cosmology ? _ to answer this question , we carry out a monte carlo simulation and use a large simulated sample to investigate the cosmographic capabilities in different scenarios . our simulations are based on the ghirlanda relation which is calibrated for @xmath11 and @xmath115 . making use of the sample in table 1 , we find @xmath117 , @xmath118 , @xmath134 rad , and @xmath135 $ ] .
we also find @xmath136 $ ] .
these restrictive conditions are imposed upon our simulations .
each simulated grb is characterized by a set of @xmath137 , @xmath138 , @xmath139 , @xmath13 and @xmath140 , where @xmath137 is the bolometric fluence .
the simulation procedure is as follows : \(1 ) we consider the lognormal distributions for the observables @xmath141 , @xmath142 and @xmath31 . from the 17 grbs , we find @xmath143 with @xmath144 , and @xmath145 with @xmath146 .
because @xmath31 is unavailable in the literature for most grbs , we take @xmath147 with @xmath148 . the observable @xmath13 is selected according to its observational distribution cut by an upper limit @xmath149 , which is shown in fig 5 .
the uncertainty of @xmath13 is ignored .
\(2 ) we also consider the lognormal distributions for the fractional uncertainties of the observables @xmath141 , @xmath142 , @xmath31 and @xmath2 . from the observed sample ,
we find @xmath150 with @xmath151 } \sim 0.16 $ ] , @xmath152 with @xmath153 } \sim 0.45 $ ] , @xmath154 with @xmath155 } \sim 0.10 $ ] , and @xmath156 with @xmath157 } \sim 0.26 $ ] , respectively .
we ensure in code that the fractional uncertainties of the observables @xmath141 , @xmath142 , @xmath31 and @xmath2 are less than @xmath158 , @xmath159 , @xmath160 and @xmath159 , respectively .
\(3 ) we simulate a grb characterized by a set of ( @xmath161 , @xmath162 , @xmath163 and @xmath13 ) according to the distributions that these parameters follow , compute its @xmath164 in the cosmic model of @xmath11 and @xmath115 , and then calculate its @xmath2 by the ghirlanda relation of @xmath165 .
\(4 ) the grb generated in step ( 3 ) follows the `` rigid '' ghirlanda relation .
we add a random deviation to each parameter , except for @xmath13 , to make this burst more realistic , i.e. , @xmath166 , @xmath167 , @xmath168 and @xmath169 , where @xmath170 is randomly taken from 0 and 1 .
\(5 ) using the parameters @xmath137 , @xmath138 , @xmath139 , @xmath13 , and @xmath171 , we compute the parameters @xmath172 and @xmath173 for @xmath11 and @xmath115 , and the quantity @xmath174 .
\(6 ) since @xmath134 rad , @xmath135 $ ] and @xmath136 $ ] are valid for the observed sample , we require that @xmath172 , @xmath175 , and @xmath176 of a simulated grb must be within the corresponding ranges .
\(7 ) repeat steps ( 3)-(6 ) to generate a sample of 80 bursts .
the half - opening angles of the simulated sample are less than 0.23 rad when @xmath64 and @xmath65 are taken from 0 to 1 , as in the observed sample .
we carry out a circular operation to achieve the simulated sample , which may be established in the _ swift _ era .
by method i , we find the typical @xmath124 and @xmath125 for the large sample decrease to 0.02 and 0.05 . this is because for the small observed sample , the dispersion of the ghirlanda relation is mainly contributed by a few `` outliers '' , while for the large simulated sample , the bursts are distributed around the `` rigid '' ghirlanda relation with a gaussian distribution ( see @xmath177 ) .
such a large sample seems to be more realistic . among several scenarios
, we only perform to what degree the simulated sample can constrain the @xmath178 parameters with method iii .
the results are shown by solid confidence intervals in fig 6 .
also shown are the constraints derived from the sn gold sample ( dashed contours ) at the same confidence levels .
the main points revealed by this figure are : ( 1 ) a large high-@xmath13 grb sample could effectively measure cosmology .
the simulated grb data are consistent with the concordance model of @xmath179 , yielding a @xmath180 . with the prior of a flat universe ,
the mass density is @xmath181 at the @xmath182 level .
( 2 ) grbs can well constrain @xmath64 rather than @xmath65 due to their high redshifts . the orientation of the elliptical contours
is almost vertical to the @xmath64 axis .
this is an advantage of grbs over sne ia for the cosmological use .
( 3 ) grbs supply complemental content to the sn cosmology .
the 157 sne ia provide evidence of cosmic acceleration at a very high confidence level .
a large grb sample would reach a similar conclusion . alone
the sn sample nearly rules out the flat universe model at @xmath6 level , but a combination of grbs and sne makes the concordance model of @xmath179 more favored and thus more agreement with the conclusion from wmap observations ( spergel et al .
2003 ) .
if the low-@xmath13 calibration is realized for grbs , then the solid confidence regions in fig 6 will become smaller so that they lie in the part of cosmic acceleration at a high c.l . in the @xmath178 plane ( not shown in this work ) .
we here present a rough estimation of the detection rate of low-@xmath13 bursts in the _ swift _ era .
taking @xmath11 , @xmath115 , and @xmath116 , the observed rate of bursts with redshift less than @xmath13 is @xmath183 where @xmath184 is the comoving grb rate density , and @xmath185 is the isotropic comoving volume element , @xmath186 ^ 2 \frac{1}{{\sqrt { \omega _ m ( 1 + z)^3 + \omega _ \lambda } } } .\ ] ] we assume @xmath187 , where @xmath188 and @xmath189 are the comoving rate densities of core - collapse supernovae and star formation , respectively ( porciani & madau 2001 ) .
we also assume the constant fraction @xmath190 , which considers the grb formation efficiency out of the core - collapse sne and the grb beaming effect .
additionally , we take @xmath191 .
the global star formation rate for the einstein - de sitter universe ( steidel et al .
1999 ) is @xmath192 thus , we derive the detective probability of low-@xmath13 grbs in the _ swift _ era ( 2@xmath784 years ) : when @xmath193 , there will be @xmath194 burst ; and when @xmath195 , there will be a few bursts .
these results agree with the present observational grb data .
in addition , taking into account possible different origins of very low-@xmath13 bursts ( e.g. grb980425 ) and high-@xmath13 bursts , it might not be valid to directly apply the low-@xmath13 calibrated relation to a high-@xmath13 sample .
however , if the cosmic evolution of the `` standard candle '' relation is ignored , then in the future one could consider a sample of grbs with redshift @xmath195 for the low-@xmath13 calibration .
as the zero - order approximation , the theoretical luminosity distance of such a grb sample can be written as @xmath196 .
at present , grbs with known redshifts are about 45 out of a few thousands , among which 17 are available to derive the ghirlanda relation .
this relation is so tight that it has been considered for the cosmological use .
although the low-@xmath13 calibration of the ghirlanda relation is not realized , the observed 17 high-@xmath13 bursts still provide interesting and even inspiring results .
we find grbs independently place a constraint on the mass density @xmath197 ( @xmath6 ) for a flat @xmath7cdm model .
grbs are becoming more and more standardized candles . through the ghirlanda relation ,
the mean scatter in grbs is a factor of @xmath133 larger than that in sne ia .
however , the disadvantage of larger scatter in grbs is compensated , to some extent , by their advantages of high redshifts and immunity to the dust extinction .
the shape of the constraints in fig 6 implies that grbs could not only measure the mass density @xmath64 but also provide complementarity to the sn cosmology .
the _ swift _ satellite is hopeful to establish a large grb sample with known redshifts , perhaps including low-@xmath13 bursts . in this sense , the grb cosmology now lies in its babyhood .
a reliable theoretical basis of the ghirlanda relation is also important for grbs as a cosmic ruler . using the small angle approximation for all bursts
, a scaling analysis gives @xmath198 ( @xmath199 ) as long as the parameters such as the spectral index @xmath200 of the distribution of accelerated electrons , the the energy equipartition factor @xmath201 of the electrons , the energy equipartition factor @xmath202 of magnetic field , the bulk lorentz factor @xmath0 , etc . , or their combinations are clustered ( zhang & mszros 2002 ; dai & lu 2002 ; wu , dai & liang 2004 ) .
this power - law relation could also be understood by the emissions from off - axis relativistic jets ( yamazaki , ioka & nakamura 2004 ; eichler & levinson 2004 ; levinson & eichler 2005 ) and the dissipative photosphere model ( rees & mszros 2005 ) .
different plausible explanations imply that this topic needs further investigations .
it should be pointed out that the ghirlanda relation is obtained under the framework of the uniform top - hat jet model .
other input assumptions include the uniform circumburst medium density @xmath31 and the constant efficiency @xmath30 of converting the initial ejecta s kinetic energy into @xmath0-ray energy release .
however , @xmath31 and @xmath30 should be different from burst to burst , and @xmath31 is variable for a burst in the wind environment ( dai & lu 1998 ; chevalier & li 1999 ) .
thus , the quantity @xmath203 in the ghirlanda relation might not be clustered for the observed bursts ( friedman & bloom 2004 ) .
new relations with as few well - observed quantities as possible are required for improvement - ray energy and the @xmath204 peak energy by considering the break time of an afterglow light curve . clearly , these three quantities are directly observed . ] . in this paper , for those bursts with unknown @xmath31
, we assumed @xmath113 as the median value of the distribution of the measured densities , together with a constant fractional uncertainty of @xmath114 ( ghirlanda et al .
2004a ; dlx04 ) .
we also treat @xmath52 and @xmath53 for all the 17 bursts .
finally , the ghirlanda relation is not valid for all the observed batse sample .
this implies that this relation may suffer from the data selection effect ( band & preece 2005 ) .
its validity will be tested by the ongoing observations of the _ swift _ mission . however , no matter whether the low-@xmath13 sample is established or not , the grb cosmology is expected to progress much in the coming years .
xd is grateful to x.l .
luo for helpful discussions .
we thank the anonymous referees for valuable comments that have allowed us to improve our manuscript significantly .
this work was supported by the national natural science foundation of china ( grants 10233010 , 10221001 , and 10463001 ) , the ministry of science and technology of china ( nkbrsf g19990754 ) , the national post doctoral foundation of china , and the research foundation of guangxi university .
allen , s. w. et al .
2003 mnras 342 , 287 amati , l. et al .
2002 , a&a , 390 , 81 amati , l. 2004 , astro - ph/0405318 andersen , m. i. et al .
2003 , gcn , 1993 atteia , j. l. 2003 , a&a , 407 , l1 barraud , c. et al .
2003 , a&a , 400 , 1021 barth , a. j. et al .
2003 , apj , 584 , l47 band , d.l . ,
matteson , j. , ford , l. , et al .
1993 , apj , 413 , 281 band , d. l. & preece , r. d. 2005 , apj , 627 , 2 bennett , c. l. et al .
2003 , apjs , 148 , 1 berger , e. et al .
2002 , apj , 581 , 981 berger , e. et al .
2003 , nature , 426 , 154 bjornsson g. et al .
2001 , apj , 552 , l121 bloom , j. s. , frail , d. a. , & sari , r. 2001 , aj , 121 , 2879 bloom , j. s. , frail , d. a. , & kulkarni , s. r. 2003 , apj , 594 , 674 bromm , v. , & loeb , a. 2002 , apj , 575 , 111 carroll , s. m. , press , w. h. , & turner , e. l. 1992 , ara&a , 30 , 499 chevalier , r. a. & li , z. y. 1999 , apj , 520 , l29 ciardi , b. , & loeb , a. 2000 , apj , 540 , 687 crew , g. b. et al .
2003 , apj , 599 , 387 dai , z. g. , & lu , t. 1998 , mnras , 298 , 87 dai , z. g. , & lu , t. 2002 , apj , 580 , 1013 dai , z. g. , liang , e. w. & xu , d. 2004 , apj , 612 , l101 ( dlx04 ) djorgovski , s. g. et al .
1998 , apj , 508 , l17 djorgovski , s. g. et al .
2001 , apj , 562 , 654 eichler , d. , & levinson , a. 2004 , apj , 614 , l13 fenimore , e. e. & ramirez - ruiz e. 2000 , astro - ph/0004176 firmani , c. , ghisellini , g. , ghirlanda , g. & avila - reese , v. 2005 , mnras , 360 , 1 ( fgga05 ) frail , d. a. , waxman , e. , and kulkarni , s. r. 2000 , apj , 537 , 191 frail , d. a. et al .
2001 , apj , 562 , l55 frail , d. a. et al . 2003 , apj , 590 , 992 friedman , a. s. & bloom , j. s. , apj accepted , ( astro - ph/0408413 ; fb04 ) fruchter , a. et al . 1999 ,
apj , 519 , l13 galama , t. j. et al .
1998 , apj , 497 , l13 ghirlanda , g. , ghisellini , g. , & lazzati , d. 2004a , apj , 616 , 331 ghirlanda , g. , ghisellini , g. , lazzati , d. & firmani , c. 2004b , apj , 613 , l13 ( gglf04 ) gou , l. j. , mszros , p. , abel , t. & zhang , b. 2004 , apj , 604 , 508 greiner , j. et al .
2003a , gcn , 1886 greiner , j. et al .
2003b , gcn , 2020 halpern , j. p. , et al .
2000 , apj , 543 , 697 hjorth , j. et al .
2003 , apj , 597 , 699 holland , et al .
2002 , aj , 124 , 639 holland , et al .
2003 , aj , 125 , 2291 holland , et al .
2004 , astro - ph/0405062 jakobsson , p. et al .
2003 , a&a , 408 , 941 jakobsson , p. et al .
2004 , a&a , 427 , 785 jimenez , r. , band , d. l. , & piran , t. , 2001 , apj , 561 , 171 klose , s. et al .
2004 , aj , 128 , 1942 knop et al .
2003 , apj , 598 , 102 kulkarni , s. r. et al . 1999 ,
nature , 398 , 389 lamb , d. q. , & reichart , d. e. 2000 , apj , 536 , 1 lamb , d. q. et al .
2004 , newa rev . ,
48 , 423 le floch , e. 2002 , apj , 581 , l81 liang , e. w. , dai , z. g. , & wu , x. f. 2004 , apj , 606 , l29 liang , e. w. & zhang b. 2005 , apj submitted ( astro - ph/0504404 ) lloyd - ronning , n. m. , & petrosian , v. 2002 , apj , 565 , 182 lloyd - ronning , n. m. , & ramirez - ruiz , e. 2002 , apj , 576 , 101 macfadyen , a.i . , & woosley , s. e. 1999 , apj , 524 , 262 masetti , n. et al . 2000 ,
a&a , 354 , 473 matheson , t. et al .
2003 , apj , 582 , l5 norris , j. p. et al .
2000 , apj , 534 , 248 panaitescu , a. , & kumar , p. 2002 , apj , 571 , 779 perlmutter , s. et al .
1999 , apj , 517 , 565 porciani , c. & madau , p. 2001 , apj , 548 , 522 piro , l. et al .
2000 , science , 290 , 955 press , w. h. et al 1999 , _ numerical recipes in fortran _ , cambridge university press , 660 - 664 price , p. a. et al .
2002 , apj , 573 , 85 price , p. a. et al . 2003a ,
apj , 589 , 838 price , p. a. et al .
2003b , nature , 423 , 844 porciani , c. & madau , p. 2001 , apj , 548 , 522 rhoads , j. e. 1999 , apj , 525 , 737 riess , a. g. et al .
1998 , aj , 116 , 1009 riess , a. g. et al .
2004 , apj , 607 , 665 rol , e. et al .
2003 , gcn , 1981 sakamoto , t. et al .
, 2004a , apj , 602 , 875 sakamoto , t. et al .
, 2004b , astro - ph/0409128 sari , r. , piran , t. , & narayan , r. 1998 , apj , 497 , l17 sari , r. , piran , t. , & halpern , j. p. 1999
, apj , 519 , l17 sazonov , s. y. et al .
nature 430 , 646 schaefer , b. e. 2003 , apj , 583 , l67 ( s03 ) schmidt , b. p. et al .
1998 , apj , 507 , 46 soderberg , a. m. et al .
nature 430 , 648 spergel , d. n. et al .
2003 , apjs , 148 , 175 stanek , k. z. et al .
1999 , apj , 522 , l39 steidel , c. c. et al . 1999 ,
apj , 519 , 1 tegmark , m. et al .
2004 , , 69 , 103501 tiengo , a. et al .
2003 , a&a , 409 , 938 vanderspek , r. et al .
2004 , aj , 617 , 1251 van den bergh , s. , & pazder , j. 1992 , apj , 390 , 34 vreeswijk , p. m. et al .
2001 , apj , 546 , 672 vreeswijk , p. m. et al .
2003 , gcn , 1785 wu , x. f. , dai , z. g. , & liang , e. w. 2004 , apj , 615 , 359 xu , d. 2005 , apj submitted ( astro - ph/0504052 ) yamazaki , r. , ioka , k. , & nakamura , t. 2004 , apj , 606 , l33 yonetoku , d. et al .
2004 , apj , 609 , 935 zhang , b. & mszros , p. 2002 , apj , 581 , 1236 grb970828 & 0.9578 & 297.7[59.5 ] & -0.70 , -2.07 & 96.0[9.6 ] & 20 - 2000 & 2.2(0.4 ) & 3.0[2.4 ] & 0.2 & 01,18,18,26,01,no & batse + grb980703 & 0.966 & 254.0[50.8 ] & -1.31 , -2.40 & 22.6[2.26 ] & 20 - 2000 & 3.4(0.5 ) & 28.0(10.0 ) & 0.2 & 02,18,18,26,27,27 & batse + grb990123 & 1.600 & 780.8(61.9 ) & -0.89 , -2.45 & 300.0(40.0 ) & 40 - 700 & 2.04(0.46 ) & 3.0[2.4 ] & 0.2 & 03,19,19,19,28,no & sax + grb990510 & 1.619 & 161.5(16.0 ) & -1.23 , -2.70 & 19.0(2.0 ) & 40 - 700 & 1.57(0.03 ) & 0.29(0.14 ) & 0.2 & 04,19,19,19,29,41 & sax + grb990705 & 0.8424 & 188.8(15.2 ) & -1.05 , -2.20 & 75.0(8.0 ) & 40 - 700 & 1.0(0.2 ) & 3.0[2.4 ] & 0.2 & 05,19,19,19,30,no & sax + grb990712 & 0.4331 & 65.0(10.5 ) & -1.88 , -2.48 & 6.5(0.3 ) & 40 - 700 & 1.6(0.2 ) & 3.0[2.4 ] & 0.2 & 06,19,19,19,31,no & sax + grb991216 & 1.020 & 317.3[63.4 ] & -1.23 , -2.18 & 194.0[19.4 ] & 20 - 2000 & 1.2(0.4 ) & 4.7(2.8 ) & 0.2 & 07,18,18,26,32,41 & batse + grb011211 & 2.140 & 59.2(7.6 ) & -0.84 , -2.30 & 5.0[0.5 ] & 40 - 700 & 1.56(0.02 ) & 3.0[2.4 ] & 0.2 & 08,20,20,26,33,no & sax + grb020124 & 3.200 & 120.0(22.6 ) & -1.10 , -2.30 & 6.8[0.68 ] & 30 - 400 & 3.0(0.4 ) & 3.0[2.4 ] & 0.2 & 09,21,21,21,34,no & hete-2 + grb020405 & 0.690 & 192.5(53.8 ) & 0.00 , -1.87 & 74.0(0.7 ) & 15 - 2000 & 1.67(0.52 ) & 3.0[2.4 ] & 0.2 & 10,22,22,22,22,no & sax + grb020813 & 1.255 & 212.0(42.0 ) & -1.05 , -2.30 & 102.0[10.2 ] & 30 - 400 & 0.43(0.06 ) & 3.0[2.4 ] & 0.2 & 11,21,21,21,35,no & hete-2 + grb021004 & 2.332 & 79.8(30.0 ) & -1.01 , -2.30 & 2.55(0.60 ) & 2 -400 & 4.74(0.14 ) & 30.0(27.0 ) & 0.2 & 12,23,23,23,36,42 & hete-2 + grb021211 & 1.006 & 46.8(5.5 ) & -0.805,-2.37 & 2.17(0.15 ) & 30 - 400 & 1.4(0.5 ) & 3.0[2.4 ] & 0.2 & 13,24,24,24,37,no & hete-2 + grb030226 & 1.986 & 97.1(20.0 ) & -0.89 , -2.30 & 5.61(0.65 ) & 2 -400 & 1.04(0.12 ) & 3.0[2.4 ] & 0.2 & 14,23,23,23,38,no & hete-2 + grb030328 & 1.520 & 126.3(13.5 ) & -1.14 , -2.09 & 36.95(1.40 ) & 2 -400 & 0.8(0.1 ) & 3.0[2.4 ] & 0.2 & 15,23,23,23,39,no & hete-2 + grb030329 & 0.1685 & 67.9(2.2 ) & -1.26 , -2.28 & 110.0(10.0 ) & 30 - 400 & 0.48(0.03 ) & 1.0(0.11 ) & 0.2 & 16,25,25,25,40,43 & hete-2 + xrf030429 & 2.658 & 35.0(9.0 ) & -1.12 , -2.30 & 0.854(0.14 ) & 2 -400 & 1.77(1.0 ) & 3.0[2.4 ] & 0.2 & 17,23,23,23,17,no & hete-2 +
|
three different methods of measuring cosmology with gamma - ray bursts ( grbs ) have been proposed since a relation between the @xmath0-ray energy @xmath1 of a grb jet and the peak energy @xmath2 of the @xmath3 spectrum in the burst frame was reported by ghirlanda and coauthors . in method
i , to calculate the probability for a favored cosmology , only the contribution of the @xmath4 relation that is already best fitted for this cosmology is considered .
we apply this method to a sample of 17 grbs , and obtain the mass density @xmath5 ( @xmath6 ) for a flat @xmath7cdm universe . in method
ii , to calculate the probability for some certain cosmology , contributions of all the possible @xmath4 relations that are best fitted for their corresponding cosmologies are taken into account . with this method
, we find a constraint on the mass density @xmath8 ( @xmath6 ) for a flat universe . in method iii ,
to obtain the probability for some cosmology , contributions of all the possible @xmath4 relations associated with their unequal weights are considered . with this method , we obtain an inspiring constraint on the mass density @xmath9 ( @xmath6 ) for a flat universe , and a @xmath10 for the concordance model of @xmath11 .
compared with the previous two methods , method iii makes the observed 17 grbs place much more stringent confidence intervals at the same confidence levels .
furthermore , we perform a monte carlo simulation and use a larger sample to investigate the cosmographic capabilities of grbs with different methods .
we find that , a larger grb sample could be used to effectively measure cosmology , no matter whether the @xmath12 relation is calibrated by low-@xmath13 bursts or not .
ongoing observations on grbs in the _ swift _ era are expected to make the cosmological utility of grbs progress from its babyhood into childhood .
|
the discovery of the accelerated expansion of the universe and the phenomenon of dark energy has brought about a new surge of interest in alternatives to einstein s general relativity in the recent years @xcite .
one of the most simple and paradigmatic of these is the jordan - brans - dicke theory where the gravitational interaction is mediated by an extra scalar degree of freedom @xmath4 in addition to the usual tensor ones @xcite .
a more generic scalar - tensor gravity ( stg ) action is characterized by two arbitrary functions , the coupling function @xmath5 in the kinetic term of the scalar field and the potential @xmath6 @xcite . in cosmology
the stg models of dark energy allow evolving effective barotropic index @xmath7 and dynamical crossing of the `` phantom divide '' @xcite which remains a curious possibility in the combined observational data @xcite and could be a sign pointing beyond the standard @xmath8cdm scenario based on general relativity .
however interesting its performance in cosmology , a viable gravitational theory must also pass the tests on local scales , e.g. , give a good account of the motions in our solar system .
a natural framework for such a check is provided by the parameterized post - newtonian ( ppn ) formalism @xcite .
recent years have produced new fascinating and precise measurements of the ppn parameters @xcite .
the classic result for the parameters @xmath0 and @xmath1 in stg with arbitrary @xmath5 but without a potential was obtained by nordtvedt @xcite .
more recently olmo @xcite and perivolaropoulos @xcite calculated the parameter @xmath0 in the case of the jordan - brans - dicke theory ( constant @xmath2 ) with a nonvanishing potential @xmath6 .
the full equations up to the second post - newtonian approximation for generic @xmath5 and @xmath6 have been worked out in ref .
@xcite , while ref .
@xcite tackles light propagation up to the same ppn order for stg with a vanishing potential .
the ppn parameter @xmath0 has also been found for scalar - tensor type theories where the scalar field has a non - standard kinetic term or possesses scalar - matter coupling @xcite .
the stg ppn issues are further discussed in the context of screening mechanisms that mitigate the effects of the scalar field @xcite , and dynamical equivalence with @xmath9 gravity @xcite .
this article fills the gap in the literature and presents the ppn parameters @xmath0 and @xmath1 in the case of generic scalar - tensor theories with arbitrary functions @xmath5 and @xmath6 .
we display the action and field equations of these theories in sec .
[ sec : stg ] and derive their post - newtonian approximation in sec .
[ sec : ppn ] . from this approximation
we obtain a set of post - newtonian equations which we solve in sec .
[ sec : solution ] . finally in sec .
[ sec : experiments ] we compare our results with measurements in the solar system and obtain constraints on the so far arbitrary functions @xmath5 and @xmath6 .
we end with a conclusion in sec .
[ sec : discussion ] .
we study scalar - tensor theories of gravitation in which gravity is described by a dynamical scalar field @xmath4 in addition to the metric tensor @xmath10 .
we focus on theories which are defined by the action @xmath11\ ] ] in brans - dicke - like parameterization in the jordan conformal frame @xcite .
we have chosen units such that @xmath12 , @xmath13 and @xmath14 is related to the dimensionful newtonian gravitational constant @xmath15 via @xmath16 the matter part of this action is given by @xmath17 $ ] , where @xmath18 collectively denotes all matter fields .
the gravitational part of the action contains two free functions of the scalar field @xmath4 : the coupling function @xmath5 and the potential @xmath6 .
a distinct scalar - tensor theory is defined by a particular choice of these functions . in the following
we will restrict ourselves to the case @xmath19 in order to avoid ghosts in the einstein conformal frame @xcite , as well as @xmath20 and @xmath21 , but otherwise leave the functions @xmath2 and @xmath3 arbitrary .
the variation of the action with respect to the metric and the scalar field yields the gravitational field equations @xmath22 \nonumber\\ & & + \frac{\kappa^2}{\psi}{g_{\mu\nu}}\frac{2\omega + 1}{2\omega + 3}v + g_{\mu\nu}\frac{\kappa^2}{2\omega + 3}\frac{dv}{d\psi}\ , , \label{eqn : metric}\\ \largesquare\psi & = & \frac{1}{2\omega + 3}\left[\kappa^2 t - \frac{d\omega}{d\psi}\partial_{\rho}\psi\partial^{\rho}\psi + 2\kappa^2\left(\psi\frac{dv}{d\psi } - 2v\right)\right]\,,\label{eqn : scalar}\end{aligned}\ ] ] where @xmath23 denotes the covariant derivative , @xmath24 is the dalembert operator and @xmath25 is the energy - momentum tensor obtained from the matter action @xmath26 . in the following section we will expand these field equations in a post - newtonian approximation
we now expand the field equations and of scalar - tensor gravity as displayed in the preceding section up to the first post - newtonian order . for this purpose
we make use of the parameterized post - newtonian ( ppn ) formalism @xcite and assume that the gravitating source matter is constituted by a perfect fluid which obeys the post - newtonian hydrodynamics . in this section
we start from this assumption and assign appropriate orders of magnitude to all terms appearing in the field equations .
the resulting equations can then be solved subsequently for each order of magnitude in the next section .
the starting point of our calculation is the energy - momentum tensor of a perfect fluid with rest energy density @xmath27 , specific internal energy @xmath28 , pressure @xmath29 and four - velocity @xmath30 , which takes the form @xmath31 the four - velocity @xmath30 is normalized by the metric @xmath10 , so that @xmath32 .
a basic ingredient of the ppn formalism is the perturbative expansion of all dynamical quantities in orders @xmath33 of the velocity @xmath34 of the source matter in a given frame of reference .
for the metric @xmath10 this is an expansion around a flat minkowski background , @xmath35 where each term @xmath36 is of order @xmath37 . in order to describe the motion of test bodies in the lowest post - newtonian approximation
an expansion up to the fourth velocity order @xmath38 is sufficient .
a detailed analysis shows that not all components of the metric need to be expanded to the fourth velocity order , while others vanish due to newtonian energy conservation or time reversal symmetry .
the only relevant , non - vanishing components of the metric perturbations are given by @xmath39 in order to determine these components for a given matter source we must assign velocity orders also to the rest mass density , specific internal energy and pressure of the perfect fluid .
based on their orders of magnitude in the solar system one assigns velocity orders @xmath40 to @xmath27 and @xmath28 and @xmath38 to @xmath29 .
the energy - momentum tensor can then be expanded in the form [ eqn : energymomentum ] @xmath41 we further assume that the gravitational field is quasi - static , so that changes are only induced by the motion of the source matter .
time derivatives @xmath42 of the metric components and other fields are therefore weighted with an additional velocity order @xmath43 . in order to apply the ppn formalism to the scalar - tensor gravity theory detailed in section [ sec : stg ]
we further need to expand the scalar field @xmath4 in a post - newtonian approximation . for this purpose
we expand @xmath4 around its cosmological background value @xmath44 , @xmath45 where we assume @xmath44 to be of order @xmath46 and the perturbations @xmath47 are of order @xmath37 .
we also need to expand the functions @xmath5 and @xmath6 around @xmath44 . a taylor expansion to the required order takes the form @xmath48 with constant expansion coefficients which we assume to be of velocity order @xmath46 .
with these definitions the field equations and in the lowest velocity order @xmath46 read @xmath49 in order for these to be satisfied we must set @xmath50 , which is a consequence of our expansion of the metric around a flat background .
the scalar field equation up to fourth velocity order then reduces to @xmath51 we further expand the rhs of the gravitational tensor field equation up to the necessary velocity order and obtain [ eqn : riccippn ] @xmath52\left(\psi^{(2)}\right)^2 + \frac{\omega_1}{2\kappa^2}\psi^{(2)}_{,i}\psi^{(2)}_{,i } \nonumber\\ & & - \left(\frac{\omega_0 + 2}{\psi_0 } + \frac{\omega_1}{2\omega_0 + 3}\right)\rho\psi^{(2)}\bigg ] + \frac{1}{\psi_0}\left(\psi^{(2)}_{,00 } + \frac{1}{2}h_{00,i}^{(2)}\psi^{(2)}_{,i}\right ) + \mathcal{o}(6)\ , , \label{eqn : r00ppn}\\ r_{0j } & = & \frac{1}{\psi_0}\left(\psi^{(2)}_{,0j } - \kappa^2v_j\rho\right ) + \mathcal{o}(5)\ , , \label{eqn : r0jppn}\\ r_{ij } & = & \frac{\kappa^2}{\psi_0}\left[\frac{(\omega_0 + 1)\rho + 2\psi_0v_2\psi^{(2)}}{2\omega_0 + 3}\delta_{ij } + \frac{\psi^{(2)}_{,ij}}{\kappa^2}\right ] + \mathcal{o}(4)\ , .
\label{eqn : rijppn}\end{aligned}\ ] ] in order to solve these equations we finally need to fix a gauge for the metric tensor . a useful choice for the class of scalar - tensor theories we consider is given by @xcite @xmath53 in this gauge we can express the ricci tensor up to the necessary velocity order as [ eqn : riccigauge ] @xmath54 this completes our derivation of the post - newtonian field equations .
we can now use equation to solve for the scalar field perturbation @xmath55 and equations and to solve for the metric perturbations @xmath56 , up to the necessary velocity orders .
this will be done in the following section .
in the previous section we have derived the field equations of scalar - tensor gravity in a post - newtonian approximation .
it is now our aim to construct a simple solution to these equations , which corresponds to the static , spherically symmetric gravitational field of a single , point - like mass .
our derivation consists of three steps .
first , we solve the field equations for the metric perturbation @xmath57 in the newtonian approximation in sec .
[ subsec : newton ] . using this result
we can then solve for @xmath58 in sec .
[ subsec : gamma ] and for @xmath59 in sec .
[ subsec : beta ] . in the remainder of this article
we restrict ourselves to the gravitational field generated by a point - like mass @xmath60 , which is described by the energy - momentum tensor with @xmath61 this simple matter source induces a static and spherically symmetric metric , which can most easily be expressed using isotropic spherical coordinates . in the rest frame of the gravitating mass
we use the ansatz [ eqn : gmunu ] @xmath62\delta_{ij } + \mathcal{o}(4)\ , , \label{eqn : gij}\end{aligned}\ ] ] where @xmath63 denotes the radial coordinate and the newtonian potential @xmath64 is given by @xmath65 in the potential @xmath66 we collect terms of order @xmath38 which are not of the form @xmath67 , such as the gravitational self - energy .
the three unknown functions we need to determine are the effective gravitational constant @xmath68 and the ppn parameters @xmath69 and @xmath70 . the latter two can be defined either as the coefficients of the effective gravitational potential @xmath71 as shown in the metric or as the coefficients @xmath72 and @xmath73 of the newtonian potential terms @xmath74 and @xmath75 .
the first definition invokes the interpretation that the measured values of @xmath0 and @xmath1 can be related to the effective gravitational potential @xmath76 , while the second definition suggests to relate the measured values of @xmath77 and @xmath78 to the newtonian potential @xmath74 of a fixed mass @xmath60 .
we choose the first definition in this article since the mass of the sun , which dominates the solar system physics , is determined from its gravitational effects on the planetary motions . for our calculation of the newtonian limit
we start by solving equation up to the second velocity order , @xmath79 for the scalar field perturbation @xmath80 .
the constant @xmath81 can be interpreted as the mass of the scalar field @xcite .
equation is a screened poisson equation , which is solved by @xmath82 the solution for @xmath83 thus takes the form of a yukawa potential . in the next step
we consider equations and up to the second velocity order . from these
we derive @xmath84 we can write the solution in the form @xmath85 where the effective gravitational constant is given by @xmath86 in order to interpret this result for @xmath87 as an effective gravitational constant we need to choose an experiment in which the gravitational interaction takes place at a constant scale @xmath88 . we can then choose units in which @xmath89 .
this corresponds to a rescaling of the cosmological background value @xmath44 of the scalar field to @xmath90 however , we can not make this choice globally , and hence can not remove the factor @xmath68 from the metric by a choice of units in which @xmath91 , as it is conventionally done in the basic ppn formalism @xcite .
this is the reason for the ambiguity in the definition of the ppn parameters @xmath0 and @xmath1 we discussed above .
we now turn our focus to the spatial components @xmath58 of the metric perturbation . from equations and
we obtain @xmath92 from this equation we can immediately read off that the solution for @xmath58 is diagonal .
we find that @xmath93 comparison with equation then yields the ppn parameter @xmath69 and we obtain @xmath94 this result agrees with a previously obtained result for a purely quadratic scalar potential @xmath6 and constant coupling function @xmath5 @xcite . in the limit @xmath95 and fixed finite @xmath96 , which implies @xmath97 , the ppn parameter @xmath0 approaches the known value @xmath98 for scalar - tensor gravity with a massless scalar field @xcite .
analogously in the limit @xmath99 and fixed finite @xmath100 we have @xmath97 but find the limiting value @xmath101 . the same value @xmath101 is also approached in the opposite limiting case of a massive scalar field with @xmath102 .
we finally calculate the ppn parameter beta as given in equation . for this purpose
we need the fourth order perturbation @xmath103 of the scalar field @xmath4 .
this can be obtained from the fourth order part of equation , which reads @xmath104 as for the second order perturbation @xmath80 we obtain a screened poisson equation , but with a different source term .
the terms involving the pressure @xmath29 and specific internal energy @xmath28 drop out because of our choice of a point mass as the source of the gravitational field .
we further neglect the term @xmath105 , which corresponds to a gravitational self - energy .
finally , the term @xmath106 drops out since we consider only static solutions .
the solution to the remaining equation is then given by @xmath107 \nonumber\\ & & + \frac{3m_{\psi}}{(2\omega_0 + 3)^2}\left(\frac{v_3}{v_2 } - \frac{1}{\psi_0 } - \frac{\omega_1}{2\omega_0 + 3}\right ) u^2(r)r\left[e^{m_{\psi}r}\mathrm{ei}(-3m_{\psi}r ) - e^{-m_{\psi}r}\mathrm{ei}(-m_{\psi}r)\right]\ , .
\label{psi4sol}\end{aligned}\ ] ] here @xmath108 denotes the exponential integral , which is defined by @xmath109 using the result ( [ psi4sol ] ) we can now determine @xmath110 to the fourth velocity order . from equations and we obtain @xmath111- \frac{\omega_1}{\psi_0(2\omega_0 + 3)}\psi^{(2)}_{,i}\psi^{(2)}_{,i } \nonumber\\ & & + \frac{m_\psi^2}{\psi_0}\left[\left(\frac{2\omega_0 + 1}{2\psi_0 } - \frac{2\omega_1}{2\omega_0 + 3 } + \frac{3v_3}{2v_2}\right)\left(\psi^{(2)}\right)^2 -h^{(2)}_{00}\psi^{(2 ) } + \psi^{(4)}\right ] \,.\end{aligned}\ ] ] again terms involving @xmath29 and @xmath28 drop out for the point - like matter source we consider , while terms involving the velocity @xmath112 drop out out since we perform the calculation in the rest frame of the matter source .
we further neglect the terms @xmath113 and @xmath114 , which correspond to gravitational self - energies and thus contribute only to the potential @xmath66 displayed in equation .
the remaining equation can then be integrated and we obtain @xmath115 where @xmath70 is given by @xmath116\,.\nonumber\end{aligned}\ ] ] it follows from the asymptotic behaviour of the exponential integral in the case @xmath117 , @xmath118 that all terms involving @xmath119 or @xmath120 fall off proportional to @xmath121 , and are thus subleading to the terms involving only @xmath96 and @xmath100 which fall off proportional to @xmath122 .
we therefore conclude that at large distances @xmath102 from the source both @xmath119 and @xmath120 may be neglected .
again we consider the three limiting cases which we already discussed for @xmath0 . in the limit @xmath95 and fixed finite @xmath96 we obtain the known result @xmath123 for a massless scalar field @xcite . the second case @xmath99 and fixed finite @xmath100 yields the limit @xmath124 .
we also find the limiting value @xmath124 in the case @xmath102 of a massive scalar field .
this result completes our solution to the post - newtonian field equations derived in section [ sec : ppn ] .
we have calculated the metric up to the first post - newtonian order as displayed in equation . from our calculation
we obtained expressions for the effective gravitational constant and the ppn parameters @xmath0 and @xmath1 .
in the next section we will show how these results can be compared to measurements of the ppn parameters , and which restrictions arise on the scalar - tensor theories we consider .
in the preceding section we have derived expressions for the ppn parameters @xmath69 and @xmath70 as functions of the expansion coefficients @xmath96 , @xmath119 , @xmath100 and @xmath120 . in this section
we discuss restrictions on these coefficients which arise from measurements of @xmath0 and @xmath1 in the solar system .
we thereby neglect the terms involving @xmath119 and @xmath120 , since their contribution to the ppn parameters at large distances is subleading to that of @xmath96 and @xmath100 , as we argued in the previous section .
the ppn parameters have been measured by various high precision experiments in the solar system @xcite . in this article
we restrict ourselves to those measurements which provide the strictest bounds on the parameters @xmath0 and @xmath1 and have a characteristic interaction distance @xmath125 .
in particular , we will use the bounds obtained from the following experiments : * the deflection of pulsar signals by the sun has been measured using very long baseline interferometry ( vlbi ) @xcite . from this @xmath0 has been determined to satisfy @xmath126 .
since the radar signals were passing by the sun at an elongation angle of 3 , the gravitational interaction distance is @xmath127 . *
the most precise value for @xmath0 has been obtained from the time delay of radar signals sent between earth and the cassini spacecraft on its way to saturn @xcite .
the experiment yielded the value @xmath128 .
the radio signals were passing by the sun at a distance of @xmath129 solar radii or @xmath130 . *
the most well - known test of the parameter @xmath1 is the perihelion precession of mercury @xcite .
its precision is limited by measurements of other contributions to the perihelion precession , most importantly the solar quadrupole moment @xmath131 .
the current bound is @xmath132 . as the gravitational interaction distance
we take the semi - major axis of mercury , which is @xmath133 .
* the tightest bounds on @xmath1 are obtained from lunar laser ranging experiments searching for the nordtvedt effect , which would cause a different acceleration of the earth and the moon in the solar gravitational field @xcite . for fully conservative theories with no preferred frame effects
the current bound is @xmath134 .
since the effect is measured using the solar gravitational field , the interaction distance is @xmath135 .
there are more recent experiments which we will not use here since they can not be characterized by a single value @xmath125 for the interaction distance .
these include in particular combined vlbi measurements of @xmath0 at elongation angles between 5and 30 @xcite and measurements of @xmath0 and @xmath1 using ephemeris for a large number of celestial bodies in the solar system @xcite .
one may argue that also the measurements of @xmath0 listed above can not be characterized by a single interaction distance , since the distance between the sun and the radio signals varies along their path .
a rigorous treatment would therefore require a calculation of null geodesics in the solar system @xcite .
we will not enter this calculation here and instead assume that the dominant gravitational interaction occurs at the shortest distance to the sun .
the constraints on the values of the expansion coefficients @xmath96 and @xmath100 obtained from the experiments listed above can be visualized by plotting the experimentally excluded regions in parameter space , where we use coordinates @xmath136 , measured in inverse astronomical units @xmath137 , and @xmath96 .
vertical lines mark regions which are excluded by vlbi measurements , while horizontal lines mark regions which are excluded by cassini tracking .
similarly , regions with lines from bottom left to top right are excluded by the perihelion precession or mercury , while regions with lines from top left to bottom right are excluded by lunar laser ranging .
note that the bound from the cassini experiment is displayed at @xmath138 confidence level , while all other bounds are displayed at @xmath139 confidence level , as we will explain below .
.,width=377 ] figure [ fig : full ] shows the complete region of the parameter space which is excluded at @xmath138 by the cassini tracking experiment , from which the tightest bounds on @xmath0 have been obtained . as one can see from fig .
[ fig : vlbi ] even the @xmath139 bound from vlbi measurements is significantly smaller and fully contained in the region excluded by cassini .
this can also be seen for negative values of @xmath2 from fig .
[ fig : neg ] .
.,width=377 ] for @xmath140.,width=377 ] the @xmath139 bounds obtained from measurements of @xmath1 are shown in fig .
[ fig : beta ] .
one can see that they are located entirely in the region already excluded by the cassini measurement of @xmath0 .
this also holds for negative values of @xmath96 as shown in fig .
[ fig : betaneg ] .
this result is due to the fact that the bounds on @xmath1 are significantly weaker than the bounds on @xmath0 . in order to obtain stricter constraints on the parameter space from @xmath1 , experiments at higher precision
are required .
alternatively measurements of @xmath0 or @xmath1 at smaller interaction distance @xmath125 could exclude regions of the parameter space for larger values of @xmath141 .
in this article we discussed the post - newtonian approximation of scalar - tensor theories of gravity with arbitrary coupling function @xmath5 and arbitrary potential @xmath6 . in the post - newtonian limit
these functions can be characterized by four coefficients @xmath142 of the taylor expansion .
we then defined the ppn parameters @xmath0 and @xmath1 and calculated their values for a static point mass as functions of these four constants .
it turned out that the ppn parameters are not constant , but depend on the distance between the gravitating source and the test mass . in the appropriate
limits our expressions reduce to earlier results obtained for simpler cases @xcite .
we further found that for a massive scalar field the contribution of @xmath119 and @xmath120 to the values of the ppn parameters can be neglected at large distances from the source , @xmath143 .
we finally compared our results to measurements of @xmath0 and @xmath1 in the solar system and obtained bounds on the constants @xmath96 and @xmath100 .
it turned out that the strictest bounds are obtained from measurements of @xmath0 by cassini tracking , so that the current measurements of @xmath1 place no further restrictions on @xmath96 and @xmath100 . in order to obtain new bounds it would be necessary to measure the parameters @xmath0 and @xmath1 at higher precision or at shorter interaction distances between the gravitational source and the test mass .
the latter would allow probing and possibly excluding larger values of @xmath100 , corresponding to a higher mass of the scalar field .
the work presented in this article allows for various further studies .
while we have restricted ourselves to a calculation of the ppn parameters @xmath0 and @xmath1 for a static point mass , one may ask which further contributions to the post - newtonian metric arise from moving sources or sources with non - vanishing pressure , internal energy or gravitational self - energy .
a calculation of these contributions would provide insight into further ppn parameters , which are related to preferred - frame and preferred - location effects .
comparison of these parameters with their values measured in the solar system may place further bounds on viable theories of scalar - tensor gravity .
another line of investigation would be the case of strong fields provided by , e.g. , binary pulsars .
we further aim to extend our work to theories with more than one scalar degree of freedom , and thus chart the landscape of multi - scalar - tensor gravity .
the authors would like to thank andreas schrer for pointing out an error in a previous version of this manuscript .
this work was supported by the estonian research council institutional research funding grant no .
iut02 - 27 , by the estonian science foundation grant no .
8837 and by the european union through the european regional development fund ( centre of excellence tk114 ) .
mh gratefully acknowledges full financial support from the estonian research council through the postdoctoral research grant ermos115 .
y. fujii and k. maeda , _ the scalar - tensor theory of gravitation _ , ( cambridge university press , cambridge , 2003 ) ; v. faraoni , _ cosmology in scalar - tensor gravity _ , ( kluwer academic publishers , dordrecht , 2004 ) .
l. perivolaropoulos , jcap * 0510 * 001 ( 2005 ) [ astro - ph/0504582 ] ; r. gannouji , d. polarski , a. ranquet and a. a. starobinsky , jcap * 0609 * , 016 ( 2006 ) [ arxiv : astro - ph/0606287 ] ; s. nesseris and l. perivolaropoulos , jcap * 0701 * , 018 ( 2007 ) [ arxiv : astro - ph/0610092 ] ; s. capozziello , s. nesseris and l. perivolaropoulos , jcap * 0712 * , 009 ( 2007 ) [ arxiv:0705.3586 [ astro - ph ] ] ; b. boisseau , phys . rev .
d * 83 * , 043521 ( 2011 ) [ arxiv:1011.2915 [ astro-ph.co ] ] ; l. jrv , p. kuusk and m. saal , proc .
phys . math .
* 59 * ( 2010 ) 306 [ arxiv:0903.4357 [ gr - qc ] ] ; phys . rev .
d * 81 * , 104007 ( 2010 ) [ arxiv:1003.1686 [ gr - qc ] ] , phys .
b * 694 * , 1 ( 2010 ) [ arxiv:1006.1246 [ gr - qc ] ] .
a. shafieloo , v. sahni , and a. a. starobinsky , phys .
d*80 * , 101301 ( 2009 ) [ arxiv:0903.5141 [ astro - ph ] ] ; r. lazkoz , v. salzano and i. sendra , phys .
b * 694 * , 198 ( 2010 ) [ arxiv:1003.6084 [ astro-ph.co ] ] ; a. c. c. guimaraes and j. a. s. lima , class .
* 28 * , 125026 ( 2011 ) [ arxiv:1005.2986 [ astro-ph.co ] ] ; r. g. cai , q. su and h. b. zhang , jcap * 1004 * , 012 ( 2010 ) ; h. li and x. zhang , phys .
b * 703 * 119 ( 2011 ) [ arxiv:1106.5658 [ astro-ph.co ] ] ; g. -b .
zhao , r. g. crittenden , l. pogosian and x. zhang , phys .
lett . * 109 * , 171301 ( 2012 ) [ arxiv:1207.3804 [ astro-ph.co ] ] ; r. nair and s. jhingan , jcap * 1302 * , 049 ( 2013 ) [ arxiv:1212.6644 [ astro-ph.co ] ] ; n. said , c. baccigalupi , m. martinelli , a. melchiorri and a. silvestri , arxiv:1303.4353 [ astro-ph.co ] ; j. -q .
xia , h. li and x. zhang , arxiv:1308.0188 [ astro-ph.co ] .
a. fienga , j. laskar , p. kuchynka , h. manche , g. desvignes , m. gastineau , i. cognard and g. theureau , celest .
astron .
* 111 * , 363 ( 2011 ) [ arxiv:1108.5546 [ astro-ph.ep ] ] ; e. v. pitjeva and n. p. pitjev , arxiv:1306.3043 [ astro-ph.ep ]
. j. w. moffat and v. t. toth , int .
j. mod .
d * 21 * , 1250084 ( 2012 ) [ arxiv:1001.1564 [ gr - qc ] ] ; k. saaidi , a. mohammadi and h. sheikhahmadi , phys .
d * 83 * , 104019 ( 2011 ) [ arxiv:1201.0271 [ gr - qc ] ] ; o. minazzoli , arxiv:1208.2372 [ gr - qc ] .
s. capozziello and a. troisi , phys .
d * 72 * , 044022 ( 2005 ) [ astro - ph/0507545 ] ; s. capozziello , a. stabile and a. troisi , mod .
a * 21 * , 2291 ( 2006 ) [ gr - qc/0603071 ] ; phys . rev .
d * 76 * , 104019 ( 2007 ) [ arxiv:0708.0723 [ gr - qc ] ] ; phys .
b * 686 * , 79 ( 2010 ) [ arxiv:1002.1364 [ gr - qc ] ] ; t. clifton , phys .
d * 77 * , 024041 ( 2008 ) [ arxiv:0801.0983 [ gr - qc ] ] ; m. capone and m. l. ruggiero , class .
* 27 * , 125006 ( 2010 ) [ arxiv:0910.0434 [ gr - qc ] ] ; s. capozziello and a. stabile , in _ classical and quantum gravity : theory and applications _
, edited by v. r. frignanni ( chap .
2 ) [ arxiv:1009.3441 [ gr - qc ] ] ; t. s. koivisto , phys . rev .
d * 84 * , 121502 ( 2011 ) [ arxiv:1109.4585 [ gr - qc ] ] ; t. harko , t. s. koivisto , f. s. n. lobo and g. j. olmo , phys .
d * 85 * , 084016 ( 2012 ) [ arxiv:1110.1049 [ gr - qc ] ] .
|
we calculate the ppn parameters @xmath0 and @xmath1 for scalar - tensor gravity with a generic coupling function @xmath2 and scalar potential @xmath3 in the jordan conformal frame in the case of a static spherically symmetric source .
since the potential generally introduces a radial dependence to the effective gravitational constant as well as to @xmath0 and @xmath1 , we discuss the issue of defining these ppn parameters and compare our expressions with previous calculations in simpler cases .
we confront our results with current observational constraints on the values of @xmath0 and @xmath1 and thus draw restrictions on the form of the functions @xmath2 and @xmath3 around their asymptotic background values .
|
since high tunneling magnetoresistance ( tmr ) was first observed at room temperature,@xcite magnetic tunnel junctions ( mtjs ) have been a focus of interest .
extensive research has been carried out to understand and improve the properties of mtjs . a huge increase in tmr with lower junction resistance
was achieved when alo@xmath0 tunnel barriers were replaced by mgo , and this was followed by realizations of memory devices based on mtjs with mgo barriers.@xcite tunneling current in the mtj is spin polarized , which adds another dimension to the tunneling effect , and scientific attention has thus been drawn to the spin - dependent tunneling phenomenon . when a nonmagnetic ( nm ) layer is inserted between a ferromagnetic ( fm ) layer and the insulating ( i ) tunnel barrier of the mtj ,
the spin polarization of the tunneling current changes and the tmr is directly affected .
an early theoretical work predicted the oscillatory tmr as a function of the nm thickness due to the quantum well states inside the nm layer.@xcite in sputtered samples , it has been shown that an nm layer between the tunnel barrier and fm layer could be detrimental to tmr , and the tmr decreases as a function of nm thickness.@xcite these experimental results have been explained theoretically with a free electron model , and the decay of tmr was attributed to a loss of coherence in the electron propagation.@xcite different experimental results have been obtained for a crystalline nm layer inserted between the tunnel barrier and the fm layer .
et al._@xcite experimentally investigated the dependence of tmr on cu thickness in nife / alo@xmath0/cu / co junctions with samples grown by molecular beam epitaxy .
they found that the tmr decayed but oscillated as a function of cu thickness .
the oscillation period was determined by the nesting feature of the cu fermi surface .
this oscillatory tmr has been investigated theoretically based on a single - band tight - binding model , a free - electron model , and full - band calculations.@xcite many features have been explained with calculations using simple models , but direct comparisons with the experimental data are difficult because realistic electronic structures were not considered .
the full - band calculations are very useful for the description of real systems .
however they are time consuming , and sometimes it is not easy to understand the underlying physics .
furthermore , full - band calculations are usually carried out for an ideal situation , and significant discrepancies often occur between theory and experiments .
we introduce an analytical expression that describes the dependence of tmr on nm thickness for fm / i / nm / fm junctions based on full - band structures .
our approach uses the generalization of a previously described single - band case@xcite to a multiple - band case that considers the real materials .
the tmr is expressed with transmission and reflection amplitudes of single interfaces at the fermi level , and extremal wave vectors .
the full - band structures of the materials are taken into account in our proposed model , and the calculation of several transmission and reflection amplitudes with real band structures can make a direct comparison with experimental results possible .
based on our model , selection rules for the oscillation period are discussed , and we suggest that very few oscillation periods will be observed in experiments even when there are many extremal spanning vectors of the nm fermi surface .
this situation is very different than the interlayer exchange coupling in magnetic multilayers .
our model explicitly shows that tmr dependence on nm thickness is affected by the thickness of the tunnel barrier , and predicts that the tmr will go to zero as the nm thickness increases . to check the validity of our model
, we carried out numerical calculations using an envelope - function theory for several cases including nm material with a fermi surface similar to the cu(001 ) case .
although our model is calculated using the parameters at the fermi level and the extremal wave vector , it is in good agreement with the exact numerical results .
figure 1(a ) shows a schematic diagram of an mtj with an nm layer inserted between the tunnel barrier and the right magnetic layer [ fm(r ) ] .
the growth direction is taken as the @xmath1 axis , and @xmath2 is the thickness of the nm layer .
we used a frozen potential approximation , and the eigenstate of the mtj is expressed with linear combinations of the bulk states for each layer .
we assumed that the wave vector component parallel to the interface ( @xmath3 ) is conserved throughout the mtj .
a two - channel model was adopted and spin - flip scattering was ignored .
the normalized bulk solution of the material for the left magnetic layer [ fm(l ) ] is denoted as @xmath4 for a given energy @xmath5 and @xmath3 , where @xmath6 is the @xmath1 component of the wave vector , @xmath7 is the band index , @xmath8 is the spin index , and the @xmath9 @xmath10 sign is for the state traveling to the right ( left ) . similarly , the bulk solutions of the fm(r ) and nm materials are expressed as @xmath11 and @xmath12 , respectively .
multiple bands are taken into account , and @xmath13 is the number of bulk states in the left ( right ) fm layer for a given @xmath5 , @xmath14 , and spin @xmath8 .
the number of bulk states in the nm layer for a given @xmath5 and @xmath14 is denoted as @xmath15 .
the eigenstate of the mtj is written as @xmath16 where @xmath17 is the @xmath1-component of the group velocity [ @xmath18 for the corresponding bulk eigenstate , and @xmath19 , @xmath20 , and @xmath21 are coefficients to be determined from the boundary conditions .
note that the bases in eq .
( [ eq : mtj ] ) are adjusted so that the current is normalized .
the eigenstate inside the tunnel barrier is not shown here .
we define vectors @xmath22 and @xmath23 as @xmath24 and @xmath25 .
then , @xmath26 and @xmath27 are related to @xmath28 and @xmath29 by the @xmath30-matrix,@xcite @xmath31 matrix element @xmath32 ( @xmath33 ) is a kind of transmission ( reflection ) amplitude for an incoming wave from the left @xmath34 to be transmitted ( reflected ) to @xmath35 ( @xmath36 ) .
the transmission ( reflection ) amplitude for the opposite direction is given by @xmath37 ( @xmath38 ) .
we calculated the conductance @xmath39 for low bias and zero temperature from the landauer - bttiker formalism as follows @xmath40 , \label{eq : lan}\ ] ] where @xmath41 is the fermi energy .
as shown in figs .
1(b ) and ( c ) , we considered the fm(l)/i / nm and nm / fm(r ) interfaces separately , and expressed @xmath42 of the mtj with the transmission and reflection amplitudes of each separated interface .
the eigenstate of the fm(l)/i / nm system shown in fig .
1(b ) is expressed as @xmath43 where @xmath44 and @xmath45 are coefficients .
we define @xmath46 and @xmath47 , and the relation between the coefficients is expressed as @xmath48 similarly , for the nm / fm interface shown in fig .
1(c ) , the eigenstate is given by @xmath49 the vectors @xmath50 and @xmath51 are related as follows @xmath52 @xmath53 and @xmath54 of the mtj in eq .
( [ eq : rt ] ) can be expressed with @xmath55 , @xmath56 , @xmath57 , and @xmath58 in eq .
( [ eq : rtl ] ) and @xmath59 , @xmath60 , @xmath61 , and @xmath62 in eq .
( [ eq : rtr ] ) by considering the multiple reflection inside the nm .
we introduce the mean free path @xmath63 due to scattering inside the nm layer . for simplicity
, we assumed that @xmath63 is constant , although the dependence of @xmath63 on other parameters can be included in our calculation .
then , the phase - coherent part of the reflection amplitude @xmath64 is given by @xmath65 where the matrix elements @xmath66 , @xmath67 , and @xmath68 are @xmath69 , @xmath70 , and @xmath71 .
the phase - coherent part of the transmission amplitude @xmath72 can be obtained in a similar way .
because of the scattering inside the nm layer , we have @xmath73 + { \mathrm { tr } } [ { { \bf r}^c}^{\dag } { \bf r}^c ] < 1 $ ] and we need to include the diffusive part of the transport .
we assume that the transmission back to the fm(l ) layer through the tunnel barrier is much smaller than that through the nm / fm(r ) interface .
then , the @xmath74 - { \mathrm { tr } } [ { { \bf r}^c}^{\dag } { \bf r}^c ] $ ] portion contributes to the sequential transmission.@xcite adding the coherent and sequential transmissions , we have @xmath75 = 1- { \mathrm { tr } } [ { { \bf r}^c}^{\dag } { \bf r}^c ] $ ] . finally , using the properties of the @xmath30 matrix and taking the first - order term in @xmath76
, we obtain @xmath77 \cong { \rm tr } [ { { \bf t}^l}^{\dag } { \bf t}^l ] + 2 { \rm re \ , tr}[{{\bf t}_\sigma^l}^\dag { { \bf r}'}_\sigma^l { \mathbf \rho}_\sigma^r { \mathbf \tau}_\sigma^l ] e^{- \frac{2d}{\lambda}}.\ ] ] even when @xmath63 is very large , this is a reasonable approximation because the magnitude of the matrix element @xmath78 is less than 1 , and the higher - order terms are more rapidly oscillating as functions of @xmath2 and consequently contribute less to the conduction .
the conductance is given by @xmath79 , % \left ( - \frac{\partial f_l ( \varepsilon)}{\partial \varepsilon } % \right),\ ] ] where @xmath80 is the conductance of the fm(l)/i / nm junctions and the energy is set to the fermi level ( @xmath81 ) .
the conductance depends on the magnetic configurations , and we denote the conductance for parallel ( anti - parallel ) magnetization of two magnetic layers as @xmath82 .
the tmr is given by @xmath83 , where @xmath84 is @xmath85 .
here , we will show the calculation of @xmath84 , and @xmath86 can be obtained in the same way .
we define @xmath87 and @xmath88 as @xmath89 and @xmath90 , where @xmath91 ( @xmath92 ) is the majority ( minority ) spin .
@xmath84 is expressed as @xmath93 where @xmath94 and @xmath95 are @xmath96 and @xmath97 .
the summation over @xmath98 can be performed in a manner similar to the calculation of the interlayer exchange coupling in magnetic multilayers.@xcite @xmath99 and @xmath100 are rapidly changing as functions of @xmath101 .
we assume the exponential dependence of @xmath102 such that @xmath103 .
suppose that @xmath104 is an extremal point , which means @xmath105 = 0 $ ] at @xmath106 .
since the main contribution to the integral comes from the vicinity of the extremal point , we expand @xmath107 around the extremal point as follows @xmath108 where new parameters @xmath109 , @xmath110 , @xmath111 , @xmath112 , @xmath113 and @xmath114 are evaluated at @xmath115 and @xmath116 .
then , the summation over @xmath117 is carried out analytically and @xmath84 is given by @xmath118 , \label{e : delg}\ ] ] where @xmath119 is the number of the extremal points of the same kind .
the phase of the square root is taken from @xmath120 to @xmath121 .
the parameters in eq .
( [ e : delg ] ) are evaluated at the fermi level and the extremal point .
suppose we have @xmath122 at @xmath123 and @xmath124 at @xmath125 .
in general , @xmath126 is different from @xmath127 and the corresponding extremal point @xmath128 is a complex number .
this makes other parameters such as @xmath129 complex numbers , and the situation is rather complicated .
however , when @xmath126 and @xmath127 are far apart , the contribution is negligible because of small @xmath130 .
the most important case is when @xmath131 and @xmath127 coincide .
this is expected to happen often at @xmath132 due to symmetry . in this case ,
@xmath133 and @xmath134 are real and @xmath84 becomes @xmath135^{\frac{1}{4 } } { \left [ % \left ( \frac{b}{\kappa_{x}^b } \right)^2 \right]^{\frac{1}{4 } } } } % f \left [ % \frac{2(d+d _ { \varepsilon } ) \pi k_b t}{\hbar v ^f } \right ] , \label{e : delgsingle}\ ] ] where we used @xmath136 with @xmath137 , which is expected for a typical thickness of the tunnel barrier . in this case
, @xmath129 is exactly the extremal spanning vector of the nm fermi surface . the extremal spanning vector @xmath129 of the nm fermi surface gives rise to the period of the tmr oscillation . in principle , multiple periods are possible depending on the shape of the nm fermi surface .
however , the extremal spanning vectors of the nm fermi surface are the necessary condition for tmr oscillation , and the period would be observed only when the corresponding @xmath130 is large enough . except for the symmetry point in @xmath98 space ,
the extremal points of the nm fermi surface hardly coincide with the maxima point of @xmath130 of the tunnel barrier .
thus , only few periods would be observable among the possible periods from the extremal spanning vectors . compared with the oscillation of the interlayer exchange coupling for a corresponding spacer , fewer periods would be observed in tmr oscillation . moreover , as @xmath138 increases , @xmath139 decreases rapidly as a function of @xmath98 away from the maxima .
thus , the multiple periods would be more difficult to observe for a thicker tunnel barrier .
since @xmath140 is material dependent , the period of tmr oscillation could change for a different tunneling barrier . in eq .
( [ e : delgsingle ] ) , the extra phase factor arises from the square root in the denominator .
this extra phase factor is @xmath2 dependent , and the measured oscillation period will be slightly different from @xmath141 .
for instance , the period of tmr oscillation for nife / alo@xmath0/cu(001)/co is expected to be slightly smaller than @xmath141 , which is the period of the interlayer exchange coupling for fcc co / cu(001)/co multilayers .
we considered the observable periods with the assumption that the tunneling probability is maximum at @xmath132 .
then , only the oscillation for the extremal spanning vector at @xmath132 would be observed . in nife
/ alo@xmath0/cu(001)/co junctions , there are two possible periods from the fermi surface of cu along the ( 001 ) direction .
the long period is from the extremal spanning vector at @xmath132 , and the extremal vector for the short period is far away from @xmath132 .
thus , the short period is intrinsically invisible , and only the long period could be observed .
there are four possible oscillation periods in nife / alo@xmath0/cu(110)/co junctions from the cu fermi surface along the ( 110 ) direction.@xcite among them , the only observable period is the extremal spanning vector corresponding to @xmath132 .
this period is very short and would be easily wiped out by the interface roughness .
thus , it would be difficult to observe a tmr oscillation in this system .
there is only one possible period for cu(111)@xcite and this period has been observed experimentally in the interlayer exchange coupling of co / cu(111)/co multilayers.@xcite however , this period would not be observed in nife / alo@xmath0/cu(111)/co junctions because the corresponding extremal spanning vector is far from @xmath132 .
therefore , except for the long period of nife / alo@xmath0/cu(001)/co junctions , it would be difficult to observe a tmr oscillation as a function of cu thickness .
the situation for the au and ag spacers will be same as in the case of cu because the shapes of the fermi surfaces are similar .
in the fe / cr / fe multilayers , the long periods of interlayer coupling oscillation have been observed as a function of the cr thickness for the ( 100 ) , ( 110 ) , and ( 211 ) orientations.@xcite however , these long periods would not be observed experimentally in fe / alo@xmath0/cr / fe or fe / mgo / cr / fe tunnel junctions because the long periods are from the @xmath142 point of the cr fermi surface,@xcite which is away from @xmath132 for any orientation . in experiments ,
this long period has not been observed for fe / alo@xmath0/cr(001)/fe ( ref . ) and fe / mgo / cr(001)/fe ( ref . )
tunnel junctions .
usually , the long period is clearly observed because it is not eliminated by the interface roughness .
although we discussed the case that the tunneling is dominated by perpendicularly incident electrons , the analysis is similar when the tunneling probability is maximum or high away from @xmath132 . if the dependence of the tunneling probability on @xmath98 changes with a different tunnel barrier , the tmr dependence on the nm spacer will be altered accordingly .
still , the crucial criterion is whether the tunneling probability is significant or not at the point of the extremal spanning vectors of the nm fermi surface in the @xmath98 space . except special points , the chances are that the point of the maximum tunneling probability does not coincide with the position of the extremal spanning vectors of the nm fermi surface in the @xmath98 space .
thus many oscillations inferred from the nm fermi surface would not be observed .
when the tunnel barrier is extremely thin , the situation can be much different because the tunneling probability dependence on @xmath98 may change significantly .
more oscillation periods can be observed with thinner tunnel barriers . even in this case , the oscillation periods associated with relatively high tunneling probability will be observed .
without the scattering effect , @xmath84 and tmr decays as @xmath143 for a thick nm layer .
however , for thin nm layers , the decay rate is slower than @xmath143 and is affected by the tunnel barrier thickness @xmath138 , and also by @xmath144 and @xmath145 ( @xmath98 dependence of the reflection - amplitude phase factors ) .
the amplitude decays much more slowly than @xmath143 for coherent transport when @xmath2 is of the same order of magnitude as @xmath138 . in experiments ,
the tmr oscillation decays much faster than @xmath143 , which seems to be due to scattering . as the nm thickness increases
, our model predicts that @xmath84 and tmr go to zero even when the mean free path ( @xmath63 ) is very long .
we will address this point in sec .
to test the validity of our model , we carried out numerical calculations based on an envelope - function theory for several cases .
we used the same material for fm(l ) and fm(r ) , and ignored scattering .
the continuity of the wave function and the conservation of current at the interface were taken as the boundary conditions .
first , we considered the case that the dispersion relation of the nm material is the same as that of the fm material for majority spin .
the dispersion relations of the fm material and the insulator are given by @xmath146 and @xmath147 , respectively , where @xmath148 is the spin - dependent effective mass of the fm material , @xmath149 is the bare electron mass , and @xmath150 is the height of the tunnel barrier .
we set @xmath151 [ @xmath152 for the majority ( minority ) spin in the fm layer using the spin - splitting energy @xmath153 .
the effective mass of the fm material is @xmath154 for the majority spin and @xmath155 for the minority spin .
schematics for the dispersion relations of the fm , i , and nm materials are shown in fig . 2 .
under these conditions , all the traveling states in the nm layer have corresponding traveling states in the fm layer , and total reflection does not occur at the nm / fm interface for any @xmath156 .
the parameters used in the calculation are @xmath157 ev , @xmath158 ev , and @xmath159 ev . the tmr is plotted as a function of nm thickness @xmath2 for the tunnel - barrier thicknesses of @xmath160 nm and @xmath161 nm in figs .
3(a ) and 3(b ) , respectively .
the solid line represents the exact calculation , which was obtained using eq .
( [ eq : lan ] ) .
the dotted line is the result of our analytical simple model described by eq .
( [ e : delgsingle ] ) .
note that the extremal point for the nm fermi surface coincides with the maximum transmission point of the tunnel barrier at @xmath162 .
when the nm layer is thin , there is some discrepancy between the exact result and our analytical model , but the agreement improves as the nm thickness increases .
the overall trend of the nm thickness dependence is well depicted by our analytical model .
the tmr oscillates and goes to zero as the nm thickness increases .
it is also shown that the tmr dependence on @xmath2 is affected by the thickness of the tunnel barrier .
the peak points of tmr do not coincide for different tunnel barrier thicknesses .
the decay of tmr as a function of @xmath2 is faster for a thinner tunneling barrier .
it is essential to consider the @xmath163 dependence of the transmission coefficient in the calculation .
the effect of the tunnel barrier thickness on the nm thickness dependence of tmr is well described by our proposed analytical formula eq .
( [ e : delgsingle ] ) .
third , we studied the case that the point for the maximum tunneling probability does not coincide with the position of the extremal spanning vector of the nm fermi surface in @xmath173 space .
we assumed that the dispersion relation of the tunnel barrier is given by @xmath174^{1/2 } + \frac{\hbar^2 k_z^2}{2 m_0 } + v_b - v_{\| } , \label{e : b}\ ] ] where @xmath175 is the bottom energy of the tunnel barrier , @xmath176 is an energy parameter , and @xmath177 is the magnitude of the fermi wave vector for the nm material . with this tunnel barrier ,
the tunneling probability is highest at @xmath178 . for the calculation
, we used @xmath167 nm , @xmath179 ev , @xmath180 ev , and @xmath181 .
the other materials are assumed to be the same as in sec .
the dispersion relation of the fm material is given by @xmath182 with @xmath151 , @xmath183 , @xmath154 , and @xmath184 .
the dispersion relation of the nm material is the same as that of the fm material for majority spin .
note that the extremal spanning vector of the nm fermi surface is located at @xmath185 .
the parameters used in the calculation are @xmath157 ev and @xmath159 ev . the tmr dependence on the thickness of the nm layer is displayed in fig .
the solid line is the exact result and the dotted is the result of our analytical formula .
as expected , the tmr decays faster and it is almost negligible when the nm layer is thicker than about 2 nm . the inset is the transmission coefficient as a function of @xmath186 along the [ 110 ] direction for the fm / i / nm system .
it has the maximum value around @xmath187 , at which we have @xmath188 . on the other hand
, the spanning vector ( @xmath189 ) of the nm fermi surface has the longest at @xmath162 .
the maximum point of the tunneling probability does not coincide with the extremum of the nm fermi surface , and the extremal point is determined from @xmath190 = 0 $ ] .
the extremal point is @xmath2-dependent and a complex number .
it is not easy to calculate the @xmath2-dependent extremal point exactly and we used the following approximation . at the fermi level ,
@xmath6 in eq .
( [ e : b ] ) is an imaginary number ( @xmath191 ) .
we used a parabolic function of @xmath192 which was expanded in a taylor series around @xmath178 .
the calculation became much simpler , and the extremal point and the corresponding vector @xmath193 in eq .
( [ e : expand ] ) were obtained immediately as functions of @xmath2 . in the range of thin nm layers with significant tmr
, @xmath193 is shorter than @xmath194 , the extremal spanning vector of the nm fermi surface .
thus , the oscillation period in this region is longer than what is expected from the nm fermi surface .
whenever the extremal spanning vectors of the nm fermi surface are significantly away from the point of the maximum transmission coefficient in the @xmath173 space , we expect rapid decay of tmr as a function of the nm thickness .
finally , we investigated the free - electron model case , where all the effective masses are simply the bare electron mass @xmath149 .
the nm band is assumed to be the same as the majority - spin band of fm , which is commonly adopted in the theoretical calculations .
the parameters used in the calculation are @xmath157 ev , @xmath195 ev , @xmath196 ev , and @xmath160 nm . a plot of the tmr dependence on nm thickness is displayed by a dotted line in fig .
6 . in this case
, the tmr reaches a finite value when the nm thickness becomes infinite . at first sight , this seems to contradict our analytical formula eq .
( [ e : delgsingle ] ) given in sec
the reason for finite tmr at infinite @xmath2 can be explained as follows .
we obtained @xmath197 from the dispersion relation at the fermi level .
when the magnetizations of the fm layers are antiparallel , electrons with the majority spin in the fm(l ) layer do not penetrate to the fm(r ) layer for @xmath198 because @xmath6 becomes imaginary in the fm(r ) layer .
the electrons with the majority spin for @xmath199 in the fm(l ) layer are totally reflected at the nm / fm(r ) interface due to the potential step and do not contribute to conduction .
thus , @xmath200 is underestimated and gives rise to a finite @xmath84 and consequently a finite tmr , even when the nm layer is infinite , as long as the transport is coherent .
this is more pronounced for a larger @xmath153 , lower barrier height , and thinner tunneling barrier because a larger portion of electrons with the fermi energy will be completely reflected at the nm / fm(r ) interface . when deriving eq .
( [ e : delg ] ) in sec .
ii , we considered multiple reflection inside the nm layer to calculate the transmission , and we assumed that once the electrons tunnel through the tunnel barrier , most of them flow to the fm(r ) layer after multiple reflections .
however , in the ap magnetizations of fm layers of the free - electron model , the electrons with majority spin for @xmath201 in the fm(l ) layer can not penetrate into the fm(r ) layer after tunneling because of the potential step , resulting in a finite tmr for the infinite nm layer . to clarify this point ,
we calculated the tmr for the finite fm(r ) layer ; namely , the fm(l)/i / nm / fm(r)/nm junctions .
the fm(r ) layer is 1 nm thick , and the result is shown by the solid line in fig .
the fm(r ) layer behaves like a potential barrier for electrons with the majority spin and @xmath202 in the fm(l ) layer .
however , the tmr goes to zero as the nm thickness increases . when the fm(r ) layer is not too thick , the tunneling probability through this potential barrier is larger than that through the tunnel barrier ( i ) , and most of electrons with majority spin and @xmath202 in the fm(l ) layer eventually flow to the nm(r ) layer once they tunnel through the insulating barrier . even for the infinite fm(r ) layer ,
the tmr for the infinite nm layer becomes negligible as @xmath153 increases , the thickness of the tunnel barrier increases , and the barrier height increases .
this is because only a small portion of electrons with the fermi energy will be completely reflected at the nm / fm(r ) interface .
also , if the minority spin band of the fm layer is set to be the same as the nm band , the tmr goes to zero as the nm thickness increases because there is no total reflection at the nm / fm(r ) interface .
the finite tmr for the infinite nm layer is possible when a significant portion of electrons with the fermi energy have less transmission probability into the fm(r ) layer than that through the tunnel barrier .
complete reflection of electrons at the interface between two metals may occur for some given @xmath163 at the fermi level due to the mismatch of the electronic states .
however , even slight scattering would lead to the penetration of electrons into the fm(r ) layer . in this sense
, we do not expect that a large portion of electrons with fermi energy are reflected completely , in reality , at the nm / fm interface .
thus , it is unrealistic to expect a finite tmr as the nm thickness increases in experiments .
we calculated the tmr of fm / i / nm / fm tunnel junctions .
the tmr was calculated as a function of nm thickness using the landaur - bttiker formula .
multiple band structures were included and an analytical form describing the tmr was obtained .
conductance was calculated from the summation of the transmission over @xmath203 .
the transmission was obtained by considering multiple reflections between the i / nm and nm / fm interfaces .
the summation over @xmath204 was carried out analytically .
the contribution was mainly from the extremal point in @xmath205-space that was determined from the combination of the nm fermi surface and the @xmath204-dependence of the transmission coefficient of the fm / i / nm junction .
the tmr was expressed with the transmission coefficient of the fm / i / nm junction , reflection amplitudes at the nm / i and nm / fm interfaces at the fermi level , and the extremal wave vector .
many oscillation periods can be inferred from the shape of the nm fermi surface , but they can be observed only when the corresponding spin asymmetry of the transmission coefficient is significant .
we suggest that only few oscillation periods are likely to be observed in real experiments .
when the nm spacer is thin , our proposed model indicates that the decay of the tmr was slower than the inverse of the space thickness for coherent transport .
numerical calculations were performed to investigate the accuracy of the proposed formula .
an envelope - function theory was adopted , and our model was compared to the exact result .
we showed that the results of the proposed formula are in good agreement with the exact calculations .
the tmr dependence on the thickness of the nm spacer was affected by the tunnel barrier thickness , which was well described by our formula .
the numerical calculation was extended to the case with multiple extremal spanning vectors in the fermi surface of the nm spacer .
our proposed formula is in good agreement with the exact result , and only the oscillation period with significant spin asymmetry of the transmission coefficient was observed as predicted using our formula . when the tunneling probability associated with each extremal spanning vectors of the nm fermi surface is low , the tmr decays very fast as the nm thickness increases .
a free - electron case was also considered .
the nm band was assumed to be the same as the majority spin band of the fm layer , and spin - splitting in the fm layer was assumed to be rather large .
the tmr approached a finite value as the nm thickness increased .
this was because a large portion of electrons with the majority spin in the left fm layer were reflected completely at the right nm / fm interface .
when the semi - infinite fm layer on the right side was replaced by a finite layer , the tmr decayed to zero as the nm thickness increased . as long as the transmission into the fm layer was higher than that through the tunnel barrier , the tmr became zero as the thickness of the nm spacer increased .
we suggest that finite tmr for the infinite nm spacer is unrealistic in real experiments .
this work was supported by the basic science research program through the national research foundation of korea ( nrf ) , funded by the ministry of education , science and technology ( grants no .
2011 - 0005251 and no .
2012 - 0002984 ) .
|
oscillatory tunneling magnetoresistance ( tmr ) as a function of spacer thickness is investigated theoretically for a magnetic tunnel junction with a nonmagnetic layer inserted between the tunnel barrier and the ferromagnetic layer .
tmr is characterized in an analytical form , that is expressed with the transmission and reflection amplitudes of single interfaces at the fermi level , and by the extremal wave vectors .
electronic structures with multiple bands are taken into account in the derivation characterizing the tmr , and the proposed analytical expression can be directly applied to real junctions .
based on our model , the features of tmr dependence on spacer thickness are discussed , including selection rules for the oscillation period .
numerical calculations are performed using an envelope - function theory for several cases , and we show that our model is in good agreement with the exact result .
|
the na48/2 experiment at the cern sps is searching for the charge asymmetry in the decays @xmath3 and @xmath4 .
the experiment uses two simultaneous oppositely charged beams with a central momentum of 60 gev / c and a momentum band @xmath5 , propagating along the same beam line .
the decay volume is a 114 m long vacuum tank with a diameter of 1.92 m for the first 66 m , and 2.4 m for the rest .
the detected particles are dominated by the @xmath6 decay products .
charged particles are measured by the magnetic spectrometer , consisting of four drift chambers @xcite and a dipole magnet , located between the second and third chambers .
charged particles are magnetically deflected in the horizontal plane by an angle corresponding to momentum kick of 120 mev / c .
the momentum resolution of the spectrometer is @xmath7 ( @xmath8 in gev / c ) .
the magnetic spectrometer is followed by a scintillator hodoscope consisting of two planes segmented into horizontal and vertical strips and arranged in four quadrants .
a liquid krypton calorimeter @xcite is used to reconstruct @xmath9 decays .
it is segmented transversally into @xmath10 cm cells by a system of cu - be electrodes .
the calorimeter is 27 @xmath11 thick and absorbs the electromagnetic showers almost completely .
its energy resolution for gamma - quanta is @xmath12 e in gev .
the space resolution for single electromagnetic showers can be parametrized as @xmath13 cm for each transverse coordinate @xmath14 .
a neutral hodoscope consisting of a plane of scintillating fibres is installed in the liquid krypton calorimeter at a depth @xmath15 .
it is divided into four quadrants , each consisting of eight bundles of vertical fibres optically connected to photomultiplier tubes .
data have been collected in 2003 and 2004 , providing samples of about @xmath16 @xmath4 decays . here
we report the analysis results of a partial sample @xmath17 decays recorded in 2003 .
@xmath1 decays are selected by a two - level trigger .
the first level requires a signal in at least one quadrant of the scintillator hodoscope in coincidence with the presence of energy depositions in calorimeter consistent with at least two photons . at the second level ,
a fast on - line processor reconstructs the momentum of charged tracks and calculates the missing mass , assuming the decay @xmath18 of a 60 gev / c kaon flying along the beam axis .
the requirement that the mass of missed system @xmath19 is not consistent with the @xmath20 rejects the main @xmath21 background .
events with at least one charged particle having a momentum above 5 gev / c , and at least four energy deposit clusters in liquid krypton calorimeter , each above the energy threshold of 3 gev and close in time to the track , are selected for the further analysis . the distance between any two photons in calorimeter
is required to be larger than 10 cm , and the distance between each photon and the impact point of any track on calorimeter must exceed 15 cm .
fiducial cuts on the distance of each photon from the calorimeter edges and centre are applied to avoid the shower energy leakage .
the distance between the charged particle track and the beam axis at the first drift chamber is required to be larger than 12 cm . for each possible pair of photons
we assume that it originates from @xmath22 decay , and calculate the distance between the decay vertex and the front plane of calorimeter : @xmath23}}{m_{\pi^0}}\ ] ] where @xmath24 , @xmath25 are the energies of the two photons and @xmath14 are their impact point coordinates on calorimeter .
the two photon pairs with the smallest @xmath26 difference are selected as the best combination consistent with two @xmath20 mesons from @xmath1 decay , and the arithmetic average of two @xmath27 values is used as the decay vertex position .
the final event selection requires that the @xmath28 invariant mass differs from the pdg charged kaon mass by at most @xmath29 mev .
the invariant mass of @xmath30 subsystem ( @xmath31 ) in @xmath32 decays has been studied to search for possible anomaly at the threshold region : @xmath33 .
figure [ s2_1](a ) shows the @xmath31 distribution , without any acceptance correction , for the selected events .
fig.[s2_1](b ) shows the same distribution in the region close to the threshold @xmath34 .
the slope change at the threshold is clearly visible .
such an anomaly has not been observed in previous experiments .
the sudden change of slope observed in this plot suggests the presence of a threshold `` cusp '' effect from the decay @xmath35 contributing to the @xmath32 amplitude through the charge exchange reaction @xmath36 .
the presence of a cusp at @xmath37 in @xmath30 elastic scattering due to the effect of virtual @xmath38 loops has been discussed first by meiner et al .
for the case of @xmath32 decay cabibbo @xcite has proposed a simple re - scattering model describing the decay amplitude as the sum of two terms representing `` unperturbed amplitude '' and contribution from the @xmath35 decay amplitude through @xmath36 charge exchange .
the latter contribution is proportional to @xmath39 , the s - wave @xmath40 charge exchange scattering length . in the limit of exact isospin symmetry @xmath41 , where @xmath42 and @xmath43 are the @xmath44 scattering lengths in the i=0 and i=2 states , respectively .
recently cabibbo and isidori @xcite have proposed a more complete model , that takes into account all rescattering processes in two - loop approximation . in the limit of exact isospin symmetry
the amplitude depends on five s - wave scattering lengths , that can be expressed as linear combinations of @xmath42 and @xmath43 .
this model have been used in order to extract @xmath42 and @xmath43 from the fit to the data .
the decay amplitude depends not on @xmath45 only , but also on the two @xmath46 masses squared @xmath47 .
we replace them with the average values to avoid multidimensional integration in the fit procedure : @xmath48 where @xmath49 . in order to extract the decay amplitude parameters , the following function @xmath50 of the measured @xmath0 mass was fitted to the experimental @xmath51 spectrum : @xmath52 here @xmath53 is the theoretical distribution from @xcite of kaon decays over the true value of two - pion invariant mass @xmath54 .
@xmath55 is the function , describing the experimental setup resolution and acceptance , that is obtained by monte carlo simulation .
the fit by this model in the interval @xmath56 allows to extract four parameters : @xmath57 , @xmath58 , @xmath59 ( the slope ) , @xmath60 ( a quadratic term in the dalitz distribution over @xmath61 variable ) and one parameter for normalization .
it leads to @xmath62 for 146 degrees of freedom .
the better fit ( @xmath63 ) is obtained by adding to the model a term describing the expected formation of a @xmath38 atom ( `` pionium '' ) decaying into @xmath30 at @xmath64 .
the fit value for the rate of @xmath65 decays is @xmath66 , which is in good agreement with the predicted value @xmath67 @xcite .
sensitivity of the measured @xmath57 to the pionium production branching ratio is rather small : @xmath68 .
the model used do not include radiative corrections , which are particularly important near @xmath69 , and contribute to the formation of the pionium .
for this reason a group of seven consecutive bins with the width of @xmath70 , centered at @xmath64 have been excluded from the fit in one of the versions of analysis .
in such a case the measured @xmath57 value becomes lower by 0.008 , that is taken into account as the contribution to the systematical error .
there were made three independent analyses which use different event selection criteria and different monte carlo simulations to take into account acceptance and resolution effects .
two of them use a fast monte carlo code , tuned to the experimental resolutions and taking into account the geometry and kinematics . for the third analysis
the full scale geant - based @xcite simulation of the detector is implemented . in all the analyses
the extensive checks for the accordance between the simulation results and data have been made .
the difference between two simulations leads to slightly different second derivative of the acceptance as a function of @xmath31 , that does nt affect the parameter @xmath57 .
study of systematics shows that no corrections are needed .
however , the corresponding uncertainties were taken into account .
for measured @xmath57 parameter the major ones are shown in the table [ system ] . [ cols= " < , < " , ] +
a critical parameter of the models @xcite and @xcite is the ratio @xmath71 between weak amplitudes of @xmath35 and @xmath32 decays . the corresponding `` external '' uncertainty should be complemented by an additional theoretical error of @xmath72 as the result of neglecting higher - order terms and radiative corrections .
this value is compatible withing the errors with the results obtained in the e865 bnl @xcite and the dirac @xcite experiments , as well as with the chpt calculation result @xcite , @xcite ( @xmath74 ) .
it is also in good agreement with the result of fit to low energy experimantal data using dispersion relations @xcite ( @xmath75 ) . in the forthcoming stages of analysis we plan to decrease the systematical uncertainty and to take into account the isospin symmetry breaking . at tree level , omitting the one - photon exchange diagrams , isospin symmetry breaking contributions to the elastic @xmath44 scattering amplitude can be expressed as a function of one parameter @xmath76 @xcite . in particular , the amplitudes at the threshold of @xmath77 mass are corrected by a factor of @xmath78 for the processes @xmath79 , @xmath80 and @xmath81 ; by @xmath82 for @xmath83 ; and by @xmath84 for @xmath36 .
implementing of such a correction will lead to the corresponding decrease of measured @xmath57 .
k.maltman and c.e.wolfe , phys .
b393 ( 1997 ) 19 ; phys .
lett . b424
( 1998 ) 413 ; m.knecht and r.urech , nucl .
phys . b519
( 1998 ) 329 ; g.colangelo , j.gasser , b.kubis and a.rusetsky , comments on isospin breaking in @xmath44 scattering ( unpublished ) .
|
the anomaly ( cusp ) in the mass spectrum of @xmath0 from @xmath1 decay has been observed in the na48/2 experiment . using the recently developed interpretation of this effect in terms of chiral perturbative theory ,
the pion scattering parameters are measured .
the preliminary result for the pion scattering length difference , based on the part of statistics , is @xmath2 .
|
brown dwarfs ( bds ) are stellar - like objects .
the only difference from ordinary stars is that the mass is too low to bring up the central temperature to the level of stable hydrogen burning , thus the bd luminosity decreases with time . as an example , from 70 myr to 10 gyr , a 0.08 @xmath4 object at the hydrogen burning limit would decrease a factor 15 in luminosity , while a 0.06 @xmath4 would go a factor 700 ( burrows et al .
the ideal target for a bd search thus is a fairly young , nearby and rich star cluster .
the pleiades is the obvious choice in the northern hemisphere , being at @xmath7 pc and 70 - 120 myr old .
several recent authors have proposed an age above 100 myr . in this paper
120 myr is adopted .
the mean distance modulus of the cluster used is 5.53 ( see basri et al .
1996 for a discussion ) . whether bds could be
a significant part of the local dark matter is a subject of controversy .
several recent photometric surveys have found a significant drop in the luminosity function from @xmath8 to @xmath9 , leading to a turnover in the initial mass function ( imf , @xmath10 ( @xmath11 = mass ; @xmath12 = imf - index ) ) at @xmath13 0.3 @xmath4 ( see e.g. gould et al .
1996 ; tinney 1993 ) or a continued rise towards the hydrogen burning limit ( see e.g. kirkpatrick et al .
1994 ) depending on the choice of mass - luminosity relation .
kroupa ( 1995 ) showed that the difference between the nearby stellar luminosity function ( lf ) , measured by parallax and the more distant lf , measured by photometry alone , can be explained by undetected binary companions in the distant sample . from a recently derived mass - luminosity relation ( chabrier et al .
1996 ) and the well - known photometric lf ( see e.g. gould et al .
1996 ) , mera et al .
( 1996 ) concluded that the imf for low mass stars continues to rise to the hydrogen burning limit .
however the number of known field stars at the low mass end of the main sequence is small . to clearify this item
it is necessary to discover more low mass stars and bds , preferrably at a known distance and age . in 2 ,
observations , reductions and a short discussion on photometry and completeness limits is given . in
3 , the extraction of pleiades members is described .
4 discusses contamination and overall observing strategy . in
5 the implications of this paper on the imf - index is discussed and compared to other authors .
all observations for this program were obtained at the 2.56 m nordic optical telescope ( not ) , la palma .
the observed area covers 180 @xmath0 in @xmath14 and is centred at ra @xmath15 , dec @xmath16 ( j2000.0 ) .
the @xmath1 data was taken with brocam1 ( tek1k ) , operated in cassegrain focus with 0.176``/pixel and 3x3 field . for future proper motion measurements , the astrometry errors due to nonuninform pixels and possible effects of rotator position
were reduced by using four different position angles for each i - field ( 0 , 90 , 180 and 270 degrees ) exposing 5 min at each position .
debiassing and flatfield corrections were done in a standard fashion within iraf ( image reuction and analysis facility ) .
median seeing was @xmath17 '' , varying from 0.46 `` to 0.80 '' .
the @xmath18 observations were done with the arnica nicmos3 ( 256x256 ) array , which was made available at not via a collaboration with arcetri astrophysical observatory , florence .
the last third of 6 nights in aug - sep95 was used .
unfortunately the pixel size ( 0.55 " ) was not very well matched to the seeing ( @xmath19 ) and all images were undersampled .
another problem was that focus changed across the field , which due to astigmatism in the not optics led to stars of slightly different elongation across the field .
apparently this was caused by the chip not being perpendicular to the optical axis which could be seen as a slight change in pixel scale across the field .
the @xmath2 data could be treated the same way as @xmath1 , but high and rapidly varying background in @xmath3 necessitated a different approach .
the sky for each @xmath3-image was defined as the median of the four images nearest in time and subtracted from the image , which then was flatfielded by a differential flat .
a major problem in this kind of survey is to distinguish stars from distant galaxies at faint magnitudes .
the whole area was visually inspected and all objects were classified as stars , galaxies , binaries or unclassifiable ( too faint for a meaningful classification ) .
1411 objects out of 3800 were classified as stars .
since the seeing conditions were excellent it is estimated that the discrimination between stars and galaxies is reliable to @xmath20 .
since point spread functions ( psfs ) were undersampled in @xmath18 it was not suitable to use ordinary psf fitting .
for single aperture photometry it is necessary to always keep the same aperture radius .
however for the @xmath18 data centering errors and focus shifts are non - negligible , thus in order to maximize the signal to noise ratio it was desirable to change aperture radius from one object to another . a new , more robust , method ( used also in @xmath1 )
was developed , brief outlines : 1 .
all objects were measured in a series of 80 apertures , with radii ranging from 0.05 `` to 4 '' in steps of 0.05 " .
2 . from the `` best '' stars in each field a growth curve ( stetson 1990 ) was constructed , i.e. magnitude shift as function of aperture radius .
each objects curve was subtracted by that frames growth curve .
ideally this curve is completely flat , but due to centering errors , errors in the sky level etc ... that is generally not the case .
however , there is normally an almost flat part of the curve at @xmath21 fwhm .
the magnitude difference was taken as the mean level of that part .
the advantage with this method is that the right mean level will be found even if objects are differently focussed or elongated as compared to the reference growth curve .
transformation of @xmath18 instrument magnitudes to the cit system was done by observations of ir standard stars and transformation equations provided by the arnica team ( hunt et al .
@xmath1 instrument magnitudes were transformed to the kron - cousins system via observations of landolt ( 1992 ) standard stars .
typical internal errors are 0.1 mag ( @xmath22 ; @xmath23 ; @xmath24 ) and 0.02 mag ( @xmath25 ; @xmath26 ; @xmath27 ) .
the zeropoint errors are @xmath130.05 mag in j and @xmath130.08 mag in k. in i they are estimated to be less than 0.02 mag .
zeropoint errors are not included in table 1 and table 3 .
all objects recognized as binaries or stars very close to a galaxy in @xmath1 were , as a check , also measured by psf - fitting . for a few objects that were resolved in @xmath1 , but not in @xmath18 , the whole system was measured as one object in all bands , thus getting the binary system correct in the colour magnitude diagrams . in this kind of surveys
it is important to know the magnitude limit to which the survey is complete .
the number of detected stars as a function of magnitude increases exponentially and gives a straight line in a log(@xmath28 ) vs mag plot as long as the survey is complete .
the point at which the curve turns away from the straight line should be regarded as the completeness limit .
thus @xmath29 , @xmath30 and @xmath31 are the completeness limits of this survey ( magnitude error @xmath13 0.08 mag ) .
the 50 % limit is @xmath13 0.7 mag fainter .
several different kinds of errors define the zone of potential pleiades members in the colour magnitude diagram .
the maximum error in the absolute magnitude of a single pleiades member is not likely to exceed 0.2 mag because of uncertainty of its distance .
the photometric error for @xmath32 is less than 0.1 mag . for binaries ,
the worst case is considered .
two identical companions get 0.75 mag brighter than a single star of the same colour .
therefore the bright edge is raised by 0.75 mag .
the age of the pleiades is also a source of error .
basri et al .
( 1996 ) argue that the age is @xmath13 115 myr while the `` canonical '' age deduced from the upper main sequence turn - off is 75 myr . to account for the worst possible case ,
the 70 myr sequence -1.05 mag defines the bright limit , and the 120 myr model + 0.30 mag defines the faint limit .
an extension of the model colours in chabrier et al .
( 1996 ) based on the brett ( 1995 ) models was kindly provided by plez ( 1996 ) .
next generation models from allard et al .
( 1996 ) were also included .
both models show excellent agreement with the data ( fig . 1 & 2 ) .
note the clear gap between the background field stars and the pleiades sequence and that all previously known members within the field were detected with this technique . as shown in fig . 1 & 2 , 6 objects in the sample ( @xmath33 ) can be regarded as potential cluster members ( table 1 ) . of these
, 5 were previously known as proper motion members ( hambley et al .
1993 ; rebolo et al .
the new candidate ( not1 ) is very close to a much brighter star , discovered only as a consequence of the excellent seeing conditions during the observations . a finding chart and coordinates of not1 are given in fig .
its colours and magnitude indicate a binary of equally sized components , each of @xmath34 , slightly heavier than ppl15 ( basri et al .
it is , however , also consistent with being a foreground m - dwarf ( see sect .
4 ) . .photometry of potential pleiades members .
magnitude errors are internal . + [ cols="<,<,<,<",options="header " , ] the lf from gould et al .
( 1996 ) was used to estimate the number of foreground m - dwarfs for @xmath35 to @xmath36 .
thus it is not likely , although possible , that not1 is a foreground m - dwarf .
since late m - dwarfs are presumed to be rare , it is interesting that a few candidates turn up in this survey , see table 3 .
both not2 and not3 appear quite close to the pleiades zone in fig .
2 . could they be low mass pleiades bds ?
not3 was in a part of the field that was also covered by jameson & skillen ( 1989 ) .
it was found in their original data , however far too faint for astrometry of sufficiently high accuracy for a proper motion measurement .
not2 and not3 will though be observed in future runs .
+ towards this part of the pleiades the total reddening is @xmath37 mag ( stauffer & hartmann 1987 ) , which corresponds to @xmath38 mag ( cardelli et al .
thus if these stars really are m - dwarfs , @xmath39 is less than 0.1 mag even for the most distant candidate and does not affect classification significantly .
our goal was not to cover as large an area as possible , but to reach very faint magnitudes in order to investigate the presence of a population of low mass bds in the pleiades .
this survey is complete to @xmath40 , corresponding to 0.035 @xmath4 ( 120 myr ) or 0.01 @xmath4 ( 70 myr ) . in fig .
3 , lfs from recent surveys are compared to lfs deduced from 4 different imf - indices . it is clear from previous surveys that an imf - index close to or even above 0 is favored at least down to @xmath41 .
this survey , complete to @xmath29 in the pleiades domain , one magnitude fainter than any previous survey does not give one single object below @xmath42 .
thus @xmath43 ( simons & becklin 1992 ) can be rejected with 99.8% confidence , @xmath44 with 96.9% .
corresponding figures for @xmath45 is 75.1% and for @xmath46 45.0% .
the final conclusion is that if low mass bds ( @xmath47 @xmath4 ) exist in the pleiades , they follow an imf with index less steep than @xmath45 , and can not comprise a very large part of the cluster s mass . allard f. , alexander d.r . , hauschildt p.h . ,
schweitzer a. 1996 , apj submitted bahcall j.n . ,
soneira r.m .
1984 , apjs 55 , 67 basri g. , marcy g.w . , graham j.r .
1996 , apj 458 , 600 brett j.m .
1995 , a&a 295 , 736 burrows a. , hubbard w.b .
, saumon d. , lunine j.i .
1993 , apj 406 , 158 cardelli j.a . ,
clayton g.c . ,
mathis j.s .
1989 , apj 345 , 245 chabrier g. , baraffe i. , plez b. 1996 , apj 459 , l91 gould a. , bahcall j.n . , flynn c. 1996 , apj 465 , 759 hambley n.c .
, hawkins m.r.s . ,
jameson r.f .
1993 , a&as 100 , 607 hunt l.k . , migliorini s. , testi l. , et al .
1995 , aj submitted jameson r.f .
, skillen i. 1989 , mnras 239 , 247 kroupa p. 1995 , apj 453 , 358 kirkpatrick j.d . ,
mcgraw j.t . ,
hess t.r . , liebert j. , mccarthy jr , d.w .
1994 , apjs 94 , 749 landolt a.u .
1992 , aj 104 , 340 leggett s.k .
1992 , apjs 82 , 351 lucke p.b . 1978 ,
a&a 64 , 365 mera d. , chabrier g. , baraffe i. 1996 , apj 459 , l87 plez b. 1996 , private communication rebolo r. , zapatero osorio m.r . , martin e.l . 1995 ,
nature 377 , 129 rebolo r. , martin e.l .
, basri g. , marcy g.w . , zapatero osorio m.r .
1996 , apj 469 , l53 simons d.a .
, becklin e.e .
1992 , apj 390 , 431 stauffer j.r . , hartmann l. 1987 , apj 318 , 337 stauffer j.r . , hamilton d. , probst r.g .
1994 , aj 108 , 155 steele i.a . ,
jameson r.f .
, hambley n.c .
1993 , mnras 263 , 647 stetson p.b . , 1990 ,
pasp 102 , 932 tinney c.g .
1993 , apj 414 , 279 williams d.m . ,
boyle r.p . ,
morgan w.t . ,
1996 , apj 464 , 238 zapatero osorio m.r .
, rebolo r. , martin e.l .
1996 , a&a in press
|
an area large enough ( 180 @xmath0 ) to put constraints on a possible low mass brown dwarf population in the pleiades has been surveyed to very faint magnitudes in @xmath1 , @xmath2 and @xmath3 .
the completeness limit , i=21.6 , corresponds to a mass of 0.01 @xmath4 for a cluster age of 70 myr and 0.035 @xmath4 for 120 myr .
the result is consistent with previous investigations at higher masses that the brown dwarf initial mass function is a @xmath5 , or even less steep , power law .
thus low mass brown dwarfs can not contribute significantly to the pleiades mass .
one new possible pleiades member was found , mass @xmath6 @xmath4 ( age 120myr ) .
|
the recently synthesized isotypic metal - rich ternary borides li@xmath0pd@xmath1b and li@xmath0pt@xmath1b @xcite , with noncentrosymmetric crystal structures , have received considerable attention because of their superconducting properties @xcite .
the superconducting transition temperature @xmath3 of li@xmath0pd@xmath1b is about @xmath4 k , while its isotypic pt - based counterpart li@xmath0pt@xmath1b shows a @xmath3 between @xmath5 and @xmath6 k @xcite .
badica and co - workers @xcite have been able to synthesize a series of pseudo - binary solid solutions li@xmath0b(pd@xmath7pt@xmath8)@xmath1 with @xmath9 varying from @xmath10 to @xmath11 and they report having observed superconductivity in the entire @xmath9-range .
@xmath3 decreases monotonically from @xmath12 @xmath4 k as @xmath9 increases from @xmath10 , dropping to @xmath13@xmath14 k for @xmath15 .
the electronic structure of the two end compounds of the solid solution li@xmath0pd@xmath1b and li@xmath0pt@xmath1b have been discussed by chandra and co - workers @xcite .
sardar and sa @xcite have used a three - dimensional single band @xmath16-@xmath17 model to discuss the superconductivity of li@xmath0pd@xmath1b , pointing out that the superconductivity in this compound arises from the pd @xmath18__d _ _ electrons , and thus should be dominated by the strong correlation effects of narrow band pd @xmath18__d _ _ electrons .
the importance of the correlation effects of the @xmath18__d _ _ electrons for the superconducting properties of li@xmath0pd@xmath1b and li@xmath0pt@xmath1b was also mentioned by chandra _
_ @xcite . on the other hand , yokoya _ et al . _
@xcite report excellent agreement between their x - ray photoemission results for the valence band spectrum and standard band structure calculations , which can describe only weakly correlated materials .
they conclude that the correlation effects play a negligible role for the physical properties of this superconductor , and that superconductivity in this material arises from the pd @xmath18__d _ _ electrons hybridized with b @xmath5__p _ _ and li @xmath5__p _ _ electrons .
absence of strong correlation effects in li@xmath0pd@xmath1b is also indicated in the nmr measurements in this compound by nishiyama _
_ @xcite . since the pd(pt )
bands are almost full @xcite , the correlation effects are not expected to be strong . the purpose of the present work is to examine the two isotypic superconductors li@xmath0pd@xmath1b and li@xmath0pt@xmath1b from the viewpoint of electron - phonon coupling and conventional ( _ s_-wave ) superconductivity , without explicit consideration of electron correlation effects and spin - orbit ( so ) coupling . to this end
, we use the rigid muffin - tin approach of gaspari and gyorffy @xcite implemented within the linear muffin - tin orbital ( lmto ) method @xcite by gltzel _
_ @xcite and later by skriver and mertig @xcite .
the present paper is organized as follows . in section [ sec_electronicstructure ]
we discuss the electronic structure , charge transfer and bonding in the two compounds . in section [ sec_electron - phonon ]
we discuss the electron - phonon coupling .
in particular we provide estimates of the fermi - surface averaged electron - phonon matrix elements ( hopfield parameters ) , and electron - phonon coupling constants .
our results indicate that conventional electron - phonon coupling , without consideration of so effects , can describe the observed superconductivity in li@xmath0pd@xmath1b , but not in li@xmath0pt@xmath1b .
there is some similarity in the superconducting properties of li@xmath0pt@xmath1b and another recently discovered noncentrosymmetric crystal cept@xmath1si with strong so coupling @xcite .
this is discussed in section [ sec_electron - phonon ] . in section [ sec_conclusion ]
we present our conclusions .
the electronic structures of the two ternary borides li@xmath0pd@xmath1b and li@xmath0pt@xmath1b , which crystallize in the * p*4@xmath132 structure ( space group no .
212 ) with 4 formula units per unit cell @xcite , were calculated with the lmto @xcite method using the atomic sphere approximation ( asa ) .
the lattice constants used were @xmath19 and @xmath20 @xmath21 for li@xmath0pd@xmath1b and li@xmath0pt@xmath1b , respectively .
the positions of the atoms were taken to be the same as given by eibenstein and jung @xcite and used in the electronic structure calculation of chandra _ et al .
the atomic structure is shown in fig . [ fig_xyz ] .
the shortest bonds are between pd and b ( @xmath22 @xmath21 ) and li and li ( @xmath23 @xmath21 ) .
each b atom is connected to six pd atoms , and each pd atom has two b nearest neighbors . the structure of the pd and b atoms can be described as a network of connected pd@xmath1b tetrahedra @xcite . to show this , in fig .
[ fig_xyz ] we have also drawn bonds for the shortest pd - pd distances ( @xmath24 @xmath21 ) , which are however longer than some of the pd - li distances ( @xmath25 @xmath21 , not shown ) . the three - fold coordinated li atoms form their own network ( fig .
[ fig_xyz ] ) .
because the li - li distances are @xmath26 smaller than in bcc li @xcite , it has been anticipated @xcite that the li atoms transfer their @xmath5__s _ _ electrons entirely to the pd and b atoms .
the valence charge density of li@xmath0pd@xmath1b [ fig .
[ fig_rho](b ) , the total charge density is presented in fig .
[ fig_rho](a ) ] , shows indeed that there is very little electronic valence charge present near the li atoms . to see the effect of bonding it is often useful to subtract the atomic valence charge densities from the total valence charge density .
the result is shown in fig . [ fig_drho ] .
again it is clear that the network of li atoms is depleted of charge .
in addition , a strong increase in the charge density is found between the b and pd atoms , which indicates covalent bonding .
the increase in charge density between the pd atoms is less pronounced , suggesting that b - pd bonds are stronger than pd - pd bonds .
we have carried out the lmto calculation without empty spheres ( which a more rigorous calculation would have required ) using the @xmath27 basis and with the @xmath28-orbitals downfolded . even though our results are not the best possible results within the lmto - asa scheme , a comparison with more rigorous full - potential linear augmented plane waves plus local orbitals ( fp - lapw+lo ) results presented by chandra
_ et al . _
@xcite shows that our results are of acceptable accuracy .
the densities of states shown in fig .
[ fig_dos ] densities of states of ( a ) li@xmath0pd@xmath1b and ( b ) li@xmath0pt@xmath1b ( thick solid curves ) .
partial pd and pt-_d _ ( thin curves , shadings ) and b-_p _ ( dotted curves ) densities of states are also shown.,scaledwidth=80.0% ] agree very well with the results of chandra _ et al .
the agreement of our calculated densities of states with another set of fp - lapw results of yokoya _ et al . _
@xcite is also very good . in fig .
[ fig_dos ] among the various partial densities of states , only the pd(pt)-@xmath29 and b-@xmath30 densities of states are shown , since these have the largest values at the fermi level and as a result electron - phonon interaction is dominated by pd(pt ) @xmath29-@xmath28 and b @xmath30-@xmath29 scattering ( section [ sec_electron - phonon ] ) .
note that the density of states per atom is almost the same for pd-@xmath29 ( 3.34 states/(ry atom ) ) and b-@xmath30 states ( 3.23 states/(ry atom ) ) in li@xmath0pd@xmath1b . in li@xmath0pt@xmath1b
these numbers corresponding to pt-@xmath29 and b-@xmath30 states are 4.2 states/(ry atom ) and 3.72 states/(ry atom ) , respectively .
the main features of our lmto calculations that are used in estimating the electron - phonon matrix elements in section [ sec_electron - phonon ] are the atom and orbital resolved densities of states , listed in table [ table1 ] .
.partial @xmath31 and atom resolved densities of states , including both spins , at the fermi energy in states / ry for @xmath32 atom unit cells of li@xmath0pd@xmath1b and li@xmath0pt@xmath1b .
the numbers shown are for @xmath33 pd(pt ) , @xmath4 li and @xmath18 b atoms .
the total fermi level dos , @xmath34 , is @xmath35 states / ry / cell for li@xmath0pd@xmath1b and @xmath36 states / ry / cell for li@xmath0pt@xmath1b . [ cols="^,^,^,^,^,^",options="header " , ] the atom - resolved and the total hopfield parameters , @xmath37 and @xmath38 , obtained by using eqs .
( 5 ) and ( 6 ) are listed .
we also list the calculated bulk moduli of the two compounds , which can be used to obtain estimates of the debye temperatures and average phonon frequencies for these solids .
the analysis of the electron - phonon coupling and critical temperatures in the two compounds is based on the quantities listed in tables [ table1 ] and [ table2 ] .
for li@xmath0pd@xmath1b the hopfield parameter calculated according to eqs .
( 5 ) and ( 6 ) is @xmath39 ev / @xmath40 .
note that the boron contribution is only @xmath41 less than the palladium contribution .
most dominant contributions come from pd @xmath29-@xmath28 ( @xmath42 ev / @xmath40 ) and b @xmath30-@xmath29 ( @xmath43 ev / @xmath40 ) scattering . although double in number than the b atoms , all li atoms together contribute only @xmath44 to the hopfield parameter @xmath38 .
@xmath45 of the contribution of the li atoms is from the @xmath46 channel .
a few comments regarding the decomposition of the hopfield parameter into atom - resolved parts and various partial wave channels in an alloy are in order .
because certain potential parameters depend on the ratio @xmath47 ( or @xmath48 ) , where @xmath2 is the sphere radius for the atom and @xmath49 is the average wigner - seitz radius of the alloy , the atom - resolved parameters may differ for different choices of the sphere radii , all chosen within the allowable sphere overlaps .
the total hopfield parameter is expected to be independent of the choice of the sphere radii , if these are chosen to keep the asa errors as small as possible , but the partial - wave resolved parameters ( i.e , @xmath2-@xmath30 , @xmath30-@xmath29 etc . )
are dependent on the ratio @xmath47 .
ideally , the lmto calculation for the two solids li@xmath0pd@xmath1b and li@xmath0pt@xmath1b should have been carried out with empty spheres in order to reduce asa errors .
this would have given an improved value of the total hopfield parameter , including contributions from both atomic and empty spheres . despite some asa errors present in our results
, we expect the trends revealed by them to be correct . in the simplest approximation , assuming isotropic electron - phonon coupling , one can attempt to estimate @xmath50 from @xmath38 using eq . ( 1 ) , assuming that @xmath51 is the concentration average of the mass of the component atoms in the alloy and @xmath52 is the average phonon frequency for the alloy . to estimate the average phonon frequency we use the prescription of moruzzi , janak and schwarz @xcite , relating the debye temperature @xmath53 to the bulk modulus , atomic mass and the average wigner - seitz radius : @xmath54 where @xmath55 is the average wigner - seitz radius in atomic units
, @xmath56 is the bulk modulus in kbar and @xmath51 is the atomic mass .
although moruzzi _ et al . _
@xcite verified the validity of this expression for elemental metallic solids , we assume its validity for the alloy by considering @xmath51 to be the concentration average of the masses of the component atoms . from the lmto calculation of total energy versus volume we find the bulk modulus @xmath56 of li@xmath0pd@xmath1b to be @xmath57 gpa , a value consistent with the bulk moduli of elemental solids pd ( @xmath58 gpa ) , b ( @xmath59 gpa ) and li ( @xmath60 gpa ) @xcite . using @xmath57 gpa for @xmath56 in the above equation we find @xmath61 k. using the empirical estimate @xmath62 @xcite yields an average phonon frequency of @xmath24 thz .
the use of eq .
( 1 ) then yields @xmath63 , a value not large enough to explain the observed superconductivity of li@xmath0pd@xmath1b at @xmath64 k. this could have been foreseen from the small value , @xmath39 ev / @xmath40 , of @xmath38 itself .
the value of @xmath38 for niobium , which has a comparable @xmath3 of @xmath65 k , is in the range @xmath66@xmath67 ev / @xmath40 @xcite .
however , in view of the uncertainties , not only in the frequency value , but the validity of eq .
( 1 ) itself , this value of @xmath50 can not be expected to be more than a coarse approximation .
it is known that even for one - component systems the rigid muffin - tin approximation usually gives values of the hopfield parameter that are lower than those obtained via more rigorous calculations of the electron - phonon interaction ( see table ii . in ref .
@xcite ) . for the solids under consideration with a large number of optical phonon branches ,
the extent of underestimation is probably more severe .
what is learnt from the above rigid muffin - tin ( or atomic sphere ) calculation is that the contribution to the electron phonon coupling from both boron and palladium should be equally important , and that from the lithium atoms should be negligible . following the prescription of gomersall and gyorffy @xcite one could try to analyse the electron - phonon coupling by writing @xmath68 where the subscripts pd and b refer to atom - resolved quantities .
this decomposition is based on the large difference between the pd and b masses , and assumes that li atoms do not contribute significantly to the electron - phonon coupling , as supported by the lmto results .
the heavy palladium atom vibration is assumed to couple to the electrons through zone boundary acoustic phonons and the lighter boron atom vibration is assumed to couple to electrons via the optical modes .
it is reasonable to assume that in the acoustic modes only the palladium atoms vibrate , and in the optical modes only the lighter boron atoms vibrate . instead of guessing the average phonon frequencies in the above equation
, we can estimate the ratio @xmath69 by assuming that the average optical phonon frequency is , to take an example , @xmath70 times higher than the average zone boundary acoustic phonon frequency . with the values of @xmath71 and @xmath72 quoted in table [ table2 ] , we then find @xmath73 , i.e. if the above scenario holds , then the boron contribution to the electron - phonon coupling is larger than the pd contribution . for @xmath74 to be equal to or greater than @xmath75 , in this scenario ,
the average optical phonon frequency needs to be about @xmath76 times or higher than the average zone boundary acoustic phonon frequency . from the consideration of the linear specific heat coefficient @xcite , lee and pickett @xcite
have obtained @xmath77 .
with @xmath73 , this would yield @xmath78 and @xmath79 .
note that the value @xmath80 is not unreasonable , as a full - potential lmto linear response calculation by savrasov and savrasov @xcite yields @xmath81 in elemental fcc pd .
incidentally , the use of the mcmillan expression @xcite : @xmath82 with the coulomb pseudopotential @xmath83 chosen to be @xmath84 , as it is often done , @xmath85 k ( table [ table2 ] ) , and @xmath86 yields @xmath87 k. the above discussion points out that a description based on conventional ( @xmath2-wave , spin singlet cooper pairs ) electron - phonon interaction should be applicable to li@xmath0pd@xmath1b .
of course a more rigorous calculation of the electron - phonon interaction is needed .
such a calculation is likely to reveal negligible coupling of lithium atom vibrations to the electron states at the fermi level , while the vibrations of the palladium and boron atoms should couple strongly . according to yuan _
_ @xcite temperature dependence of the penetration depth in li@xmath0pd@xmath1b could be fit by a two - gap bcs model with a small ( @xmath88 k ) and a large ( @xmath89 k ) gap .
thus the claim is that li@xmath0pd@xmath1b is a two - gap bcs superconductor just like mgb@xmath0 @xcite .
the fermi surface plots presented by chandra _
_ @xcite show a fermi surface with several sheets : a central large sheet and several smaller ones originating from the four bands crossing the fermi level .
a two or multi band superconductivity is thus conceivable in this material just as in mgb@xmath0 with two distinct sheets in the fermi surface @xcite .
however , @xmath3 may still be dominated by a single band @xcite , particularly in the absence of strong interband scattering .
the inadequacy of the above approach to describe the electron - phonon coupling in li@xmath0pt@xmath1b becomes clear from the much larger value of the hopfield parameter obtained for this compound and listed in table [ table2 ] .
note that the unusually large contribution from boron is partly due to larger difference between the boron sphere radius @xmath2 and the average wigner - seitz radius @xmath49 .
a calculation including empty spheres to reduce asa errors would have changed the numbers in table [ table2 ] somewhat , without changing the results qualitatively . even with the reduced value of @xmath90 ( table [ table2 ] ) , the much larger value of @xmath38 would yield a value for @xmath50 larger than that for li@xmath0pd@xmath1b , implying a @xmath3 that is actually higher than in li@xmath0pt@xmath1b .
this is in clear contradiction to the experimental observation that @xmath3 decreases monotonically as the pt concentration increases from zero to unity the most serious drawback of the above analysis is the neglect of so coupling , which is more pronounced in li@xmath0pt@xmath1b than in li@xmath0pd@xmath1b .
so coupling , in the absence of the center of inversion symmetry in these two compounds , can result in substantial splittings of the spin degenerate bands considered in this work . in a recent submission , lee and pickett @xcite discuss so coupled bands in both compounds : splitting of the spin degenerate bands is found to be as large as @xmath91 mev and @xmath92 mev at the fermi level in li@xmath0pd@xmath1b and li@xmath0pt@xmath1b , respectively .
the so splitting of @xmath91 mev or @xmath93 k in li@xmath0pd@xmath1b is less than the estimated debye temperature @xmath94 k , the cut - off energy of the electron - phonon interaction .
in li@xmath0pt@xmath1b the so splitting is @xmath92 mev or @xmath95 k , an order of magnitude larger than the estimated debye temperature @xmath96 k. it is clear that so coupling would strongly alter the symmetry of the cooper pairs , and the nature of the superconducting state in li@xmath0pt@xmath1b @xcite .
yuan _ et al .
_ @xcite have claimed evidence of line of nodes in the energy gap in li@xmath0pt@xmath1b .
such a line of zeros in the energy gap is also seen in cept@xmath1si @xcite , another noncentrosymmetric crystal with strong so splitting of the spin degenerate bands @xcite .
the existence of the line of nodes is not a symmetry - imposed requirement and remains to be explained for both li@xmath0pt@xmath1b and cept@xmath1si .
cept@xmath1si is a heavy fermion superconductor ( @xmath97 k ) , so the mechanism of the pair formation is unclear .
li@xmath0pt@xmath1b is most probably an electron - phonon superconductor , where the electron - phonon coupling needs to be determined from the so coupled electron wavefunctions and is expected to differ strongly from that determined from the spin degenerate wavefunctions .
the results presented in this work are not rigorous , and should only be used as a guide to understand the relative strengths of coupling of the vibration of the different atoms to the electron states at the fermi level .
based on these results , a description of superconductivity assuming conventional @xmath2-wave electron - phonon coupling and via a single electron - phonon coupling parameter appears possible in li@xmath0pd@xmath1b .
the inclusion of so coupling in the calculation of electron - phonon interaction should lead to improvement in results .
for li@xmath0pt@xmath1b the so effects are an order of magnitude stronger and do need to be included in the calculation of the electron - phonon interaction in order to explain the lower @xmath3 in this compound . the trends and the conclusions reached for li@xmath0pd@xmath1b in this work should be tested via rigorous calculation of the phonon frequencies and electron - phonon interaction .
this is particularly important in view of the large number of phonon branches in this material .
the results of this paper indicate that only the modes where the boron and the palladium atoms vibrate should couple to the electrons
. the electron - phonon coupling should be equally strong for the modes where essentially the pd atoms or the b atoms vibrate . according to our analysis
the modes involving the vibrations of the li atoms should show negligible coupling to electron states at the fermi level .
this is also suggested by our finding that almost the entire valence charge of lithium is transferred to the pd - b complex ( section [ sec_electronicstructure ] ) .
recent work of lee and pickett @xcite also supports this conclusion .
first principles linear response calculations of the phonons and electron - phonon coupling to verify the results of the present work are currently in progress . due to the strong so coupling in li@xmath0pt@xmath1b the properties of the superconducting state is expected to deviate strongly from that described by spin - degenerate wavefunctions @xcite . however , superconductivity should still be of the electron - phonon type ( possibly also spin - singlet type , as suggested by lee and pickett @xcite ) . the experimental result that @xmath3 decreases monotonically in solid solutions
li@xmath0b(pd@xmath7pt@xmath8)@xmath1 as the pt concentration increases from zero to @xmath11 indicates that the mechanism of electron pair formation remains unchanged , i.e. of the electron - phonon type .
financial support for this work was provided by natural sciences and engineering research council of canada .
skb gratefully acknowledges helpful discussion with o. jepsen , b. mitrovi and k. v. samokhin .
u. eibenstein and w. jung , j. solid state chem .
* 133 * ( 1997 ) 21 .
k. togano , p. badica , y. nakamori , s. orimo , h. takeya and k. hirata , phys .
* 93 * ( 2004 ) 247004 .
p. badica , t. kondo , t. kudo , y. nakamori , s. orimo and k. togano , cond - mat/0404257 .
s. chandra , s. mathi jaya and v.c .
valsakumar , cond - mat/0502525 .
m. sardar and d. sa , cond - mat/040168 .
m. nishiyama , y. inada and g .- q .
zheng , phys .
b * 71 * ( 2005 ) 220505(r ) ; cond - mat/0506191 v1 .
t. yokoya , t. muro , i. hase , h. takeya , k. hirata and k. togano , phys .
b * 71 * ( 2005 ) 092507 .
p. badica , t. kondo and k. togano , j. phys .
jpn * 74 * ( 2005 ) 1014 .
gaspari and b.l .
gyorffy , phys .
* 28 * ( 1972 ) 801 .
andersen , phys . rev .
b * 8 * ( 1975 ) 3060 ; o.k .
andersen , o. jepsen , m. sob in _ electronic structure and its applications _ , edited by m. yossouff , lecture notes in physics , v. 283 ( springer , berlin , 1987 ) p. 157 ; o.k .
andersen , o. jepsen , and d. gltzel , in _ highlights of condensed matter theory _ , edited by f. bassani _ et al . _ ( north - holland , amsterdam , 1985 ) , p. 59 ; h.l .
skriver , _ the lmto method _ ( springer , berlin , 1984 ) .
d. gltzel , d. rainer and h.r .
schober , z. phys .
b * 35 * ( 1979 ) 317 .
skriver and i. mertig , phys . rev .
b * 32 * ( 1985 ) 4431 .
samokhin , e.s .
zijlstra and s.k .
bose , phys .
b * 69 * ( 2004 ) 094514 ; * 70 * ( 2004 ) 069902(e ) . w.l .
mcmillan , phys .
* 167 * ( 1968 ) 331 .
gomersall and b.l .
gyorffy , j. phys .
f * 3 * ( 1973 ) l138 .
klein and d.a .
papaconstantopoulos , phys .
* 32 * ( 1974 ) 1193 .
papaconstantopoulos and b.m .
klein , phys .
* 35 * ( 1975 ) 110 .
klein , w.e .
pickett , d.a .
papaconstantopoulos and l.l .
boyer , phys .
b * 27 * ( 1983 ) 6721 .
t. jarlborg , a. junod and m. peter , phys .
b * 27 * ( 1983 ) 1558 .
mazin and s.n .
rashkeev , phys .
b * 42 * ( 1990 ) 366 .
d. rainer , prog . low temp .
x ( 1986 ) 371 .
papaconstantopoulos , l.l .
boyer , b.m .
klein , a.r .
williams , v.l .
morruzzi and j.f .
janak , phys .
b * 15 * ( 1977 ) 4221 .
moruzzi , j.f .
janak and k. schwarz , phys .
b * 37 * ( 1988 ) 790 .
c. kittel , _ introduction to solid state physics _ , 7@xmath98 edn .
( john wiley , 1996 ) , p. 59 .
bose , o.v .
dolgov , j. kortus , o. jepsen and o.k .
andersen , phys .
b * 67 * ( 2003 ) 214518 .
k. togano and h. takeya , unpublished results communicated to w.e .
pickett ( cond - mat/0507105 v1 ) .
lee and w.e .
pickett , cond - mat/0507105 v1 .
savrasov and d.y .
savrasov , phys .
b * 54 * ( 1996 ) 16487 .
p.b . allen and b. mitrovi ,
_ solid state physics _ , edited by h. ehrenreich , f. seitz and d. turnbull ( academic , new york 1982 ) , vol .
37 , p. 1 .
yuan , d. vandervelde , m.b .
salamon , p. badica and k. togano , cond - mat/0506771 .
s. soma _ et al .
_ , nature ( london ) * 423 * ( 2003 ) 65 .
i.i . mazin and v.p .
antropov , physica c * 385 * ( 2003 ) 49 .
b. mitrovi , j. phys .
: condens .
matter * 16 * ( 2004 ) 9013 .
j. kortus , i.i .
mazin , k.d .
belashchenko , v.p .
antropov and l.l .
boyer , phys .
lett . * 86 * ( 2001 ) 4656 .
y. kong , o.v .
dolgov , o. jepsen and o.k .
andersen , phys .
b * 64 * ( 2001 ) 020501(r ) .
b. mitrovi , eur .
j. b * 38 * ( 2004 ) 451 .
gorkov and e.i .
rashba , phys .
* 87 * ( 2001 ) 037004 .
frigeri _ et al .
* 92 * ( 2004 ) 097001 .
sergienko and s.h .
curnoe , phys .
b * 70 * ( 2004 ) 214510 .
k. izawa _ et al .
_ , phys .
* 94 * ( 2005 ) 197002 .
|
electronic structure of recently discovered isotypic ternary borides li@xmath0pd@xmath1b and li@xmath0pt@xmath1b , with noncentrosymmetric crystal structures , is studied with a view to understanding their superconducting properties .
estimates of the fermi - surface averaged electron - phonon matrix element and hopfield parameter are obtained in the rigid ion approximation of gaspari and gyorffy [ phys .
rev .
lett .
* 28 * ( 1972 ) 801 ] .
the contribution of the lithium atoms to the electron - phonon coupling is found to be negligible , while both boron and palladium atoms contribute equally strongly to the hopfield parameter .
there is a significant transfer of charge from lithium , almost the entire valence charge , to the b - pd(pt ) complex .
the electronic structure and superconducting properties of li@xmath0pd@xmath1b , thus , can be understood from the viewpoint of the compound being composed of a connected array of b - pd tetrahedra decoupled from the backbone of li atoms , which are connected by relatively short bonds .
our results suggest that conventional @xmath2-wave electron - phonon interaction without explicit consideration of so coupling can explain qualitatively the observed @xmath3 in li@xmath0pd@xmath1b .
however , such an approach is likely to fail to describe superconductivity in li@xmath0pt@xmath1b . and electron - phonon coupling , hopfield parameter , rigid ion and muffin - tin approximations , linear muffin - tin orbitals 74.70.-b , 74.62.fj , 74.25.kc , 71.20.-b
|
the early stages of evolution of low - mass stars are relatively well characterised by the formation of a system with a central protostar , accreting material from a rotating accretion disk at scales of a few hundreds of astronomical units ( au ) , and simultaneously ejecting a collimated outflow with the presence and crucial role of magnetic fields .
accretion and mass - loss processes , two closely related mechanisms , govern the formation of low - mass stars ( e.g. , girart , rao & marrone 2006 ; mckee & ostriker 2007 ; machida , inutsuka & matsumoto 2008 ; armitage 2011 ; williams & cieza 2011 ) .
spatio - kinematical studies of jets and herbig - haro ( hh ) objects show that outflows from low - mass young stellar objects ( ysos ) are non - steady , but presenting variability in the ejection velocity as well as pulsed events , probably related to recurrent instabilities in the accretion disks ( e.g. , zinnecker , mccaughrean & rayner 1998 ; reipurth & bally 2001 ; estalella et al . 2012 ) . with respect to the formation of massive stars ( @xmath10 10 m@xmath7 ) , there are several examples showing the presence of massive disk - protostar - jet systems at scales of thousand au ( in some cases with magnetic fields oriented parallel to the collimated outflows ) , indicating that stars at least up to @xmath220 m@xmath7 form via an accretion disk as low - mass do ( e.g. , patel et al . 2005 ; jimnez - serra et al . 2007 ; torrelles et al . 2007 ; davies et al . 2010 ; carrasco - gonzlez et al .
2010 , 2012a ; vlemmings et al . 2010
; fernndez - lpez et al . 2011
; maud & hoare 2013 ) .
furthermore , variation in the ejection velocity of outflows as well as remarkable short - lived episodic ejection events ( few tens of years ) have been reported towards massive ysos through radio continuum ( e.g. , mart et al 1995 ; curiel et al .
2006 ) and h@xmath0o maser spatio - kinematical observations ( e.g. , torrelles et al . 2001 , 2003 ; surcis et al .
2011a , b ; chibueze et al . 2012
; sanna et al . 2012 ; kim et al . 2013 ; trinidad et al .
this indicates that outflows in massive ysos are also non - steady , probably due to instabilities in the accretion processes as in the case of low - mass ysos .
mainly because most of the observational efforts have been concentrated in low- and high - mass ysos , little is known on what happens in intermediate - mass protostars , with only a few examples showing clear disk - yso - jet systems at scales of a few hundred au ( e.g. , palau et al .
2011 ; carrasco - gonzlez et al .
2012b ) , but up to now without any observational information on the magnetic field and outflow kinematical behaviour at these small scales .
observations of the first stages of the evolution of intermediate - mass ysos at small scales are therefore important to fill that gap , and thus to have a more complete vision about the processes of star formation throughout a more continuous range of masses . in this paper
, we present multi - epoch very long baseline interferometry ( vlbi ) h@xmath0o maser observations towards intermediate - mass ysos to study the spatio - kinematical distribution of their masers with @xmath20.5 mas resolution .
this kind of study has previously been very useful in massive star - forming regions to detect outflow activity through the presence of h@xmath0o maser emission , allowing the identification of new centres of star formation , some of them associated , as mentioned before , with short - lived episodic ejection events with kinematic ages of a few tens of years . with the present work , we now extend this study to intermediate - mass star - forming regions .
the optically visible herbig be star lkh@xmath1 234 , with luminosity @xmath210@xmath11 l@xmath7 and mass @xmath25 m@xmath7 , is one of the brightest sources of the ngc 7129 star - forming region , located at a distance of 0.9 kpc ( herbig 1960 ; strom et al .
1972 ; bechis et al . 1978
; fuente et al . 2001 ; trinidad et al . 2004 ; xu et al . 2013 ) .
high angular resolution near- and mid - infrared images obtained by kato et al .
( 2011 ) reveal , in addition to lkh@xmath1 234 , a cluster of eight yso candidates within a distance of @xmath210 arcsec from lkh@xmath1 234 ( named by kato et al . as objects
b , c , d , e , f , g , nw1 , and nw2 ) .
very large array ( vla ) observations show five radio continuum sources in a region of @xmath25 arcsec ( vla 1 , vla 2 , vla 3a , vla 3b , lkh@xmath1 234 ; trinidad et al .
2004 ) , some of them with mid - infrared counterparts : vla 2 ( nw2 ) , vla 3a+3b ( nw1 ) , and lkh@xmath1 234 ( see fig .
nw2 and nw1 , are not seen in the j , h , and k - band images , but they are bright in the @xmath12 , @xmath13 , and mid - infrared bands , indicating that they are highly embedded ysos ( kato et al .
maser emission is detected towards this region , mainly associated with vla 1 , vla 2 ( nw2 ) , and vla 3a+3b ( nw1 ) ( tofani et al .
1995 ; umemoto et al . 2002
; trinidad et al . 2004 ; marvel 2005 ; see fig .
1 ) . several outflows emerging from this region have been also observed , e.g. : i ) a jet observed in [ s ii ] , extending to the south - west ( p.a . @xmath2
252@xmath14 ) up to distances of @xmath240 arcsec ( @xmath236,000 au ) , without evidence of a counter - jet ( ray et al .
1990 ) ; ii ) an infrared jet seen in v=10 s(1 ) h@xmath0 emission , extending towards the south - west ( p.a . @xmath2
226@xmath14 ) up to distances of @xmath210 arcsec ( @xmath29,000 au ) , also without any evidence of a counter - jet ( cabrit et al .
the overlapping of these two outflows within a small area of the sky makes it very difficult to clearly distinguish between various candidate driving sources .
what we know , however , is that all the different studies exclude that the herbig be star lkh@xmath1 234 is driving any of the outflows of this star - forming region , but that the main outflow activity centres are probably the ysos vla 2 ( nw2 ) and vla 3a+3b ( nw1 ) ( trinidad et al . 2004 ; marvel 2005 ; kato et al .
2011 ) . following the nomenclature of trinidad
et al . ( 2004 ) , in this paper we will refer to the region containing all this cluster of ysos as the star - forming lkh@xmath1 234 region " . in section 2 , we present the multi - epoch vlbi h@xmath0o maser observations towards the star - forming lkh@xmath1 234 region and the main results , in particular the discovery of a very young , compact bipolar h@xmath0o maser outflow associated with vla 2 ( nw2 )
. the implications of these results are discussed in section 3 , while the main conclusions of this work are presented in section 4 .
the @xmath15 h@xmath0o maser transition ( rest frequency = 22235.08 mhz ) was observed with the very long baseline array ( vlba ) of the national radio astronomy observatory ( nrao ) towards the star - forming lkh@xmath1 234 region at three epochs ( 2001 december 2 , 2002 february 11 , and 2002 march 5 ) , during an observing on - source time of @xmath25 hours per epoch .
a bandwidth of 8 mhz sampled over 512 spectral line channels of @xmath20.21 km s@xmath3 width and centred at @xmath16 km s@xmath3 was used .
the data were correlated at the nrao pete v. domenici science operations center ( soc ) .
calibration and imaging was made using the astronomical image processing system ( aips ) package of nrao .
delay and phase calibration was provided by the sources 3c345 , 3c454.3 , bl lac , b1739 + 522 , and b2007 + 777 , while bandpass corrections were determined through observations of 3c345 , 3c454.3 , and bl lac .
the size of the synthesised beam was @xmath20.4 mas for the three epochs .
self - calibration of the uv data was made through a strong point - like maser persisting in all three epochs .
this maser , with flux density @xmath210 jy and radial velocity @xmath17 km s@xmath3 , has absolute coordinates @xmath1(j2000.0 ) = 21@[email protected]@xmath20 , @xmath21(j2000.0 ) = 66@[email protected]@xmath23 ( @xmath24 0.05@xmath23 ) , and is spatially associated with the vla 2 source .
this maser was also used for a first preliminary coordinate alignment of the three epochs of observations , which helped us to the identification of the h@xmath0o maser emission in the region .
we detected h@xmath0o maser emission towards the sources vla 1 , vla 2 , and vla 3a+3b ( fig .
2 ) . for each epoch
, we obtained individual data cubes of 8096@xmath258096 pixels @xmath25 512 velocity channels , and pixel size of 0.1 mas , for three fields , centred on each of the radio continuum sources .
the rms noise level of the individual velocity channel maps depends on the intensity of the masers in each velocity channel , and ranges from @xmath27 to 25 mjy beam@xmath3 . for the identification of the maser spots in the region ( by maser spot we mean emission occurring at a given velocity channel and distinct spatial position )
, we adopted a signal - to - noise ratio ( snr ) threshold of @xmath10 10 .
all the maser spots detected in the region , most of them unresolved with the beam size of @xmath20.4 mas , were fitted with two - dimensional gaussian profiles , obtaining their positions , flux density , and radial velocity . from the snr and the beam size ,
we estimate that the 1@xmath26 accuracy in the relative positions of the maser spots within each epoch is @xmath27 0.02 mas ( beam/[2@xmath25snr ] ; meehan et al . 1998 ) .
o maser spectra observed with the vlba towards vla 1 ( top ) , vla 2 ( middle ) , and vla 3a+3b ( bottom ) in 2002 february 11 .
these spectra were obtained by adding components from the clean images of the different regions with the task ispec ( aips ) . for vla 1 , h@xmath0o maser emission
is only detected in this particular epoch .
for vla 3a+3b , weak emission ( @xmath20.09 jy ) is also detected at @xmath28 km s@xmath3 in epoch 2001 december 2.,width=313 ]
|
we report multi - epoch vlbi h@xmath0o maser observations towards the compact cluster of ysos close to the herbig be star lkh@xmath1 234 .
this cluster includes lkh@xmath1 234 and at least nine more ysos that are formed within projected distances of @xmath210 arcsec ( @xmath29,000 au ) .
we detect h@xmath0o maser emission towards four of these ysos .
in particular , our vlbi observations ( including proper motion measurements ) reveal a remarkable very compact ( @xmath20.2 arcsec = @xmath2180 au ) , bipolar h@xmath0o maser outflow emerging from the embedded yso vla 2 .
we estimate a kinematic age of @xmath240 yr for this bipolar outflow , with expanding velocities of @xmath220 km s@xmath3 and momentum rate @xmath4 @xmath5 @xmath6 m@xmath7 yr@xmath3 km s@xmath3@xmath8/@xmath9 , powered by a yso of a few solar masses .
we propose that the outflow is produced by recurrent episodic jet ejections associated with the formation of this yso .
short - lived episodic ejection events have previously been found towards high - mass ysos .
we show now that this behaviour is also present in intermediate - mass ysos .
these short - lived episodic ejections are probably related to episodic increases in the accretion rate , as observed in low - mass ysos .
we predict the presence of an accretion disk associated with vla 2 .
if detected , this would represent one of the few known examples of intermediate - mass stars with a disk - yso - jet system at scales of a few hundred au .
[ firstpage ] masers stars : formation ism : individual objects : lkh@xmath1 234 ism : jets and outflows .
|
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.